Learning with Less Effort: Efficient Training and Generalization in (Multi-)Robot Systems

The growing demand for automation has sparked interest in multirobot systems that can handle complex tasks through collaboration. While these systems offer advantages in speed, coverage, and capability compared to single robots, getting multiple robots to learn and coordinate effectively remains challenging-training robots requires extensive effort on data and computation, and learned policies often struggle to generalize beyond training conditions. My research addresses two fundamental challenges in multi-robot learning: reducing the training effort required and improving generalization to reduce policy retraining. First, we propose methods to make training more efficient-using human-drawn sketches rather than teleoperated demonstrations for manipulation tasks, and utilizing individual robot demonstrations instead of joint multirobot ones for learning collaborative behaviors. Second, we develop techniques to help learned policies adapt to new scenarios without retraining-introducing frameworks that maintain coordination under different observation conditions and enable effective information sharing across varying initial states. My work aims to create more practical and adaptable multi-robot systems that can be efficiently trained and deployed across diverse real-world settings.