Building an Adaptive Inspection Robot System to Improve the Quality of Power Transmission Line Operation and Maintenance in Mountainous Areas

This is the authors' abstract. We don't add key points for this paper.

In view of the problems of low efficiency of manual inspection and prominent safety hazards caused by the wide distribution of power transmission lines in mountainous areas, complex terrain, and frequent extreme weather, this paper proposes an adaptive inspection robot system based on the fusion of multimodal perception and dynamic decisionmaking. First, a modular reconfigurable mechanical structure is designed to achieve dynamic posture adjustment and stable obstacle crossing in complex terrains such as steep slopes and ice-covered wires. Secondly, a multi-level perception network is constructed, and the online identification of typical defects such as insulator damage and hardware corrosion is achieved through a lightweight YOLO-Transformer hybrid model. Finally, a dynamic path planning algorithm based on deep reinforcement learning (Deep Reinforcement LearningDynamic Path Planning, DRL-DPP) is proposed to autonomously generate risk avoidance strategies and optimal inspection routes. Experiments show that the system has an accuracy rate of about 9 3. 0 % in defect detection in a simulated mountain environment, and the inspection efficiency is about 2.93 times higher than that of manual inspection. The average response time of dynamic path planning is about 0.81 seconds, and the response time distribution shows a good normal trend. The comprehensive test results verify that the designed system has excellent environmental adaptability, task execution stability and intelligent decision-making capabilities in complex environments, providing practical technical support for the intelligent operation and maintenance of power transmission lines in mountainous areas.