Emergence of Scenario-Appropriate Collaborative Behaviors for Teams of Robotic Bodyguards

摘要

We are considering the problem of controlling a team of robotic bodyguards protecting a VIP from physical assault in the presence of neutral and/or adversarial bystanders in a variety of scenarios. This problem is challenging due to the large number of active entities with different agendas and dynamic movement patterns, the need of cooperation between the robots as well as the requirement to take into consideration criteria such as social norms in addition to the main goal of VIP safety. We show how a multi-agent reinforcement learning approach can evolve behavior policies that outperform hand-engineered approaches. Furthermore, we propose a novel multi-agent reinforcement learning algorithm inspired by universal value function approximators that can learn policies that exhibit appropriate, distinct behavior in environments with different requirements.