Behavior-oriented Lunar Rover Control System Based on Multi-agent Architecture
- Paper number
IAC-07-A3.I.A.16
- Author
Prof. Ju Hehua, Chinese Society of Astronautics, China
- Coauthor
Mr. Pan Haining, Beijing University of Technology, China
- Year
2007
- Abstract
Firstly a mechanism architecture for lunar rover is briefly introduced in this paper. In allusion to it, a multi-agent control system architecture is presented based on fuzzy and behavior-oriented theories. This control system is realized in VxWorks operating system for rover's motion planning. The agent mentioned in this paper is defined as the essential entitiy that can sense and act on its environment, also including the ability of decision -making. Multi-agent system is a parallel one that can be carried out by the multi-task principle in VxWorks. By optimizing the tasks of monitoring, teleoperation and motion planning with VxWorks multi-task coordination, the speed and stability of system are both greatly enhanced. The basic behaviors include moving to goal, obstacle avoidance and so on. Reinforcement learning is introduced to optimize these behaviors. A learnable presentation structure of behavior is also presented to deal with the presentation issue that how to describe a modified or new behavior appropriately caused by behavior learning. Finally, the system is carried out on the rover platform and the experimental results show that the proposed technique is feasible.
- Abstract document