Anand Kumar Mishra

Design, Simulation, Fabrication and Planning of BioInspired Quadruped Robot
A Thesis
Presented to
The Academic Faculty
by
Anand Kumar Mishra

In Partial Fulfillment
of the Requirements for the Degree of
Master of Technology
in
Mechatronics Engineering
Indian Institute of Technology Patna

[May 2014]
Copyright Anand Kumar 2014
INDIAN INSTITUTE OF TECHNOLOGY PATNA

Evaluation Committee Approval
of a Thesis submitted by
ANAND KUMAR MISHRA

The thesis entitled Design, Simulation, Fabrication and Planning of BioInspired Quadruped Robot has been read, found satisfactory and approved by
the following DPPC committee members.
Name: 1___________________,
Coordinator
School of Engineering/Basic Sciences
Date: _________________
Name: 2 ___________________, Supervisor
Date: _________________
Name: 3 ___________________, Co-Supervisor
Date: _________________
Name: 4 ___________________, Member
Date: _________________
Name: 5 ___________________, Member
Date: _________________
Name: 6 ___________________, Member
Date: _________________
Name: 7 ___________________, Member
Date: _________________
Name: 8____________________, Member
Date: _________________
ii
Dedicated
To
My beloved Master, My father,
mother, Sapana, Shipra and
Akanksha
ACKNOWLEDGEMENTS
I would like to express my sincere gratitude to my thesis advisor Dr. Atul Thakur. He
has given me excellent guidance and advice over the years here at IIT Patna. His
guidance helped me in all the time of research and writing this thesis.
Besides my advisor, I would like to thank the DPPC; Dr. Karali Patra, Dr.
Somnath Sarangi, Dr. Somnath Roy, Dr. M.K. Khan, Dr. Manabendra Pathak, Dr. S.S
Panda and Dr. Mayank Tiwari, for their encouragement, insightful comments, and
hard questions. I would like to convey my sincere thanks to all the faculty members in
the department of Mechanical and Electrical Engineering at IIT Patna.
My special thanks to the Center for Artificial Intelligence and Robotics,
DRDO for providing me an opportunity for internship and thereby enhancing my
skills and knowledge.
I would like to express my most sincere gratitude and deep regards to my
Co-guide Mr. Sartaj Singh, Sc E, Centre for Artificial Intelligence and Robotics, for
his guidance, monitoring and constant encouragement throughout the course of
project period.
I am very thankful to Mr. P. Murali Krishna, Sc E, Mr. Babu Jadhav, Sc D,
Mr. K Ramachandran, Sc C, Mr. Dharmendra Kumar Patel, Sc B at CAIR for their
valuable information and guidance through various stages, which helped me in
completing my internship project.
I also take this opportunity to express a deep sense of gratitude to Dr. Dipti
Deodhare, Sc G for granting me an opportunity to carry out the project at CAIR,
DRDO.
I thank my labmates Amrutesh and Delip and also I thank my friends
Sherbahadur, Nitin, Kapil and Amit for all the fun we have had in the last two years.
Last but not the least I would like to thank my family: my loving mother and
father, my sisters and Akanksha Tripathi who has always loved, supported, and
believed in me.
(Anand Kumar Mishra)

iii
Certificate
This is to certify that the thesis entitled DESIGN, SIMULATION, FABRICATION
AND PLANNING OF BIO-INSPIRED QUADRUPED ROBOT, submitted by
ANAND KUMAR MISHRA to Indian Institute of Technology Patna, is a record of
bonafide research work under my supervision and I consider it worthy of
consideration for the degree of Master of Technology of this Institute. This work or a
part has not been submitted to any university/institution for the award of
degree/diploma. The thesis is free from plagiarized material.
________________________________
________________________________
Supervisor
Date: ______________
iv
Declaration
I certify that
a. The work contained in this thesis is original and has been done by me under the
general supervision of my supervisor/s.
b. The work has not been submitted to any other institute for degree or diploma.
c. I have followed the institute norms and guidelines and abide by the regulation as
given in the Ethical Code of Conduct of the institute.
d. Whenever I have used materials (data, theory and text) from other sources, I have
given due credit to them by citing them in the text of the thesis and giving their details
in the reference section.
e. The thesis document has been thoroughly checked to exclude plagiarism.
Signature of the Student

Roll No: 1211MT02
TABLE OF CONTENTS
LIST OF FIGURES
viii
LIST OF TABLES
xii
LIST OF SYMBOLS
xiii
LIST OF ABBREVIATIONS
xiv
ABSTRACT
1.
INTRODUCTION
1.1
2.
Introduction to legged robots

1.1.1
Brief review of legged robots
1.1.2
Advantages of legged robots
1.1.3
Disadvantages of legged robots
1.1.4
Application of legged robots
1.2 Introduction to bio-inspired robotics
1.3 Introduction to quadruped robot
1.4 Objective
1.5 Organization of thesis
LITERATURE REVIEW
2.1
Overview
10
2.2
Snake robot
10
2.2.1
Active cord mechanism series
10
2.2.2
Amphibots
11
2.3 Legged robots
12
2.3.1
Scout
12
2.3.2
Takken-II
12
2.3.3
KOLT
13
2.3.4
Kenken-II
14
2.3.5
Big Dog
14
2.3.6
ELIRO-II
15
2.3.7
Shelley-RHex
16
2.3.8
Salamander
17
vi
3.
DESIGN METHODOLOGY OF QUADRUPED ROBOT

3.1 Design description of legged robot
18
3.2 Design of alligator inspired robot
19
3.2.1
Biological inspiration
19
3.2.2
Robot design
21
3.3 Design description of mammalian inspired robot

4.
26
KINEMATIC AND DYNAMIC ANALYSIS AND TRAJECTORY PLANNING

OF QUADRUPED ROBOT
4.1 Introduction
27
4.2 Kinematic modeling of alligator inspired robot
27
4.2.1
Frame assignment
27
4.2.2
D-H parameters
28
4.2.3
Forward kinematics
28
4.3 Jacobian: velocity analysis of alligator inspired robot
30
4.4 Trajectory planning for quadruped leg
30
4.4.1 Straight line trajectory of quadruped leg with via points
32
4.4.2 Cubic trajectory of quadruped leg with via points
33
4.5 Dynamic analysis of quadruped robot
34
4.6 Kinematic analysis of mammalian inspired robot
38
4.6.1 Frame assignment
38
4.6.2 D-H parameters
38
4.6.3 Forward kinematics
38
4.6.4 Inverse kinematics
40
4.7 Jacobian: velocity analysis of mammalian inspired robot
43
4.7.1 Linear and angular velocity from leg frame
43
4.7.2 Linear and angular velocity from body frame
44
4.8 Dynamic analysis of mammalian inspired quadruped robot
45
5. GAIT PLANNING AND STATIC STABILITY ANALYSIS OF QUADRUPED

ROBOT
5.1 Introduction
50
5.2 Gate synthesis
51
5.2.1 Crawl gait
52
5.2.2 Trot gait
54
vii
5.2.3 Bound gait
55
5.3 Stability analysis of quadruped robot

5.3.1 Introduction
55
55
5.3.2 Static stability margin and longitudinal stability margin

for crawl gait
56
5.3.3 Algorithm developed for SSM for crawl gait
58
6. SIMULATION OF QUADRUPED ROBOT

6.1 Introduction
60
6.2 Control architecture for simulation
62
6.3 Multibody simulation of alligator inspired robot
63
6.3.1 MBD simulation of leg
63
6.3.2 MBD simulation of whole body
64
6.4 MBD simulation of mammal inspired robot

6.4.1 Leg simulation
67
6.4.2 Dynamic leg simulation
69
6.4.3 Whole body simulation
70
6.5 Finite element analysis of quadruped robot
8.
74
6.5.1 Leg simulation
75
6.5.2 Body simulation
75
6.6 Stability analysis of mammal inspired robot

7.
67
76
FABRICATION AND ASSEMBLY OF QUADRUPED ROBOT

7.1 Fabrication of alligator inspired robot
77
7.2 Assembly of alligator inspired robot
79
7.3 Fabrication and assembly of mammalian inspired robot
81
7.4 Modular joint descriptions
81
7.4.1 Construction
81
7.4.2 Absolute encoder
83
CONTROL ARCHITECTURE AND EXPERIMENTAL STUDY

8.1 Control plan
84
8.2 Motion control of alligator inspired robot
86
8.2.1 Hardware descriptions for motion control

8.3 Motion control of mammalian inspired robot
8.3.1
Hardware descriptions for motor control

viii
87
89
90
8.3.2
Part descriptions and specifications
90
8.3.3
Software descriptions for motion control
93
8.4 Experimental study of alligator inspired robot

9.
102
CONCLUSION AND FUTURE WORK

9.1 Conclusion
103
9.2 Future scope of work
104
References
ix
LIST OF FIGURES
Figure No.
Page
Figure 1.1 (a) Sprit wheeled robot sticking on sandy surfaces (courtesy to
NASA) (b)Big Dog robot is moving on sandy surfaces (courtesy
to Boston Dynamics) (c) Sketch of legged machine obtained
Rygg from the planar drawing in (Artobolevsky, 1964) (d)
Mechanical horse patented by U.S. Patent No. 491,927 dated
Feb. 14,1983 (e) The Adaptive Suspension Vehicle (ASV)
(photograph courtesy of Professor Waldron) (f) PV-II walking
robot (courtesy of Professor Hirose)
Figure 1.2 (a) Leg design of crocodiles and alligator (b) Leg design
mammalian
Figure 1.3 (a) Model of Robot in SolidWorks(b) Model of Robot in Pro/E

Figure 2.1 (a)Amphibious
Fukushima
snake-like
robot
"ACM-R5"(courtesy
to
Robotics Lab, Japan) (b) Amphibot 1 snake robot
(courtesy to Bio robotics lab, EPFL) (c) Amphibot 2 snake robot

(courtesy to Bio robotics lab, EPFL)
12
Figure 2.3 Scout robot (courtesy to CIM lab, McGill University)
12
Figure 2.4 Takken II (courtesy to Professor Hiroshi Kimura and group)
13
Figure 2.5 KOLT robot (courtesy to bio-robotics lab Stanford and Ohio
University)
13
Figure 2.6 KenKen II (courtesy to bio-robotics lab Tohoku University)
14
Figure 2.7 BigDog (courtesy to Boston Dynamics)
15
Figure 2.8 ELIRO-II (courtesy to Kyungpook National University)
16
Figure 2.9 Shelley-RHex (courtesy DARA challenge project)
17
Figure 2.10 Salamander robot (courtesy to Bio robotics lab, EPFL)
17
Figure 3.1 Design approach
18
Figure 3.2 (a) Type of limb and segments of alligators leg (b) Anatomy of
pelvic limbs (c) Anatomy of pectoral limbs
19
Figure 3.3 Design inspiration
20
Figure 3.4 Scaled legged of alligator-inspired robot on body length
21
Figure 3.5 Designed CAD model in Pro/E
22
Figure 3.6 Pro/E design alligator-inspired robot
23
Figure 3.7 Free body diagram
24
Figure 3.8 Selected motor
25
Figure 3.9 Quadruped robot
26
Figure 4.1 Frame assignment of the alligator-inspired robot
28
Figure 4.2 Velocity vector of neighbouring link
30
Figure 4.3 Trajectory planning with via point
31
Figure 4.4 Straight line trajectory
32
Figure 4.5 (a) Frame assignment on body (b) Frame assignment on leg
38
Figure 5.1 Quadruped gaits (a) Top view of a robot (b) Graph of leg
sequences (c)(h) sequences
51
Figure 5.2 Quadruped gait types
52
Figure 5.3 (a) Gait sequencing of crawl gait (b) Gait diagrams of crawl
gait (c) Successive gait pattern of a two-phase discontinuous
gait
53
Figure 5.4 Trot gait sequences (b): Gait diagram of trot gait sequences
54
Figure 5.5 (a) Bound gait sequences (b) Gait diagrams of bound gait
sequences
55
Figure 5.6 Support polygon and different static stability margin
56
Figure 5.7 Static stability margin
57
Figure 5.8 Quadruped in static mode all legs on ground
58
Figure 5.9 Geometric view of quadruped when it lifts its 1st leg
58
Figure 5.10 Geometric view of quadruped when it lifts its 2nd leg
59
Figure 6.1 (a) Simplified CAD model of mammal (b) Simplified CAD
model of alligator (c) Defined joint and motion in Adams(d)
Plant generated by Adams(e) Simulink model of close loop
control for simulation
62
Figure 6.2 Control strategy for quadruped simulation
63
Figure 6.3 (a) CAD model of leg imported in Adams (b) Desired angular
position achieved (c) Torque required for robot leg
Figure 6.4 (a) CAD model of alligator-inspired robot imported in Adams
(b) Desired angular position achieved by Knee joint(c) Desired
angular position achieved by Hip joint(d) Distance travelled by
xi
64
the robot (e) Speed achieved (f) Angular velocity variation of

joint1 (g) Acceleration variation of the robot(h) Reaction force
from the ground(i) Power consumption required each motors
67
Figure 6.5 (a) CAD model of mammal-inspired leg imported in Adams (b)
Torque required for robot leg at joint 1 (c) Torque required for
robot leg at joint 2 (d) straight line trajectory of 3dof leg with
via point (e) For straight line trajectory of 3dof leg with via
point
69
Figure 6.6 (a) Model based control in joint space for dynamic analysis
quadruped robot (b) Torque control simulation results
70
Figure 6.7 (a) CAD model of mammal -inspired robot imported in Adams
Figure 6.8
71
(a) Crawl gait sequences (b) Angular position variation form

the each joint of legs (c) Velocity achieved in the crawl gait
implementation
(d)
Torque
required
in
crawl
gait
implementation by robot joints (e) Reaction forces from the

ground crawl gait implementation
72
Figure 6.9 (a) Velocity achieved in the trot gait implementation (b) Torque
required in trot gait implementation by robot joints (c) Reaction
forces from the ground crawl gait implementation
73
Figure 6.10 (a) Velocity achieved in the bound gait implementation (b)
required in bound gait implementation by robot joints
74
Figure 6.11 (a) Maximum shear stress and von misses stress of leg FEA
simulations (b) Maximum shear stress and von misses stress of
whole body FEA simulations
75
Figure 6.12 Static stability simulation of the robot

Figure 7.1 (a) Left-hand image is that of the targeted robot leg design
prototype developed in Pro/ENGINEER Wildfire 4.0 whereas
right-hand image is that of actual robot limb fabricated in 3D
Rapid Prototype Printer using ABSplus in ivory at about 1.6 kW
(RMS) (b) Part required for robot assembly; (i) Hons require for
motor coupling (ii) End of the leg part fabricate in 3D printer
(iii) Screws and nut required (iv) Front body part (v) Rear body
xii
76
part (vi) Leg of the robot (vii) Battery required (viii) Leg
coupling
79
Figure 7.2 (a) Assembled modular joint (b) Assembled modular joint
section model
83
Figure 8.1 The coordinate frame, BODY, used here for kinematic analysis
is the same as that used for standard analysis of any
vehicle/locomotion dynamics
85
Figure 8.2 Control priority level for robot
85
Figure 8.3 Flow chart communication for master, slave card and PC of
quadruped
86
Figure 8.4 (a) Proteus simulation of shield (b) Shield developed front view
(c) Shield back view (d) Arduino UNO
88
Figure 8.5 The algorithm to move the alligator forward is summed up as a

flowchart. The actual amount of actuation to be given and the
time delay for the different actuations involved was based on
gait diagram described earlier. This flowchart was generated
using Flow Breeze, a flowchart automation add-in for MS-Excel
89
Figure 8.6 Flow chart communication for master, slave card and PC of
quadruped
89
Figure 8.7 (a) Slave card (b) Motor
92
Figure 8.8 (a) PWM generation (b) Encoder testing (c) Servo loop based
motion control (d) Servo loop and trajectory based motion
control
101
Figure 8.9 (a) Single axis motion control hardware (b) Motion control
implemented
102
Figure 8.10 (a) Robot speed vs delay time (b) Robot speed vs Hip angle
Figure 9.1
amplitude
102
Current status of my research work
103
xiii
LIST OF TABLES
Figure No.
Page
Table 3.1 Alligators lengths and masses
20
Table 3.2 Design specifications of modular body robot
22
Table 3.3 Modular robot servo motor specifications
25
Table 3.4 Design specification mammal-inspired robot
26
Table 4.1 D-H Parameters
28
Table 4.2 D-H Parameters of mammal leg
38
Table 7.1 Material specification
77
Table 7.2 Component required
77
Table 8.1 Requirement for single axis motion control
90
xiv
LIST OF SYMBOLS
Over all efficiency
Twist angle
Link length
Link offset
Torque of joints or motors
Link rotation
Leg position at the end
Time for one point to next point for trajectory

