Vision-Guided Dual-Arm Humanoid Robotic Disassembly of End-of-Life 18650 Lithium-ion Battery Packs

2026-06-06

AI Summary

This work proposes a vision-guided, fully automated disassembly method for fixture-free, retired 18650 battery packs with unknown initial poses, employing a dual-arm humanoid robot. Relying solely on generic parallel grippers, RGB-D sensing, and a pretrained grasp detector, the approach achieves precise extraction of all 21 cells in arbitrary orientations through a closed-loop wrist-mounted camera correction scheme and a dual-arm mid-range cooperative support-and-transfer strategy. The system eliminates the need for external fixtures by integrating perception-driven learning with filtering mechanisms to handle pose uncertainty and extends the effective workspace of the dual arms, thereby overcoming the reliance of existing systems on fixed clamping or specialized tools. Experiments demonstrate an end-to-end disassembly success rate of 8 out of 10 trials, a cell localization root-mean-square error of 2.4 mm, and an average processing time of 6.0 minutes per pack, offering a practical fixture-free disassembly module for industrial-scale battery recycling.

Abstract

The growing volume of retired lithium-ion battery packs from electric vehicles and portable electronics calls for automated disassembly that is safe, flexible, and selective down to the individual cell. Existing robotic systems, however, mostly assume known pack poses, external fixtures, or specialised tooling, leaving fixture-free cell-level disassembly under pose uncertainty largely unsolved. This paper presents a vision-guided dual-arm pipeline that disassembles a 21-cell 18650 pack from an arbitrary initial pose using only general-purpose parallel-jaw grippers, RGB-D sensing, and a pre-trained grasp detector. Pose uncertainty is absorbed by a learn-and-filter perception stack with discrete look-and-move wrist-camera corrections, while a mid-task support transfer between the two arms extends the effective workspace without any external clamp. The pipeline achieves an 8/10 end-to-end success rate, a cell-localisation root-mean-square error of $2.4$\,mm, and a mean cycle time of 6.0\,minutes per pack, providing a practical, fixture-free building block for industrial battery recycling.

Problem

• fixture-free disassembly
• pose uncertainty
• cell-level disassembly
• lithium-ion battery recycling
• humanoid robotics

Innovation

• vision-guided robotics
• dual-arm manipulation
• fixture-free disassembly
• pose uncertainty
• battery recycling