| With the rapid development of robot technology and intelligence,mobile robots have been widely used in work scenarios such as substation inspections and hazardous environment reconnaissance,replacing manual operations and reducing safety hazards and labor costs.In order to better meet the needs of quickly deploying wheeled inspection robots for inspection in unknown and complex environments.This paper focuses on the problem of autonomous exploration of mobile robots in order to better meet the needs of quickly deploying wheeled inspection robots for inspection in unknown and complex environments.We propose an improved two-stage viewpoint planning method based on clustering and traveling salesman planning,which can achieve autonomous exploration in various unknown and complex indoor and outdoor environments.We first present the inspection robot system framework,analyzing the robot motion model and sensor model,including laser radar.We complete the calibration of robot parameters such as speed parameters and coordinate system transformation parameters,including vehicle,laser radar,and 6-axis IMU coordinate system.We also provide detailed implementation plans for the simulation environment experiment platform and wheeled robot experiment platform.Next,we propose a robot autonomous exploration method based on DSVP and specifically analyze the two stages in this algorithm.In the exploration stage,the random tree grown by the dynamic expansion RRT is used as the local path planning,and the tree nodes of the random tree are used as viewpoints.During the path planning process,the viewpoints are screened,and the point with the maximum gain among all the viewpoints is selected as the local target point.However,this method’s exploration direction is too random and may not work in certain special structures,and the selection of the local target point within the local range is not necessarily optimal.In the relocation stage,the global target point closest to the current robot position is selected for navigation.The DSVP method ignores the spatial distribution information of all unexplored sub-regions in the overall space and the spatial distribution information of target points in local and global exploration.To address these problems,we propose an improved two-stage viewpoint planning method based on clustering and traveling salesman planning.We specifically analyze the clustering method,traveling salesman planning method,and mapping and localization algorithms used in the exploration algorithm and provide detailed optimization and implementation plans for the two stages.Finally,we conduct autonomous exploration and inspection experiments based on the proposed method.We complete parameter settings for the simulation platform based on different environments and conduct comparative experiments between the two methods in five different types of simulation environments.We also conduct testing and analysis in two real environments and apply the method to the autonomously developed remote inspection monitoring system and local monitoring system.The experiments demonstrate that the proposed method has significant practical value. |