Acceleration due to gravity
Linear velocity of the links

Linear velocity at center of links
Angular velocity of joint
Angular acceleration of joint
Inertia force
Moment due to inertia
Statics moment on the links
Static force on each link
Angular velocity relative to links
Angular acceleration relative to links
Duty factor
Mass matrix
Coriolis force matrix
Gravitational force component matrix
Stride length
Stroke length
No. of legs
yaw movement
Linear acceleration
xv
LIST OF ABBREVIATIONS
CAD
Computer aided design
CAN
Controller area network
COM
Centre of mass
DOF
Degree of freedom
FEA
Finite element Analysis
LxWxH
Length x Width x Height
PID
Proportional integral derivative
PWM
Pulse width modulation
MBD
Multibody dynamics
SSM
Static stability margin
PR
Priority resister
D-H
Denavit-Hartenberg
MSS
Mechanical system simulation
PMMA
Polymethy methacrylate
ABS
Acrylonitrile butadiene styrene
LLC
Low level control
MLC
Medium level control
HLC
High level control
CNA
Contra lateral and non-adjacent
COG
Center of gravity
SLSM
Static longitudinal stability margin
CSP
Conservative Support Polygon
SCLSM
Crab Longitudinal Stability Margin
xvi
ABSTRACT
This work reports design, simulation, fabrication, and planning of bio-inspired

quadruped robot. This dissertation deals with two types of bio-inspired quadruped
locomotion patterns, namely, mammalian and reptilian. Legged locomotion is one of
the most successful locomotion patterns found in the Nature. Quadruped walking in
many mammals and reptiles have made them very successful in surviving against
tough environments such as uneven terrains. Nature evolved legged locomotion over
half a billion years. It should be noted that the biological evolution favoured legged
locomotion instead of wheeled locomotion in spite of wheeled locomotion being
faster. This is because more than half of Earths landmass has highly rough terrain
and can be traversed by legged rather than wheeled locomotion. We thus take
inspiration from the Nature to develop legged robots that can traverse on rough
terrains and has advantage over wheeled robots.
We have developed alligator-inspired robot at the Mechatronics laboratory, IIT
Patna. The robot has four legs and each leg has two revolute joints representing the
hip (yaw) and knee (roll) respectively, which are actuated by servo motors, thereby
imparting eight degrees of freedom to the robot. The main contributions of this part
of the thesis include following.
Design and modeling of the robot based on multidisciplinary design approach
incorporating CAD modeling, kinematic and dynamic modeling, multibody
simulation and finite element analysis of the robot. We used Pro Engineer for
CAD modeling, Adams, Nastran and Adams Flex for multibody and FEM
simulation.
Fabrication of the designed components of the robot using a CO2 laser cutting
machine and 3D printer. We developed a 3D leg design that has 2D components
to enable manufacturing on a laser cutting machine to reduce fabrication cycle
time.
Integration of electronics such as microcontroller, sensor, servomotors, drivers,
etc.
xvii
Synthesis of crawl and trot gaits. We implemented the gaits in simulation

followed by on the physical robot. The maximum speed attained by the developed
robot is 0.51 BL/sec.
The Mammal-inspired robot known as Robotic mule is being developed at the
Center for Artificial Intelligence and Robotics, DRDO Bangalore where I was
working as an intern for seven months during my project work. This robot is expected
to carry heavy weapons and food over rocky terrains for the soldiers and civilian
purposes. Robotic mule has 4 legs and each leg has 3 degrees of freedom (one roll and
two pitch motions). Each motion is actuated by high torque DC motor with gear ratio
991 and the motor has an inbuilt encoder for positional feedback. The motion control
of the joints is based on forward feedback control. The load carrying capacity of the
robot is less than 10 kg. My contribution in this project includes the following:
Derivation of D-H parameters, frame assignment, kinematic and dynamic
analysis, simplified CAD modeling using SolidWorks.
Multibody dynamic simulation using Adams which helps to validate and optimize
the robot design and mechanism. I performed both kinematic and dynamic
simulation of the robot and implemented a proportional (P) controller.
Gait synthesis such as crawl, trot and bound gait. Algorithms were first
implemented in simulation followed by on actual robot.
Development and testing of static stability margin algorithm in the simulation.
Development of single axis motion control using dsPIC30F4011 series
microcontrollers.
xviii
CHAPTER 1
INTRODUCTION
1.1.
Introduction to legged robot
According to Raibert, only half of the earth's land mass is currently accessible by
wheeled or tracked vehicles, with a much larger fraction being accessible to legged
creatures. Legged robots often offer some advantage over wheeled robots
[1]
. For
example, wheeled or tracked vehicles are restricted by the roughness of the terrain
which they must traverse (see Figure 1.1(a), (b)).
Figure 1.1 (a) Sprit wheeled robot sticking on

sandy surfaces (courtesy to NASA)
Figure 1.1 (b) Big Dog robot is moving on

sandy surfaces (courtesy to Boston Dynamics)
Wheeled robots require a continuous path of support. On the other hand, legged
robots can traverse highly uneven terrains as they require point contacts between the
legs and the terrain. In addition, legs can serve as an active suspension system helping
to decouple terrain variations from body movements thus smoothing locomotion.
Some of the legged robot design challenges include incorporation of leg and
body compliance, multi-sensor integration, vibration resistance, dynamic and static
stability, onboard power supply. Control of legged robots also present a unique
challenge of gait planning for various terrain types. This is because of high
nonlinearity and discontinuous nature of the dynamics of the legged robots.
1.1.1. Brief review of legged robots

Simple walking mechanisms were developed to study legged locomotion for toys that
move on limited type of flat terrains. Researchers started to study about the walking
mode of the legged creatures and recoded their motion. Afterwards, to study and
improve the walking mechanism the biological gaits and their mathematical
formulation were derived. Thus, stability measurements and gait generation
algorithms were developed based on exact locomotion patterns found in the Nature.
Improvements were later made to adapt to specific legged robots. In considering the
history of walking machines we can see from 18th century researchers have developed
many legged robot. Some of the examples are shown in Figure 1.1(c) to Figure 1.1 (f).
First robot legged mechanism was built around 1870 by Chebyshev in Figure 1.1 (c).
Similarly in 1893 mechanical horse was built shown in Figure 1.1 (d).
Due to their specific advantage over wheeled robot currently people are doing many
researches on the legged robot [2].
Figure1.1 (c) Design of legged machine obtained
Figure 1.1 (d) Mechanical horse patented by Rygg
from the 2D drawing in (Artobolevsky, 1964)
U.S. Patent No. 491,927 dated Feb. 14, 1893
Figure 1.1 (e) The Adaptive Suspension Vehicle (ASV)

(photograph courtesy of Professor Waldron)
Figure 1.1(f) PV-II walking robots

(photograph courtesy of Professor Hirose)
1.1.2. Advantages of legged robots

There are several advantages of legged machine over the traditional tracked vehicles
and the advantages are explained below [3].
a. Mobility
Mobility of a legged robot is more than a wheeled robot due to their Omni directional
nature. Legged robots can change the direction without any hindrance of the body and
only change in leg positions are required and wheeled robot have less maneuverability
than legged robots[2],[3].
b. Overcoming obstacles
A legged robot can overcome obstacles that are at a lower level than the maximum
ground clearance, just by stepping on them. On the other hand, a wheeled robot can
only overcome obstacles with heights of up to half of the wheel radius.
c.
Active suspension
Active suspension is much better in legged robot due to natural property of

suspension however in wheeled robot a suspension is required. Active suspension
property come with the adjustment of leg position to maintain the body height same as
according to terrain nature.
d. Energy efficiency
Efficiency of heavy legged vehicle is better than the wheeled vehicle as suggested by
the Hutchinson in 1940 and it was proved by Bekker through the experiment, the
efficiency was much better on irregular terrain by legged vehicle over wheeled or
tracked vehicle.
e. Natural terrain
As Raibert suggested that only half of the land mass can be travelled by the wheeled
vehicle and wheeled vehicle is required expensive flat surfaces while legged vehicle
can go any type of surfaces such highly rugged, soft muddy etc.
1.1.3. Disadvantages of legged robots
Complex mechanisms are required to make robust legged vehicle.
Control is one of the challenging problems.
Gait planning for the highly rugged terrain is very difficult task.
Stability is one of the major issue in legged robotics.
1.1.4. Application of legged robots

1. Military applications
2. Inspection of Nuclear Power Plants
3. Land, Submarine and Planetary Exploration
4. Forestry and Agricultural Tasks
5. Forestry and Agricultural Tasks
6. Help for Disabled People
7. Study of Living Creatures
8. Civil Projects
9. Humanitarian De-mining
10. Support for AI Technique
1.2.
Introduction to bio-inspired robotics
Legged locomotion is one of the most successful locomotion patterns found in the
Nature. Evidence of first locomotion among biological organisms is around 585
million years old. This means that it took the evolution process around 585 million
years to evolve various forms of walking gaits. Quadruped walking in many mammals
and reptiles have made them very successful in surviving against tough environments
such as uneven terrains. In comparison, humans invented wheel-based locomotion
around 4000 years ago. Nature evolved legged locomotion instead of wheels because
more than half of Earths landmass cannot be traversed by wheels, even today. It is
thus the imperative for the roboticist to learn the design and gait patterns from the
phylogenetic analysis of biological organisms and try to mimic them as closely as
possible to improve the performance of existing robots.
In this report, design and simulation of an alligator-inspired and mule inspired
robot have been accomplished. Robot exhibits a particular type of reptilian
locomotion and mammalian locomotion. Alligator locomotion is generally considered
as an intermediate step in the evolutionary paradigm of vertebrate locomotion
[3], [4]
At one extreme, are the amphibians and lizards who are natural sprawlers. Their limbs
are held laterally to the body. At the other extreme, are the mammals and dinosaurs
exhibiting erect locomotion posture (ELP). Humans exhibit ELP too. In ELP, limbs
are held directly under the body. Reptilian locomotion posture of crocodiles and
alligators is considered to be an intermediate locomotion posture between these two

extremes as a kind of transition from sprawl-to-erect [5].
Another interesting feature of alligator locomotion that has come to light is that
alligators still use sprawl to negotiate muddy lands at low speeds while high walk is
activated on drier lands at higher speeds. This transition between different gaits as a
function of speed is a general locomotion feature [6], [7], [8] .
Most reptiles can actually adapt to an amphibious habitat. Thus, imitating a reptile
can help in development of robots that can perform both terrestrial as well as aquatic
locomotion. But building robot designs inspired by biological counterparts can enable
us in imparting critical ethological functionality to robots. Moreover, it has been
studied that artificial intelligence can be developed more efficiently in bio-inspired
compliant designs [9], [10].
Success of biological designs lies in the optimal cooperative and coordinated
interplay among the following factors [11].
a.) Availability of optimal energy generation, storage and conversion mechanisms.
b.) Ability of memorising, storing, processing, internalising and passing information.
c.) Complaint and robust structural features that can adapt to various environment.
This is what has inspired the fields of biomimetics, biokleptics and bio-inspiration.
However, exact mimicking of reptilian gait (biological gait in general) is plagued with
following challenges:a.) Mechanical complexity of nature is overwhelming. Fabrication of intricate
skeleton-muscular system, which gives enormous agility to animals, is very
difficult to replicate using existing manufacturing technology.
b.) Imitating self-replicating biological units like cells cannot be fabricated in
laboratory easily.
c.) Energy generation and storage mechanism in biological systems are not
understood to an extent that it can be copied in robots.
Therefore, in the light of the above-mentioned motivations and challenges, many
researchers have proposed the use of biological inspiration instead of exact replication
of biological structures [12], [13], [14].
A lot of research has been done in the field of walking bio-inspired robots
mimicking biological counterparts. The bio-inspiration in case of mobile robots
5
comes largely from tetra pods - both reptiles as well as mammals (See Figure 1.2 (a)
and (b)
Figure 1.2 (a): Leg design of crocodiles and alligators
Figure 1.2 (b):.Leg design of mammalian
Note the ball-and-socket joint provided at the knee and hip about which rotatory
actuation takes place which results in locomotion. The approximate perpendicular
orientation of the coordinate frames attached to the hip and knee joints is the salient
feature of all members of the order Crocodilia. Note also the approximate parallel
orientation of the coordinate frames attached to the hip and knee joints. For walking
robots, both static and dynamic walking as well as both active and passive walking
has been extensively researched
[15]
. Interest in legged walking robots over
wheeled/tracked robots generated mainly because of

a.) More efficient traversal of difficult terrains,
b.) Nature is the best designer hence just bio-mimic the design handed over to us
by evolution,
c.) More robust and durable if designed on the principles of compliant
mechanism, say by using flexible links, and
d.) More efficient motion planning can take place as a wider range of
manoeuvrability is made available.
1.3.
Introduction to quadruped robot
In this thesis, we are focusing on quadruped robots, which belong to the broader class
of legged robots as described earlier. In quadruped robots there are four legs, each leg
should have minimum two degrees of freedom. However, some quadruped robots
may have higher degree of freedom legs too. Quadruped robots can walk and run but
standing still it must be passively stable. The main challenge of quadruped robot is to
maintain stability by actively shifting body weight during the gait.
The quadruped robots described in this thesis are inspired from two types of animals,
namely, mammals and reptiles. Mammalian type locomotion has an advantage that
they can move faster due to geometrical configuration of leg joints. Some examples of
mammalian locomotion based robot include ABIBO robotic dog, robotic cheetah etc.
Another advantage of mammalian type four-legged robots is an ease in gait transition.
Mammal inspired robot thus have a potential to serve as useful platform for research
in Human-robot interaction (see Figure 1.3 (a)). Reptile-inspired robots have
relatively slower speed but higher static stability (see Figure 1.3 (b)). Quadruped
robots can be used for civilian and military purpose to carry heavy loads.
Figure 1.3 (a) Model of Robot in SolidWorks
1.4.
Figure 1.3 (b) Model of Robot in Pro/E
Objective
The thesis has been carried out partly at IIT Patna and CAIR, DRDO Laboratory. The
objectives of each part of the thesis are outlined below.
IIT Patna
Kinematics analysis of alligator-inspired quadruped robot.
CAD modeling of alligator-inspired quadruped using Pro\E based alligator

anatomy
Kinematic, dynamic modeling and trajectory planning
Derivation of gait sequences and simulation using Adams and Matlab.
Analysis of stability of alligator-inspired quadruped both static as well as dynamic.
Embedded control software for alligator-inspired quadruped.
Finite element analysis of alligator-inspired quadruped.
Complaint body design and simulation of alligator-inspired quadruped.
Development of alligator-inspired quadruped with complaint in the body.
Speed and stability analysis on actual setup.
7
CAIR, DRDO Laboratory

Kinematics analysis of mammalian-inspired quadruped robot.
CAD modeling of mammalian-inspired quadruped using SolidWorks and

kinematics simulation with Adams and Matlab.
Dynamics analysis of mammalian-inspired quadruped robot with 3 DOF legs.
Derivation of gait sequences and simulation using Adams and Matlab.
Analysis of stability of mammalian-inspired quadruped both static as well as
dynamic.
Design single axis motion control for mammalian-inspired quadruped
Embedded Control Software for mammalian-inspired quadruped
1.5.
Organization of the report
The outline of the thesis is as follows.

