Multi-agent based Robot Control
Japanese version
Overview
We have been developing a distributed robot control system based on a multi-agent framework in which flexible robot behaviors emerge through interactions of autonomous agents.
Background
Conventional robot control based on a centralized, monolithic architecture has been difficult to handle an unknown, dynamically changing real world. A promising way to cope with this problem is to build a robot control system by coordinating multiple functional agents, each of which independently controls a robot's sensors or actuators. In this distributed control framework, flexible behaviors can be achieved in such a way that each agent partially copes with the uncertainties of the real world. Moreover, a robot can evolve motion by adding new agents for new sensors or actuators. We call this control scheme multi-agent based robot control and have constructed an experimental system with a 6-axis manipulator.
Topics
Last modified on Tue Oct 6 17:22:16 1998
