Research On Topology-aware Based Car-following And Lane-changing Planning

The vehicle can obtain the location and operating conditions of other surrounding vehicles by using topology-aware technology.According to the aware information,the vehicle can reasonably control its own operating state and drive with the surrounding vehicles to form a driving queue.However,the previous car-following and lane-changing models mostly used a single vehicle as the modeling object,and did not consider the state of the surrounding vehicles.Even if the multi-vehicle network is used as the modeling object,due to the large scale,node mobility,weak scalability and other conditions,the evolution of the traffic flow will be difficult to describe.To this end,according to the problems that the vehicle may encounter in the process of different road conditions,divides the problem of car-following and lane-changing planning into three sub-problems,which will be comprehensively integrated on the basis of in-depth research,and design optimized models and solutions to achieve collaboration between vehicles.The main research contents and conclusions include:(1)A vehicle queue is formed based on topology-aware technology,and two teaming schemes are proposed: alliance chain teaming strategy and team size control strategy.The former mainly considers the influence on the team formation caused by different initial states,communication delays and communication topologies.Vehicles and roadside units complete the team formation by building an alliance chain.The latter assumes that the vehicle freely chooses its constant driving speed.Calculates the connection probability between vehicles to obtain the fleet capacity range through vehicle load estimation and effective capacity model.Finally,theoretical analysis and simulation experiments prove that the proposed method can effectively form a queue,and at the same time can effectively control the vehicle network load.(2)A vehicle lane changing strategy is designed,which is divided into three parts.The first part analyzes the lane-changing decision of the vehicle when the car is driving on the road,and uses Logit model to calculate the probability of vehicle lane change;the second part establishes the obstacle repulsion domain to determine the initial orientation angle of the vehicle,and further predict the trajectory range;the third part combines the above two kinds of evidence to determine the initial resultant direction,and predicts the trajectory clusters of the lane-changing vehicles under normal driving behavior.The simulation results show that the algorithm can evaluate the lane change trend in different traffic scenarios,and plan a smooth motion trajectory cluster,and select an appropriate target path among them,so that the vehicle can drive smoothly and safely.(3)The vehicle-following strategy models and algorithm implementations are proposed.The improvement is based on Gipps model by introducing three key parameters such as vision angle,lateral offset angle,and lane-changing decision probability of the preceding vehicle,two car-following models are created by considering the lateral shift and the situation of the front car preparing lane changing.Theoretical analysis and simulation experiments prove that both improved models can reproduce the car-following behavior.Compared with the original Gipps model,the two improved models perform better in simulating acceleration changes and vehicle deceleration behavior,and make it easier to stabilize traffic flow.