Page 299 - Introduction to AI Robotics

282

a. 7 The Hybrid Deliberative/Reactive Paradigm
b.
Figure 7.11 Two robots using TCA: a.) Xavier, a RWI robot at Carnegie Mellon Uni-
versity (photograph courtesy of Reid Simmons) and b.) Dante (photograph courtesy
of NASA Ames Research Center).

given several jobs to drop, Prodigy can prioritize and optimize the sched-
ule. Once the current task has been established, the Path Planning layer is
engaged. Navigation is handled by a Partially Observable Markov Decision
Process (POMDP, pronounced “pom D P”) module which determines what
the robot should be looking for, where it is, and where it has been. As with
the relationship between strategic and tactical behaviors in SFX, the Obsta-
cle Avoidance Layer takes the desired heading and adapts it to the obstacles
CURVATURE-VELOCITY extracted from the evidence grid virtual sensor. The layer uses a curvature-
METHOD velocity method (CVM) to factor in not only obstacles but how to respond with
a smooth trajectory for the robot’s current velocity.
The table below summarizes TCA in terms of the common components
and style of emergent behavior:

TCA
Sequencer Agent Navigation Layer
Resource Manager Navigation Layer
Cartographer Path-Planning Layer
Mission Planner Task Scheduling Layer
Performance Monitoring Agent Navigation, Path-Planning, Task-Scheduling
Emergent behavior Filtering