Chapter 2 discusses literature review of the bio-inspired robot. A survey work of
snake robots, swimming robots and terrestrial locomotion based robot is briefly
described.
Chapter 3 discusses of the design methodology of the quadruped robot. Design of
quadruped robot describes CAD modeling based on imitation from biology, the
design leg ratio of robot is based on anatomy of actual alligators and torque
calculation and selection motors has been done.
Chapter 4 discusses a kinematic, dynamic modeling of quadruped robot. The chapter
will explain the frame assignment, D-H parameters, forward kinematics analysis
velocity Jacobian analysis and dynamic analysis has been done. For robot motion
smooth the cubic and straight line trajectory has been derived.
Chapter 5 reports a gait planning and stability analysis. Gait planning of quadruped is
based on mammalian-inspired and reptilian inspired robot. Graphical representation of
trot, crawl and bound has been shown and static stability margin analysis is done in
this chapter.
Chapter 6 discusses multibody, FEA simulation of quadruped robot. Simulation of
alligator inspired robot and mammalian inspired robot is separately done.
Chapter 7 reports fabrication and assembly of alligator-inspired robot and modular
joint assembly of mammal-inspired robot.
Chapter 8 discusses motion control and control architecture of the quadruped robot.
The chapter will elaborate the single axis motion controller which consist the
8
hardware and software part both and 3 layer control architecture is used for alligator
inspired quadruped robot
Chapter 9 contains a summary of important conclusions and scope for future work in
the area of bio-inspired quadruped robots.
CHAPTER 2
LITERATURE REVIEW
2.1. Overview
This chapter presents a review of the state of art in the area of bio-inspired robotics.
This review investigates the current methods of terrestrial locomotion based and
amphibious robots. It will cover the scope and application of these robots.
2.2. Snake robots

Implementing these advantages in a robot is a hard task but a close approximation is
quite possible. Snake inspired robots have come a long way since their inception. The
concept of serpentine motion and its advantages were recognized by Shigeo Hirose
who pioneered in this research. Since then many engineers conceptualized a lot of
ideas in making the snake robots. James Hopkins in 2009 developed a general
classification of these robots based on the material published till that date.
2.2.1. The Active Cord mechanism series:

ACM 3: This model was created in the initial stages of the snake robot research.
It consisted of 20 links with passive wheels actuated solely with the help of the
revolute joints between the links [16].
ACM-R3: The key difference between ACM 3 and ACM R3 is that the latter had
a scope to move in 3D. That is, the robot had consecutive passive wheels as well
as the revolute joints placed orthogonally. This model was significantly lighter
than ACM 3
ACM R4: R4 builds on R3 by replacing passive wheel with active ones. This
change enables the robot to traverse rough ground with relative ease [16], [17].
ACM R5: This is an amphibian design to increase the versatility of the robot. R5
uses paddles on the passive wheels to acquire anisotropic friction both on ground
and water. But the technology of the time could solve the problems that impeded
the realization of practical snake robots [18].
10
Figure 2.1(a) Amphibious snake robot "ACM-R5"(courtesy to Fukushima Robotics Lab, Japan) [18]
grip onto the next robot. Thus the self sufficient modules form a chain and
accomplish tasks like climbing the stairs or crossing a ditch.
2.2.2. Amphibots:
i.
Amphibot 1: This is a modular robot developed in the quest to find new robot
designs which could exhibit dextrose locomotion. Each link was made to be slightly
buoyant and the center of mass of each link was purposefully kept slightly lower than
the geometric centre of mass so that during idle conditions, the snake could float and
obtain vertical orientation [19].
Figure 2.1(b) Amphibot 1 snake robot (courtesy to Bio robotics lab, EPFL) [19]
ii.
Amphibot 2: This is an advanced version of Amphibot 1. The design was
simplified and new powerful motors were equipped. An electronic suite was
introduced which could handle up to 127 segments theoretically. The robot used
removable wheel sets to achieve ground motion [20], [21].
11
Figure 2.1(c) Amphibot 2 snake robot (courtesy to Bio robotics lab, EPFL) [20]
2.3. Legged robots
2.3.1. Scout
Martin Buehler in 1990s, at Ambulatory Robotics Lab (ARL) of McGill University
started developing the Scout robot series. In 1999 he presented, a quadruped robot
Scout II (Figure 2.3) which was dynamically stable and had a very simple mechanical
design. This robot had one active rotational joint at each leg located at the hip with the
help of which it was able to rotate the leg in the sagittal plane. The leg is divided into
two parts-upper and lower legs. These parts are connected by a spring to form a
compliant prismatic joint. It weighed around 27kg and has 0.55m x 0.48m x 0.27m
(LxWxH). After some year, Scout II, with a forward velocity of up to 1.3 m/s, was
able to perform a stable bounding gait [23], [24], [25].
Figure 2.3: Scout robot (courtesy to CIM lab, McGill University) [23]
2.3.2. Takken II
Hiroshi Kimura and colleagues had developed a quadruped robot Patrushand, later
Tekken series. These robots are a central pattern generators (CPG) based robot and
have joint reflexes. It is mainly used to study controllers which are biologically
inspired. In Figure 2.4, Tekken II is shown; it is a small quadruped robot having
12
length of 0.3m length and weight of 4.3kg weight. Mechanical springs are adding
compliance to the joints whereas it is actuated by electric motors [26], [27].
Figure 2.4 Tekken II (courtesy to Professor Hiroshi Kimura and group) [27]
2.3.3. KOLT
Kenneth Waldron and his group at Stanford University in collaboration with Ohio
State University had developed the KOLT robot shown in Figure 2.5. The quadruped
robot has a weight of about 80kg and has the following dimensions: 1.75m x 0.6m x
0.8m (LxWxH). Robot has electric actuation with mechanical springs. In running
experiments, the robot was attached to a boom that permitted free motion in the plane
and it was found that the robot performed stable trotting with 1.1m/s on a treadmill
[28], [29], [30], [31]
Figure 2.5: KOLT robot (courtesy to bio-robotics lab Stanford and Ohio University) [29]
13
2.3.4. KenKenII
In 1998, a hydraulic quadruped robot was developed by Sang-Ho Hyon with his
colleagues at Tohoku University (Japan). They had first built the monopod robot
KenKenI and later a biped version KenKenII (Figure 2.6), but an actual quadruped
robot was never constructed. Nevertheless, they managed to achieve impressive
results [32], [32], [33].
Figure 2.6 KenKen II (courtesy to bio-robotics lab Tohoku University) [33]
2.3.5. Big Dog

Boston Dynamics Corporation was founded, as a spin-off from the MIT, in 1992 by
Marc Raibert and some of his colleagues. The initial company focus was on software
for human simulations, such as DI-Guy, which at that time was being used for
military applications. In 2005 however they presented the first version of their
quadruped robot called Big Dog (Figure 2.7).The main goal of the project was the
development of a mechanical mule with the following properties
Autonomous power
Capability of carrying heavy payloads
Outdoor operational
Having static and dynamic mobility
Fully integrated sensing for mobility
Able to jump over a 1m ditch, climb 45 (100%) slopes, run at 5m/s, and carry over
50kg payload.
14
Figure 2.7 BigDog (courtesy to Boston Dynamics) [35]
The robot presented in 2005 (BigDog 2005) was 1m in length and 0.3m width, and
had a weight of about 90kg. It had four legs each having four DOF: hydraulic
cylinders are powered by three active rotational joints and one passive linear joint in
the foot based on a pneumatic spring [34], [35], [36].
2.3.6. ELIRO-II
Kyungpook National University came up with a robot gait built to incorporate
discontinuous twisting at the waist. Many natural walking gaits incorporate a central
DOF at the spine, which increases speed and stability, especially while turning. A
systematic approach was taken and following assumption were made that is, the robot
would be on flat terrain, parallel to the ground, with mass less legs, a uniform
mass of the body, and a single degree of freedom along the z-axis at the center as
the waist joint. From these set of assumptions, the kinematics was determined,
nothing especially that in a non rigid body an optimal hip position can be created.
Based on these kinematic observations some additional features were drawn, the
stride increases drastically from a rigid body, maneuverability in unfavorable
situations was greatly improved and the general gait stability was improved. In gait
planning, ten time steps were used with the center of gravity shifted at varying
intervals; the bending angle for the center and the leg order were varied. Eventually
the conclusions from this were used in a quadrupedal robot with 3 DOF pantograph
legs called ELIRO-II. This robot achieved stable turning on flat ground and it is noted
15
that with implementation of a 3 DOF, stability on rough terrain could similarly be

achieved [37].
Figure 2.8 ELIRO-II (courtesy to Kyungpook National University) [37]
2.3.7. Shelley-RHex
Shelley is a branch of RHex. It is designed from a mechanical perspective with the
goal of amphibious movement without any increase in weight or size. To accomplish
this, the frame of RHex was replaced with a closed shell formed from carbon fiber
into a curved shape for hydrodynamics and impact resistance, with Neoprene to
dampen impacts and vibrations. Some waterproofing issues arose from the power
switch, antenna, access panel and motor connection which were fixed by various
solutions such as adding clamping force and waterproof material, encasing the entry
port with silicone and O-rings. Shelley was able to achieve surface swimming. In the
water, the AQUA can achieve swimming speeds up to 1.0 m/s or approximately 1.54
BL/sec. It achieves swimming motion by creating small oscillations. If it engages all
six legs, it has the capability to control five degrees of freedom directly: surging,
heaving, pitch, roll and yaw are all possible. Phase offsets are used to change between
gaits with amplitude offsets used for pitch and roll and a change in amplitude scale
factor between the sides used for yaw half circle [38].
16
Figure 2.9 Shelley-RHex (courtesy DARA challenge project)[38]
2.3.8. Salamander
The salamanders robot is amphibious robot which can perform walking and
swimming gaits. It uses primary actuator for the center spine. It swims by body
undulation to create a wave for the propagation while in walking it uses terrestrial
gaits with the help of its leg. By using some assumptions, a robot was created with ten
actuators, four for the legs and six for the body. Using a central pattern generator
comprised of 16 to 32 oscillators as the gait basis. Relationships between each were
defined by matrices of weights and modeled in formulae representing the phase and
amplitude of any given oscillator as well as the positive signal representing the central
burst action. In creating this robot, it was shown to be consistent with the actual
salamanders movement in both walking and swimming. Ultimately it is noted that
the central pattern generator concepts used here would be very helpful in general
robotics, given that the dimensionality of a control problem may be reduced [39].
Figure 2.10: Salamander robot (courtesy to Bio robotics lab, EPFL) [39]
17
CHAPTER 3
DESIGN METHODOLOGY OF QUADRUPED ROBOT

3.1.
Design descriptions of legged robot
The design and development of bio-inspired quadruped robot is based on the

multidisciplinary design approach. This approach focuses on CAD modeling,
kinematic and dynamic modeling, multibody simulation, finite element analysis of the
robot. This is concurrent engineering approach to do research for academic and R&D
for prototype modeling shown in Figure 3.1. This research helps to utilize the
concurrent engineering concepts in design and development of area of robotics
researches.
In this approach we developed a CAD design based on alligator anatomy and
with the help of multibody dynamics (MBD) simulation we validated our design and
mechanism. For making MBD simulation more realistic the simulation environment
parameters are very close to the actual system such as material properties of robot
body, legs and track. Simulation is based on the kinematics and dynamics analysis,
trajectory planning and gait planning using forward feedback control algorithms.
With MBD simulation we got the forces and torques at the joint and body of
robot that is further used in FEM analysis of the leg of the robot.
CAD design
Design
optimization
Kinematic and
dynamic modeling
Control
Gait Planning
Robot
Dynamic simulation
Fabrication
FEM analysis
and simulation
Figure 3.1: Design approach
Using FEM, we validated the design to determine whether the model can withstand
loading under different type of terrains. In addition, the concurrent multibody
simulation and FEM analysis helped us in optimizing the design. Next sections will
cover the design of alligator-inspired robot and mammal inspired robot.
18
3.2.
Design of alligator-inspired robot
3.2.1. Biological inspiration

Alligator limbs are mainly categorized in two main parts, namely, pectoral and pelvic
limbs (see Figure 3.2(a)). Pectoral limb is front (anterior/Fore) leg and pelvic limb is
rear (posterior/hind) of legs of the alligators. Each leg of alligator has 3 DOF
movements and contains 3 segments, namely, proximal, middle and distal. Figure 3.2
(b, c) shows the anatomy of legs consisting pectoral limb depicting humerus, ulna,
and manus length and pelvic limb depicting femur, tibia and pes length of segments.
Figure 3.2 (a): Type of limb and segments of alligators leg
Figure 3.2 (b): Anatomy of pelvic limbs
Figure 3.2 (c): Anatomy of pectoral limbs
We developed the equivalent design inspired by the alligator, however we have not
mimicked robot as actual alligators (see Figure 3.3) because of the leg design of
alligators are very complex and fabrication and material selection of a same structure
is very challenging so we developed simplified model of leg design. We developed
19
the leg on the scaling factor of the body length, mass of the alligators. Researchers
have done the scaling analysis on the basis of these parameters, see Table 3.1
Figure 3.3: Design inspiration
We have developed the symmetrical legged design of our robot. Here pectoral limb
design and pelvic limb design is same whether it is not same as in alligators that is
basically categorized in hip limb and knee limb. Hip and knee limb lengths of robot
are 0.38th of the body length of the robot. The mass of the robot is 0.7kg and body
length of the robot is 0.26 m. Legged and body thickness of the robot is 3mm, (see
Figure 3.4) [40]..
Table 3.1: Alligators length and masses
SN
STL
TL
PL
HL
UL
ML
PVL
FL
TBL
PL
12
0.27
0.33
0.104
0.038
0.028
0.037
0.131
0.04
0.037
0.052
0.52
15
0.29
0.33
0.104
0.042
0.028
0.035
0.135
0.044
0.039
0.05
0.58
11
0.29
0.35
0.109
0.042
0.031
0.04
0.14
0.044
0.037
0.054
0.67
13
0.31
0.38
0.119
0.044
0.033
0.041
0.156
0.049
0.042
0.061
0.75
14
0.32
0.38
0.122
0.044
0.033
0.04
0.152
0.048
0.041
0.054
0.86
0.47
0.55
0.167
0.068
0.047
0.062
0.219
0.075
0.06
0.087
3.20
0.51
0.56
0.148
0.072
0.055
0.061
0.233
0.078
0.022
0.096
3.50
Specimen No. =SN, Ulna length (m) = UL, Manus Length (m) =ML, Pelvic limb length =PVL
Femur length (m) =FL, Tibia length (m) =TBL, Pes length (m) = PSL, Mbody(Kg) =m, Length snout to
tail base (m) = STL, Tail length (m) =TL, Pectoral Limb length (m) =PL, Humerus length (m) =HL
20
Figure 3.4: Scaled legged of alligator-inspired robot body length
3.2.2. Robot design

The alligator-inspired robotic platform has four legs to support the body and assist
in its locomotion. Each leg is a simple 2-link rigid body mechanism and has two
servos. One servo is provided for each hip and knee joints. Ankle joint has been
ignored because our focus is on studying the salient feature that distinguishes
mammalian gait from reptilian gait: orientation of rotating axes of hip and knee
with respect to each other. The hip servo provides hip-yaw and knee servo
provides knee roll. This is very different from mammalian arrangement. For
mammals, the hip servo and knee servo both provide pitching action. The purpose
of body in our platform at present is to contain the Electronic Control Circuit (ECC) as
well as impart passive roll stabilisation. Servos are housed outside the body in slots
created for each limb. This design is have modular structure where is body beaked
into two parts. Robot has given passive compliances (springs) which will move
the body in vertical direction and constraint into the horizontal direction and
developed robot with compliance in the body which will be help to the robot make
more energy efficient and it will help to robot for taking turn. CAD model and
design specification of robot is shown in the Figure 3.5 and Table 3.2.
21
Figure 3.5: Designed CAD model in Pro/E
Table 3.2 Design specifications of modular body robot

S.
No.
1
Characteristics
Value
Mrobot
700g
Mservo+ MArduino + Mbattery
300g
Mul, Mul - Mservo
15 g, 15 g
Mll, Mll - Mservo
15 g, 15 g
Width
100 mm
Length
260 mm
Limb length
100 mm
Distance from front hip joint to front of robot
40 mm
Distance from back hip joint to front of robot
40 mm
10
Centroid coordinate
(50,130,-110)mm
11
Hip height from ground
100 mm
12
Maximum speed Vmax
0.133m/s
13
Speed in Body length per second, BL/s
0.51BL/s
14
Spring stiffness
3940 N/m
14
Gait
Trot , crawl
A. CAD design of modular structure based alligator-inspired quadruped

robot
I designed alligator inspired robot using Pro\E software. CAD design contains the
all basic information of robot such as, joint type, screw, nut design, equivalent
motor design, leg design and body design of the robot Figure 3.6
22
Figure 3.6: Pro\E design alligator-inspired robot with Isometric view, Front view, Side view and
Top view
A. Motor selection and calculation

For calculation of motor torque at each joint of the robot leg. We have considered
robot is statically stable and all the load of the robot is acting at centroid of the body.
We assumed that robot in the worst position where it requires maximum torque to
stand and stretched his leg in maximum of his length. The reaction force of the robot
is vertical to ground of the surfaces
23
Figure 3.7: Free body diagram
(3.1)
For torque at joint 1
Here only friction will come into the picture so the net torque at joint1 will be
(3.2)
(3.3)
Net torque at joint 1 =
(3.4)
For torque at joint 2

Here only reaction from the ground will come into the picture so net torque at
joint will be
(3.5)
Net torque at joint 1 =
For flat surface
(3.6)
, and
And from MBD simulation in section 6.3.2 we are getting reaction forces
From equation (3.2)
24
=15 N
For Torque at Joint 2

Here only reaction from the ground will come into the picture so net torque at
joint will be
From equation (3.6)
=20o
Here
At 90
B. Motor specification for modular robot
Figure 3.8 Selected motor

