CH 07

CHAPTER SEVEN
TRAJECTORY PLANNING
Introduction to Robotics: Analysis, Control,

Applications, Third Edition. Saeed B. Niku. ©
2020 John Wiley & Sons Ltd. Published 2020 by
Introduction to Robotics: Analysis, Control, Applications, Third Edition.
Saeed B. Niku.
© 2020 John Wiley & Sons Ltd. Published 2020 by John Wiley & Sons Ltd.
This solution manual may not be copied, posted, made available to students,
placed on any outlet, any other electronic media system, and the Internet
without prior expressed consent of the copyright owner.
This material may not be used except in conjunction with the “Introduction to
Robotics: Analysis, Control, Applications”

Introduction
 In this chapter we discuss the joint-space and Cartesian-

space trajectory planning.
 We also discuss the methods that can be used for achieving
desired velocity and accelerations limits and maximum
values and blending motion portions.

Path versus Trajectory
 A path is the collection of positions and orientations that a

robot (or its joints) goes through between different
locations.
 A trajectory is the time history of the locations and
orientations that the robot (or its joints) goes through
between different points.
 A trajectory in the time history of a path; therefore, it
includes how the joint values change in time.

Joint-Space versus Cartesian-Space
 Cartesian-space describes the location and orientation of

robot’s end effector in space.
 Joint-space is the description of joint values in time. The
collection of joint-space values creates the Cartesian-space
values.
z
x
y

Cartesian-Space Trajectory Problems
 Robot running into itself

 Sudden changes in orientation while making a motion.
x
y

Basics of Trajectory Planning
 Assume joints moves at their maximum value.

 The resulting motion is non-uniform and unpredictable.
Time
y 5 4 3 (sec)
 
6 2
B 0 20 30
1 30 40
1
2 40 50
A
3 40 60
4 40 70

5 40 80

o x

Basics of Trajectory Planning: Cont.
 Normalizing the joints to start and stop together improves

the motion.
 The path is unpredictable.
y Time
6 5 4 (sec)
 
3
B 2 0 20 30
1 1 24 40
A 2 28 50
3 32 60
 4 36 70
5 40 80

x

 Planning a desired path (such as a straight line) improves

the outcome.
 Acceleration or velocities may be unachievable.
y Time
6 (sec)
 
5
B 4 0 20 30
3
2 1 13 56
1
2 15 69
A
3 19 78
 4 28 82
5 40 80

x

 Dividing the path (becomes a trajectory) into variable

portions to control velocities and accelerations improves
the outcome.
 Boundary velocities and accelerations are specified.
#  
y
8 7 1 20 30
6
B 5 2 16 40
4
3 21 3 15 50
A 4 13 59
5 15 70
 6 25 80
7 32 81
8 37 80

x
Blending of Motion Portions
 To prevent stop/start regimes at every point, motion portions may be

blended together.
 There may be a need for additional via points to ensure passage
through a desired location.
y B y B''
C C B
B'
A A
x x
y y D
B B
B' E
C C
A A
Introduction to Robotics: Analysis, Control, x x

Trajectory Planning in Joint-Space
 It is more desirable and more practical to plan a trajectory

in joint-space.
 We will use the joint-space trajectories to create a motion
that is specified in Cartesian-space.
 In joint-space trajectory planning, it is the motions of each
individual joint that is planned based on some regime
(higher-order polynomials, controlled acceleration limits,
blended motions, and so on).
 Each joint’s motion is planned individually based on the
values calculated from the inverse kinematic equations.

Third-Order Polynomial Trajectory Planning
 Initial and final positions and velocities are known; 4

values
 (ti )  i  (t f )   f
 (t )  0
i  (t )  0
f
 4 unknowns may be calculated; therefore, a third-order

polynomial.
2 3
 (t )  c0  c1t  c2t  c3t
 (t )  c1  2c2t  3c3t 2
Third-Order Polynomial Trajectory Planning: Cont.
 Substitute the initial and final conditions into these equations to get:
 (ti )  c0   i
2 3
 (t f )  c0  c1t f  c2t f  c3t f
 (t )  c  0
i 1
 (t f )  c1  2c2t f  3c3t f  0
2
 Or in matrix form:
  i  1 0 0 0   c0 
   0 1 0 0   c1 
 i  
 f  1 t f tf 2 t f  c2 
3
    
 f  0 1 2t f 3t f 2   c3 
Third-Order Polynomial Trajectory Planning: Cont.
 From the inverse kinematic equations the boundary values

are determined.
 Each joint is driven based on the third-order polynomial.
 The resulting motion follows the desired trajectory.

