| In many industrial fields such as electric power,coal mine,and chemical industry,the traditional on-site inspection work still adopts the manual inspection method of fixed time and fixed route,which has the defects of low efficiency,high repetition and high work danger.With the development of technology,the autonomous inspection mode of mobile robots has been greatly developed,which can effectively solve the defects of the above manual inspection,and is a typical application and research hotspot of autonomous navigation and path planning research.Taking the wheeled robot and ROS(Robot Operating System)system as the development platform,this study proposed the design and development of the outdoor inspection robot system based on lidar mapping and MGA(Multi-Object Genetic Algorithm)algorithm,which realizes the autonomous navigation function of the outdoor inspection robot,significantly improves the safety of the planning path,and reduces the path length and planning time.The specific research content of the thesis is as follows:(1)In terms of the construction and preprocessing of navigation maps,the thesis proposes an obstacle detection mechanism,which first reduces the number of point clouds by downsampling of voxel rasters.Secondly,the random sample consensus(RANSAC)algorithm is used to extract and segment the ground point cloud to avoid misclustering.The Euclidean clustering method based on K-d tree nearest neighbor search was used again to cluster obstacle point clouds.Finally,based on PCA principal component analysis method,a three-dimensional bounding box of obstacles is generated on the autonomous navigation map to ensure the safety of autonomous navigation.(2)In terms of path planning,the thesis improves the traditional genetic algorithm and proposes to improve the genetic algorithm MGA algorithm.In this algorithm,an adaptation function including obstacle avoidance safety is constructed to obtain the path safety selection index,and the detection operator and the deletion operator are introduced at the same time,the former can detect and delete duplicate redundant nodes,thereby improving the evolution efficiency and reducing the planning time.The latter can replace the polyline segment with a straight line segment without obstacles more reasonably in the path planning,effectively reducing the path length.Through simulation experiments,compared with RRT* and traditional genetic algorithms,the MGA algorithm proposed in the thesis has significantly improved the safety of obstacle avoidance,and the path length and planning time have also been greatly improved.(3)On the basis of the above research results,the thesis completes the construction of an autonomous wheeled mobile robot platform based on lidar,and carries out corresponding outdoor autonomous navigation experiments.The experimental results show that on the point cloud map constructed by Li DAR,the detection rate of obstacles in the passable road range is high,and the path planning function of mobile robot based on MGA algorithm is realized,and the obstacle avoidance safety is significantly improved,and the path length and planning time are also greatly improved. |