Table 3.3 Modular robot servo motor specification
Required Pulse
3-5 Volt Peak to Peak Square Wave
Operating Voltage
4.8-6.0 Volts
Operating Temperature Range
-10 to +60 Degree C
Operating Speed (4.8V)
0.2sec/60 degrees at no load
Operating Speed (6.0V)
0.16sec/60 degrees at no load
Stall Torque (4.8V)
5.5 kg/cm
Stall Torque (6.0V)
7 kg/cm
Weight
41g
Dimensions
41x20x36mm
3.3. Design Description mammalian inspired robot

Quadruped robot is being developed in CAIR. Purpose of this robot is to carry the
heavy loads for civilians and military. According to proposed design robot can carry
the 10 kg load. Robot has 4 legs and each leg has 3 DOF freedom means total degree
of freedom of robot is 12. Each leg of quadruped has 10 kg weight and has 3 joint.
25
Material used for structure is aluminum. The total weight of the robot is 58kg
approximately without consideration of power supply. Design of robot structure is
modular and flexible. No permanent joint (e.g. Welding or riveting) is used in
mechanical structure. Controller of the robot is kept into the universal joint and it has
used 12 slave controllers for 12 joints and all controllers are directly connected to
master controller. The design of robot was already developed by the team and I
contributed the simulation and control of part of the development of the robot which
is discussed in next section. CAD model of robot design and specification shown as
follows (see Figure 3.9 and Table 3.4)
Figure 3.9: Quadruped robot

Table 3.4 Design specification mammal-inspired robot
S.
No.
1
2
5
6
7
8
9
10
12
13
14
Characteristics
Value
Mrobot
Mmotor+ Mextra
Width
Length
Leg, link 1 length
Leg, link 2 length
Leg, link 3 length
Centroid coordinate
Maximum speed Vmax
Body length per second, BL/s
Gait
20kg
1.5 g
360 mm
720 mm
80 mm
400mm
400mm
(-360,-180,-720)mm
4 m/s
1.66 BL/s
Trot , crawl, bound
26
CHAPTER 4
KINEMATIC, DYNAMIC ANALYSIS AND

TRAJECTORY PLANNING OF QUADRUPED ROBOT
4.1.
Introduction
The mechanism of walking robot is very complex due to many DOF. Kinematics and
dynamic modeling are requiring before simulation. Kinematics model describes joint
variables, leg position and velocities however dynamics model helps to determine the
torque and forces at the joint, body and ground. This chapter contains kinematic,
dynamic analysis, trajectory planning and Jacobian analysis of the quadruped robot.
Kinematic analysis is covered forward kinematics and inverse kinematics of robot. To
define the robot kinematics first we need to derive D-H parameters using geometrical
configurations of robot and for dynamic analysis and robot smooth motion, velocity
Jacobian and trajectory planning are required. In the following section kinematic and
dynamic modeling of alligator inspired robot and mammal inspired robot has been
discussed [42].
4.2.
Kinematic modeling of alligator-inspired robot
In this section we discussed the kinematic modeling of robot. It consists of frame

assignment of leg and body, D-H parameters analysis and forward kinematics of
alligator inspired robot.
4.2.1. Frame assignment

Frame assignment of the robot gives the position and orientation in joint space and
Cartesian space because this is the arbitrary references and totally depends on the
observer reference frame. Global reference is used for body of quadruped and local
reference frame for leg and 3 frames we used for each joint according to D-H
convention (see Figure 4.1). Our convention is always considered the Z axis is in
vertical direction (outward to body) and joint motion.
27
Figure 4.1 Frame assignment of the alligator-inspired robot
Proper color coding is done for representing each axis of frames, Red is used for z
axis, green for y axis and blue for the y axis. I have assumed the direction of each
frame is same due to reduce the extra calculation except body frame.
4.2.2. D-H parameters

DH parameters of quadruped robot are defined from geometrical configuration and
frame references [41]
Table 4.1: D-H Parameter
S.N
Twist
angle
Link
length
90
Link offset
Link length
4.2.3. Forward kinematics

Forward kinematics analysis is done using Table 4.1 of the robot. Here we do the
transformation of leg from local 0th frame to 3th frame (see Figure 4.1). First
transformation for first link form 0th frame to 1st frame for joint 1
28
(4.1)
Similarly transformations form 1st frame to 2nd frame for joint 2.
(4.2)
Transformations form 2nd frame to 3rd frame for joint 3.
(4.3)
Final transformation from 0th frame to 3rd frame for quadruped leg from leg frame
(4.4)
(4.5)
(4.6)
Similarly the transformation for remaining three legs will be same because we have
taken the assumption frame assignment of all leg is same for reducing the calculation.
That means there is no need to calculate the transformation of the all legs.
(4.7)
(4.8)
(4.9)
29
4.3.
Jacobian: velocity analysis alligator-inspired robots
In this section we discussed the Jacobian matrix for all joints of each leg. Linear
velocity can describe by a point while angular velocity can be a body. Jacobian is
always associated with the angular velocity and describes the linear velocity. We
considered the 0th frame of robot leg is as a reference. Our calculation is done on
arbitrary links like {i}. Hence,
and
is the linear velocity of the origin of link frame {i}
is the angular velocity of link frame {i}. The velocities of link i+1 will be that
of link i, plus whatever new velocity components were added by joint i+1 (see the
Figure 4.2).
Figure 4.2: Velocity vector of neighbouring link [41]
Angular and linear velocity can be written as

(4.10)
(4.11)
After differentiation of equation 4.7, 4.8, and 4.9 we get the Jacobian matrix from 3th
frame w.r.t to 0th frame.
(4.12)
4.4. Trajectory planning for quadruped leg
Leg movement describes by using trajectory planning of the robot in three
dimensional spaces. Here, trajectory refers to a time history of position, velocity and
acceleration for each degree of freedom. When we simulate legged of quadruped
without their trajectory planning. We found leg is moving randomly with sudden
30
change in velocity in the Cartesian/Joint space. That creates vibration so our

requirement is to make robot motion smooth with our desired way point and for actual
robotics platform it should follow a particular trajectory which helps to make easier
the balancing of robot. For trajectory planning of leg is important for kinematics
based multibody simulation, here we are doing trajectory planning for legged
quadruped robot with and without via point. Quadruped has three DOF leg so robot is
moving in the Cartesian space in all 3 directions. This section discusses trajectory
plan with via point as follows.
Figure 4.3: Trajectory planning with via point
Assumptions for formulation of equations

We are considering one point between starting and end point of the robot motion.
Position and velocity at t=0,
(see the Figure 4.4)

,
,
(4.13)
,
,
(4.14)
,
31
(4.15)
4.4.1. Straight line trajectory of quadruped leg with via point
Figure 4.4: Straight line trajectory
General equation for straight line is
-segment 1 t=0 to
-- segment 1 t=
(4.16)
to
(4.17)
From eq. 4.13, 4.14, 4.15 put the value in eq. 4.17, 4.16. We get
,
(4.18)
,
(4.19)
(4.20)
(4.21)
(4.22)
,
32
(4.23)
4.4.2. Cubic trajectory of quadruped leg with via point
Segment 1 t=0 to
Segment 1 t=0 to
(4.24)
(4.25)
From eq. 4.13, 4.14, 4.15 put the value in eq. 4.24, 4.25. We get. We get each cubic
will be evaluated over an interval starting at t=0 and end t=
and i=1, 2
(4.26)
After solving the equation 4.26 we get
(4.27)
Put the value of eq. 4.27, 4.26 in 4.25, 4.24

33
Similarly for y and z
4.5. Dynamic analysis of alligator-inspired leg

Robot dynamics describes the forces and moment relation between robot motion that
is specifically described as following
Robot location and its derivatives, velocity and acceleration.
Forces and torques applied at the robot joints.
Dimensional parameters of the robot leg such as link length, mass and inertia.
Forces and moments propagate through the kinematic chain from one leg to another,
and therefore dynamic coupling exists. For dynamic analysis there are two methods
one is Lagrangian methods and other is Newtons and Eulers method. Here my
analysis is based on Newtons Euler approach. For dynamic modeling quadruped
robot, we have derived the torque of each joint.
Propagation of rotational and linear acceleration of leg
(4.28)
(4.29)
34
Velocity about center of gravity of the link
(4.30)
Value of angular and linear joint acceleration from eq. 4.28, 4.29
For link0
(4.31)
(4.32)
For link 1, i= 0;
(4.33)
(4.34)
For link 2, i= 1;
(4.35)
(4.36)
Form equation 4.30 we can write the acceleration

For center of link 1, i=2;
(4.37)
For center of link 2, i=2;

K=
35
(4.38)
Having computed the linear and angular accelerations of the mass center of each link,
we can apply the NewtonEuler equations to compute the inertial force and torque
acting at the center of mass of each link. Thus we have
(4.39)
(4.40)
Here
inertia, m mass of the
link
Inertial force and torque for link 1

Put the value of
from eq. 4.37, 4.34

(4.41)
(4.42)
Inertial force and torque for link 2

Put the value of
, from eq. 4.38, 4.36,

(4.43)
(4.44)
Net acting force and moment on link1

(4.45)
(4.46)
36
Net acting force and moment on link2

(4.47)
(4.48)
After getting the forces and moments about all direction. We calculated get the torque
about each joint
(4.49)
Torque about first joint
(4.50)
Torque about 2nd joint

(4.51)
General dynamics equation is

(4.52)
, M is mass matrix, V is centrifugal and Coriolis force matrix, G is gravity
force matrix.
37
4.6. Kinematic analysis of mammalian-inspired robot

Kinematic modeling of mammalian inspired robot is same as reptilian inspired robot
only difference is mammalian have 3 DOF leg configurations so the analysis of the
mammal is more computational and complex than reptilian. This section discusses on
D-H parameters, frame assignment, forward kinematics and inverse kinematics
4.6.1. Frame assignments

Frame convention is same as discussed in previous section and assign frame is
described as follows (see Figure 4.5 (a), (b))
Figure 4.5 (a): Frame assignment on body
Figure 4.5 (b): Frame assignment on leg
4.6.2. D-H parameters

Table 4.2 D-H Parameter of mammal leg
+90
are joints angles of the quadruped leg for joint 1 and joint 2 and joint 3 D,
L1, L2 are link lengths of the robot for joint 1 and joint 2 and joint 3.
38
4.6.3. Forward kinematics

With the help of D-H parameters I calculated forward kinematics for leg and body of
the robot.
A. Forward transformation of leg from local reference frame
Here we do the transformation of leg from local 0th frame to 4th frame.
Transformations are briefly described for 0th frame to 4th frame Appendix A, A.1 to
A.4. Final transformation is
(4.53)
,
,
(4.54)
Similarly the transformation for remaining three legs will be same because of
assumption for frame assignment of all leg is same as one which is reducing the
calculation.
B. Generalized transformation from body frame to leg frame
There will be one rotation about y axis for transformation of body to leg frame and
three translations Px, Py and Pz. Rotation from y axis will be same for each leg due to
assumption but the translation will be different for all legs. Magnitude of the
translation will be same for the legs only direction will be different.
about y axis and generalized transformation
is rotation
from body to leg frame that is

(4.55)
(4.56)
39
=is a angle of rotation about y axis .this matrix will be same for all leg
transformation.
=distance from the center of the body to leg frame in x direction
=distance from the center of the body to leg frame in y direction
=distance from the center of the body to leg frame in z direction
,
is different for each leg then transformation of body frame to leg frame for
each leg describe briefly in Appendix A. General transformation is

(4.57)
4.6.4. Inverse kinematics

Inverse kinematics is used to find the joint angle at given desired position. It helps to
find the possible solution for desired point.
A. Inverse kinematics from leg frame
Inverse kinematics is to find the joint angle of the legs. Leg of quadruped has many
solutions for each angle. In the legged robot the all the solution may not be desirable.
So we need an inverse kinematic analysis of the robot. For getting solution from
inverse kinematic equation of the robot, there are many methods available like,
graphical methods, algebraic method and pseudo inverse method. Here I prefer the
algebraic method. From algebraic method we assume a new matrix of 4x4
th
thats
th
equal to 0 frame to 4 frame transformation.

(4.58)
Multiply the inverse of 1st transformation matrix in both side of the equation 4.53
(4.59)
Compare the value of both side of equation 4.49 we found the three equation independent
with
(4.60)
(4.61)
(4.62)
After solving the equations 4.60, 4.61, 4.62 we can find the all joint angle of the leg 1
40
(4.63)
(4.64)
Take square of equation 4.64, 4.62 and add them we found the third joint angle
(4.65)
(4.66)
(4.67)
Multiply with equation 4.62 with 4.6 and equation 4.64 with 4.67, we found the 2nd
joint angle
(4.68)
Here we have two solutions for joint 3 and two solutions for joint 2. Inverse kinematics for
all leg of quadruped is same.

B.
Inverse Kinematics from body frame
In this case all the inverse kinematics solution for all leg should not be same. Here I
am going discuss inverse kinematics for only one leg (leg 1). Method of finding a
solution of leg joint is same as earlier discussed in above sections. Here I also
preferred the algebraic method. From algebraic method we assume a new matrix of
4x4 thats equal to pth frame to 4th frame transformation.
(4.70)
Multiply the inverse of 1st transformation matrix in both side of the equation 4.70
(4.71)
Compare the value of both side of equation (4.71). we found the three equations
independent with r
41
(4.72)
(4.73)
(4.74)
(4.75)
Divide the equation 4.72 and 4.73. We get the first joint angle of the quadruped
(4.76)
Where
b=
or
or
a=
or
b=
or
or
a=
or
,
Here we found the two solution of the joint 1 for 1st leg and take square of the 4.72
and 4.74 and add these equations.
(4.78)
=S-d
(4.79)
(4.80)
Take square of the equation 4.74 and 4.80 and add these equations it we get the joint
angle 3
(4.81)
Multiply the equation 4.80 by P1 and equation 4.74 by P2 then we get the third joint
angle
(4.82)
42
From the body frame we got the 6 number of solution two solution for each joint angle. In
the 6 solution there may be chance occur the redundancy
4.7.
Jacobian: velocity analysis mammalianinspired robot
This section discusses Jacobian matrix for all joints of each legs. Here we have
Jacobians for leg and body frames.
4.7.1. Linear velocity and angular velocity form leg frame

From the equation 4.10 and 4.11
For link0
(4.83)
(4.84)
For link 1
(4.86)
(4.87)
For link 2
(4.88)
(4.89)
For link 3
(4.90)
(4.91)
For end point of link3 and frame 4
There is no joint at the 4th frame then
43
(4.92)
(4.93)
Jacobian for quadruped leg from 0th frame to 4th frame
(4.94)
(4.95)
(4.96)
(4.97)
4.7.2. Jacobian: linear velocity and angular velocity form body

frame
All the joint velocities are independent with the frame so all the joint velocity will be
same from the body frame and leg frame. Only difference comes when we take the
joint velocity form body frame
one rotation matrix will be multiply with
transformation matrix the

(4.100)
We know the joint velocity of 4th frame w.r.t to its own frame from equation 4.74 then
velocity from body frame is
(4.101)
44
(4.102)
4.8.
Dynamic analysis mammalian inspired legs
In the dynamic analysis is same as the explained in section 4.5. Here we are writing
the direct equations. Putting all the values in equation 4.28 and 4.29 we get the
angular acceleration and linear acceleration of all links and joints
For link0
(4.103)
For link 1
(4.104)
For link 2
(4.105)
For link 3
(4.106)
45
(4.107)
For end point of link3 and frame 4
There is no revolute joint at the 4th frame then
(4.108)
But linear velocity will be different due to the angular velocity of joint 3 at 4th frame
is
(4.109)
After getting the accelerations on each joints then we derived the acceleration at
center of the links. From the equation 4.30 we get all the acceleration of center of
each links.
For center of link 1
(4.110)
46
(4.111)
(4.112)
Dynamic force is derivation is based on Newton-Euler equation from equation 4.39,

4.40; we get the all force and moment from each link of the robot
Static force and moment calculation
All the static forces are known because we know force acting the end of the leg. But
we are not considering any inertial force and moment here. Here we are assuming the
force is acting on the leg contact is in all the direction of the Cartesian space that
is
, ,
and moment is zero.
Static force and moment for joint 4
(4.113)
Static force and moment for joint
(4.114)
47
(4.115)
(4.116)
Now net force acting on the each link is

