Research On Intelligent Car Decision Method Based On Multiagent Behavior Prediction

In today’s society,under the background of energy shortage,environmental pollution,artificial intelligence and the transformation of automobile industry,vehicle intelligence is developing rapidly.Driverless vehicle is an important part of intelligent vehicle,intelligent vehicle decision-making is one of the core technologies of driverless vehicle.The behavior prediction of people,vehicles and other intelligent units around the intelligent vehicle is the key to improve the degree of personification of decision-making.In this paper,multi-agent interaction is considered for behavior prediction,and on this basis,the intelligent vehicle decision-making method is studied.Firstly,the motion characteristics and interaction of pedestrians and vehicles in mixed traffic flow are analyzed,and the social force model of each intelligent unit is established under the joint action of target driving force,repulsive force of moving objects around and environmental boundary force,which is the basis of behavior prediction.The effectiveness of the model is verified by simulation experiments.Secondly,aiming at the scene of road collision avoidance,the intelligent vehicle active collision avoidance system is established.By analyzing the operation characteristics of the driver,the overall scheme of the system is given.Based on the longitudinal braking process,the safety distance models for vehicles and pedestrians are established respectively,and the calculation method of expected acceleration is given.Aiming at the collision avoidance decision-making,using the collision time distance and the minimum braking distance as the judgment index,two collision avoidance methods,braking deceleration and steering lane change,are designed.Then,based on the social force model of each intelligent unit,a behavior prediction method considering the interaction of pedestrians and vehicles in the mixed traffic environment is proposed,which is introduced into the decision-making process,and a collision avoidance decision model of intelligent vehicle based on finite state machine is established,which can effectively predict the trajectory change caused by the interaction of intelligent units,so as to reduce the risk of accidents.Finally,the prescan / Simulink co simulation platform is built to establish different types of mixed traffic scenes including pedestrian,vehicle and road information.The simulation results of intelligent vehicle collision avoidance decision system with and without behavior prediction are compared and analyzed to prove the effectiveness of the proposed method.