Sharif University of Technology - CEDRA

Robot Motion Trajectory Generation

Hey! Where Should I go?

Sharif University of Technology - CEDRA

Robot Motion Trajectory Generation

We shall study Methods to: Compute a Trajectory that describes the desired motion of a manipulator in space Trajectory is the time history of position, velocity, and acceleration for each joint (i.e. degrees-offreedom). Trajectory Planning means the determination of the actual trajectory, or path, along which the robot end effector will move in Cartesian Space.
Sharif University of Technology - CEDRA

Robot Motion Trajectory Generation

In executing a trajectory, a manipulator moves from its initial position to a desired goal (final) position. However, the motion may be:
1. Smooth (nice and secure) 2. Rough (vibrations and damaging)

Trajectory in Cartesian Space

Sharif University of Technology - CEDRA

Robot Motion Trajectory Generation

In general, the task of a robot manipulator is defined by a sequence of points which are denoted as end points and are stored in the robots control computer.

Trajectory in Cartesian Space

Motions of a robot manipulator is described as the motions of the Tool Frame {T} relative to the Station Frame {S}.
Sharif University of Technology - CEDRA

Robots Motion Classifications

Point-to-Point (PTP) Robotic Systems: The robot moves to a numerically defined location and then the motion is stopped. Then, the end-effector performs the required task with the robot being stationary. Upon completion of the task, the robot moves to the next point and the cycle is repeated. (Ex: Spot Welding Operations) Therefore; in PTP robots, the path of the robot and its velocity while traveling from one point to the next is not important!
Trajectory in PTP Cartesian Robotic System

2-End point

1-Starting point X

Sharif University of Technology - CEDRA

Robots Motion Classifications

Example: Consider a simple case in am which each joint of a PTP robot must move to its end point f as fast as possible without exceeding a maximum -am admissible acceleration am and a maximum velocity Vm. Desired Joint Trajectory consists of 3-segments: 1. [0, t1]; based on max. acceleration 2. [t1, t2]; based on max. velocity 3. [t2, tf]; based on max. deceleration
Typical acceleration, velocity and position of an axial motion in PTP Robot.

a t

Vm

f 2

1 0 0 t1 t2

tf

Sharif University of Technology - CEDRA

Robots Motion Classifications

a Lets assume that the time durations of acceleration am and deceleration segments are equal:{t1 = tf - t2}. The control program must calculate the two switching times based on the initial and final joint values (0 =0 -am and f ) and Vm and am. Joint Trajectory during: 1st Segment: 2nd Segment: 3rd Segment:
(t ) = 2 + Vm (t t 2 )
a t2 a t2 (t ) = m , 1 = m 1 , Vm = amt1 2 2

V Vm

(t ) = 1 + Vm (t t1 ) , 2 = 1 + Vm (t 2 t1 ) ,
am (t t 2 ) a t , f = 2 + Vmt1 , 2 2
2 2 m 1

f 2

From above equations we can write: f V V and t1 = m , t f = m + f = Vmt 2 t 2 = f

Vm am am
Typical acceleration, velocity and position of an axial motion in PTP Robot.

Vm

1 0 0 t1 t2

tf

Sharif University of Technology - CEDRA

Robots Motion Classifications

Continues Path (CP) Robotic Systems: In this system the robot tool performs the task while the axes of motion are moving (both robot and the tool are moving simultaneously, and the speed of each joint can be controlled independently). (Ex: Arc Welding Operations) Therefore; in CP robot operations, the position of the robots tool at the end of each segment , together with the ratio of axes velocities, determine the generated trajectory. Ex: variations in the velocity of the arc welding result in a nonuniform weld seam thickness (i.e. an unnecessary metal builtup or even holes)

Sharif University of Technology - CEDRA

Robot Motion Trajectory Generation

Motions of a robot manipulator is described as the motions of the Tool Frame {T} relative to the Station Frame {S}. To move the manipulators Tool from {Tool}i to {Tool}f , there exists an infinite number of trajectories.
{Tool}

f i

Trajectory in Cartesian Space

i: initial position of the tool f: final position of the tool

Sharif University of Technology - CEDRA

10

Robot Motion Trajectory Generation

To provide more details about the desired trajectory, we sometimes define some via points (intermediate points) throughout the path. via points
{Tool}
Trajectory in Cartesian Space

f i i: initial position of the tool f: final position of the tool t

Path Points: a set of Via points plus Initial and Final points.
Sharif University of Technology - CEDRA