Example:
 Going from 30 to 75 degrees in 5 sec.
 Acceleration needed may be too high.
80
Position
60
40
20
Velocity
0
Acceleration
-20
0 1 2 3 4 5
Seconds

Fifth-Order Polynomial Trajectory Planning
 Initial and final positions, velocities, and accelerations are

known; 6 values
 6 unknowns may be calculated; therefore, a fifth-order
polynomial.
 (t )  c0  c1t  c2t 2  c3t 3  c4t 4  c5t 5
 (t )  c1  2c2t  3c3t 2  4c4t 3  5c5t 4

(t )  2c2  6c3t  12c4t 2  20c5t 3

Fifth-Order Polynomial Trajectory Planning: Cont.
 Substitute the initial and final conditions into these

equations and solve for the constants.
 Use the equation to run each joint based on the data
calculated from the inverse kinematic equations.

Example:
 i  30 i  0 / sec i  5 / sec 2
 Given:
 f  75  f  0 / sec f  5 / sec 2
80
Position
60
40
20
Velocity
0
Acceleration
-20
0 1 2 3 4 5
Seconds

Linear Segments with Parabolic Blends
 Run the joints at constant speed between the initial and final locations.
 Equivalent to a first-order polynomial, where the velocity is constant,
and acceleration is zero.
 Requires infinite accelerations to create instantaneous velocities at the
boundaries.
 Blend linear segment with parabolic sections at the beginning and at
the end of the motion segment to create a continuous position and
velocity
f
B
Position
i  
A
i
0 tb t tf-tb tf Time
Linear Segments with Parabolic Blends: Cont.
 Assume initial and final positions are  i and  f at

times ti = 0 and tf and parabolic segments are
symmetrically blended at blending times tb and tf - tb
1
 (t )  c0  c1t  c2t 2
2
 (t )  c1  c2t
(t )  c
2

 Substitute boundary conditions to get:

 (t  0)   i  c0  c0   i
 
 (t  0)  0  c1  c1  0
   
 (t )  c2  c2  
 and
1 2
 (t )   i  c2 t
2
 (t )  c2 t
(t )  c2
 adding constant acceleration for the linear segment we get:
1 2
 A   i  c2 tb
2
 A  c2 tb  
 B   A   ((t f  tb )  tb ))   A   (t f  2tb )
B   A  
 f   B  ( A   i )
  0
f

 The necessary blending time can be calculated as:
i   f   t f
tb 

 Maximum velocity can be calculated as:
 max  2( f   i ) t f
 Final parabolic segment is symmetrical with the initial
parabola, but with a negative acceleration
Linear Segments with Parabolic Blends and Via
Points
 To prevent stop-and-go motion, blend them together.

 The final conditions for one section is the initial condition
for the next.
 Make sure maximum allowable accelerations are not
exceeded

Higher-order Polynomials
 When more segments are present, more information is

available.
 Higher-order polynomials may be used, but more difficult to
solve higher-order polynomials:
 (t )  c0  c1t  c2t 2    cn 1t n 1  cnt n

 Use combinations of lower-order polynomials for different
segments of the trajectory and blend them together to satisfy
all required boundary conditions.
 Can use a 4-3-4 trajectory, a 3-5-3 trajectory, and a 5-cubic
trajectory to replace a 7th-order polynomial
Higher-Order Polynomials: Cont.
 In a 4-3-4 trajectory we use:

 a 4th-order polynomial to plan a trajectory between the initial point
and the first via point (e.g. lift-off),
 a 3rd-order polynomial between two via points (e.g. lift off and set-
down),
 and a 4th-order polynomial between the last via point (e.g. set-
down) and the final destination.
 A 3-5-3 trajectory is planned between the initial and the
first via point, between the successive via points, and
between the last via point and the final destination.