(4.117)
(4.118)
(4.119)
Torque about first joint
(4.120)
Torque about 2nd joint

(4.121)
48
Torque about 3rd joint

(4.122)
+
+
49
CHAPTER 5
GAIT PLANNING AND STATIC STABILITY ANALYSIS

5.1.
Introduction
In order to make the mechanisms walk it is essential to perform a leg and body motion
sequence called gait. Since many years gait has been researched however, for a long
time the results functioned on surfaces with very favourable conditions: flat and level
terrain. The observation, comprehension and mathematical formulation of ordinary
natural gaits were the main focus on legged locomotion and it was found that many of
these motions are periodic in nature.
They categorized the motion into two way periodic and non periodic (free gaits). It
was ultimately found that the motion pattern of the quadruped robot is periodic in
nature. Another characteristic of this type of gaits is continuous gait, in which the
body moves with constant motion while all legs move simultaneously. For an uneven
terrain, mammals and insects generally use a special kind of continuous gait called
wave gait, at a low speed.
Many researchers have developed the mathematical formulation of continuous gait of
quadruped robots. Researchers had defined the duty factor, stroke length and phase
for legged of quadruped robots. In 1968 and 1989 (McGhee and Frank, 1968; Zhang
and Song, 1989) formulated a mathematical model so that they could describe the
sequence of leg and derive the model to perform crab and circular gaits, in case of flat
terrain conditions. They had also developed, in a specific uneven terrain, the
mathematical formulation for continuous gaits.
It was found that the mammal and insects change their gait pattern on sharp
irregular terrains, thus they defined it as a secure gait and it is characterized by the
sequential motion of legs and body [42]. The body is pushed forward/backward with all
of the feet properly positioned on the ground and then leg is moved with all other
three legs and body remaining stationary, these gaits are called discontinuous gaits.
For causing intermittent body motion which is beneficiary to a real legged machine:
they are very easy to implement and they provide longitudinal stability margin which
is greater as compared to wave gaits. Also, they could achieve a faster velocity than
wave gaits.
50
5.2.
Gait synthesis
In this section only I will consider the standard gait sequence which nowadays people
are using for four legged robots. The dictionary meaning of gait is a method of
moving on foot. In the field of legged locomotion, a gait is defined as a repetitive
pattern of foot placements (Todd, 1985). More precise definition given by Song and
Waldron (1989), as follows.
A gait is defined by the time and the location of the placing and lifting of each
foot, coordinated with the motion of the body in its six degrees of freedom, in
order to move the body from one place to another. McGhee and Frank were
focused on the mathematical formulation of gait sequences. These researchers focused
on finding the quadruped gaits to maintain static stability. For this study, McGhee
introduced a notation called the event sequence. An event is defined as a foot
placement or a foot lifting. Quadruped results for event sequence in only six event
sequences (see Figure 5.1 (b)) and illustrated in Figure 5.1 (c)(h). In this example,
with the movement of foot 4, the locomotion cycle starts. Leg numbers are indicated
in Figure 5.1 (a). For numbering the legs we use front to rear, we use even numbers
for right legs and odd numbers for left legs [2].
Tomovic (1961) defined, in case of creeping gait, that it is an n-legged robot and
every support pattern involves at least n1 contact points. Hence, gaits in Figure5.1
are creeping gaits. It has three feet in support. Creeping gaits may be either singular or
non-singular.
Figure 5.1: Quadruped gaits (a) top view of a robot (b) graph of leg sequences (c)(h) sequences[2]
51
Based on existing research on animal locomotion, we can categorize the gaits into two
types, namely, periodic and non periodic (see the Figure 5.2).
Figure 5.2: Quadruped gait types
5.2.1. Crawl gaits

The crawl gait is defined as the quadruped animals lift their one leg at time and all
rest should be contact with the ground. This is most stable gait of the quadruped
robot. This is a low velocity gait of quadruped animals and it is known as the standard
gait. This is also called as the creeping gaits or the crawl gait. There is no universally
accepted definition of standard gaits but crawl and creeping gaits are synonyms for
quadrupeds. It is defined as when of an n-legged robot as a creeping gait and every
support pattern involves at least n1 contact points [43], [2].
In generating discontinuous-periodic gaits for quadrupeds, certain aspects should be
considered:
If one leg in its support phase reaches the rear limit of its workspace (kinematic
limit), this leg should change to the transfer phase to be placed at its front
kinematic limit.
The body is propelled forward with all legs on the ground. After a body motion,
at least one leg should stay in its rear kinematic limit to perform a transfer
phase into the next leg motion.
The leg that is contra lateral and non-adjacent (CNA) to the present transfer leg
should be placed at such a point that after the placement of the transferred leg,
the COG stays on the other side of the line connecting the CNA leg with the
52
transfer leg (see Figure 5.4 (c)) In this way, it will be possible to lift another leg
while maintaining the machines stability.
The sequence of legs should be periodic; this will allow several locomotion
cycles to be joined to follow a path.
Graphical representation of gaits sequences has shown in the Figure 5.5(b) and here
we for crawl gait duty factor .75. Minimum value of duty factor for static stable gaits
is .75 for four legged robots
Figure 5.3 (a): Gait sequencing of crawl gait
Figure 5.3 (b): Gait diagrams of crawl gait
Figure 5.3 (c): Successive gait pattern of a two-phase discontinuous gait [2]
53
5.2.2. Trot Gait

The trot is a gait in which diagonally opposite pairs of feet are alternatively lifted,
swung forward, and again placed on the ground. Twice during each stride, the body is
ballistic and without support. In the case of some larger animals, this ballistic phase
amounts simply to the feet being dragged along the ground, nonetheless the feet are
not supporting the body during this phase of motion. Figure 5.4 (a), Figure 5.4 (b) shows
an illustration of a horse in a trotting gait [44], [2]. .
Figure 5.4 (a): Trot gait sequences
Figure 5.4 (b): Gait diagram of trot gait sequences
5.2.3. Bound Gait

The bound is a running gait used by a few small quadrupeds such as squirrels,
rodents, and dogs. In the bound, support alternates between pairs of legs, with the fore
and hind limbs acting in unison to thrust the body forward (See Figure 5.5(a), (b))
shows a Siberian souslik in a bound gait. As can be observed from this Figure, there is
approximately an 180o phase difference between the sets of front and back legs. All
gaits discussed so far, the bound has the shortest gait period, allowing for frequent
interactions of the legs with the ground. This makes the bound well suited for obstacle
avoidance and for providing propulsion. Simulation analysis first conducted by
Murphy and later validated by Neishtadt and Li showed that certain simple planar
quadruped bounding models are always passively stable for dimensionless body
inertia values of less than one and duty factor is always less than 0.5[2].
54
Figure 5.5(a) Bound gait sequences
5.3.
Figure 5.5(b) Gait diagram of bound gait sequences
Stability analysis of quadruped robot
McGhee and Frank in 1968 had provided the first definition for static stability for
ideal walking robot, which eventually had led to the research on walking-robot
stability. An ideal robot is statically stable if the horizontal projection of its center of
gravity (COG) lies inside the support pattern. For An ideal robot it is important to
have a mass less legs, and system dynamics are assumed to be absent. The idea of
static stability was inspired by insects which use their massless legs to simultaneously
support their body during walking and provide propulsion. Hence, in order to move
the body while maintaining balance, their sequence of steps is arranged to ensure
static stability. The motion of these quadrupeds can be extended only to even terrain,
because the stability criterion is based on the zero moment point which is valid for
that kind of surface only. The mechanical simplicity of these designs makes the robot
useless for real applications. Research on legged locomotion was meant for utilized its
advantage for industrial and social purposes, such trot and gallop will not be meant
for those applications [45].
5.3.1. Introduction
McGhee and Frank in 1968 had proposed the first static stability criterion for an ideal
walking robots at constant speed along a constant direction and over flat, even terrain
.As per the Center of Gravity Projection Method the vehicle is statically stable if the
horizontal projection of its COG lies inside the support polygon (defined as the
convex polygon formed by connecting footprints). Later McGhee and Iswandhi, in
1979 applied this to uneven terrain (by redefining the support polygon as the
horizontal projection of the real support pattern). The static stability margin (SSM)
was defined for a given support polygon as the smallest of the distances from the
55
COG projection to the edges of the support polygon. However, it is complex to

calculate the equation for SS. Zhang and Song (1989) proposed the Longitudinal
Stability Margin (SLSM), defined as the smallest of the distances from the COG
projection to the front and rear edges of the support polygon along the machines
longitudinal axis. SLSM is a go approximation to SSM, and it is simpler to calculate.
Assume polygon and different static stability margins of a walking robot as a nonideal vehicle and inertial effects arise during acceleration, the use of the crab
longitudinal stability margin (SCLSM) defined by Zhangand Song, 1990 will be more
convenient. SCLSM is the smallest of the distances from the COG projection to the
front and rear edges of the support polygon along the machines motion axis. Figure
5.6 shows SSM, SLSM and SCLSM for a given support polygon. Mahalingham et al.
(1989) define the Conservative Support Polygon (CSP) as a subset of the support
polygon, in order to limit the motion of the COG projection to guarantee system
stability in the case of failure of any of the supporting legs. However, the usage of the
CSP is limited to machines with six or more legs using a crawl gait [2].
Figure 5.6: Support polygon and different static stability margins [2]
5.3.2. Static stability margin mathematical formulation for crawl

gait
and
can is depends on the geometry of the robot but It does not depends on
the height of the robot. For calculating the
and
we need to use a simple
geometry of straight line. With the help of straight line equation we can calculate the
and
by taking vertical and horizontal component of line which is
perpendicular to the diagonal line 2, 3.

56
2
6
0
5
Figure 5.7: Static stability margin
When first leg is in air then slop of the diagonal line (1, 2)
(5.1)
Here we assume the center of gravity of the robot is at center of the body. When robot
will walk position of center of gravity of will be change but with the help of sensor
we can get new position of the centroid.
Distance (P) =
(5.2)
Form the Figure 5.7 the line 5. 6 are perpendicular to the line 3, 2. So we can write
(Slope of line 3, 2) * (slope of line 5, 6) = -1
Assume point (
(5.3)
) for line 5, 6 which is passing to the line 2, 3 then we can write

= -1
(5.4)
Then stability margin and longitudinal stability margin is

=
(5.6)
(5.7)
The above stability criteria are all based on geometric concepts; SSM, SLSM and
SCLSM are independent of COG height and consider neither kinematic nor dynamic
parameters. Intuitively speaking, the stability of a non ideal walking machine should
depend on those parameters.
57
5.3.3. Algorithm developed for SSM for quadruped crawl gait

In the section 5.3.2 is briefly defined the static stability margin. With the help of
definition we can measure the stability margin of quadruped robot. Here we are
making mathematical formulation for crawl gait for stability measurement. In the
stability measurement algorithm two cases occurs (see the figure 5.9 and 5.10).
Figure 5.8: Quadruped in static mode all legs on ground
Case1. If robot lifts its first leg or fourth leg

In this case robot lifts his first leg and other three legs make the contact to the ground.
With help of three legs it forms a triangle. For making robot stable the center of
gravity should be inside the triangle if it going outside that means system is unstable.
Here we know the coordinate points of three legs. We assume the point is outside of
the triangle and we are taking the diagonal as reference. If the point is outside the
diagonal side then we can say robot is unstable. To measure the stability margin first
we will derive the equation of the diagonal line. Equation of the diagonal line is
(5.8)
Figure 5.9: Geometric view of quadruped when it lifts its first leg
When robot will lift first leg the CG of robot will be shift. If the cg point is satisfying
the equation that means CG lying on the diagonal line (3, 2), if it is outside of the
58
triangle means right side of the diagonal line (3, 2) we will get so positive value of
stability margin and if it is under the triangle or left side of the diagonal line (3, 2)
stability margin will be negative.
(5.9)
(5.10)
(5.11)
Case2. If robot lifts its second leg or 3rd leg
All the process of calculation is same as explained in case 1 and only difference is that
in this case stability margin will be measured from the outside the polygon
Figure 5.10: Geometric view of quadruped when it lifts its 2nd leg
59
CHAPTER 6
SIMULATION QUADRUPED ROBOT

6.1.
Introduction
This chapter will be discussing about the multibody dynamic simulation (MBD), FEA
analysis of quadruped robot. This simulation chapter will consists of leg simulation of
robot, full body simulation for mechanism testing and torque, speed analysis, energy
analysis, force analysis, stability and FEA analysis of leg and joint of the robot. The
simulation is performed for alligator and mammal inspired robot. For multibody
simulation of I used the Adams multibody dynamics software platform which is
integrated by the Matlab and Pro\Engineer. The simulation steps for MBD analysis
are as following:
Step 1- Initially for simulation of robot we should have simplified CAD model
software (Pro\E, Solidworks etc.) and multibody simulation software (Adams, Webot
etc.) and Matlab. For our project work I used Pro\E 4.0 and Adams view 2012 and
Matlab 2010 b. We did forward kinematics, Jacobian analysis and trajectory planning
therefore prepared simplified model of quadruped robot.The model dimension is same
as the actual robot and kinematic analysis is based the geometry of the actual robot.
All the conditions (mass length, boundary condition, friction, coefficient of
restitution) are near to real value then simulation results will be more realistic and
useful. Simplified models of quadruped are shown below (see Figure 6.1(a), (b))
Figure 6.1(a): Simplified CAD model of mammal
Figure6.1 (b): Simplified CAD model of alligator
Step 2- Import the CAD model into Adams in parasolid format then define all the
properties such as mass, Poisson ratio and Youngs modulus.
Step 3- Define the joints as in developed design of quadruped robot and motion which
will be act as actuators.
60
Figure6.1 (c): Defined joint and motion in Adams
Step 4- After applying the joint and motion we had to define the joint variables into
the Adams. These variables will be defined as an input and output. The input variable
will be angular velocity and output variable will be an angular position.
Step 5- We used Adams control which was directly integrated by the Matlab and the
We prepared the Matlab code using kinematics for each leg that is based on joint
space and Cartesian space and with using Adams control we had generated control
plant which is a bridge between Adams and Matlab(see the Figure 6.1(d))
Step 6- In this step Adams control plat was integrated Matlab using Matlab simulink
model.
Figure 6.1 (d): Plant generated by Adams
61
Step 7- We developed P controller in Matlab and developed implemented the gait

analysis applies the control strategy and gait generation for our robot. We used the
Crawl gait, trot gait and bound gait for quadruped robot.
Figure6.1 (e): Simulink model of close loop control for simulation
6.2.
Feedback control architecture for simulation
Motors used for quadruped robot has the high gear ratio and very high damping value
and I am using velocity as input to the motor and acceleration is constant. Here I used
proportional controller because integral and derivative control is not playing very
much important role for angular position / velocity control. Control block diagram is
shown in Figure 6.2.
Figure6.2: Control strategy for quadruped simulation
62
6.3.
MBD simulation of alligator-inspired robot
In this section the simulation of alligator-inspired quadruped robot is described which

will cover simulation of leg and full body simulation alligator inspired quadruped
robot. Simulated model was created to validate, optimize the design and mechanism
of robot.
6.3.1. MBD simulation of leg

Before doing whole body simulation of the robot, I started with single leg simulation
which helped to design of legged parameters such a leg ratio, required torque for
selection of motors and maximum speed achieved by the robot. Simulation of
alligator inspired quadruped legs have same procedure which is briefly described in
previous section. Here whole simulation was run for 5 sec in Matlab. The input
angular velocity about z and output is angular position about z axis. Defined motion
and joint is shown in Figure 6.3(a).
Figure 6.3(a): CAD model of leg imported in Adams
The whole simulation was run the Intel core i3 processor, 2.40 Ghz CPU. Simulation
results for input and output variable shown in following Figures in joint space. The
simulation result shows the desire value of angular position of the robot is 45o (see the
Figure 6.3(b)) by using P controller and required torque by the leg 1 and leg2 is .25
N-mm (See the Figure 6.3(c)).
63
Figure 6.3(b): Desired angular position achieved
Figure 6.3(c): Torque required for robot legs
6.3.2. MBD simulation of whole body

This section describes the whole body simulation of alligator inspired quadruped
robot. MBD simulation should be near to real parameters which helps to reduce the
reality gap between the simulation and actual. In simulation, materials are defined as
same as material used for fabrication of the robot. Results of simulation showed the
speed, acceleration achieved by the robot, total energy consumption, torque and
reaction force getting from the ground. In the simulation two gait has been
implemented i.e. crawl gait for walking of alligator robot, trot gait for getting
maximum speed of the robot. Simulation showed the maximum speed achieved by the
designed robot is .57 BL/sec. Simulation results of the forces, velocities and power
consumption (see the Figure 6.4(a)-(h))
64
Figure 6.4(a): CAD model of alligator-inspired robot imported in Adams
Figure 6.4(b): Desired angular position achieved by Knee joint
Figure 6.4(c): Desired angular position achieved by Hip joint
Figure 6.4(d): Distance travelled by the robot
65
Figure 6.4(e): Speed achieved
Figure 6.4(f): Angular velocity variation of joint1
Figure 6.4(g): Acceleration variation of the robot
Figure 6.4(h): Reaction force from the ground
66
Figure 6.4(i): Power consumption required by each motors