11

Robot Motion Trajectory Generation

In executing a trajectory, it is generally desirable to smoothly move the manipulator from its initial position to a desired final position. For Smooth Motion: We need to have position and velocity to be continues functions of time. A Smooth Function: A function which is continues and has a continues first derivative.
Sharif University of Technology - CEDRA

Trajectory in Cartesian Space

12

Robot Motion Trajectory Generation

Joint Space Schemes: Methods of path generation in which path shapes are described in terms of functions of joint angles.
Each path point (via points + initial and final points) is specified in terms of a desired position & orientation of the Tool frame {T}, relative to the Base frame {B}. {T}i In Cartesian Space a set of via points may be used to take the Tool frame from initial state to a final state. Each via and path point is converted into a set of desired joint angles by applying inverse kinematics. A Smooth Function is defined for each of n-joints that pass through the via and path points. The time duration required for each motion segment is the same for each joint, so that all joints will reach the via point at the same time.
Sharif University of Technology - CEDRA

{T}f

{i}

{f} {B}

Joint Space Schemes

13

Robot Motion Trajectory Generation

Joint Space Schemes:
Each via and path point is converted into a set of desired joint angles by applying inverse kinematics. {Tool} frame = {vector }

via
{T}f

{T}i {by Inverse Kinematics} = a set of {is and fs} Generate a Smooth Function through all points. In between via points, the shape of the path, although rather simple in joint space, is complex if described in Cartesian space. Joint Space schemes are easiest to compute, since there is usually no problem with singularities of the mechanism.
Joint Space Schemes Sharif University of Technology - CEDRA

{i}

{f} {B}

14

Robot Motion Trajectory Generation

Cubic Polynomials:
{T}f

By applying inverse kinematics. {Tool}i = {Tool}f = {is } {fs }

{T}i

We need to define a smooth function for each joint {i} whose value at ti is i, and at tf is f There exist many smooth functions for (t)= ?...
Joint Space Schemes

{f} {B}

Sharif University of Technology - CEDRA

15

Robot Motion Trajectory Generation

Cubic Polynomials:
For Smooth joint motion, we need to impose at least 4-constraints on (t) as:
(t) f

(0) = i = 0 (t f ) = f &(0) = 0 &(t f ) = 0

Continues in Velocity: Start and Stop at Zero Velocities

i= 0 ti=0

Many path shapes t tf

We need to define a smooth function for each joint whose value at ti is i, and at tf is f There exist many smooth functions for (t)= ?...

Sharif University of Technology - CEDRA

16

Robot Motion Trajectory Generation

Cubic Polynomials: A Cubic Polynomials satisfies the above constraints.

(t ) = a0 + a1t + a2t 2 + a3t 3

(0) = 0 = a0 (t f ) = f = a0 + a1t f + a2t 2 + a3t 3 f f &(0) = 0 = a1 &(t f ) = 0 = a1 + 2a2t f + 3a3t 2 f

(4-Coefficients)
a0 = 0 a1 = 0 a2 = 3 ( f 0 ) t2 f 2 ( f 0 ) t3 f

a3 =

Sharif University of Technology - CEDRA

17

Robot Motion Trajectory Generation

5th Order Polynomials:
If we add the acceleration constraints at the beginning and end of the path; then we have 6-Constraints. If acceleration is continues, then vibration is diminished (i.e. if ai = af = 0) at the start and the stopping points. Hence, a 5th Order Polynomial would be sufficient to define the joint motion.

(t) =a0 +a1t +a2t2 +a3t3 +a4t4 +a5t5

Sharif University of Technology - CEDRA

(6-Coefficients)

18

Position, Velocity, and Acceleration Curves

f (t)

Position
0 t0=0
ai = af =0

tf

&

Velocity
ai = af =0 vi = vf =0

t0=0

& &
Zero Slope t0=0
ai = af =0

tf

Acceleration
tf t

Sharif University of Technology - CEDRA

19

For intermediate (via) points at which f you dont want the robot to stop, velocity constraints at the end of cubic polynomial is no longer zero.

