Research On Expressway Lane Changing Model For Autonomous Driving Vehicles

With the rapid development of autonomous driving technology,the further improvement of vehicle traffic efficiency has also created a greater space for exploration.Traditional micro-traffic models have problems such as ignoring vehicle delays,simplifying the lane-changing process,and not being able to guarantee no collision when the vehicle in front stops suddenly.It is difficult to provide effective theoretical support for the driving behavior of self-driving vehicles using new hardware and technologies.Expressway is currently the most important vehicle driving scenario,and car-following and lane-changing are the most basic and common driving behaviors of vehicles,and the lane-changing operation is based on the car-following logic of the vehicle.Aiming at the above problems and scenarios,this thesis proposes a lane-changing model for autonomous vehicles based on the MOBIL lane-changing decision model.First of all,based on the existing vehicle control logic and distance strategy,a carfollowing model considering time delay and no collision is established based on the optimal model.The improved model realizes the balance between vehicle distance and speed.Model local stability and queue stability.The model is simulated in three typical scenarios of car-following,emergency stop of the vehicle in front and stoppage,and compared with the Gipps model,which is also a collision-free car-following model,the results show that the SLC model proposed in this thesis has better speed and stability and comfort,it is more suitable for subsequent lane changing scenarios.Secondly,to balance the amount of simulation calculation and the need for authenticity,select the vehicle kinematics model,and simulate the lane-changing process of the vehicle based on the preview target lane method,and describe the whole process of lane-changing execution more completely.Compared with common lane-changing path planning methods such as Bezier curves and sinusoidal acceleration curves,the lanechanging execution process in this thesis has higher implementability,compatibility and lower computational cost under the premise of meeting the lane-changing requirements such as curvature.It is more suitable for the completion of simulation experiments.Thirdly,combined with the improved car-following model,the existing MOBIL lane-changing decision-making model is improved,and the expected acceleration and actual acceleration in the acceleration benefits are distinguished,and the benefit weight distribution of the vehicle behind this lane and the vehicle behind the target lane is optimized..At the same time,considering driving safety and avoiding invalid lane changes,additional lane changing requirements are added: when the vehicles around the ego vehicle are changing lanes,the ego car is not allowed to change lanes.Then,a relatively standard highway lane-changing scene and test indicators and parameters were formulated,and Matlab was used to analyze the human driver’s carfollowing model,the Gipps car-following model,the SLC car-following model proposed in this thesis,and the MOBIL improved lane-changing model.The results show that the lane-changing model proposed in this thesis has the highest passenger comfort under the premise of realizing no collision,and the loss zone at a speed of about 20% of the traffic flow increases.There is more than 200% increase in mileage for a single lane change.The density-flow simulation results indicate that the lane-changing strategy itself has room to play only at low vehicle densities.Finally,summarize the research content of this thesis,put forward the insufficiency of the research content and the prospect of the future direction of improvement.