DESKFZG561

MECHANISMS & ROBOTICS

Kinematics of ManipulatorsLecture 10
The inverse kinemaTic modeling of
Manipulator

Books :

1. Robotics and Control,

Mittal R. K. & Nagrath I. J, TMH
2. Fundamentals of Robotics: Analysis and
Control, Robert J., Schilling, Prentice Hall
1

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse KInematIcs
In the direct kinematics, methods has been discussed to derived the kinematic
equations for any manipulator to know the position and orientation of the end
effector an 0Tn , for given values of joint link displacements.
Equation specifies the end effector frame{n} relative to the base frame {0}, for
the n DOF manipulator.

The joint displacements (q1,q2 .qn) that leads to the end effector to a certain
position and orientation T can be found by solving the kinematic model equations
for unknown joint displacement, the location of the end effector is achieved.

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse KInematIcs
In inverse kinematics problem of the serial is computationally expansive and
generally takes a very long time in the real time control of manipulators. Tasks to
be performed by a manipulator are in the Cartesian space, whereas actuators
work in joint space.
Cartesian space includes orientation matrix and position vector. However, joint
space is represented by joint angles. The conversion of the position and
orientation of a manipulator end-effector from Cartesian space to joint space is
called as inverse kinematics problem.
Definition : The determination of all possible and feasible sets of joint variables,
which would achieve the specified position and orientation of the manipulators
end effector with respect to the base frame.
We must know a) The end effector position and orientation for the instantaneous
location of each joint. b) The joint displacements required to place the end
effector in a new location c) corresponding motion of each joint, which will
produce the desired tool tip motion.
3

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse KInematIcs
The position and orientation of the end
effector is collectively referred as
configuration of the end effector
transformation matrix T represents it.

Configuration of space or Cartesian Space : The configuration of the end

effector is represented by three position component as displacement along three
orthogonal axes of base frame and three rotations about the base frame axes
representing a six dimensional space is Configuration space.
Configuration , or position and orientation, of the end effector is a function of
joint displacement variables q1,q2 .qn

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse KInematIcs

For an n DOF manipulator the set of n joint displacement variables represented

by a n x1 vector.
The set of all n x1 joint displacement vectors generates the joint vector space or
joint space.
The mapping shows the relationship between the Cartesian space and joint
space representation of a manipulators end effector position and orientation.
The inverse kinematics is the determination of the set of positions and
orientations in Cartesian space that are reachable by the origin of the end
effector frame as the joint displacements vector q ranges over the joint space.
5

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

I n v e r s e K i n e m a t i c s - From Position to Angles

A Simple Example
Revolute and
Prismatic Joints
Combined

Finding :

y
arctan( )
x
More Specifically:

(x , y)

y
arctan 2( )
x

arctan2( y/x) specifies that its in

the first quadrant

Finding S:
X

S (x 2 y 2 )
6

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse Kinematics of a Two Link Manipulator

(x , y)

Find:

l2

l1

Given: l1, l2 , x , y
1, 2

Redundancy:
A unique solution to this problem
does not exist. Notice, that using the
givens two solutions are possible.
Sometimes no solution is possible.
(x , y)

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Inverse KInematIcs
Two main solution techniques - 1) Analytical 2) Numerical methods.
In Analytical type, the joint variables are solved analytically according to given
configuration data.
In Numerical methods type of solution, the joint variables are obtained based on
the numerical techniques.
Two approaches in Analytical method: Geometric and Algebraic solutions.
Geometric approach is applied to the simple robot structures, such as 2DOF planar manipulator or less DOF manipulator with parallel joint axes.
Algebraic approach is applied for the manipulators with more links and whose
arms extend into 3 dimensions or more the geometry gets much more tedious.
There are difficulties to solve the inverse kinematics problem when the
kinematics equations are coupled, and multiple solutions and singularities exist.
Mathematical solutions for inverse kinematics problem may not always
correspond to the physical solutions and method of its solution depends on the
robot structure.
8

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

The Geometric Solution

l2

(x , y)