 We must solve for 4 coefficients for a third-order polynomial, 5 for a

fourth-order polynomial, and 6 for a fifth-order polynomial.
 Both 4-3-4 and 3-5-3 trajectories require solving for a total of 14
coefficients. For the 4-3-4 trajectory, the unknown coefficients are in
the form:
 (t )1  a0  a1t  a2t 2  a3t 3  a4t 4

 (t )2  b0  b1t  b2t 2  b3t 3
 (t )3  c0  c1t  c2t 2  c3t 3  c4t 4

1. Initial position of is known.

2. Initial velocity may be specified.
3. Initial acceleration may be specified.
4. Position of the first via point is known, and is the same
as the final position of the first fourth-order segment.
5. The first via point's position is the same as the initial
position of the third-order segment for continuity.
6. Continuous velocity must be maintained at the via point.
7. Continuous acceleration must be maintained at the via
point.
8. Position of a second (and other) via point is specified and is

the same as the final position of the third-order segment.
9. The position of the second (and other) via point is the same
as the initial position of the next segment for continuity.
10. Continuous velocity must be maintained at the next via
point.
11. Continuous acceleration must be maintained at the next via
point.
12. Position of destination is specified.
13. Velocity of the destination is specified.
14. Acceleration of the destination is specified

 The result is:
1 0 0 0 0 0 0 0 0 0 0 0 0 0 
1   a 
  0 1 0 0 0 0 0 0 0 0 0 0 0 0   0
 a
 1  0 0 2 0 0 0 0 0 0 0 0 0 0 0   1
1   2 3 4   a2 
  1  1 f 1 f 1 f 1 f 0 0 0 0 0 0 0 0 0   
 2  0 0 0 0 0 1 0 0 0 0 0 0 0 0   
a3
 2     a4 
  0 1 2 1 f 3 1 f
2
4 1 f
3
0 1 0 0 0 0 0 0 0  b 
0  2   0
 0  0 0 2 6 1 f 12 1 f 0 0 2 0 0 0 0 0 0  b 
1
  2 3  
 3  0 0 0 0 0 1 2 f 2 f 2 f 0 0 0 0 0   b2 
  0 0 0 0 0 0 0 0 0 1 0 0 0 0   b3 
 3    
 0  0 0 0   c0 
2
0 0 0 0 1 2 2 f 3 2 f 0 1 0 0
 0  0 0 0 0 0 0 0 2 6 2 f 0 0 2 0 0   c1 
     
 4  0 0 0 0 0 0 0 0 0 1 3 f 3 f
2
3 f
3
 3 f   c2 
4
   2 3
 c 
 4  0 0 0 0 0 0 0 0 0 0 1 2 3 f 3 3 f 4 3 f   3 

 4  0 0 0 0 0 0 0 0 0 0 0 2 6 3 f
2 c 
12 3 f   4 

 Or simply:
    M C 
C    M   
1
 This can be used to run a joint based on a 4-3-4

polynomial.
 Similar results may be obtained for 5-3-5 or 5-cubic
regiments.

Cartesian-Space Trajectories
 All the regiments for joint-space can also be used here.

 We can do the following in a loop to run a robot in
Cartesian-space:
 Increment the time by t  t  t .
 Calculate the position and orientation of the hand based on the
selected function for the trajectory,
 Calculate the joint values for the position and orientation through
the inverse kinematic equations of the robot,
 Send the joint information to the controller,
 Go to the beginning of the loop.

Straight-Line Trajectory
 Straight-line motions between points are the most

practical trajectories for industrial applications.
 Blending the motions for multiple destinations (such as
via points) is also very common.
 For a straight-line trajectory, the transformation between
the initial and final positions and orientations must be
calculated and divided into small segments.

Straight-Line Trajectory: Cont.
 The total transformation R between the initial

configuration Ti and final configuration Tf can be
calculated as follows:
T f  Ti R
Ti 1T f  Ti 1Ti R
R  Ti 1T f

 At least three different alternatives may be used to

convert this transformation into small segments:
 1. Use large number of segments for smooth motion. This, in
reality, creates a large number of differential motions [3]. Using
the equations developed for differential motions in Chapter 3,
we can relate the position and orientation of the hand frame at
each new segment to the differential motions, the Jacobian, and
joint velocities by:
D  JD and D  J 1D
dT    T
Tnew  Told  dT

 2. The transformation between the initial and final locations R can be

decomposed into a translation and two rotations. The translation
involves moving the origin of the frame from the initial position to the
final position. The first rotation is to align the hand frame to the desired
orientation, and the second rotation is to rotate the hand frame about its
own axis to the final orientation. All three transformations are executed
simultaneously.

