Page 216 - Designing Autonomous Mobile Robots : Inside the Mindo f an Intellegent Machine

Instilling Pain, Fear and Confidence

3. Disorientation
4. Diminished fitness

For example, after being in an accident we will tend to drive less aggressively for some
period. Even after many years, if we find ourselves in the same location or situation
in which the accident occurred, we may become less confident and drive more defen-
sively. Finally, if we feel somehow incapacitated or realize we may have missed a turn, we
will have a similar reaction.

Effects of virtual confidence
The proper responses of an autonomous robot to a loss of confidence are therefore rel-
atively clear. While a robot remains in a state of diminished confidence, it should:

1. Reduce speed as a percentage of programmed speed
2. Reduce planning deceleration to begin slowing further away from obstacles

3. Reduce acceleration
4. Reduce torque limits on drive servos

By doing these things, we give the robot more time to sense stealthy obstacles, and
we assure that if there is an incident it will be less severe than it would otherwise
have been. It is important to remember that the impact force of our robot is propor-
tional to its weight, but proportional to the square of its velocity! If we have these
various control parameters in an accessible form, and not as constants , then modify-
5
ing them should be a simple matter.

Going back to our example from Figure 13.1, let’s assume that we have experienced
problems on the DA path segment and thus stopped using the GE-ED-DA path in
favor of the GE-EB-BA route. Now let’s assume that we experience a problem along
EB that makes this route more costly (risky) than the original GE-ED-DA route, but
not as costly as going all the way around the GE-EF-FC-CB-BA route. If we had
weighted our penalty points higher, then perhaps the long route would have been
the best remaining approach, but assume this is not the case. We are going to go
back to using the GE-ED-DA route, but we need to reflect the increased risk of this
route by reducing the robot’s confidence.

5 See Chapter 6.

199