\
6.4.
MBD simulation of mammal-inspired quadruped robot
Multibody simulation of mammal-inspired quadruped has been done in CAIR lab

DRDO. The simulation platform is based on Adams and Matlab. The simulation is
mainly focused to implement the kinematic, dynamic analysis and walk gait and trot
gait and bound gait. This section is discussed about the legged simulation with
trajectory; dynamics based leg simulation and whole body simulation.
6.4.1. Leg simulation

Quadruped robot has 3 DOF leg one is roll motion and two are pitch motion and due
to pitch motion mammal can go faster than reptilian. Simulation procedure is briefly
discussed in the section 6.1. CAD model was directly imported into the Adams and
material property assigned as selected material in the design which is aluminum. The
helped to predication of joint motion and torque requirement for selection of
actuators. In the leg simulation trajectory plan has been implemented for smooth
motion the leg. Simulation results such as trajectory plan of leg, velocity and torque
are shown in as following (see Figure 6.5(a)-(e))
Figure 6.5(a): CAD model of mammal-inspired leg imported in Adams
67
Figure 6.5(b): Torque required for robot leg at joint 1
Figure 6.5(c): Torque required for robot leg at joint 2
Note: In the plot red colored line is our desired and blue color line is our actual
trajectory and green color is our error in x desire and x actual.
Figure 6.5(d): For straight line trajectory of 3dof leg with via point
68
Figure 6.5(e): For straight line trajectory of 3dof leg with via point
6.4.2. Dynamic leg simulation

Objective of dynamics based legged simulation is to design a torque control actuator.
Dynamics based control will have less position error that kinematics based control.
This control architecture will help to design very precise robots which can help in
dynamic stability of the quadruped robot and some industrial robot which will be used
for polishing and welding application. Control strategy is non linear control based
algorithms. The control strategy is law partition based control for neglecting non
linearity (see Figure 6.6(a) to 6.6(b))
Figure 6.6 (a): Model based control in joint space for dynamic analysis quadruped robot [41]
69
Figure 6.6 (b): Torque control simulation results
6.4.3. Whole body simulation

This section will describe the whole body simulation of mammal-inspired quadruped
robot. The simulation platform was developed in august 2013 and Used high speed
process, Xenon quad core, 3.4 GHz processor. For Mammal inspired robot
simulation three gaits are implemented, crawl, trot and bound. Maximum speed
achieved by the robot was 1.736 BL/sec which was found in bound gait. With the
help of simulation we conclude highest torque is required in crawl gait.
Figure 6.7(a): CAD model of mammal -inspired robot imported in Adams
A. Crawl gait implementation in simulation

70
Figure 6.8(a): Crawl gait sequences
Figure 6.8(b): Angular position variation form the each joint of legs
Figure 6.8(c): Velocity achieved in the crawl gait implementation
71
Figure 6.8(d): Torque required in crawl gait implementation by robot joints
Figure 6.8(e): Reaction forces from the ground crawl gait implementation
B. Trot gait
Figure 6.9(a): Velocity achieved in the trot gait implementation
72
Figure 6.9(b): Torque required in trot gait implementation by robot joints
Figure 6.9(c): Reaction forces from the ground crawl gait implementation
C. Bound gait
Figure 6.10(a): Velocity achieved in the bound gait implementation
73
Figure 6.10(b): Torque required in bound gait implementation by robot joints
6.5.
Finite element analysis of quadruped robot
Finite element analysis (FEA) offers excellent modeling capabilities for individual
components of robot for estimating stresses and strains. Rigid body mechanical
system simulation (MSS) efficiently analyzes large motions of complex systems and
can be used to generate component loads for FEA. Objective of this simulation is to
validate the design and find out stresses and strains at failure point which helps to
select the material of the robot and parameterize the design in the respect of inertia,
loads, and geometry of the robot. The robot simulation is based on Adams Flex,
Nastran and Patran. After doing simulation in Adams we got the reaction forces which
I gave as input as force for FEA analysis.
6.5.1. Leg FEA simulation

Here I did the FEA simulation for leg and got stresses at the joint of each leg. From
the Figure 6.10 (a) we see the stress at joint at zero time is high due to instant load at
the leg end. Factor of safety of designed leg is
FOS=
(6.1)
For shear stress
74
Figure 6.11(a): Maximum shear stress and von misses stress of leg FEA simulations
6.5.2. Body FEA simulation

Objective of body FEA analysis is to know how much factor of safety is coming
which will help to optimization of design and material selection of the robot. Factor of
safety of designed body is
Figure 6.11 (b): Maximum shear stress and von misses stress of whole body FEA simulations
75
6.6.
Stability analysis of mammal-inspired robot
Static stability margin algorithm is briefly discussed in previous chapter. An

algorithm was implemented in Adams to measure the static stability margin of the
quadruped robot.SSM is basically used in walking gait of the quadruped robot. In the
implemented algorithm has assumed the margin at center of gravity of the robot is
zero. Stability margin achieved in mammal inspired quadruped robot is approximately
90 mm.
Figure 6.12: Static stability simulation of the robot
76
CHAPTER 7
FABRICATION AND ASSEMBLY AND CONTROL OF

QUADRUPED ROBOT
7.1. Fabrication of alligator-inspired robot
This chapter will discuss about the fabrication methodology and assembly of
quadruped robot. Material selected for robot is Delrin, PMMA sheets and ABS. FEA
simulation concluded that selected material have enough strength to carry the loads of
the robot. Alligator-inspired robot has 4 legs, body, Arduino shield and Arduino
board. Mostly material used for the fabrication of robot is PMMA. The specification
of the material and component requirement of the robot is as following.
Table 7.1 Material specification
PMMA
DELRIN
ABS
Young modulus (MPa)

Poisson ratio
3200
.29-.40
3120
.35
3200-3500
.35
Density (g/cm3)
1.17
1.4
1.04
Ultimate tensile stress (MPa)
70
91
50-65
Ultimate compressive stress(MPa)
124
96
65
Shear stress (MPa)
42
66
--
Table 7.2 Component required

S No.
1
Parts name
PMMA,Delrin ,3 MM sheets
Quantity
2
2
3
4
5
6
7
8
Servo motors
Arduino board
PCB board
Heat sink
Voltage regulator IC(7085)
Dry cell rechargeable battery (2 Amph) 12 V
Dry cell rechargeable battery 9 v
10
1
2
10
10
1
1
The fabrication of leg and body part of robot has been done on laser cutting machine.
In the design of alligator-inspired robot, body of the robot was divided into two parts,
front and rear. These parts are connected through link with one revolute joint and a
fixed joint. Robot has 4 legs and each leg has two limbs which are based on actual
limb arrangement of alligator, upper and lower limbs were fabricated using same
machine. But unlike real alligators having oblique arrangement of limbs, our robot has
77
its limbs inclined perpendicular to each other. As of now, any 2D design for leg will
suffice. But each leg also has to house its own servo motor which is almost of the
same size as the limb. To accommodate the servo, a sideways protrusion has to be
provided to house of the servo, thus transporting the limb design into 3D domain.
Figure 7.1(a): Left-hand image is that of the targeted robot leg design prototype developed in
Pro/ENGINEER Wildfire 4.0 whereas right-hand image is that of actual robot limb fabricated in 3-D
Rapid Prototype Printers using ABSplus in ivory at about 1.6 kW (RMS)
But manufacturing of any 3D part requires not only a lot of time but is also expensive.
Most commonly used 3D Rapid Prototyping Printers requires considerable time as
in:a) Melting of thermoplastic material ( in our case, ABSplus in ivory)
b) Manufacturing time
c) Removal of support material (in our case, SR-30 soluble) by dissolution in
reagent
To save on resources, the 3D design was transformed into 2D by simply breaking the
upper and lower limbs into two 2D shapes which will fit in via slot-fitting mechanism.
Thus, instead of 3-4 hours, our job could be done in few minutes now in a lasercutting or water-jet cutting machine. To sum up, an extra link had to be provided at
each limb to house the servo. Hence for two limbs at each leg, four rigid links were
used. The parts are fabricated by using CO2 laser cutting machine [46].
78
(i)
(ii)
(iv)
(vi)
(iii)
(v)
(vii)
(viii)
Figure 7.1(b): Part required for robot assembly; (i) Hons require for motor coupling (ii) End of the leg
part fabricate in 3D printer (iii) Screws and nuts required (iv) Front body part (v) Rear body part
(vi)Leg of the robot (vii) battery required (viii) Leg coupling
7.2.
Assembly of the alligator-inspired quadruped robot
After fabrication of the parts of the robot then I started assembly of the robot.
Assembly of robot is very simple and modular. Step for assembly is as following
Step-1 The robot body parts (front and rear parts) connected through a link. One link
which is connected with front part of the body and a revolute joint and other end of
link which is connected through a rear part of the body will be connected through
fixed joint.
79
Step-2 Put the servos on the body of the robot and it will be known as hip servos
for actuation for yaw motion of alligator-inspired robot.
Step-3 Assemble the hip leg limb with servo housing then connects with the hip
servos. Hip leg limb will be connected with the knee servos.
Step-4 Connect the knee servos with knee leg limbs for roll motion.
Step-5 Put the based area on the body to to keep Arduino and shield.
80
7.3.
Fabrication and assembly of mammal-inspired robot at CAIR
Quadruped is being developed at CAIR and their people are doing manufacturing and
assembly of quadruped robot. I was involved in assembly and fabrication of modular
joint of quadruped robot and I was tested a gaits of quadruped robot. Detail
description of modular joint in following section.
7.4.
Modular Joint Description
Quadruped robot has 12 joint. Weight of legs of the quadruped are very heavy so it
generating the high bending and redial load on the each joint of the legs. For reducing
these effects we developed the joints of the each modular leg. Motors have limited
capability to bear these loads in uneven terrain or long journey.
7.4.1. Construction
Basic advantage of modular joint is it has no effects of the radial load and bending
moment. In the construction of the modular joint we separated mainly into two parts.
One part contains the sleeve and other part contains the bearing housing. Sleeve of the
modular joint is fixed to body and legged bracket. In the sleeve we put the spacer, and
then we put the motor in it and fixed the motor with sleeve by putting bolts. For
making our joint more reliable and robust we put the slave card in back side of the
sleeve. Motor controller is fixed such a way I/O of motor connects easily. For suitable
motor wiring we connected it with help of the brackets. And front view of the sleeve
we put the absolute encoder to measure the motor position from the zero. For
connecting absolute encoder we used the two spur gear one we put on the shaft of the
81
motor and other we fixed with potentiometer of the encoder. We placed both gears in
measuring the reference position which means that when motor rotates, the absolute
encoder rotates with help of the spur gear. In the universal joint other part is bearing
house and front Cap. The bearing house is used to keep two bearings and a Circlips to
lock the bearings and motor. Now we put the sleeve in the bearing house and bearing
house. Front cap has the slot which helps to lock it with the motor shaft by key and
we fixed the front cap on the bearing housing by nut and bolt. Now when motor shaft
will rotate bearing house without putting any radial load on the joint. It is shown in
the Figure (7.2 (a), (b)).
Figure 7.2 (a): Assembled modular joint section model
82
Figure 7.2 (b): Assembled modular joint
7.4.2. Absolute encoder

In this section I will explain why we use this additional encoder while we have inbuilt
encoder in the motor. Motor encoder contains incremental encoder so when we give
the power to the motor it takes position as zero from where it starts. That is why we
will not get the reference point for the motor position. If we have not reference point
for the encoder so it will not possible to define the references joint position of the
robot. For getting reference point we use the external encoder. When we powered the
motor it takes the references zero then it will calculate the further angular position of
the joint the define references hence absolute encoder being employed to get the
absolute position. When we will get the absolute position we give this reference to
motor encoder then absolute encoder will not be very much meaningful.
83
CHAPTER 8
CONTROL ARCHITECTURE AND EXPERIMENTAL

STUDY OF QUADRUPED ROBOT
8.1.
Control Plan
Any hardware implementation in robotics requires clarity of the controls it employs.

And controls tend to be multi-layered for complex systems. But here focus was on
gait design, so basic 3-layered control architecture was used. However it must be
pointed out here to avoid future confusion that when planning/control algorithms are
implemented on the alligator-robot, the nature and layers of control will become more
complex.
Firstly, there are lower-level controls (LLCs) in the form of angular positions
of the eight servo motors. Each leg has 2 DOF one due to servo in upper limb for
hip-yaw and another due to servo in lower limb for knee-roll, thus summing up to a
total of 8 DOF for the quadruped with regard to LLCs. Thus the eight LLCs are the
eight servo angles. Each LLC is associated with a range of angles it can turn to and
the control set. The control set is the usual 2-tuple: clockwise and counter-clockwise
rotation by 1 degree.
Secondly, the turning of servo motors controls the hip yaw and knee roll at each
leg. These 4 pairs of angular orientations form the middle-level controls (MLCs). In
the next section, where the kinematics model of the joints and limbs has been
explicitly derived, the MLCs play a pivotal role in the analysis because MLCs map
directly to the robot configuration which is the higher-level control. Like LLCs, each
MLC is associated with a range of angles and corresponding control set. The control
set is still the 2-tuple: clockwise and counter-clockwise rotation by 1 degree. The only
difference will be the phase-shift between LLCs and MLCs, i.e., a servo angle at one
of the upper limb of 37 degree may correspond to a hip yaw of 15 degree depending
on the initial arrangement.
Finally, there are higher-level controls (HLCs) in the form of surge, sway and
yaw. For the robot, choose its body reference point anywhere, say, the centroid of the
rectangular body. Then set up a body coordinate frame (see Figure 8.1).
84
Figure 8.1: The coordinate frame, BODY, used here for kinematic analysis is the same as that used for
standard analysis of any vehicle/locomotion dynamics
Then the 3 DOFs are obvious: surge, sway and yaw. Each HLC is associated with a
range of values it can take and the control set. For surge, the range is theoretically [0,
The robot does not operate in the sway direction directly but only due the
combined effect of surge and yaw. This phenomenon is summed up in the equations
below.
(8.1)
=r
(8.2)
The symbols have usual meaning (Fossen, 2011).However, the range for sway is
theoretically unrestricted. Yaw is also theoretically unrestricted as it is the amount by
which the robot turns. And for convenience of manoeuvrability, no restrictions were
placed. Thus,
LLC: 8 servo
angles
[0,
(8.3)
(-,)
(8.4)
[0,2 ]
(8.5)
MLC: 4 pairs
of hip yaw
and knee
roll
HLC: surge,
sway, and
yaw of body
centorid
Figure 8.2: Control priority level for robot
85
However, due to discretisation of control set related to the three HLCs determined
by the respective geometric/kinematic as well as kinetic constraint on forces and
moments that can be developed, etc., not all configurations are attainable. To control
locomotion, signals sent to the eight servo parameters via Arduino Uno get
transformed into a unique 3-tuple of [x y
]T depending on its initial configuration
and, of course, the signal command. This transformation takes place via the middlelevel controls (MLCs) of hip yaw and knee roll.
Below is a flowchart representation of the algorithm was implemented to move the
alligator forward.
Summing up, there are 8 controls and 3 degree-of-freedom s for our robot and
hence this is theoretically an over-actuated system because control to degree-offreedom ratio, 8/3>1. But, the exact nature of actuation of the system varies over time
due to:a.) Presence of sufficient coupling between upper and lower limb of each leg,
b.) Symmetricity and periodicity constraints laid down by the gait diagram.
c.) Inertia of rigid body and lack of compliance.
d.) Wear and tear.
e.) Manual modification of robot body.
f.) Change of operating environment of robot.
Thus, the nature of actuation of a robot is not simply a function of its control to
degree-of-freedom ratio; it also relies on other factors mentioned above.
8.2.Motion control of alligator-inspired quadruped robot
Motion control of quadruped robot is based on the servo feedback control. Design
architecture of motion control is shown in the Figure 8 servos are directly connected
by the Arduino shield and Motor shield is directly connected to Arduino board. This
is 3 layer based control which is briefly described in the previous section.
86
Figure 8.3: Flow chart communication for master, slave card and PC of quadruped
8.2.1. Hardware descriptions for motion control