Robot Motion Trajectory Generation (t)

(0) = 0 (t f ) = f &(0) =& 0 &(t f ) = & f

4-Constraints

t0=0

t tf

&

& f
2 3

(t) = a0 + a1t + a2t + a3t

&0
t =0 tf

0 Sharif University of Technology - CEDRA

20

Robot Motion Trajectory Generation

(t) = a0 + a1t + a2t 2 + a3t 3
4-Coefficients
a0 = 0 & a =
1 0

a2 =

3 2 & 1 & ( f 0 ) 0 f t2 tf tf f 2 1 & & ( f 0 ) + 2 ( f + 0 ) 3 tf tf

a3 =

Several Ways to specify the desired velocities at the via points:

By the user in terms of Cartesian linear and angular velocities (mapped into joint velocities by Jacobian Inverse) The system automatically choose The system automatically choose such that the acceleration is continues
Sharif University of Technology - CEDRA

21

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
Another choice of path shape is to move Linearly from initial joint position 0 to the final joint position f. (t) Linear Function Results in:
f

1. Velocity to be discontinuous at the beginning and end of motion, 2. Acceleration to be discontinuous 0

t t 0= 0 (Linear Function) tf

Undesirable (Rough Motion)

Sharif University of Technology - CEDRA

22

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
To avoid rough motion, we can add a Parabolic Blend to the linear function at each end.
(t)

Adding Parabolic blends f & with constant acceleration(&b ) results in smooth change in half velocity. Linear function and two parabolic functions are splined together so b t 0 that the entire path is continuous t =0 tb thalf tf--tb tf 0 in Position and Velocity. (Linear Function with Parabolic Blends)

(t ) = a0 + a1t + a2t 2
Sharif University of Technology - CEDRA

23

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
Assume both blends have the same time duration, and note that velocity at the end of the blend must be equal to the velocity of the linear segment.
For the Linear segment:

& &(tb ) = &btb =

h b
t h tb

(t)

= ( slope speed )

For the Blend region:

& &b = Cons tan t

Integrating using I.C.:

half

& &b = &bt + C1 0

&b (0) = 0 b (0) = 0

b 0 t0=0

t tb thalf tf--tb tf

0 (Linear Function with Parabolic Blends) 0 1 && 2 b = bt + C1t + C2 2 Sharif University of Technology - CEDRA
24

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
1 && 2 b = btb + 0 at tb , and let: t=2th 2
Combining the above relations, we get:

& &btb =

h b
th tb

f +0 1 && 2 btb 0
2 1 t tb 2 2

& & &btb2 &bttb + ( f 0 ) = 0

Where: t : Desired duration of motion

& &b :

Acceleration acting during the blend region

Sharif University of Technology - CEDRA

25

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
Given any f, 0, and t, we can follow any of the paths & given by choices of &b and tb which satisfy equation *. .

& & &btb2 &bttb + ( f 0 ) = 0

& You may pick &b and solve for tb ?

t tb = 2

& & &b2t 2 4&b ( f 0 ) & 2&b

4( f 0 ) t2

& & & &b2t 2 4&b ( f 0 ) &b

Sharif University of Technology - CEDRA

26

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
& You may pick & and solve for tb ?
b

t tb = 2

& & &b2t 2 4&b ( f 0 ) & 2&b

4( f 0 ) t2

& & & &b2t 2 4&b ( f 0 ) &b

= Equality means linear segment has zero length (Path is composed of two blends with equivalent slope). > As the acceleration & increases, the blend region becomes & b & , we are back to shorter and shorter. In the limit when& b the simple linear interpolation case.
Sharif University of Technology - CEDRA

27

Robot Motion Trajectory Generation

Linear Function with Parabolic Blends:
Adding via points:
(t) f

0 t t0=0
(Linear Function with Parabolic Blends)

tf

Sharif University of Technology - CEDRA

28