Page 197 - Introduction to AI Robotics

180
5 Designing a Reactive Implementation

Figure 5.10 A Pioneer P2-AT with a forklift arm suitable for picking up soda cans.
(Photograph courtesy of ActivMedia, Incorporated.)

to N, then the robot is “close enough” to the trash can and the robot can drop
the can.
There are two problems with the behavior table. The first is that it doesn’t
show the sequence, or flow of control, clearly. The second is how did the
designer come up with those behaviors? This is where a FSA is particularly
helpful. It allows the designer to tinker with the sequence and represent the
behavioral design graphically.
Fig. 5.11 shows a FSA that is equivalent to the behavior table. The FSA
may be clearer because it expresses the sequence. It does so at the cost of not
showing precisely how the sequence would be implemented and encour-
aging the designer to create internal states. A programmer might imple-
ment two wander behaviors, one which is instantiated by different releasers
and terminates under different conditions, and two move-to-goal behaviors.
Many designers draw and interpret FSA as carrying forward previous re-
leasers. For example, the correct transition from Grab Trash to Wander For
Trash can is FULL and NO_BLUE, but a designer may be tempted to label the
arrow as only NO_BLUE, since to get that state, the gripper had to be FULL.
This is a very dangerous mistake because it assumes that the implementation
will be keeping up with what internal state the robot is in (by setting a vari-