Continual Policy Distillation of Reinforcement Learning-based Controllers for Soft Robotic In-Hand Manipulation

arXiv:2404.04219v1 Announce Type: cross
Abstract: Dexterous manipulation, often facilitated by multi-fingered robotic hands, holds solid impact for real-world applications. Soft robotic hands, due to their compliant nature, offer flexibility and adaptability during object grasping and manipulation. Yet, benefits come with challenges, particularly in the control development for finger coordination. Reinforcement Learning (RL) can be employed to train object-specific in-hand manipulation policies, but limiting adaptability and generalizability. We introduce a Continual Policy Distillation (CPD) framework to acquire a versatile controller for …

abstract adaptability applications arxiv benefits challenges continual control cs.ai cs.lg cs.ro development distillation flexibility grasping impact manipulation nature object policy reinforcement reinforcement learning robotic solid type world