To design a motion control I used Arduino UNO board which has inbuilt
Atmega256 controller and to control 8 servos from Arduino is very difficult
because of voltage and current distribution should be uniform and current should
be same for each servo if it is not happening then we will not able to synchronize
the gait properly. Component requirement of motion control is discussed in
previous chapter.
a. Arduino shield
Circuit diagram of Arduino shield is developed in Proteus 7.0 and simulation of
circuit has been done for testing of current distribution. Shield contains 8 voltage
regulator IC (7805), heat sink and PCB board, switch and mail female connectors.
Figure 8.4(a): Proteous simulation of shield
87
Figure 8.4(b): Shield developed front view
Figure 8.4(c): Shield back view
b. Arduino UNO
Arduino is used as a tool, which is used to sense and control the real world in a better
way than your computer simply. It is platform which is easily available on the
internet. Its schematic is also given on the internet. Any expert can modify the
circuitry or code of this software this is open to all software so called open-source
software.
Arduino can detect the physical real world by using different type of sensors
and control the surroundings by controlling actuators. This board is based on the
ATmega 32/16 and programmed using Arduino programming language. Arduino can
connect with other programming language by serial port and program directly can be
written in that language and can be burn on the Arduino board.
Figure 8.4(d): Arduino UNO
c. 7805 IC description
Voltage regulator IC has been design for very vast application for high voltage IC can
draw the 1.5 A output current. IC enables itself on min 5 volt. 7805 has 3 pins one one
88
is used for ground and two others are input and output voltage. Each of these
regulators can deliver up to 1.5 A of output current. It can carry the current but due to
this it generates lot of for which we have used heat sink. IC will work for 34V, 20 A
current.
d. Control algorithm
The trot gait is implemented for robot motion and algorithms is described below in
detail. The key to understanding the gait algorithms is that the notation RF_HS
represents Right Front Hip Servo and so on for Left Hind and Knee Servos.
Figure 8.5: The algorithm to move the alligator forward is summed up as a flowchart. The actual
amount of actuation to be given and the time delay for the different actuations involved was based on
gait diagram described earlier. This flowchart was generated using Flow Breeze, a flowchart
automation add-in for MS-Excel.
8.3.
Motion control of mammal inspired quadruped robot
3-channel Input Capture (IC) Module and for motion control of quadruped robot we
will be based on the slave and master control. Here in the Figure 8.6 we can see the
flow of communication to motion and the controller. Computer will be connected to
the master controller by serial communication and all slave will be communicated to
the master using CAN bus.
89
Figure 8.6: Flow chart communication for master, slave card and PC of quadruped
8.3.1. Hardware Description for motion control

To design a motion control only I designed for single axis. I reproduced single axis
motion control card (slave card) in OrCAD. Single axis motion control includes
design of slave card for joint motor and master card for coordination of all slave joint
and interface to the host computer. Slave and master card coordination bus would be
CAN bus and host communication is through RS232 serial bus. For motion control I
will do position and velocity control of the high gearing PMDC motor. Motion will be
open loop and close loop both. In the open loop motion control, it will be based on
using interrupts timer and PWM and in the close loop control will be based on the
servo loop error and motion will be followed by a particular trajectory (cubic,
quadratic or quintic).Closes loop control will give the positional feedback with the
help of encoder sensor. For designing a motion control parts are required are
Table 8.1 Requirement for single axis motion control
Item No.
1
2
3
4
5
6
7
8
Parts required for single axis motion

control
PIC micro controller
Motor driver
MPLab software (Compiler etc.)
PIC Programmer IDC 3
Power supply
Motor (PMDC, Servo Motor)
Oscilloscope
Multi meter
90
QTY.
1
1
1
1
1
1
1
1
8.3.2. Parts descriptions and specifications

Parts descriptions is as following
A. Controller card
For designing single axis motion control hardware I used the PIC series 8 bit 18F
microcontroller. Microcontroller is PIC 18F2431. It has 28 pins microcontroller.
Figure 8.7: (a) Slave card
B. Motor
For single axis motion control I used maxon series Swiss made PMDC motor. Its
torque output is 30N/m. Motor contains three basic parts planetary Gear head,
brushless motor, and encoder.
Planetary Gear
Gear box is used to increase the motor torque. Gear train mechanism is based on spur
and epicyclical planetary gear heads which is matched with motor shaft. It provides
very high torque which mostly used for industrial and R&D purposes. The gear ratio
of the robot is 394:1 and motor weight is 960g.The torque output from the motor is
30Nm and efficiency is 68 %.
Brushless Motor
Motor is brushed motor with powerful permanent magnets. It has used iron less rotor.
It advance motor based on current technology which is compact, with low inertia
drives, Motor torque is .895Nm and speed is 2800rmp at24 Volt, 1 Amps. It has 92 %
load efficiency.
Encoder
High precision encoders, DC tachometers and resolvers with a high signal resolution
are mounted exclusively on motors with through shafts for resonance reasons. The
assembly requires adjustment to the motors and may only be done in the delivery
91
plant. Encoder is Encoder MR 256-1024 3 channel and count per turn is 256 and max
speed is 37500rpm.
Figure 8.7: (b) Motor
C. Motor driver
For controller the motor the driver is required. In this motion control hardware I used
LMD series motor driver. I used LMD18200 driver. Features, applications,
descriptions and pin diagram given as following
Features
a. Numerically Controlled Machinery
b. No Shoot-Through Current
c. Thermal Warning Flag Output at 145C
d. Thermal Shutdown (Outputs Off) at 170C
e. Delivers Up to 3A Continuous Output
f. Operates at Supply Voltages Up to 55V
g.
Low
(ON) Typically 0.33 per Switch at 3A
h. TTL and CMOS Compatible Inputs

i. Shorted Load Protection
j. Internal Clamp Diodes
k. Internal Charge Pump with External Bootstrap Capability
Descriptions
The LMD18200 is basically a 3A H-Bridge which is designed for motion control
applications. In order to built the device we use a multi-technology process which
combines bipolar and CMOS control circuitry with DMOS power devices on the same
monolithic structure. Ideal for driving DC and stepper motors; the LMD18200
accommodates peak output currents up to 6A. An innovative circuit which facilitates
low- low-loss sensing of the output current has been implemented.
92
Applications
a. Computer Printers and Plotters
b. DC and Stepper Motor Drives
c. Position and Velocity Servomechanisms
d.
Factory Automation Robots
e. Numerically Controlled Machinery
8.3.3. Software description of single axis motion

For developing single axis motion control first we should have better understanding of
PWM, registers configuration, PWM control at deferent duty cycle, interrupt, and
timer. For developing motion control I will follow the following steps.
1. Internal resister configuration
2. Led blinking
3. PWM generation
4. PWM generation with duty cycle
5. Open loop motion control using interrupt , timer and PWM
6. Encoder configuration
7. Closed loop control using encoder
8. Closed loop control based on servo loop and trajectory loop
9. Use Master card and Host computer for motion control
10. Can bus interfacing and RS232 interfacing with master slave and PC
A. Internal resister configuration for PIC 2431 microcontroller
To start with microcontroller first we have define some basic configurations, without
this microcontroller will not be work properly. These configurations are like oscillator
selection power mode and watch dog timer mode or register mode. Writing of the
configurations is depends on the type of compiler. Here I used Microchip C18
compiler. With C18 compiler I have define the controller configuration.
#include<p18f2431.h>
#pragma conFigure OSC =HS
#pragma conFigure PWRTEN = OFF
#pragma conFigure BOREN = OFF
#pragma conFigure WDTEN = OFF
#pragma conFigure LVP = OFF
93
B. PWM Generation with PIC 2431 /dsPIC4012 microcontroller

The power control PWM module simple makes it easier to generate multiple,
synchronized pulse width modulated (PWM) outputs, use in the control of motor
controllers and power conversion application. In particular, the following power and
motion control applications are supported by the PWM module:
Three-phase and Single-phase AC Induction Motors
Switched Reluctance Motors
Brushless DC (BLDC) Motors
Uninterruptible Power Supplies (UPS)
Multiple DC Brush Motors
PWM can be generated with two ways one is PWM with fixed duty cycle and other is
variable PWM duty cycle. For PIC2431/ RA0-5 are for fixed duty cycle and RC2 is
for variable duty cycle. For defining the different duty cycle first we have to define
the PR resister and CCPR resisters the mathematical formulations is given as
PWM period = ((PR2+1)*4*
*TMR2 *Prescaler value)
PWM duty cycle = (CCPR1L:CCP1CON<5:4>) *Tosc* Prescale )
(8.1)
(8.2)
Algorithm for PWM generation

I took the PWM at minimum frequency 2 kHz and oscillator frequency is 2 MHz and
prescalar value is 16 the PR2 resister value is 156.
94
Figure 8.8: (a) PWM generation
C. Servo loop and interrupt based motion control

The basic elements required to implement a DC Servo motor controller are: Joint
Level Control (implemented in the slave).
Motor Feedback:
Incremental optional encoder is used, which has slotted infrared emitter detector pairs.
Disk rotation in IR pulses which must be handled-pulse counting, timing and
quadrature yields the disks angular position, velocity and direction respectively.
Servo Loop:
P, PI or PID control algorithm can be used to maintain the position and speed control.
Motor Input:
Motor can be given input from a digital or PWM output from the controller. Higher
Level Control (implemented in the master/PC) Overall robot control mechanism.
95
Algorithms for encoder testing
Figure 8.8: (b) Encoder testing
D. Motor Position control

The P-control algorithm is used to maintain the position and speed control. The basic
concept of servo loop involves error calculation and determination of the absolute
value of the PWM duty cycle. The error is calculated by subtracting the current
position from the target position. The current position is updated each time by reading
the encoder count. The timer2 post-scalar is used to have a servo update rate at a
different frequency than the PWM output. The absolute value of the duty cycle is
calculated by multiplying the error with the proportionality constant .This duty cycle
is used to generate the PWM which is given as the input to motor. The value of the
proportionality constant depends upon the dynamics of the system and is given from
the PC along with the target position.
96
Algorithms for servo loop based motion control
97
Figure 8.8: (c) Servo loop based motion control
98
a. Algorithm for Servo Loop and Trajectory Based Motion Control

Start
VREGH
CAP1BUF
T2CON = 124
PR2 = 255
TMR2IE = 1;
TMR2IP = 0;
av=0;error=0,error_p=0;
error_i=0;No_rotation;
Yes
No
LED
joint_move_linear
(360);
99
End
ENCODER_ISR
Interrupt
high
TRISC=0
TRISB=0
TRISA=0
QEICON=60
MAXCNT=65536
POSCONT=0
MAXCANT=POSCANT
IC2QEIF=1
UPDOWN
=1
T0CON = 0x88;
TMR0H = (655362500)>>8;
TMR0L = (655362500) & 0xFF;
TMR0IF = 0;
TMR0IE = 1;
TMR0IP = 0;
LED Toggle
IC2QEIF=0
No. of rotation
count++
IC2QEIF=0
IC2QEIE=1
IC2QEIP=1
IPEN=1
GIE=1
PIE=1
100
No. of
rotation
count-- -
4
Trajectory
Update
3
SERVO_LOOP
pwm,
long int tf = 2000;

av = av0 +
(longint)((avfav0)*t*1.0/tf);
t++;
if
(t>tf)
t =
tf;
error_p
kp = 0.6
kd = 0
ki = 0.0001
enc_cnt=POSCNT+65536*no. rotation
error_p=error
error=av-enc_cnt
error_i=error_i+error
pwm=kp*error+kd(error-error_p)+error_i*ki
t = tf;
pwm>1023
pwm=1023
joint_move
_linear()
pwm<1023
pwm=-1023
av0 = ((long int) POSCNTH

<< 8) + ((long
int)POSCNTL) +
no_of_rotation*65536;
Pwm<0
((long int)
(jv*(169.0/9.0)*
4*256/360.0))
pwm=-pwm
RC0=1
CCPR1L=pwm>>2
DC0B=pwm&0b11
101
RC0=0
2
SERVO_ISR
Interrupt Low
TMR2IF=1
3
SERVO
LOOP
TMR20F=0
TMR0IF=1
4
Trajectory
Update
TMR0IF = 0;
TMR0H = (65536-2500)>>8;
TMR0L = (65536-2500) & 0xFF;
Figure 8.8: (d) Servo loop and trajectory based motion control
102
After implementation of the servo loop and trajectory loop for single motor I
implemented different gaits on actual robot. I used dynamixel servomotors motion
control the robot. I implemented the crawl, trot and bound gaits on quadruped robot.
Figure8.9 (a): Single axis motion control hardware
Figure8.9 (b): Motion control implemented
0.6
0.6
0.5
0.5
Speed (BL/sec)
Speed (BL/sec)
8.4. Experimental study of alligator-inspired robot

Alligator inspired robot has been developed and we did some experiment on speed
achieved by the robot to see the effect of angle of knee and hip. Maximum speed
achieved by the robot is 0.51BL/sec. The following Figure 8.10(a), (b) shows the
effected of the angle and delay time. Here these plots are showing the variation of
with delay time between next step of robot which reducing with increasing in delay
time and plot Figure 8.10(b) which shows the speed is decreasing when decreasing
the hip angles.
0.4
0.3
0.2
0.1
0.4
0.3
0.2
0.1
0
0
10
15
Delay time(Sec)
Figure 8.10(a): Robot speed vs delay time
20
40
60
Hip Angle Amplitude(Deg)

Figure 8.10(b): Robot speed vs Hip angle amplitude
103
CHAPTER 9
CONCLUSION AND FUTURE SCOPE

