Research On Planning Method For Drop Point Of Quadruped Robot Base On Map Reconstruction

Quadruped robots are flexible in movement and can travel through complex terrain to reach difficult places.They also can complete some special tasks,such as disaster relief,mine clearance and so on.At present,the research on autonomy of mobile robots has been very popular.For quadruped robot to have autonomous ability,it is necessary to build accurate map with rich information based on environment sensing sensor,which can plan the path to the target in real-time online and select the appropriate landing point.Based on the above tasks,this paper mainly completed the following tasks:(1)The point cloud data are obtained after the environmental information collected by the lidar,and the obstacle grid is identified by the point cloud data,and a series of terrain elements are obtained,including terrain curvature,terrain normal vector and so on.(2)The path planning algorithm is improved based on dynamic window method.The improved algorithm is introduced from three aspects: robot motion model,path enumeration method and path evaluation method.Compared with the dynamic window method,the enumeration of the improved algorithm is more intuitive,and the enumeration can be adjusted actively to coordinate the efficiency and accuracy of the algorithm.The evaluation index mainly includes the presence of obstacles,the degree of fit with the global planning path(to avoid falling into local minimum)and the distance from the target.(3)foot drop planning based on walk gait of quadruped robot.First,the stability of the robot during walking is ensured by designing the walking method.Then,the indexes for evaluating the landing points are proposed,including terrain curvature(degree of concavity and convexity),terrain gradient,robot stride(which determines the speed of the robot),and the consistency between the robot direction and the local planning path.