MATLAB LECTURE - Trang 16

9/16/2015

12

Nguyen Q. Hoang, Nguyen V. Khang

45

nonlinear equations

more unknowns than equations

Given

( )

n

t

 

q

( , )

0

f x q

Application of pseudo-inverse: Inverse kinematics

( ) [ ,

]

m

E

E

t

x

r O

determined

Number of solutions

Finite number of solutions

Infinite number of solutions

Methods: analytical and numerical

Problems

Nguyen Q. Hoang, Nguyen V. Khang

46

Method based on the pseudo-inverse of jacobian matrix

( )

( )

d

or

dt

f q

q

x

J q q

x

q

1

1

1

1

2

1

2

...

( )

...

...

...

...

...

n

m

m

m

n

f

f

f

q

q

q

f

f

f

q

q

q

f

J q

q

system of m linear equations, with n unknown

1

2

min

T

J

q Wq

introduction the functional

W : positive weighting matrix

Solution

1

1

1

( )[ ( )

( )]

T

T

q W J q J q W J q

x

1

( )[ ( )

( )]

T

T

q

J q J q J q

x

J x

if W is identical matrix

1

( )[ ( )

( )]

T

T

J

J q J q J q

Application of pseudo-inverse: Inverse kinematics

0

( )

(0)

t

t

dt

q

q

q

Nguyen Q. Hoang, Nguyen V. Khang

47

Application of pseudo-inverse: Inverse kinematics

Blockdiagram of Inverse kinematics

x

( )

W

J q

1/s

Solving for q

0

(0)

x

q(t)

(0)

q

q

Drawback of this diagram :
accumulated errors due to rounding and integral method makes
don’t satisfy the constraint equations, so the end-effector moves out the
given trajectory.

( ), ( )

t

t

x

q

( ( ), ( ))

0

t

t

f x

q

Nguyen Q. Hoang, Nguyen V. Khang

48

Application of pseudo-inverse: Inverse kinematics

Method of Error feedback

,

( )

( )

e

Ke

e

x

f q

e

x

J q q

Ke

( )

J q q

x

Ke

(

)

q

J x

Ke

t

( )

0

e

Null space of jacobian matrix

( )

J q

o

(

) (

)

q

J x

Ke

E

J J z

By putting

o

z

the redundancy of manipulator can be exploited:

- advoiding obstacles
- advoiding impact with the joint limitations
- advoiding singular configurations