Development, Planning and Control of an Autonomous Mobile Manipulator for Power Substation Live-Maintaining

In recent decades, there has been a surge in the development of diverse robotic systems aimed at the maintenance of substation equipment. With advances in machinery manufacturing, artificial intelligence and computers, automatic robots have a promising prospect in autonomous maintenance for substation. But, at present, extensive research and applications of robotics are focused on inspection robots and less on autonomous maintenance robots. Autonomous maintenance robots can replace manual operations with large volumes and high risks in harsh environments, capabilities that inspection robots do not have. However, performing complex maintenance tasks for robots in substation scenarios with many uncertainties is quite challenging. Autonomous maintenance robots require precise positioning and navigation of the target position, robust detection of power equipment, and fast motion planning of redundant manipulators in open-type substations. We propose a complete solution for the autonomous maintaining operation of a mobile manipulator in an open-type substation. First, a multimodal framework for semantic visual simultaneous localization and mapping (VSLAM) and a global path planning method based on A-star algorithm and time-elastic-band (TEB) algorithm are proposed for autonomous navigation. Second, a YOLOV8-based detection method for power equipment is described. Third, a linear-variational-inequality-based primal-dual neural network (PDNN) is designed to plan the motion of the manipulator for the maintenance tasks. Finally, an autonomous mobile manipulator is designed to validate the proposed method. The experiment results demonstrate effectiveness of the robot in the common substation equipment maintenance tasks.