Efficient Self-Supervised Data Collection for Offline Robot Learning

A practical approach to robot reinforcement learning is to first collect a large batch of real or simulated robot interaction data, using some data collection policy, and then learn from this data to perform various tasks, using offline learning algorithms. Previous work focused on manually designing the data collection policy, and on tasks where suitable policies can easily be designed, such as random picking policies for collecting data about object grasping. For more complex tasks, however, it may be difficult to find a data collection policy that explores the environment effectively, and produces data that is diverse enough for the downstream task. In this work, we propose that data collection policies should actively explore the environment to collect diverse data. In particular, we develop a simple-yet-effective goal-conditioned reinforcement-learning method that actively focuses data collection on novel observations, thereby collecting a diverse data-set. We evaluate our method on simulated robot manipulation tasks with visual inputs and show that the improved diversity of active data collection leads to significant improvements in the downstream learning tasks.

Authors

• Shadi Endrawis
• Gal Leibovich
• Guy Jacob
• Gal Novik
• Aviv Tamar

Keywords

• Visualization
• Automation
• Conferences
• Reinforcement learning
• Grasping
• Data collection
• Robot learning
• Learning Task
• Robot Manipulator
• Real Robot
• Offline Learning
• Neural Network
• High-dimensional
• Value Function
• Supervised Learning
• State Space
• Intrinsic Motivation
• Unique Conditions
• Collection Of Datasets
• Reward Function
• Markov Decision Process
• Target Network
• Reinforcement Learning Algorithm
• Simulation Domain
• Intrinsic Rewards
• Reinforcement Learning Agent
• Replay Buffer
• State Visit
• Beginning Of Episode
• High-dimensional State Space
• Random Exploration
• Data Collection Methods