Using the Law of Cosines:

c 2 a 2 b 2 2ab cos C
2

( x 2 y 2 ) l1 l2 2l1l2 cos(180 2 )
cos(180 2 ) cos( 2 )
2

l1

Using the Law of Cosines:

sin B
sin C

b
c
sin 1
sin(180 2 )
sin( 2 )

l2
x2 y2
x2 y2
1 1
y
arctan 2
x
9

WILP BITS PILANI , PUNE CENTER

x 2 y 2 l1 l2
cos( 2 )
2l1l2

x 2 y 2 l12 l2 2

2 arccos

2l1l2

Redundant since 2 could be in the

first or fourth quadrant.

Redundancy caused since 2 has two possible

values

l sin( )
y
2
1 arcsin 2
arctan 2
x 2 y2
x

Prof Milind Ramgir

The Algebraic Solution

2

l2

(x , y)

c 1 cos 1
c 1 2 cos( 2 1 )
(1) x l1 c 1 l 2 c 1 2

l1

(2) y l1 s 1 l 2 sin

1 2

(3) 1 2

(1) 2 ( 2 ) 2 x 2 y 2

l1 c 1 l 2 (c 1 2 ) 2 2 l1l 2 c 1 (c 1 2 ) l1 s 1 l 2 (sin 1 2 ) 2 2 l1l 2 s 1 (sin 1 2 )

l1 l 2 2 l1l 2 c 1 (c 1 2 ) s 1 (sin 1 2 )
2

l1 l 2 2 l1l 2 c 2
2

Only Unknown
2

x y l1 l 2
2 arccos
2 l1l 2

10

WILP BITS PILANI , PUNE CENTER

Note:
cos(ab) (cosa)(cosb) (sina)(sinb)
sin(ab) (cosa)(sinb) (cosb)(sina)
Prof Milind Ramgir

x l1 c 1 l 2 c 1 2
l1 c 1 l 2 c 1 c 2 l 2 s 1 s 2

cos(ab) (cosa)(cosb) (sina)(sinb)

c 1 ( l1 l 2 c 2 ) s 1 ( l 2 s 2 )

sin(ab) (cosa)(sinb) (cosb)(sina)

y l1 s 1 l 2 sin

1 2

l1 s 1 l 2 s 1 c 2 l 2 s 2 c 1
c 1 ( l 2 s 2 ) s 1 ( l1 l 2 c 2 )
c1

s1

We know what 2 is from the previous

slide. We need to solve for 1 . Now
we have two equations and two
unknowns (sin 1 and cos 1 )

x s1 ( l 2 s 2 )
( l1 l 2 c 2 )

x s1 ( l 2 s 2 )
y
( l 2 s 2 ) s 1 ( l1 l 2 c 2 )
( l1 l 2 c 2 )

11

Note:

Substituting for c1 and simplifying

many times

1
x l 2 s 2 s 1 ( l1 2 l 2 2 2 l1 l 2 c 2 )
( l1 l 2 c 2 )
y ( l1 l 2 c 2 ) x l 2 s 2
x2 y2
WILP BITS PILANI , PUNE CENTER

y ( l1 l 2 c 2 ) x l 2 s 2
1 arcsin
x2 y2

Prof Milind Ramgir

MANIPULATOR WORK SPACE

The workspace is defined as the volume of space in which the manipulator is
able to locate the end effector.
The work space gets specified by the existence of or nonexistence of solutions to
the inverse problem.
Reachable Work Space (RWS)- The region that can be reached by the origin of
the end effector frame with at least one orientation.
If a particular point in the work space can be reached by only one orientation,
the manipulability of the end effector is very poor and it is not possible to do any
practical work satisfactorily with just one fixed orientation.
Look for solution for the points in workspace, which can be reached by more
than one orientation.
Dexterous Work Space (DWS)- The space where the end effector can reach
every point from all orientation. It is either smaller (Subset) or same as the
reachable workspace.
12

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

MANIPULATOR WORK SPACE

2 DOF Planar manipulator with links L1 and L2.
RWS is plane annular space with radii r1 = L1 + L2 and r2 = L1 - L2
The DWS for this case is null.
Inside the RWS there are two possible orientations of the end effector for a given
position, while on the boundaries of RWS, end effector has only one possible
orientation.
For the special case of L1 = L2 , the RWS is a circular area and DWS is a point at
the center.

