Design And Implementation Of Substation Intelligent Inspection Robot Based On ROS

With the rapid development of science and technology,robots have become a research focus in many fields,especially in the maintenance and inspection of substations in the power industry.The traditional manual inspection method is inefficient and high-risk,and the development of intelligent inspection robots has become an important means to solve this problem.This article is based on this background and mainly conducts research in the following areas:Firstly,in terms of robot hardware system design and ROS software system design,this article establishes a hardware system composed of a main controller,depth camera,laser radar,inertial sensor,and encoder,and uses the ROS system to realize communication and cooperation among the internal components of the robot.The components in this hardware system can work together to achieve the robot’s functions such as movement,perception,operation,and communication,providing important technical support for substation equipment inspection and monitoring.Secondly,for substation environments with few indoor features,many indoor features,and outdoor environments,this article studies SLAM algorithms based on 2D laser radar SLAM,2D laser radar and RGB-D camera data fusion SLAM,and 3D laser radar SLAM.Traditional Cartographer SLAM algorithms have been improved,and comparative analyses have been conducted in real and Gazebo simulated environments.Different SLAM algorithms have been applied to different environments to achieve accurate positioning and navigation,improving the efficiency and safety of inspection tasks.Then,this article analyzes three different path planning algorithms,namely Dijkstra algorithm,A algorithm,and Hybrid A* algorithm,and conducts simulation environment tests in Gazebo.Through analyzing their advantages and disadvantages and comparing simulation effects,Hybrid A* algorithm is ultimately chosen as the substation path planning algorithm.Finally,in order to enable the robot to detect illegal vehicles and pedestrians in time during the inspection process,the YOLOv5 algorithm was adopted to realize 2D plane vehicle and pedestrian detection,and the model identification accuracy was verified in KITTI public data set.Although the detection result was good,due to the limited 2D plane information,it could not fully meet the actual needs.The YOLOv5 algorithm was used to realize smoke and flame detection,as well as to identify key targets in the substation environment.In order to verify the superiority of 3D object detection over 2D object detection,a 3D object detection network based on key point feature pyramid is designed in this paper.The results show that the average detection accuracy of the proposed algorithm is better than that of the common models for automobile targets.Finally,3D object detection network is selected as the robot detection network for vehicles and pedestrians.The detection algorithm is embedded into the robot to realize accurate real-time inspection recognition task.The experimental results show that both the algorithm used in this paper and the improved algorithm have good results.In the complex environment of the substation,the robot has the characteristics of accurate drawing construction,high detection efficiency,and convenient deployment and can handle various complex environments and tasks.