Cooperative Reinforcement Learning: Brief Survey and Application to Bio-insect and Artificial Robot Interaction

摘要

In this paper, we address our on-going research called BRIDS that is a trial to formulate bio-insect and artificial robot interaction based on reinforcement learning. First we briefly review relevant research works in this field. Since cooperation among mobile agents is a key technology, we examine cooperative reinforcement learning. Specifically we will briefly review the concept of area of expertise in cooperative learning area. In fact bio-insect and artificial robot interaction has been studied in Leurre. Our research, however, has a key difference from Leurre in that it is to drive bio-insect by coordination of a group of mobile robots towards a desired point. In author's best knowledge, driving bio-insect towards a desired point based on fully autonomous coordination of multiple mobile agents has not been studied in existing publications.