自律分散ロボットにおける多様な政策と代替性に関する研究

抄録

In multi-agent system constructed by learning-agent, even homogeneous agents with same hardware would produce diversity in their decision-making rules after learning. In this study, we introduced a measurement to describe the similarity between probabilistic decision-making rules of distributed autonomous robots. The experiment results showed the measurement can be used to describe the behavior similarity.