Plan-Based Control for Autonomous Soccer Robots Preliminary Report

摘要

Robotic soccer has become a standard "real-world" testbed for autonomous robot control. This paper presents our current views on the use of plan-based control mechanisms for autonomous robotic soccer. We argue that plan-based control will enable autonomous soccer playing robots to better perform sophisticated and fine tuned soccer plays. We present and discuss some of the plan representations that we have developed for robotic soccer. Finally, we outline extensions of our plan representation language that allow for the explicit and transparent specification of learning problems within plans. This extended language enables robot controllers to employ learning subplans that can be reasoned about and manipulated.