Goal-Auxiliary Actor-Critic for 6D Robotic Grasping with Point Clouds
arXiv:2010.00824
Abstract
6D robotic grasping beyond top-down bin-picking scenarios is a challenging task. Previous solutions based on 6D grasp synthesis with robot motion planning usually operate in an open-loop setting, which are sensitive to grasp synthesis errors. In this work, we propose a new method for learning closed-loop control policies for 6D grasping. Our policy takes a segmented point cloud of an object from an egocentric camera as input, and outputs continuous 6D control actions of the robot gripper for grasping the object. We combine imitation learning and reinforcement learning and introduce a goal-auxiliary actor-critic algorithm for policy learning. We demonstrate that our learned policy can be integrated into a tabletop 6D grasping system and a human-robot handover system to improve the grasping performance of unseen objects. Our videos and code can be found at https://sites.google.com/view/gaddpg .
References in corpus (14)
- Adam: A Method for Stochastic Optimization
- Continuous control with deep reinforcement learning
- QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulation
- Leveraging Demonstrations for Deep Reinforcement Learning on Robotics Problems with Sparse Rewards
- Hindsight Experience Replay
- Visual Reinforcement Learning with Imagined Goals
- Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics
- Data-efficient Deep Reinforcement Learning for Dexterous Manipulation
- Reinforcement and Imitation Learning for Diverse Visuomotor Skills
- Goal-conditioned Imitation Learning
- Data-Efficient Learning for Sim-to-Real Robotic Grasping using Deep Point Cloud Prediction Networks
- Learning to Reach Goals via Iterated Supervised Learning
- Visual Imitation Made Easy
- Reactive Human-to-Robot Handovers of Arbitrary Objects