We consider robot action planning to be the computational process of
generating and revising high-level robot control programs based on
foresight. Our research goal is to equip autonomous robot controllers
with robot action planning capabilities that enable them to perform
better than they possibly could without having these capabilities. Our
work on robot action planning concentrates on three aspects.
-
Methods for robot action planning (such as Probabilistic,
Prediction-based Schedule Debugging);
-
Realistic models for symbolically predicting concurrent reactive
robot behavior;
-
Runtime plan adaptation for autonomous robots;
