A CNN-Based Fast Picking Method for WEEE Recycling

Recycling of Waste Electrical and Electronic Equipment (WEEE) is challenging due to the high variety in both design composition products this waste stream. Therefore, targeted demanufacturing appropriate material sorting processes are crucial, but remain often a manual thus labor-intensive costly activity. Despite first steps towards automation these can be noticed, many them sorting, very few at product level. In perspective, one key tasks automated classification, picking manipulation WEEE products. presented research, closed-loop grasp planning method proposed for random The grasping combines Convolutional Neural Network-based quality prediction Closed-Loop control, named CNNB-CL, find optimized region unknown objects dense clutter. A large-scale dataset generated CNN training, containing over 2.3 million synthetic grasps, their corresponding qualities evaluated by simulations using 3D models. 7-layer light-weighted quantitatively estimates each candidate. addition, CNNB-CL utilizes feedback from force-torque sensor detect adjust strategy. Various empirical experiments prove that finds an clutter 2,000 candidates within 0.3 s achieves average success rate 92% different clutters. Furthermore, speed as 1,400 mm/s small (< 300 g) with FANUC Delta robot (M-2i 3SL). fast-picking ability reveals its potential application (pre-) processing WEEE.