 3. The transformation between the initial and final locations R can be
decomposed into a translation and one rotation about an axis q. The
translation involves moving the origin of the frame from the initial position
to the final position. The rotation is to align the hand frame to the final
desired orientation. Both transformations are executed simultaneously
o'
q
a'
o n
a n'
o n
a
y
x
Continuous path example
 The result of successive motions with continuous-path
on and off
Starting point
Starting point
Without blending Without blending
With blending With blending

Continuous Trajectory Recording
 Some operations are too complicated or too intricate to be generated by straight
lines or other higher-order polynomials.
 It is possible to teach the robot how to move, record the motions, and later replay
the motions and execute them.
 An operator may move the robot in the same fashion required to accomplish a
task in real time by either releasing the joint-brakes and physically moving the
robot, or by moving the joints of a robot model.
 In either case, the joint values are continuously sampled in time and recorded.
 Later, the robot is operated by playing back the sampled data.
 This technique is simple and requires little programming or calculations.
 All motions must be accurately performed, sampled, and recorded for accurate
playback.
 Every time a part of the motion needs to be changed, the robot must be
programmed again.
 This technique is difficult for large, heavy robots.


CH 07

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CH 07

Uploaded by

Copyright:

Available Formats

CHAPTER SEVEN

Introduction to Robotics: Analysis, Control,

Introduction to Robotics: Analysis, Control,

 In this chapter we discuss the joint-space and Cartesian-

Introduction to Robotics: Analysis, Control,

 A path is the collection of positions and orientations that a

Introduction to Robotics: Analysis, Control,

 Cartesian-space describes the location and orientation of

Introduction to Robotics: Analysis, Control,

 Robot running into itself

Introduction to Robotics: Analysis, Control,

 Assume joints moves at their maximum value.

Introduction to Robotics: Analysis, Control,

 Normalizing the joints to start and stop together improves

Introduction to Robotics: Analysis, Control,

 Planning a desired path (such as a straight line) improves

Introduction to Robotics: Analysis, Control,

 Dividing the path (becomes a trajectory) into variable

 To prevent stop/start regimes at every point, motion portions may be

Introduction to Robotics: Analysis, Control, x x

 It is more desirable and more practical to plan a trajectory

Introduction to Robotics: Analysis, Control,

 Initial and final positions and velocities are known; 4

 4 unknowns may be calculated; therefore, a third-order

 From the inverse kinematic equations the boundary values

Introduction to Robotics: Analysis, Control,

Introduction to Robotics: Analysis, Control,

 Initial and final positions, velocities, and accelerations are

 (t )  c0  c1t  c2t 2  c3t 3  c4t 4  c5t 5

 (t )  c1  2c2t  3c3t 2  4c4t 3  5c5t 4

Introduction to Robotics: Analysis, Control,

 Substitute the initial and final conditions into these

Introduction to Robotics: Analysis, Control,

Introduction to Robotics: Analysis, Control,

 Assume initial and final positions are  i and  f at

Introduction to Robotics: Analysis, Control,

 Substitute boundary conditions to get:

 adding constant acceleration for the linear segment we get:

Introduction to Robotics: Analysis, Control,

 The necessary blending time can be calculated as:

 To prevent stop-and-go motion, blend them together.

Introduction to Robotics: Analysis, Control,

 When more segments are present, more information is

 (t )  c0  c1t  c2t 2    cn 1t n 1  cnt n

 In a 4-3-4 trajectory we use:

Introduction to Robotics: Analysis, Control,

 We must solve for 4 coefficients for a third-order polynomial, 5 for a

 (t )1  a0  a1t  a2t 2  a3t 3  a4t 4

Introduction to Robotics: Analysis, Control,

1. Initial position of is known.

8. Position of a second (and other) via point is specified and is

Introduction to Robotics: Analysis, Control,

 This can be used to run a joint based on a 4-3-4

Introduction to Robotics: Analysis, Control,

 All the regiments for joint-space can also be used here.

Introduction to Robotics: Analysis, Control,

 Straight-line motions between points are the most

Introduction to Robotics: Analysis, Control,

 The total transformation R between the initial

Introduction to Robotics: Analysis, Control,

 At least three different alternatives may be used to

Introduction to Robotics: Analysis, Control,

 2. The transformation between the initial and final locations R can be