Autonomous positioning and navigation of a quadruped robot for power transmission inspection integrating IMU and visual SLAM

In response to the difficulty of accurately identifying and recognizing complex scenes in power transmission inspection sites for quadruped robots, which leads to poor autonomous positioning and navigation accuracy of robots, a power transmission inspection quadruped robot autonomous positioning and navigation method that integrates IMU and visual SLAM is proposed. Firstly, the original point cloud is segmented, and IMU pre integrated pose is used to remove the nonlinear motion distortion of the original point cloud. The obtained point cloud is then subjected to line and surface feature extraction. Then match the line and surface features of adjacent frames, and fuse the initial pose obtained by IMU pre integration in the hierarchical pose estimation process. Finally, based on visual SLAM optimization constraints on the pose, keyframes are matched with local maps, and factor graph optimization is used to achieve accurate SLAM for power transmission inspection, realizing autonomous positioning and navigation of the quadruped robot for power transmission inspection. The experimental results show that the proposed method has high autonomous positioning and navigation accuracy, high attitude control accuracy, and short time consumption for the quadruped transmission inspection robot. It provides technical reference for intelligent perception and safety inspection of the quadruped transmission inspection robot.