A Digital Twin-driven Intelligent Inspection Robotic System for Elevator Buffer in Confined Hazardous Space

Zhiyong Hu, Xuan Wang, Qingyuan Guo, Jing Bian, HongChen Liu, Jinkang Hu

Reviewer: ~Qingyuan_Guo1

Keywords: Digital Twin; Confined Space Robotics; Motion Planning; Visual Defect Detection; Elevator Safety; Autonomous Mobile Robot

Abstract: Elevator buffer is the final physical safety barrier of vertical transportation systems, whose regular and reliable inspection is mandatory to prevent catastrophic squatting and roof-rushing accidents.However,traditional manual inspection in narrow,dim,and high-risk elevator pit confined space suffers from extreme safety hazards, low efficiency, and strong subjective judgment deviation. Existing robotic inspection solutions lack a closed-loop virtual-real interaction framework based on digital twin (DT), leading to poor adaptability in complex unstructured pit environments, slow convergence of manipulator path planning, and insufficient generalization of defect detection models. To address these challenges, this paper proposes a full-process digital twin-driven intelligent inspection robotic system for elevator buffer autonomous inspection. First, a 1:1 high-fidelity digital twin model of the elevator pit is established with a four-layer architecture, including geometric twin, physical twin, behavioral twin, and rule twin, realizing real-time bidirectional mapping between virtual and physical spaces. On this basis, the digital twin framework is adopted to enable three core functions: prior map-based autonomous navigation and obstacle avoidance of the mobile robot, virtual pre-planning and real-time optimization of manipulator motion in confined space, and DT sample-enhanced visual defect detection and ranging. Finally, both virtual simulation experiments and physical prototype verification are conducted. The results show that our method achieves 92.3% defect detection mAP@0.5 and ±6.8mm ranging accuracy, while improving single-unit inspection efficiency by 67% compared with manual inspection, which verifies the effectiveness and superiority of the proposed digital twin-driven framework for confined space robotic inspection.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 1