cs.ROMar 31, 2026

Industrial-Grade Robust Robot Vision for Screw Detection and Removal under Uneven Conditions

Authors: Tomoki Ishikura, Genichiro Matsuda, Takuya Kiyokawa, Kensuke Harada

Organizations: Manufacturing Innovation Division, Panasonic Holdings Corporation, 2-7 Matsuba-cho, Kadoma, Osaka, Japan · Department of Systems Innovation, Graduate School of Engineering Science, The University of Osaka, 1-3 Machikaneyama, Toyonaka, Osaka, Japan · Industrial Cyber-physical Systems Research Center, The National Institute of Advanced Industrial Science and Technology (AIST), 2-3-26 Aomi, Koto-ku, Tokyo, Japan

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

As the amount of used home appliances is expected to increase despite the decreasing labor force in Japan, there is a need to automate disassembling processes at recycling plants. The automation of disassembling air conditioner outdoor units, however, remains a challenge due to unit size variations and exposure to dirt and rust. To address these challenges, this study proposes an automated system that integrates a task-specific two-stage detection method and a lattice-based local calibration strategy. This approach achieved a screw detection recall of 99.8% despite severe degradation and ensured a manipulation accuracy of +/-0.75 mm without pre-programmed coordinates. In real-world validation with 120 units, the system attained a disassembly success rate of 78.3% and an average cycle time of 193 seconds, confirming its feasibility for industrial application.

Figures & tables

Figure 1: Overview of the automated disassembly system.
Figure 2: Required manipulation precision of ±0.75\pm 0.75 mm for screw removal within the workspace ( 900×400×750900\times 400\times 750 mm, width ×\times depth ×\times height).
Figure 3: Associated challenges in the proposed system.
Figure 4: Screws existing in actual recycling factory.
Figure 5: Proposed method for detecting screws with variable visual conditions.
Figure 6: Test Evaluation Gauge (TEG) used for the preliminary evaluation. The gauge replicates the size of an air conditioner outdoor unit and provides controlled pose variations (in-plane rotation ±4∘\pm 4^{\circ} and tilt ±6∘\pm 6^{\circ} ).
Figure 7: Example of the screw detection process. (a) Input RGB-D image. (b) Probability map. (c) Detection result (yellow markers indicate recognized screw centers).
Figure 8: The calibration jig and the driver-hand.
Figure 9: A scene of calibration.
Figure 10: The acquired calibration dataset.
Figure 11: Target workpieces.
Figure 12: Robotic disassembly for actual air conditioner outdoor unit.
Figure 13: Exceptional detection failure cases: (a, b) dirt/spider webs; (c, d) severe deformation/cracking.
Figure 14: Robotic disassembly for actual microwave oven.
More in this topic

Robotics

Robotics studies the design and control of machines that perceive, move, and act in physical environments.

See all 735 papers

Explore similar work

CardsList

  1. Robotic Agentic Platform for Intelligent Electric Vehicle Disassembly

    Mar 19, 2026Zachary Allen, Max Conway, Lyle Antieau +2Robotic ManipulationLanguage Model-Based Control

    Department of Computer Science, University of Colorado, Boulder

  2. ROBOCYCLE: Autonomous Dual-Arm Robotic Manipulation and Coordination for Recycling Applications

    Jul 3, 2026Rubén de J. Hilario-Cruz, Jesus A. García-González, Enrique Coronado +1Robotic GraspingBimanual Robotic Manipulation

    Tecnologico de Monterrey, School of Engineering and Sciences, Department of Computer Science, Ave. Eugenio Garza Sada 2501, Monterrey 64849, NL, Mexico. · National Institute of Advanced Industrial Science and Technology (AIST) Tokyo Waterfront, 2 Chome-3-26 Aomi, Koto City, Tokyo 135-0064, Japan.

  3. SMR-Net:Robot Snap Detection Based on Multi-Scale Features and Self-Attention Network

    Mar 1, 2026Kuanxu HouObject DetectionRobotic Assembly