3 DOF
redundant Planar
manipulator with links L1 , L2 and
L3 with (L1 + L2 ) > L3 , the
RWS is a circle of radius (L1 +
L2 + L3 ) , while DWS is a circle
of radius (L1 + L2 - L3 ).

13

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

MANIPULATOR WORK SPACE

RWS of an n DOF manipulator is the geometric locus of the points that can be
achieved by the origin of the end effector frame as determined by the position
vector of direct kinematic model.
To locate the tool at an arbitrary position with an arbitrary orientation in 3D space
a minimum of 6 DOF are required. So that RWS may almost approach RWS.
Work space is characterized by the mechanical joint limits in addition to the
configuration and the number of DOF of the manipulator.
Work Space is specified assuming a full 3600 rotational joint range for each
revolute joint, but motion may be less than 3600.
Work space range is severely limited for prismatic joints , due to mechanical
constraints.
These limitations reduce the work space and shape of the work space may not be
similar to the ideal case.
14

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

MANIPULATOR WORK SPACE

To understand the effect of mechanical joint limits on the work space , consider
2-DOF planar manipulator with L1 > L2 and joint limit 1 and 2 such that
- 600 1 600 and - 1000 2 1000
For these limits, considering 1 = 2 = 0 as home position, annular work space gets
severely limited. The work space, obtained geometrically, is not annular any more,
rather it has a complex shape and is defined by contour ABCDEFA
Factors deciding
the workspaceDOF, Manipulators
configuration, Link
Lengths, and
allowed range of
joint motions.

15

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

SOLVABILITY OF INVERSE KINEMATIC MODEL

Inverse kinematic model is of nonlinear simultaneous equations, involving
transcendental ( harmonic sine and cosine) functions.
The number of simultaneous equations are more than the number of
unknowns. Therefore the equations may be mutually dependent.
Number of possible solutions may be multiple or may not exists for the given
end effector position and orientation.
Existence of Solutions
If the desired point P lies outside the reachable workspace- no solution
exists.
If p is within reachable workspace, but lies within the dexterous workspace
orientation may not be realized.
If the wrist has fewer than 3 DOF to orient the end effector, then certain
classes of orientation are not realized.
16

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

SOLVABILITY OF INVERSE KINEMATIC MODEL

Existence of Solutions
cont.
The last row of the matrix 0Tn is always constant, it can yield a maximum of
12 simultaneous equations, which are nonlinear algebraic equations involving
transcendental functions in n unknowns ( the joint variables).
Nine equations from the rotation matrix involve only three unknowns
corresponding to the orientation of the end effector expressed by Euler's angles
or roll-pitch-yaw angles and three displacement vector. i.e. six independent
constraints in n unknowns.

17

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

SOLVABILITY OF INVERSE KINEMATIC MODEL

Existence of Solutions
cont.
Therefore to have all position and orientation solution, the number of DOF n
must at least match the number of independent constraints. n 6
In addition to six independent constraints equations , the tool position and
orientation must be such that the limits on the joint motions are not violated.
Other than 6 DOF the solution is complex. E.g. when DOF is less than 6 the
manipulator can not attain the general goal position and orientation in 3D
space. Mathematically over determined case of finding less than six unknown
from six equations . When DOF is more than 6, under determined case of
finding more than six unknown from six independent non linear simultaneous
equations.

