Research On Dynamic Obstacle Avoidance Algorithm Based On Lidar

Autonomous navigation is a key step towards robot intelligence and is an important factor in giving robots the ability to sense and act.Path planning as well as obscuration avoidance algorithms are essential for self-determined autonomous navigation robots to find a safe and collision-free optimal route in an unstructured dynamic space.This thesis concentrates on the diagnosis and research of LIDAR-based dynamic obstacle avoidance algorithm.By processing the 3D LiDAR point cloud data,the accurate localization of obstacles is accomplished;and on this the manual potential field method is used to conduct dynamic obstacle avoidance,while improving the traditional manual potential field method for the three shortcomings that existence.The main research contents of this thesis are as follows:(1)On the basis of completing the joint calibration of LIDAR and robot IMU,the compensation of motion distortion is performed for LIDAR.At the meantime,the multi-sector plane fitting method with the variance value introduced,the RANSAC algorithm,and the least squares method are used to construct the optimal ground plane model to efficiently remove ground points and accurately establish the environmental map.(2)In consideration of the problems of unreachable target,poor dynamic environment adaptability and local minimal value of the traditional artificial potential field method,we improve the artificial potential field method by submitting the method of adding distance factor,decomposing the repulsive force,constructing the adaptive effective rang of obstacles,and fusing with A*algorithm to construct the general repulsive force rotation evaluation function in the repulsive potential field function.The simulated experiments and the verification of the robot platform algorithm show that the enhanced dynamic obstacle avoidance algorithm can be successfully used to achieve the target location in complex environments,which has solved the problems of unreachable target,poor adaptability to dynamic environment and local minimal value of the traditional artificial potential field method.At the moment,in comparison with other algorithms,the improved algorithm plans a path with small angle of turn and reduced path search time,which allows the robot to achieve a more efficient and smooth dynamic obstacle avoidance in complex environments.