DexPBT: Scaling up Dexterous Manipulation for Hand-Arm Systems with Population Based Training

2023/05/20 by Aleksei Petrenko, Arthur Allshire, Petrenko, Aleksei +7 · 3 citations
Computer Science · #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Reinforcement Learning in Robotics #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2305.12127

openalex publication_date 2023/05/20 · openalex created_date 2023/05/24 · openalex updated_date 2026/07/28

Abstract

In this work, we propose algorithms and methods that enable learning dexterous object manipulation using simulated one- or two-armed robots equipped with multi-fingered hand end-effectors. Using a parallel GPU-accelerated physics simulator (Isaac Gym), we implement challenging tasks for these robots, including regrasping, grasp-and-throw, and object reorientation. To solve these problems we introduce a decentralized Population-Based Training (PBT) algorithm that allows us to massively amplify the exploration capabilities of deep reinforcement learning. We find that this method significantly outperforms regular end-to-end learning and is able to discover robust control policies in challenging tasks. Video demonstrations of learned behaviors and the code can be found at https://sites.google.com/view/dexpbt

Cited by

• Time as a Control Dimension in Robot Learning
• Interactive Shaping of Granular Media Using Reinforcement Learning
• Deep Sensorimotor Control by Imitating Predictive Models of Human Motion

Related

• Learning Complex Dexterous Manipulation with Deep Reinforcement Learning\n and Demonstrations
• DeXtreme: Transfer of Agile In-hand Manipulation from Simulation to Reality
• Dexterous Manipulation with Deep Reinforcement Learning: Efficient, General, and Low-Cost
• Towards Human-Level Bimanual Dexterous Manipulation with Reinforcement Learning
• DexVIP: Learning Dexterous Grasping with Human Hand Pose Priors from Video