Obstacle Avoidance Control Of Autonomous Vehicle Based On Moving Horizon Optimization

The car parc keeps rising nowadays with the rapid development of automobile in-dustry. Meanwhile, a series of problems have emerged such as increasing trafc pressure,road congestion, and frequent trafc accidents. To deal with this situation, intelligenttransportation system has come into being. As an emerging technology of ITS, the au-tonomous vehicle has been the research hotpot both in domestic and at abroad. Andif autonomous driving can be applied to vehicles, the efciency of trafc system will begreatly improved, and the trafc accidents will be reduced or even eliminated, that is, theresearch of autonomous vehicles is of great significance. Since the trafc environment isconstantly changing, obstacles or pedestrians will be unavoidably encountered by the au-tonomous vehicles. Therefore, how to take approaches to avoid from obstacles efectivelyis the main research of this paper.The vehicle state and the driving environment change constantly. At the same time,a variety of inherent dynamics constraints and physical constraints exit in the vehiclesystem. In addition, some other various constraints of trafc rules is necessary to en-sure the safety and stability. All these make obstacle avoidance of autonomous vehiclebecome a constrained multi-objective optimization problem. To solve this problem, thispaper presents a research of obstacle avoidance for autonomous vehicle based on movingoptimization.Vehicle driving strategy for autonomous obstacle avoidance is proposed based onthe definition of safe driving distance. Then the vehicle obstacle avoidance problemis decomposed into a series of decision-making problems of basic sub-operations, suchas the left lane change, cruise, and right lane change. The reference trajectory whichcontains the states of host vehicle and controlled variable to be optimized is obtainedbased on the parameterized trajectory idea considering kinematic equations of obstaclevehicle, dynamics equations of the host vehicle. To make the predicted trajectory ofthe host vehicle track this reference trajectory, the model predictive control algorithmis designed. With this algorithm, the cooperative control for trajectory planning andtracking of autonomous vehicle is implemented in the obstacle avoidance problem. Asthe vehicle in the lane change process often causes overshoot and leads to accidents, the constraint model predictive control algorithm containing safety factors is designed in orderto reduce the vehicle overshoot during lane change operation so as to ensure the stabilityand safety of the vehicle. The proposed algorithm is validated using the high-fidelityvehicle dynamics simulation model, through a variety of simulation conditions with thechanges of the host vehicle longitudinal velocity and road adhesion coefcient and soon. With the united autonomous driving strategy of left lane changing, overtaking lanecruising and right lane changing, the designed obstacle avoidance control algorithm isverified to efectively control the vehicle to avoid obstacles, and to ensure the stabilityand security, satisfying the requirements of autonomous vehicle motion control.This article provides a new idea for autonomous vehicle obstacle avoidance control,but there are still some work needing to be further improved. This paper only considersthe lateral motion control, longitudinal velocity planning and control will be the nextcontent to study.