Bin Picking
Bin picking, the automated task of selecting and removing objects from a bin of jumbled items, aims to improve efficiency and reliability in various industries, particularly manufacturing and warehousing. Current research focuses on improving the robustness and speed of bin-picking systems through advancements in 6D pose estimation (using techniques like keypoint prediction and deep Hough voting), multimodal object detection and segmentation (leveraging RGB-D data and redundancy for reliability), and efficient grasp planning (incorporating dynamic manipulation, adaptive grippers, and online tool selection). These improvements are crucial for expanding the applicability of robotic automation to complex, unstructured environments and increasing the productivity of automated processes.
Papers
SD-Net: Symmetric-Aware Keypoint Prediction and Domain Adaptation for 6D Pose Estimation In Bin-picking Scenarios
Ding-Tao Huang, En-Te Lin, Lipeng Chen, Li-Fu Liu, Long Zeng
MOTPose: Multi-object 6D Pose Estimation for Dynamic Video Sequences using Attention-based Temporal Fusion
Arul Selvam Periyasamy, Sven Behnke