Research On The Car-Following Model Of Autonomous Vehicles Under Ice And Snow Weather

With the rapid development of artificial intelligence,the proportion of vehicles on the road with intelligent car-following capabilities had gradually increased,but traffic flow is significantly different due to the complex and diverse driving environment.In snow and ice weather,the road surface is slippery compared to normal weather,and snowflake interference reduces the sensor performance of autonomous vehicles.The vehicle struggles to maintain continuous high-precision perception of all elements,time,and space,resulting in unstable carfollowing and the risk of collision.Theoretical and practical implications of studying the carfollowing characteristics of autonomous vehicles in ice and snow weather include enriching existing traffic flow theories and improving the safety of autonomous vehicles in winter.This paper investigates the stability and safety of car-following in ice and snow weather using autonomous vehicles as the research object.The following are the main works and outcomes:(1)Analyze the overall impact of snow and ice weather on the following behavior of autonomous vehicle,then use the road friction and perception error as the car-following action coefficient and load it into the improved IDM model.Based on the Waymo database,a comprehensive and efficient vehicle trajectory processing method is designed,including data selection,preprocessing,tracking pair extraction,noise removal,and so on,to obtain tracking segment data that conforms to the characteristics of the Time Headway of autonomous driving in ice and snow weather.In addition,the genetic algorithm was used to calibrate the model parameters.Statistical methods are employed to check the reasonableness of the model calibration results,the Wilson and the Lyapunov stability criterion is used to determine the carfollowing stability interval,proving the model’s validity.(2)Three car-following states of normal,acceleration,and deceleration are designed based on longitudinal speed and acceleration changes of autonomous vehicles in snow and ice weather,and three car-following scenario of road friction change,perceptual error change,both change are constructed in combination with real weather conditions.Each scenario consists of one or more car-following states.The simulation experiment was carried out using the SUMO simulation platform,and the findings of the microscopic traffic flow parameters of each scenario were compared.The longer the perception error time,the greater the degree of deviation of the system from the steady state,and the acceleration-following ratio deceleration,it was discovered that the worse the ice and snow weather,the more unstable the car-following queue,the longer the perception error time,the longer the perception error time,the greater the degree of deviation of the system from the steady state,and the acceleration car-following ratio deceleration.Acceleration car-following is more sensitive to changes in speed and acceleration,while deceleration car-following create more serious traffic flow shocks than acceleration carfollowing.(3)A safety car-following evaluation technique is provided in order to quantitatively analyze the security of car-following.The weighted normalization method is used to combine the two risk indicators of time-to-collision(TTC)and safety field strength(SFS)to create a comprehensive risk evaluation model.To assess the danger of acceleration car-following state,deceleration car-following state,and simulation scenarios.The findings suggest that accelerating and following is riskier than decelerating,and when the perception error surpasses 3s,there’s a chance of a collision;autonomous vehicles are most risky in snow and ice,followed by heavy snow,and quite safe in light snow.