18

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

SOLVABILITY OF INVERSE KINEMATIC MODEL

Multiple Solutions
There may exist multiple solutions in Inverse kinematics problem. We have to
choose the best possible solution.
Factors a) If more than two joints are parallel, the number of solutions multiply.
e.g. 2 DOF planar arm with a wrist with 1 DOF will have two sets of possible
values of joint displacement ( 1 2 ) and ( 1 2 ) for the same end effectors (P)
position and orientation. ( 1 2 ) solution is elbow up [Preferred] and ( 1 2 ) is
elbow down position. Two solution are obtained because the axes of two
consecutive revolute joints of the arm are parallel.

b) If there is existence of trigonometric

functions in the equations . The
harmonic nature of sine and cosine
functions gives same magnitude for
angles in multiple of radians.

19

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

SOLVABILITY OF INVERSE KINEMATIC MODEL

Multiple Solutions
b)

c)

d)

If there is existence of trigonometric functions in the equations . The harmonic

nature of sine and cosine functions gives same magnitude for angles in multiple of
radians. e.g. yaw , pitch and roll motions of the RPY wrist for two sets of joints
displacements ( 1 2, 3 ) and ( 1 2 3 ) with 1 = 1800 + 1 , 2 = 1800 2 , 3
= 1800 + 1 will lead to the same orientation of the wrist.
The number of non zero joint link parameters and the range of joint motion allowed.
The number of ways to reach a certain goal is directly related to the number of nonzero
link parameters. e.g. A completely general rotary jointed 6 DOF manipulator with all
six ai 0 , up to sixteen solutions are possible.
The number of DOF. e.g. 6 DOF manipulator will have infinitely many solutions. A
manipulator with more DOF than the necessary is kinematically redundant manipulator.
SCARA 1 redundant DOF in horizontal plane because only two joints (2DOF) are
needed to establish nay horizontal position.
Redundant manipulator can be
useful in avoiding obstacles or
reaching inaccessible locations

20

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Solution techniqueS
Two Approaches
Closed form

Numeric Solutions

Joint displacements are determined as

Explicit functions of the position and
Orientation of the end effector.
Analytical Algebraic or kinematic
approach

Iterative algorithms ( Newton-Raphson

method) used.

Faster method

Computationally intensive and slower

Guarantee convergence to the

correct solution

Do not Guarantee convergence to the

correct solution in singular and
degenerate cases

May not be possible for all kinds

of structure
21

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Closed form solution

Condition The condition for a 6 DOF manipulator to possess closed form
solutions is that either its three consecutive joint axes intersect or its three
consecutive joint axes are parallel.
The kinematic equations under these conditions can be reduced to algebraic
equations of degree less than or equal to four for which closed form solutions
exist.
Approach to solve the equations- Inverse transform, Screw algebra, and
Kinematic ( none is general)
Another technique to reduce the complexity is dividing the problem into two
smaller parts- the inverse kinematics of arm and the inverse kinematics of wrist.
The solution to each of 3 DOF is obtained separately and combined by coinciding
the arm end point frame with wrist base frame to get the total manipulator
solution.

22

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir

Guidelines to obtain Closed form solutions

As the matrix equality implies element by element equality 12 equations are obtained. To find
the solution for n joint displacement variables
a. Look for equations involving only one joint variable. Solve and get joint variable
solutions.
b. Next look for pairs or set of equations, which could be reduced to one equation in one
joint variable by application of algebraic and trigonometric identities.
c. Use arc tangent(Atan2) function instead of arc cosine or arc sine functions. The two
argument Atan2 (y, x) function returns the accurate angle in the range of
-
by examining the sign of both y and x and detecting whenever either x or y is zero.
d. Solution in terms of the elements of the position vector components of 0Tn are more
efficient than those in terms of elements of rotation matrix, as latter may involve solving
more complex equations.
e. In the inverse kinematic model, the right hand side of eqn is known, while the left hand
side has n unknowns ( q1,q2,q3,qn) Left hand side is product of n link transformation
matrices. 0Tn = 0T1 1T2 2T3 . n-1Tn = T Use inverse transform approach to find q1 to
qn or qn to q1 either pre or post multiplication 0T1 and n-1Tn
23

WILP BITS PILANI , PUNE CENTER

Prof Milind Ramgir