Interaction-limited Inverse Reinforcement Learning
This addresses the challenge of efficient learning in scenarios where teacher availability or access to learning dynamics is restricted, though it appears incremental as it builds on existing IRL methods with specific training improvements.
The paper tackles the problem of accelerating inverse reinforcement learning when teacher-learner interactions are limited, proposing two training strategies (CIRL and SPIRL) that enable faster training compared to random teachers or batch learners in simulations and real robot tasks.
This paper proposes an inverse reinforcement learning (IRL) framework to accelerate learning when the learner-teacher \textit{interaction} is \textit{limited} during training. Our setting is motivated by the realistic scenarios where a helpful teacher is not available or when the teacher cannot access the learning dynamics of the student. We present two different training strategies: Curriculum Inverse Reinforcement Learning (CIRL) covering the teacher's perspective, and Self-Paced Inverse Reinforcement Learning (SPIRL) focusing on the learner's perspective. Using experiments in simulations and experiments with a real robot learning a task from a human demonstrator, we show that our training strategies can allow a faster training than a random teacher for CIRL and than a batch learner for SPIRL.