An Integrated System for WEEE Sorting Employing X-ray Imaging, AI-based Object Detection and Segmentation, and Delta Robot Manipulation

TL;DR

The paper tackles the safety and throughput challenges of removing batteries from diverse WEEE by introducing an autonomous sorting pipeline that fuses dual-energy X-ray imaging with AI-based detection (YOLO) and segmentation (U-Net), and a Delta robot for precise pick-and-place on a moving conveyor. A position-estimation module translates image coordinates into real-world locations with future-state prediction, while a NVIDIA Isaac Sim-based digital twin validates and refines robot trajectories. The integrated system demonstrates high detection performance, reliable device localization, and smooth robotic manipulation, achieving fully autonomous battery sorting in both simulation and real hardware contexts. This approach promises safer, faster, and more scalable recycling operations by reducing human exposure and increasing throughput.

Abstract

Battery recycling is becoming increasingly critical due to the rapid growth in battery usage and the limited availability of natural resources. Moreover, as battery energy densities continue to rise, improper handling during recycling poses significant safety hazards, including potential fires at recycling facilities. Numerous systems have been proposed for battery detection and removal from WEEE recycling lines, including X-ray and RGB-based visual inspection methods, typically driven by AI-powered object detection models (e.g., Mask R-CNN, YOLO, ResNets). Despite advances in optimizing detection techniques and model modifications, a fully autonomous solution capable of accurately identifying and sorting batteries across diverse WEEEs types has yet to be realized. In response to these challenges, we present our novel approach which integrates a specialized X-ray transmission dual energy imaging subsystem with advanced pre-processing algorithms, enabling high-contrast image reconstruction for effective differentiation of dense and thin materials in WEEE. Devices move along a conveyor belt through a high-resolution X-ray imaging system, where YOLO and U-Net models precisely detect and segment battery-containing items. An intelligent tracking and position estimation algorithm then guides a Delta robot equipped with a suction gripper to selectively extract and properly discard the targeted devices. The approach is validated in a photorealistic simulation environment developed in NVIDIA Isaac Sim and on the real setup.

Figures (8)

• Figure 1: The developed WEEE sorting system. (1) The conveyor belt, (2) the X-ray transmission imaging system, (3) the WEEE devices and (4) the suction gripper attached to (5) a delta robot. a) X-ray imaging module, b) Delta robot, c) System in the NVIDIA Isaac Sim simulation environment.
• Figure 2: Class distribution in training and validation datasets
• Figure 3: Advanced X-ray imaging, device segmentation and battery detection on a tablet, a power bank and a laptop battery. a) HE X-ray image, b) Device segmentation using U-Net model, c) Battery detection on the reconstructed X-ray image, with annotations of the devices' center, d) Localization for future position estimation of devices with batteries. The width and height of the image correspond to a specific plane on the conveyor belt.
• Figure 4: Detailed model of THL Delta robot in NVIDIA Isaac sim.
• Figure 5: The green trajectory represents a 3D Pi-shaped path, where the intermediate points maintain a constant offset $h$. The purple point is determined as the median of the two intermediate points. A curvature factor $\alpha$ is applied to adjust the intermediate points accordingly. Using this median point, the minimum acceptable values are computed and highlighted in yellow. This approach ensures that the generated trajectory (blue) remains constrained within the boundaries defined by the Pi-shaped path.