A Framework for Robotic Bin Packing with a Dual-Arm Configuration

Abstract

In recent years, dual-arm robots have been widely studied for their high redundancy in many dexterous manipulation tasks, such as assembly, handling, and re-grasping. As a result, dual-arm robots are expected to be a potential solution for many logistics and manufacturing scenario. In this work, we propose a framework for a logistics application with a dual-arm robot, the Robotic Bin Packing (RBP). Through a two-stage visual perception and joint planning approaches, boxes with various dimensions can be successfully picked, re-grasped, and then placed at an expected pose. In addition, a heuristic approach is exercised offline for finding efficient re-grasping poses of the robot. The experimental results show that the execution time of manipulation can be improved through the proposed approach. We believe that this framework can provide practitioners with a feasible solution for bin packing issues in logistics.