Adaptive Team Behavior Planning using Human Coach Commands

In its operating life, an agent that needs to act in real environments is required to deal with rules and constraints that humans ask to satisfy. The set of rules specified by the human might influence the role of the agent without changing its goal or its current task. To this end, classical planning methodologies can be enriched with temporal goals and constraints that enforce non-Markovian properties on past traces. This work aims at exploring the application of real-time dynamic generation of policies whose possible trajectories are compliant with a set of Pure-Past Linear Time Logic rules, introducing novel human-robot interaction modalities for the high-level control of strategies for multiple agents. For proving the effectiveness of the proposed approach, we have carried out an evaluation on a partially observable, unpredictable, and dynamic scenario: the RoboCup soccer competition. In particular, we exploit human indications to condition the robot’s behavior before or during the time of the match, as happens during human soccer matches.