Research On Autonomous Exploration System Of Coal Mine Rescue Robot

The autonomous exploration of coal mine rescue robots refers to the process in which rescue robots make autonomous decisions in the unknown environment after a catastrophe through the equipped sensors,decide the next execution path,and construct a catastrophic environment map in real time.It is an important research issue in the field of robotics.The underground environment of the coal mine after the disaster is complex and changeable,and there is a risk of secondary explosion.Research on a coal mine rescue robot that replaces rescuers can enter the scene as soon as possible to detect the damage of the underground roadway and the location of the trapped people.The golden rescue time is of great significance to reduce the secondary injury of personnel.Therefore,it is of great significance to study the autonomous exploration system of coal mine rescue robots.The conventional robot autonomous exploration system is designed for ideal environments like the field and campus,whereas the coal mine catastrophe environment is complex,such as the underground roadway environment texture is weak,the road surface has bumpy road conditions,many self-repeating scenes,and the activity space is narrow,etc.,which poses challenges to the robot autonomous exploration system.This thesis investigates the autonomous exploration system of rescue robots in coal mines,the main work is outlined as follows:(1)Aiming at the problems of weak roadway texture,many road bumps,many self-repeating scenes,and long roadways with different narrowness in coal mine catastrophic environment,a lidar SLAM algorithm based on GICP registration algorithm is proposed.Firstly,a point cloud adaptive filtering method based “width” is proposed to maintain a relatively stable number of points to be matched;Then,based on the GICP registration algorithm,scan-to-scan registration and scan-to-map are performed on the current frame point cloud to obtain the front-end odometry;Finally,the loop closure method is improved by the fusion of key frame pose and Scan Context descriptor which can improve the accuracy of loop detection,and combines the backend factor graph optimization algorithm to obtain a globally consistent robot pose estimation.(2)Aiming at the problem that the limited activity space of underground environment leads to the limitation of robot activities,a graph-based autonomous exploration algorithm for rescue robots is proposed.Firstly,in the local autonomous exploration method,viewpoints are sampled around the robot and a local graph is constructed according to collision constraints and distance constraints,and the path in the local graph is extracted by the heuristic A* algorithm,which improves the path extraction efficiency in local autonomous exploration;Then,the path obtained by the local autonomous exploration method is improved using the path boosting algorithm.Finally,a global autonomous exploration framework is proposed,which maintains the growth of the global graph during the robot exploration process,which can guide the robot to the unknown area combined with the global frontiers.(3)A complete autonomous exploration system for mobile robots is established and the experimental verification is conducted.Firstly,the autonomous exploration system is systematically implemented.Then,the merits of the proposed lidar SLAM algorithm are evaluated in comparison to those of the LEGO-LOAM and FAST-LIO algorithms in the nebula odometry dataset;Finally,the lidar SLAM algorithm and the autonomous exploration algorithm proposed in this thesis are combined and deployed in the mobile robot.Finally,the superiority of the autonomous exploration algorithm proposed in this thesis is reflected compared with the NBVP autonomous exploration algorithm.