Implementation Of Path Planning And Obstacle Avoidance Algorithms Of Automatic Parking Systems

With the steady progress of economic construction,national car ownership is increasing year by year.Due to the high threshold of parking technology and insufficient construction of urban parking lots,"parking difficulty" has become an arduous problem that needs to be solved in the urbanization process.Autonomous parking technology can realize the automatic parking and one-touch summoning of intelligent vehicles under the condition of the unmanned takeover,which can effectively alleviate traffic congestion and parking difficulties.Therefore,this paper takes the autonomous parking system planning algorithm as the research object and divides the autonomous parking process into the cruise phase and the park phase,and studies the global planning of the cruise phase,the local planning of the cruise phase,the parking trajectory planning of the intelligent vehicle in the parking environment.The main research contents are as follows:(1)Autonomous parking system architecture design and model analysis.Determine the overall architecture of the modular autonomous parking system,and build the planning algorithm framework for the cruise phase and parking phase respectively.Complete the kinematic analysis of the simplified model of the vehicle based on the Ackermann steering principle.Compare the typical environment map and propose a layered map containing a static layer,dynamic obstacle layer,expansion layer,and centerline layer based on the grid map.A parking simulation scenario is built using the ROS operating system to verify the algorithm’s performance.(2)Global planning implementation for the cruise phase of the autonomous parking system.The Manhattan distance is used as the heuristic node cost,and the parent and child node orientation deviation term is introduced into the actual cost of nodes to complete the traditional A* algorithm improvement.The improved A*algorithm can significantly reduce the number of global initial path turns in the case of similar lengths.The key points with large initial path turning angles are extracted,and the minimum turning radius arc fitting is performed at the turning points.The global path smoothing process is completed by using cubic B-splines,and the ROS simulation experiments show that the global planning algorithm can quickly plan smooth paths with continuous heading and curvature in compliance with traffic rules.(3)Local obstacle avoidance implementation for the cruise phase of the autonomous parking system.Optimize the state sampling process of traditional OpenPlanner obstacle avoidance algorithm by double-layer local path cluster design and segmented linear variation of longitudinal sampling distance of path clusters with vehicle speed,and the curvature cost is introduced into the evaluation function in addition to the centerline cost,transition cost,and obstacle cost to realize the improvement of the path screening process.The local planning takes the global path as the reference line and uses the improved OpenPlanner obstacle avoidance algorithm to screen the optimal local path,and the ROS simulation experiments verify the feasibility of the algorithm.(4)Hierarchical planning framework for the parking phase of the autonomous parking system is built.The parking planning is divided into two parts: path generation and trajectory planning,and the adaptive sampling step strategy is introduced into the successor node expansion process to achieve the improvement of the traditional Hybrid A* algorithm,which can significantly reduce the parking path-solving time.The parking trajectory planning is modeled as an optimal control problem with the objective function including time and control input terms and constraints including vehicle kinematic constraints,state and control volume constraints,edge value constraints,and collision constraints.The improved Hybrid A* algorithm parking path is used as the initial value,and the parking trajectory solution is completed with the help of nonlinear optimization libraries such as IPOPT.The simulation results have verified the effectiveness and applicability of the hierarchical parking planning algorithm under different types of garages.The thesis has 60 figures,9 tables and 83 references.