A Homogeneous Robot Group Approach to Cooperative Carrying Problems

摘要

A distributed autonomous approach to adaptive system design is investigated through Cooperative Carrying Problem (CCP), in which a homogeneous group of autonomous moving robots have to cooperate in carrying a load. A robot's action generator is designed using two Reinforcement Learning (RL) units in order to cope with the dynamics in CCP. The employed strategy is that each robot is learning its behavior as it were in a static environment, by expanding the state space of the main RL unit using the prediction information of the nearest future that is generated by the supplimentary RL unit. Computer simulations with two robots are conducted in order to illustrate the effectiveness of the proposed approach. Cooperation through autonomous task decomposition is observed in the achieved stable behavior.