Research On Path Planning Algorithm Of Mobile Robot In Complex Environment

With the development of artificial intelligence and automation technology,mobile robot applications have become increasingly wide-ranging.However,with this expansion comes an increase in the complexity of the environment,making it difficult for traditional path planning algorithms to effectively satisfy all constraints.The A* algorithm is a commonly used algorithm in path planning.This thesis addresses the shortcomings of the A* algorithm in path planning in multi-scale maps and multi-obstacle environments,and improve its planning ability.This thesis further verified our proposed improvements through simulation and experimentation.The specific research content mainly includes the following aspects:(1)To address the low efficiency of the A* algorithm,this thesis employed the Chebyshev distance as the cost distance evaluation criterion and introduced dynamic attenuation through a normal distribution.This thesis then reconstructed the cost function with an adaptive weight for the heuristic function.The directional effect around the initial node was enhanced,reducing unnecessary node calculations,therefore shortening the path search time and improving efficiency.(2)In response to the issue of redundant nodes still appearing in the path node linked list after improving the A* algorithm,this thesis designed a redundant node filtering algorithm based on the collision relationship between the lines connecting each planned node and obstacles.This algorithm filters out the critical nodes and further reduces the number of path nodes.(3)To address the issue of non-smooth path planning in the A* algorithm,this thesis designed local target points based on the critical nodes retained from the previous step.This thesis then integrated a dynamically windowed algorithm with an added safety distance factor to guide the mobile robot to segmentally conduct local path planning.This approach transforms the turning path into a motion path that samples velocity,improving both the smoothness and safety of the planned path and making it more consistent with robotic kinematics.(4)The present study employs simulation and experimental validation to evaluate the effectiveness of the EA* algorithm in planning paths through environments with multiple obstacles.Using Matlab,this thesis conducted multiple simulations in such environments,which demonstrated that our proposed EA* algorithm outperforms A* algorithm,Dijkstra algorithm,and the CA* algorithm reported in the literature,in terms of the number of search nodes and search time.This thesis further improved the algorithm by incorporating the dynamic window algorithm,resulting in the EAD algorithm.The EAD algorithm outperforms the previous EA* algorithm in terms of path smoothness,while also showing improved capability in avoiding special traps and improving safety compared to the dynamic window algorithm.To test the performance of the proposed algorithm in realworld scenarios,this thesis integrated the ROS system to build a mobile robot platform.RVIZ visualization was employed to conduct experiments in critical areas of the map.Results demonstrated that the proposed EAD algorithm is effective in guiding the robot to navigate successfully to the designated locations.These findings provide further support for the practical validity of the proposed algorithm.