9.1.
Conclusion
The work describes the design, modelling, simulation and fabrication of an alligatorinspired quadruped robot. Multidisciplinary design approach is applied to develop
quadruped robot. After MBD and FEA simulation we have concluded our design
robot is in safe at applied loads and material selection and robot mechanism is correct
for deferent gaits and terrains. Also designed and implemented was the gait plan of
the developed robot based on high gait pattern of alligators.
Research in gait design can be engaged in two directions; either explore design
optimisation to match moving characteristics between a robotic alligator and real
alligators as closely as possible for the same gait diagram or given a scaled alligator
model whose locomotion is inspired by alligators anatomy, design a gait diagram that
mimics different moving characteristics of real alligator as closely as possible. The
former idea was experimented with but focus was laid on the latter idea.
The reptilian robot was tested to move robustly on soils as well as pavements. The
successful application of gait showed that indeed efficient, controlled and guided
walking could be achieved on flat terrains of different nature.
Amphibot
KOLT
Cheetah
Alligator
Big Dog
HyQ
Tekken II
Aibo
Scout
Chetah-cub
Quadruped
2020
2015
2010
2005
2000
1995
1990
1985
1980
1975
0.55
0.6
0.8
0.52
1.8
2.4
3.2
2.33
6.9
2.8
ETH Standford MIT
IIT P
BD
IIT
KIT
JP
McGill
EPFL
MIT
Figure 9.1 our current status of research
104
9.2.
Future scope of work
Design modeling and simulation has been done and a platform is developed. Some is
require to such
Dynamic stability analysis and simulation
Experimentation and testing for static stability measurement
Free gait algorithms is need to done for highly rugged terrains
Compliant analysis and speed measurement is also the future scope
Design modification of alligator inspired robot to achive more than 1 BL/sec
speed
105
References
[1] M.H. Raibert Legged Robots That Balance MIT Press, 1986.
[2] Pablo Gonzalez de Santos, Elena Garcia and Joaquin Estremera Quadrupedal
Locomotion An Introduction to the Control of Four legged Robots by SpringerVerlag London Limited 2006
[3] F.E. Fish, P.B. Frappell, R. V. Baudinette, P.M. Macfarlane, Energetics of
Terrestrial Locomotion of the Platypus Ornithorhynchus Anatinus, The Journal of
Experimental Biology ,2001, 204, pp.797803.
[4] R.W. Blob, A.A. Biewener Mechanics of limb bone loading during terrestrial
locomotion inthe green iguana (iguana iguana) and American alligator (Alligator
Mississippiensis), The Journal of Experimental Biology,(2001), 204, pp.10991122.
[5] J.M. Parrish, The Origin of Crocodilian Locomotion, Paleobiology, 1987, Vol.
13, No. 4, pp. 396-414.
[6] S.M. Reilly, J.A. Elias, Locomotion in alligator mississippiensis: kinematic
effects of speed and posture and their relevance to the sprawling-to-erect paradigm,
The Journal of Experimental Biology, 1998, 201, pp. 25592574.
[7]J.R. Hutchinson, V. Allen, K.T. Bates, Z. Li, Linking the evolution of body
shape and locomotor biomechanics in bird-line archosaurs, Nature, 2013, 497
(7447):104-7.
[8] J. S. Willey, A. R. Biknevicius, S. M. Reilly, K. D. Earls,The tale of the tail:
limb function and locomotion mechanics in Alligator mississippiensis, Journal of
experimental biology, 2004, 207(3), 553-563.
[9] D. Floreano, C. Mattiussi, Bio-Inspired Artificial Intelligence: Theories,
Methods, and Technologies. The MIT Press, 2008.
[10] J. F. Vincent, O. A. Bogatyreva, N. R. Bogatyrev, A. Bowyer, A. K. Pahl,
Biomimetics: its practice and theory. Journal of the Royal Society Interface, 2006,
3(9), 471-482.
[11] A.E. Rawlings, P.J. Bramble, S.S. Staniland, Innovation through imitation:
biomimetic, bioinspired and biokleptic research. Soft Matter, 2012, 8, 6675-6679.
[12] B. Webb, T. Consilvio, Biorobotics 2001 Mit Press.
[13] K. Williams, Amphibionics: Build Your Own Reptillian Robot(1 ed.).
McGraw-Hill, Inc., New York, NY, USA, 2003, pp. 191-269.
106
[14] T. McGeer Passive dynamic walking. Int. J. Rob. Res. 9, 2 (March 1990), pp.
62-82.
[15] Q. Huang, K. Yokoi, S. Kajita, K. Kaneko, H. Arai, N. Koyachi, K. Tanie,
"Planning walking patterns for a biped robot," Robotics and Automation, IEEE
Transactions on , vol.17, 2001, no.3, pp.280,289.
[16] Mori, Makoto, and Shigeo Hirose. "Development of active cord mechanism
ACM-R3 with agile 3D mobility." In Intelligent Robots and Systems, 2001.
Proceedings. 2001 IEEE/RSJ International Conference on, 2001, vol. 3, pp. 15521557.
[17]Kouno, Kentarou, Hiroya Yamada, and Shigeo Hirose. "Development of ActiveJoint Active-Wheel High Traversability Snake-Like Robot ACM-R4. 2." Journal ref:
Journal of Robotics and Mechatronics 25, 2013, no. 3 559-566.
[18] Takaoka, Shunichi, Hiroya Yamada, and Shigeo Hirose. "Snake-like active wheel
robot ACM-R4. 1 with joint torque sensor and limiter." In Intelligent Robots and
Systems (IROS), 2011 IEEE/RSJ International Conference on, 2011, pp. 1081-1086.
[19] A. Crespi, A. Badertscher, A. Guignard, and A.J. Ijspeert: Swimming and
crawling with an amphibious snake robot, Proceedings of the 2005 IEEE International
Conference on Robotics and Automation, 2005, pp 3035-3039.
[20] A. Crespi and A.J. Ijspeert: AmphiBot II: an amphibious snake robot that crawls
and swims using a central pattern generator, Proceedings of the 9th International
Conference on Climbing and Walking Robots, 2006, pp 19-27.
[21] Yu, Shumei, Shugen Ma, Bin Li, and Yuechao Wang. "An amphibious snakelike robot with terrestrial and aquatic gaits." In Robotics and Automation (ICRA),
2011 IEEE International Conference on, 2011, pp. 2960-2961.
[22] Harada, Kanako, Sheila Russo, Tommaso Ranzani, Arianna Menciassi, and
Paolo Dario. "Design of Scout Robot as a robotic module for symbiotic multi-robot
organisms." In Micro-NanoMechatronics and Human Science (MHS), 2011
International Symposium on, 2011, pp. 511-513.
[23] Drenner, Andrew, Ian Burt, Tom Dahlin, Bradley Kratochvil, Colin McMillen,
Brad Nelson, Nikolaos Papanikolopoulos et al. "Mobility enhancements to the scout
robot platform." In Robotics and Automation, 2002. Proceedings. ICRA'02. IEEE
International Conference on, vol. 1, 2002, pp. 1069-1074.
107
[24]Stoeter, Sascha A., Ian T. Burt, and N. Papanikolopouslos. "Scout robot motion
model." In Robotics and Automation, 2003. Proceedings.2003.
[25] Harada, Kanako, Sheila Russo, Tommaso Ranzani, Arianna Menciassi, and
Paolo Dario. "Design of Scout Robot as a robotic module for symbiotic multi-robot
organisms." In Micro-Nano Mechatronics and Human Science (MHS), 2011
International Symposium on, 2011 pp. 511-513.
[26] Kimura, H., Fukuoka, Y., and Cohen, A. H. (2007). Adaptive dynamic walking
of a quadruped robot on natural ground based on biological concepts.International
Conference on, 2003, vol. 1, pp. 90-95
[27] Fukuoka, Y., and H. Kimura. "Dynamic locomotion of a biomorphic quadruped
Tekkenrobot using various gaits: walk, trot, free-gait and bound." Applied Bionics
and Biomechanics 6, 2009, no. 1,pp 63-71.
[28] Li, Yibin, Bin Li, Jiuhong Ruan, and Xuewen Rong. "Research of mammal
bionic quadruped robots: a review." In Robotics, Automation and Mechatronics
(RAM), 2011 IEEE Conference on, 2011 pp. 166-171.
[29]Zhou, Xiaodong, and Shusheng Bi. "A survey of bio-inspired compliant legged
robot designs." Bioinspiration & biomimetics 7, 2012: no. 4.
[30] Chen, Xianbao, Feng Gao, Chenkun Qi, and Xianchao Zhao. "Spring parameters
design to increase the loading capability of a hydraulic quadruped robot." In
Advanced Mechatronic Systems (ICAMechS), 2013 International Conference on,
2013, pp. 535-540.
[31] Estremera, Joaquin, and Kenneth J. Waldron. "Leg Thrust Control for
Stabilization of Dynamic Gaits in a Quadruped Robot." In Romansy 16, 2006, pp.
213-220. Springer Vienna.
[32] Zhou, Xiaodong, and Shusheng Bi. "A survey of bio-inspired compliant legged
robot designs." Bioinspiration & biomimetics 7, 2012, no. 4.
[33] Hyon, S. H., Emura, T., and Mita, T. (2003b). Dynamics-based control of a onelegged hopping robot.J. Syst. Control Eng.,2003, 217 (2):8389.
[34] Li, Yibin, Bin Li, Jiuhong Ruan, and Xuewen Rong. "Research of mammal
bionic quadruped robots: a review." In Robotics, Automation and Mechatronics
(RAM), 2011 IEEE Conference on, 2011, pp. 166-171. IEEE, 2011.
International Journal of Robotics Research, 26:475490.
108
[35]Boston Dynamics Corp. (2008). Bigdog overview. (Online)

http://www.bostondynamics.com/img/ BigDogOverview.pdf.
[36]Boston Dynamics Corp. (2010). http://www.bostondynamics.com.
[37] C. Prahacs, A. Saunders, M.K. Smith, D. McMordie, M. Buehler, Towards
Legged Amphibious Mobile Robotics, Journal of Engineering Design and
Innovation, 2005, vol. 1.
[38] IRIS AQUA 2004 Photos, [online],
http://quintessence.cim.mcgill.ca:8080/AQUA/photos/index_html
[39] A.J. Ijspeert, A. Crespi, D. Ryczko, and J.M. Cabelguen, From Swimming to
Walking with a Salamander Robot Driven by a Spinal Cord Model, Science,
From Swimming to Walking with a Salamander Robot Driven by a Spinal Cord
Model, 2007, vol. 315, no. 5817, pp. 1416-1420
[40] Vivian Allen, Ruth M. Elsey, Nicola Jones, Jordon Wright and John R.
Hutchinson, Functional specialization and ontogenetic scaling of limb anatomy in
Alligator mississippiensis, Journal of anatomy, 2009, pp 423-445
[41] JJ craig Introduction to Robotics Mechanics and Control 3rd edition, 2004, by
Pearson publication
[42] Gonzalez de Santos, P., Jimenez, M., and Armada, M. Dynamic effects in
statically stable walking machines Journal of Intelligent and Robotic Systems, 1998,
23(1), pp 7185. [43] Gonzalez de Santos, P. and Jimenez, M. Generation of
discontinuous gaits for quadruped walking machines Journal of Robotic Systems,
1995, 12(9), 599-611
[44] R. McN. Alexander Elastic Mechanism in Animal Movement Cambridge
UniversityPress, Cambridge, 1988.
[45] S. Yazdekhasti, F. Sheikholeslam, M. Ghayour, Stability analysis of biped
robot with direct control of zero moment point. The 2nd International Conference
on Computer and Automation Engineering (ICCAE), 2010, Vol. 2, pp. 528-532
[46]Shaurya Shriyam, Anand Mishra, Debashish Nayak, Atul Thakur Design,
Fabrication and Gait Planning of Alligator-inspired Robot International Journal of
Current Engineering and Technology, 2014, pp.567-575.
109
PUBLICATIONS
Published Paper
S. Shriyam, A. Mishra, D. Nayak and A. Thakur Design, Fabrication and Gait
Planning of Alligator-inspired Robot International Journal of Current Engineering
and Technology, 2014, 567-575, 2277 - 4106
Mishra A.K., Raina R., Yadav S.B., Sarangi S., Mathematical Modeling of
Electromagnetic Levitation Ball using Bond graph in 1st international and 16th
national conference on machine and mechanism (iNaCoMM-2013), IIT Roorkee,
India, 18-21 Dec 2013.
Augustine J., Mishra A.K. and K. Patra, Mathematical Modeling and Trajectory
Planning of a 5 Axis Robotic Arm for Welding Applications, Proceedings of the
3rd International Conference on Production and Industrial Engineering (CPIE
2013), NIT Jalandhar, India, 29-31 March, 2013.
Accepted abstract
A. Mishra and A. Thakur, Multidisciplinary Design Approach and Fabrication of
Alligator-Inspired Robot in 5th International & 26th All India Manufacturing
Technology, Design and Research (AIMTDR-2014), IIT Guahati, India 12-14 Dec
2014.
Accepted Paper
Mishra A.K, Kumar R., Sarangi S., Mathematical Modeling of Electromagnetic
Levitation Based Active Suspension System in 2nd International Conference on
Mechanical,
Automotive
and
Materials
Engineering
(CMAME
2014)
Singapore,26th -28th May 2014.

Poster Presented
A. Mishra, A. Thakur, Multidisciplinary Design Approach and Fabrication of
Alligator-Inspired Robot, Poster Presentation, Research Scholar Day, IIT Patna,
India, 8th March, 2014.
110
Appendix. A
1. Kinematics transformation of mammalian-inspired
Transformation the first link form 0th frame to 1st frame for joint 1
For Leg 1
(A.1)
Similarly transformations form 1st frame to 2nd frame for joint 2.
(A.2)
Transformations form 2nd frame to 3rd frame for joint 3.
(A.3)
Transformations form 3rd frame to 4th frame for end point of the leg.
(A.4)
2. Leg convention from body frame

A. LEG 1
From Figure 4.5 (a), (b) we obtain the value of ,
=
=
111
and
=
= 900
we can find the transformation from Pth frame to 0th frame is
Whole
transformation from body frame pth to 4th frame is

(A.5)
is be same for all legs.
B. LEG 3
=
=
=
= 90
C. LEG 2
=
=
=
= 900
D. LEG 4
=
=
=
= 900
3. Dynamic modelling
Having computed the linear and angular accelerations of the mass center of each link,
we can apply the NewtonEuler equations to compute the inertial force and torque
acting at the center of mass of each link. Thus we have
112
(A.6)
Here
inertia, m mass of the
link
Inertial Force and Torque for Link 1

(A.7)
(A.8)
(A.9)
Having computed the forces and torques acting on each link, we now need to
calculate the joint torques that will result in these net forces and torques being applied
to each link.Net force on the link will be equal to the initial force. So we can write the
net inertial force will be equal to be net static force acting joint of the link this is
similar to the moment also. We can write it
(A.10)
(A.11)
113
Name-Anand Kumar Mishra

Email:anand.iitpatna@gmail.com
Mob; 72092362643
Career Objective
To pursue a career path that provides learning opportunities in my areas of interest
and meaningful avenues for adding value to the organization, self and the society.
Education
Pursuing Master of Technology in Mechatronics Engineering from Indian Institute
of Technology, Patna with an aggregate 9.11 CGPA (till 3rd semester).
Bachelor of Technology in Mechanical Engineering in 2012 from Gautam Buddha
Technical University, Uttar Pradesh with an aggregate 72.8%.
Experience
Working as Teaching Assistant since 24th July, 2012 to present at Indian
Institute of Technology, Patna.
Worked as Project assistant at Center for Artificial Intelligence and Robotics,
DRDO Lab Bangalore form 18th June 2013 to 16th January 2014.
Area of Interest
Multibody Dynamics simulation (Adams View with control, durability, vibration
and Flex), Kinematics, Dynamics and control of robots, Mechanism and Machines,
Design and Modeling of Micro and Nano electromechanical Devices, Automotive
Mechatronics, Modeling of Mechatronics Systems.
Technical Proficiency
Basic CAD Softwares- AutoCAD, Pro/Engineer, Solidworks
Basic FEM software Patran, Nastran
Microcontrollers AVR series (Atmega 8 and 16) and PIC series(2431,
dspic30F4011) and Arduino Uno board
Basic Multibody Dynamics softwares-Adams, Working Model
Basic Computational and simulation software-Matlab, Simulink
Basic C and PLC programming
Data Acquisition software National Instruments LabVIEW
Bond graph Modelling-symbols
Basic Circuit Designing- Circuit wizard ,OrCAD, Proteus
Manufacturing skills- Rapid prototyping(3D printer), Co2 Laser cutting machine,
CNC machines
Mentorship
Mentoring several undergraduate students and postgraduate students for the Mobile Robotics
Project during my M.tech final at IIT Patna in spring 2014.
Achievement and Co curricular Activities
Highest CGPA in M.tech (9.8/10) in 3rd Semester.

Silver Medal (2nd Rank) for Best B.tech Project Award in mechanical department Design and Development of Robotic Arm based on Haptic Technology.
GATE 2012 with 98 percentile.
GATE 2011 with 95.45 percentile.
Many prizes in robotics field at IITs and State level institutes. Developed a rescue
robot at IIT Kanpur technical event and developed Stair climber robot and mesh
climber robot, snake robot and robotic war event.
114

Anand Kumar Mishra - 1211MT02 - 2012 Batch PDF

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Anand Kumar Mishra - 1211MT02 - 2012 Batch PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

Design, Simulation, Fabrication and Planning of BioInspired Quadruped Robot

Indian Institute of Technology Patna

Copyright Anand Kumar 2014

INDIAN INSTITUTE OF TECHNOLOGY PATNA

ANAND KUMAR MISHRA

Name: 2 ___________________, Supervisor

Name: 3 ___________________, Co-Supervisor

Name: 4 ___________________, Member

Name: 5 ___________________, Member

Name: 6 ___________________, Member

Name: 7 ___________________, Member

Name: 8____________________, Member

(Anand Kumar Mishra)

Signature of the Student

Introduction to legged robots

Brief review of legged robots

Advantages of legged robots

Disadvantages of legged robots

Application of legged robots

1.2 Introduction to bio-inspired robotics

1.3 Introduction to quadruped robot

1.5 Organization of thesis

Active cord mechanism series

2.3 Legged robots

DESIGN METHODOLOGY OF QUADRUPED ROBOT

3.2 Design of alligator inspired robot

3.3 Design description of mammalian inspired robot

KINEMATIC AND DYNAMIC ANALYSIS AND TRAJECTORY PLANNING

4.2 Kinematic modeling of alligator inspired robot

4.3 Jacobian: velocity analysis of alligator inspired robot

4.4 Trajectory planning for quadruped leg

4.4.1 Straight line trajectory of quadruped leg with via points

4.4.2 Cubic trajectory of quadruped leg with via points

4.5 Dynamic analysis of quadruped robot

4.6 Kinematic analysis of mammalian inspired robot

4.6.1 Frame assignment

4.6.2 D-H parameters

4.6.3 Forward kinematics

4.6.4 Inverse kinematics

4.7 Jacobian: velocity analysis of mammalian inspired robot

4.7.1 Linear and angular velocity from leg frame

4.7.2 Linear and angular velocity from body frame

4.8 Dynamic analysis of mammalian inspired quadruped robot

5. GAIT PLANNING AND STATIC STABILITY ANALYSIS OF QUADRUPED

5.2 Gate synthesis

5.2.1 Crawl gait

5.2.2 Trot gait

5.2.3 Bound gait

5.3 Stability analysis of quadruped robot

5.3.2 Static stability margin and longitudinal stability margin

5.3.3 Algorithm developed for SSM for crawl gait

6. SIMULATION OF QUADRUPED ROBOT

6.2 Control architecture for simulation

6.3 Multibody simulation of alligator inspired robot

6.3.1 MBD simulation of leg

6.3.2 MBD simulation of whole body

6.4 MBD simulation of mammal inspired robot

6.4.2 Dynamic leg simulation

6.4.3 Whole body simulation

6.5 Finite element analysis of quadruped robot

6.5.1 Leg simulation