Research On Terrain Adaptive Navigation Technology Based On Hierarchical Cost Map

In recent years,with the improvement of sensor application technology and in-depth research of related algorithms,the sensing positioning and path planning ability of mobile robots is enhanced,and the autonomous navigation function of robots is more perfect and mature,and has been widely used in daily life and scientific research work.However,in some common and specific terrain environments,conventional autonomous navigation system do not seem to make more reasonable action plans.The main research content of this topic includes the design of detection scheme,construction of cost map,construction and testing of navigation system for terrain with slope characteristics and multi-pavement environment.1.When the conventional laser navigation system is faced with terrain with slope features such as barrier-free passage and stairs,the obstacle layer of the cost map will misjudge it as a no-pass area,resulting in that the robot cannot complete the humanized patrol work when the chassis is allowed to pass.To solve this problem,a navigation method based on hierarchical cost map is proposed and implemented.Firstly,multi-line laser point cloud information is used to detect slopes,and a dynamic cost map containing slope costs is generated.Based on the cost map,the path planner realizes uphill and downhill movements.After that,the update system of map and reference coordinate system is designed to realize 3D navigation in multi-plane environment.2.When surface types such as grass,cobblestones,dirt,etc.are in common view of navigation,the robot should avoid trampling on grass or driving on pavement with a worn chassis.However,the conventional laser navigation system is unable to identify the road surface type,resulting in the robot blindly driving on the road against the will of human.To solve this problem,the visual instance segmentation algorithm is used to identify the pavement,set the pavement type pass grade,map the segmentation results of the pavement to be avoided into the reference coordinate system of the actual robot,and build the cost map layer of pavement priority.The cost map allows robots to plan routes in environments with better road conditions.Finally,corresponding navigation strategies are proposed for various special road conditions to improve the robustness of the system.3.In a variety of simulation and practical scenarios with slope characteristics and pavement types,the navigation system proposed in this paper is tested experimentally.The success rate of slope detection and the passing rate of navigation are tested experimentally in the terrain environment with slope features.In the environment with multiple pavement types,the success rate of pavement identification and the feasibility of navigation strategy were tested.At the same time,the time of the related algorithms of the above two systems is tested.Both of them meet the updating frequency of the cost map and ensure the real-time navigation.The experimental results show that the navigation method proposed in this paper solves the problem of path planning for slope terrain and makes the robot’s navigation actions more reasonable in the environment of multiple pavement types.