Pseudo-Labeling and Contextual Curriculum Learning for Online Grasp Learning in Robotic Bin Picking
This addresses the challenge of dynamic adaptation in robotic grasping for industrial applications, though it appears incremental as it builds on existing methods.
The paper tackles the problem of online grasp learning for robotic bin picking by introducing SSL-ConvSAC, which combines semi-supervised and reinforcement learning to adapt to novel scenarios like unseen objects and camera perspectives, showing promise in real-world evaluations with a physical robot arm.
The prevailing grasp prediction methods predominantly rely on offline learning, overlooking the dynamic grasp learning that occurs during real-time adaptation to novel picking scenarios. These scenarios may involve previously unseen objects, variations in camera perspectives, and bin configurations, among other factors. In this paper, we introduce a novel approach, SSL-ConvSAC, that combines semi-supervised learning and reinforcement learning for online grasp learning. By treating pixels with reward feedback as labeled data and others as unlabeled, it efficiently exploits unlabeled data to enhance learning. In addition, we address the imbalance between labeled and unlabeled data by proposing a contextual curriculum-based method. We ablate the proposed approach on real-world evaluation data and demonstrate promise for improving online grasp learning on bin picking tasks using a physical 7-DoF Franka Emika robot arm with a suction gripper. Video: https://youtu.be/OAro5pg8I9U