An Examination of Behavioral Strategies in Human Agents Utilizing Active and Passive Approaches.

This research focuses on human cooperative behavior and aims to create robots that can communicate and cooperate with humans using both passive and active action strategies. In human-to-human collisions, when strategies cannot be inferred from the actions of the opponent, they sometimes act on their own to indicate their own strategies. In previous robotics research, robots have often acted following the strategies of human participants, and only a few robots have acted on their own to indicate their strategies. Previous work has resulted in a Meta-Strategy that utilizes two action strategies obtained through agent simulation. However, since human Meta-Strategy may vary between individuals, this study conducted a collision avoidance experiment in a grid space to analyze whether subjects' actions change when faced with agents using active and passive giving way strategies. Results indicate that some subjects adapt their actions in response to the agent's strategy, while others exhibit consistent behavior regardless of the agent's strategy. These differences in behavior may be attributed to variations in Meta-Strategies. The study discusses the possibility of creating agents that can switch between active and passive strategies, assuming a human Meta-Strategy.