Control in Robotics and Automation

4 EVENT-BASED PLANNING AND CONTROL FOR MULTIROBOT COORDINATION

23

1,4

1.4

,.,q

0

<

8 100

5O

o

-50

-100

................................... i: ............................... 1

L ....

~

5 l o -15o 0

time

(sec)

..... c;...;.., ..................................................

....... 'i."! :', "'" ~

~ ,, ,

....

5

time (sec)

10

1.5

.~ 0.5

<

~ s 10

time (sec)

FIGURE 1.8

Straight line path tracking based on a time-optimal plan.

l __ ~ !

I ~ !

tl

~, 0.5 -.r ............... , .....................

,- o

;.rd

-1

[- J

o 5 o

time (sec)

solid line: X

dash line: Y

dot

line: Z

ability to handle an unexpected event. It significantly improves the safety and reliability of

the robotic system.

Figure 1.13 presents the results of a similar experiment for a circular path.

The preceding experimental results indicate that the performance of the event-based

planning and control scheme is comparable to that of the time-based motion planning and

control scheme.

It is even better.

The important point, however, is that it provides a natural

reference base for sensor-based planning and control.

4 EVENT-BASED PLANNING AND CONTROL FOR MULTIROBOT COORDINATION

4.1 Introduction

An important issue in multirobot systems is coordinated control. To achieve intelligence of

multirobot systems, it is essential to develop a proper planning and control scheme for

coordination.

Multirobot coordinated control has been a research subject for several years. Various

coordination schemes have been proposed. In [23] and [49], the master-slave coordination

scheme was proposed. The hybrid position-force control theory was extended to multiarm

coordinated control [36-38]. Control algorithms for multiarm object handling that take into

24 CHAPTER 1 / SENSOR-BASED PLANNING AND CONTROL FOR ROBOTIC SYSTEMS

500 i !

......

200

,,-- ~

e" "" .s"

'*" ,,. i

.0- .....

................ . .................................. ! ................

: i :

....... i ................. i ~ .............

..... :::.:-..-...,-,: ............... i .~ :.,'.'.-.....,., ....... i"

.......................

o

0

u -500

<

-1000

0 5 10 15 20

time (secs)

100 " ..................................

,,t,,

, ....] ..... ~ .... -~,-~ .........

,,

.,~ ,,,.,

: . ~,~

: i

'" ",! ",

0 '.": .......... :'~ ....... "

9 ~,

-100 ,

......

-200 ~

o 5 lO 15 2,9

time (s~s)

0

-2

0

200 ~

!

,. , . :

............... ~ .............. -. ............ § ............... ~ 100 .............. - ............... - ................. , ..........

i .S : : f~ ' ~ ' " Ji i :i

-100 ""~ ...........

*

................. : ................. i ............

-200

5 I0 15 20 0 5 I0 15

time (sees) time (s~s)

solid line: X

dash line: Y

dot

line: Z

FIGURE 1.9

Two-circle tracking based on a minimum-energy plan,

t r

=16s.

account the object dynamics and achieve simultaneous position and internal force control

appear in [24], [33], [40], [41], and [42]. The coordination of a multifingered robot has also

been widely discussed in [44], [-45], [-46], and [47]. In a multirobot system, redundancy

becomes even more important. The related results can be found in [53], [54], [55], [56], [57],

and [58]. Dual-arm situations have been intensively investigated in 1-43], [48], [-50], [52],

and [63]. An experimental evaluation of master-slave and hybrid position-force control

schemes was presented in [51].

In this section, issues in multirobot rigid-object handling are discussed. First, a new

event-based motion reference for a multirobot system is introduced. Then time- and

energy-optimal motion plans are obtained on the basis of this new motion reference. A

general task space is defined. Based on the nonlinear feedback technique, the multirobot

system including the robots' joint motor dynamics is linearized and decoupled with respect

to the general output defined in the general task space. Then a task projection operator is

introduced. It projects the general output to a controllable subspace, that is, to the actual

task space for each individual robot. Finally, experimental results for a dual-arm coordina-

tion task are presented.

The ultimate goal is to develop an intelligent planning and control scheme for multiarm

coordination that can be conveniently implemented in a distributed computing architecture.