Research On Path Planning Of Intelligent Logistics Vehicle In Drivable Area

Intelligent logistics vehicles have the characteristics of scene and low speed,and are the focus of exploring the development of autonomous driving today.As a research platform of automatic driving technology,intelligent logistics vehicle has both practical value and research value.Planning module is an important part of intelligent logistics vehicle,which is responsible for planning a path from the starting point to the target point in the working range.Unknown obstacles often appear in the working range,therefore,how to balance the requirements of optimization and real-time and find a safe and feasible path is a great challenge for the intelligent logistics vehicle planning module.In this paper,aiming at the campus express delivery scenario,the path planning algorithm of intelligent logistics vehicle is studied.The goal of path planning algorithm in this paper is to plan a feasible path without collision risk in unknown dynamic environment and guide the intelligent logistics vehicle to to the target point safely.First of all,based on the requirements of algorithm research,this paper completes the preparation of the upstream and downstream parts of the planning module.Combined with the map theory,the global grid map of a campus dormitory area is constructed as the input environment of the path planning algorithm;Then,the vehicle kinematics model is established,and the tracking controller is constructed by combining with the Model Predictive Control algorithm,which can evaluate the trackability of the output path of the planning module and verify whether the path meets the driving requirements of the real vehicle.On this basis,this paper studies the A* algorithm applied to the global path planning module and the Rapid-Exploiting Random Tree(RRT)algorithm applied to the local path planning module.According to their shortcomings in the application scenario,the optimization scheme is proposed.For A* algorithm,by increasing the direction of node expansion,optimizing the cost of nodes in real time,and establishing a two-way search process,an improved A* algorithm is proposed to improve the path topology of the original algorithm and increase the probability of path optimality.For RRT algorithm,by designing linear bias coefficient,adding corner constraint and modifying the search tree in real time,an improved RRT algorithm is proposed to improve the efficiency and path feasibility of the RRT algorithm.Based on the above two methods,this paper proposes an improved hierarchical path planning algorithm.Firstly,the improved A* algorithm is used to complete the global optimal path planning.Then,combined with the local target selection strategy and obstacle avoidance rules designed in this paper,the improved RRT algorithm is used to complete the path replanning task in the local dynamic environment.At last,the trajectory optimization is completed by using B-sample-spline curve and a final path meeting the driving requirements of real vehicles is output.To verify the effectiveness of the algorithm,the simulation experiment of hierarchical path planning algorithm is completed by combining the grid map of campus dormitory area and MPC controller.Through the analysis of path tracking deviation,local obstacle avoidance and time-consuming data,the effectiveness and feasibility of the hierarchical path planning algorithm in this paper are verified.