Mission Planning And Trajectory Generation For Intelligent Vehicles Based On Lane-Level Map

The mission planning and trajectory generation based on lane-level map for intelligent vehicle is to plan the route and the trajectory without traffic flow disturbance based on the driving task and the geographic information.It belongs to the global planning of the intelligent vehicle and is the key technology for achieving the vehicle intelligence.In this paper,the lane-level map for lane-level mission planning and trajectory generation is studied,and a three-layer hierarchical architecture that is compatible with traditional road network is designed.The intermediate layer is established for achieveing the connection between the road layer and lane layer.The geometric representation method of the road network based on the lane-level control points is studied,and the generation strategy of control points under the controllable threshold is designed by CHS spline curve.The storage strategy of attribute data and geometric data for the hierarchical network model is introduced.Based on this road network model,a regional map is completed,and the information extraction strategy of mission planning and trajectory generation is given.Based on the hierarchical road network model,the lane-level cost model for mission planning is established including the lane-changing cost and turning cost in the intersections.The methods of hierarchical mission planning and direct mission planning aimed at long distance and short distance missions respectively are desigend.In the longdistance scenario,the road-level planning considering the intersection cost and dynamic traffic information is carried out at first,and next the topological node group corresponding to the lane level is generated by using the topological relation of the road network based on the road-level planning result,and then an algorithm for obtaining lanelevel mission planning is acheived.For the short distance mission,the direct lane-level mission planning method is designed.Taking into account the computational complexity of the lane level network,the topology optimization strategy is studied and the cost is updated accordingly,and next in the post-optimization network,the whole road network search is designed,and then the lane-level planning strategy is executed based on the A* algorithm.The effectiveness of the proposed methods are verified by the path finding results in the lane-level simulation network and the real regional map,and further the efficiency of these two methods are compared.According to the results of mission planning,the trajectory generation method from mission planning to traffic-free reference path is studied.The vehicle trajectory is divided into two parts: road and intersection.Under the premise of ensuring the first-order continuity of the curvature,the fifth-order polynomial trajectory model considering the optimal index of vehicle dynamics and kinematics is established.The strategy of trajectory generation inside roads and intersections based on the vehicle driving behavior is designed,and taking into account the efficiency of computing,the trajectory generation strategy for long distance mission is given.Based on the above research,lane-level trajectory with global continuity based on the lane-level mission is performed in the simulation road network and the real road network and further validaty of the trajectory generated is analyzed in Carsim.