Research And Design Of Vehicle Trajectory Planning Method Based On Human Vehicle Motion Situation Deduction

An important factor of road traffic congestion is that the short-distance active driving is too strong during daily vehicle driving,and efficient trajectory planning is not achieved.The driver only makes comprehensive decisions according to the relative position,motion state and driving habits of all detectable road participation elements within the visual range,which cannot achieve beyond visual range perception.At the same time,drivers cannot efficiently predict the movement trend of other road participation elements in local areas,resulting in the coincidence of drivers’spatiotemporal trajectories and uneven traffic flow distribution.At the same time,with the development of edge computing technology and vehicle networking technology,it is possible for agent vehicles to generate trajectory planning suggestions under the global traffic situation of intersections with the help of edge computing nodes.Therefore,this paper proposes a vehicle trajectory planning method based on the deduction of human vehicle motion situation.The prediction of human vehicle motion situation is very important for agent vehicle trajectory planning.The prediction accuracy will affect the results of vehicle trajectory planning.However,in the existing relevant studies,the long-term time-dependent relationship in the historical trajectory sequence is not handled well,and the impact of adjacent agents on it,that is,the impact of social spatial interaction,is not considered in place.For this reason,this paper proposes a social relationship potential grid network WR-SRPG model based on the forward rhythm,which considers the forward rhythm and personality characteristics of agents,and can effectively pay attention to different representation information from different positions of the trajectory.At the same time,it has the ability of remote self-attention,and can effectively capture the social dynamic information exchanged between agents.The experimental results show that the proposed model improves the two indexes of ADE and FDE by 37.8%and 38.3%respectively compared with Social GAN in ETHУucy data set.In order to solve various problems in the traditional trajectory planning algorithm,make full use of the global traffic information other than vehicles obtained by vehicle road coordination,and break through the memory less and social interaction of the traditional algorithm.In this paper,a value iteration algorithm framework of expansion potential field based on the deduction of man vehicle motion situation is proposed to solve the above problems.Through the static obstacle expansion potential field function and human vehicle movement situation expansion potential field function established in this paper,the output results of WR-SRPG model are integrated into the value reward map in a potential way.Finally,the final trajectory planning suggestion is generated by using the expansion potential field value iteration algorithm.Simulation results show that this method can make the agent vehicle effectively perceive the social interaction results between people and vehicles and the whole traffic movement situation at the intersection.