End-to-End Learning-Based Autonomous Operation Robot for Substation Switchgear

In recent years, the rapid advancement of intelligent robotics technology has led to the widespread integration of robots across various industries. Within the power sector, the live operation of substation switchgear presents significant safety risks and demands specialized technical expertise. To enhance operational safety and efficiency, the development of autonomous robotic systems has become imperative. This study focuses on the design and development of an autonomous switchgear operation robot, specifically tailored to meet the operational demands of substation switchgear. The research introduces an innovative “one system, three tasks” autonomous operation framework. At the system level, the robot incorporates an automated tool-switching mechanism that enables efficient transitions between multiple operation tools. At the task level, the study proposes a novel multi-task perception algorithm for autonomous switchgear operations, which integrates target detection, posture regression, and force point determination into a unified end-to-end deep learning network. This comprehensive approach facilitates precise autonomous manipulation of various switchgear components, including buttons, knobs, and handcarts. Experimental evaluations demonstrate that the developed embodied intelligent system, characterized by its seamless software-hardware integration, exhibits stable and efficient performance, effectively meeting the operational requirements for autonomous substation switchgear operations.