Learning structured reactive navigation plans from executing MDP navigation policies

Autonomous robots, such as robot office couriers, need navigation routines that support flexible task execution and effective action planning. This paper describes XFRMLEARN, a system that learns structured symbolic navigation plans. Given a navigation task, XFRMLEARN learns to structure continuous navigation behavior and represents the learned structure as compact and transparent plans. The structured plans are obtained by starting with monolithic default plans that are optimized for average performance and adding subplans to improve the navigation performance for the given task. Compactness is achieved by incorporating only subplans that achieve significant performance gains. The resulting plans support action planning and opportunistic task execution. XFRMLEARN is implemented and extensively evaluated on an autonomous mobile robot.

• Type: Conference
• Year: 2001
• Where: AGENTS
• Authors: Michael Beetz, Thorsten Belker
• Keywords: AGENTS 2001, Security Privacy, Symbolic Navigation Plans, Task Execution, flexible Task Execution