Research On Lane Keeping Control Of Intelligent Vehicle Based On Model Predictive Control

In recent years,with the rapid development of technology,intelligent vehicles have gradually become a new type of transportation tool that replaces traditional cars with unique advantages,and lane keeping systems(LKAS)have gradually been widely used.Research on lane maintenance is often limited to a single operating condition,only achieving autonomous obstacle avoidance control through a single longitudinal or transverse control,which has the disadvantage of insufficient rationality in actual operating conditions.Therefore,this article focuses on the control strategy of intelligent vehicles when encountering obstacles during lane keeping,and conducts decision-making and strategy analysis for different operating conditions.The aim is to design a lane keeping controller with the functions of accommodating pedestrians,following vehicles,and changing lanes to avoid obstacles.Accurate detection of lane lines is a prerequisite for lane maintenance.To achieve real-time lane line detection,this article uses the Nano Det framework to recognize lane lines,process lane information in the vehicle coordinate system,and establish a cubic polynomial lane centerline.The camera’s image is converted into real-time lane centers.Establish a Frenet coordinate system to describe the lateral motion of vehicles relative to the center of the lane,and plan the path of vehicles in the Frenet coordinate system.In order to make the proposed strategy have active safety obstacle avoidance ability and better rationality,this article conducts a deep analysis of pedestrian characteristics and yielding behavior,in order to design a strategy for intelligent vehicles to yield pedestrians;Design the optimal following distance based on driving experience and laws and regulations;Design the optimal lateral obstacle avoidance control curve through curvature and collision constraints.Calculate the target position in the Frenet coordinate system based on the expected strategy,and fit the expected trajectory through a fifth degree polynomial to obtain a smooth and continuous position,speed,and acceleration curve that meets the driving strategy.Design a lane keeping controller based on model predictive control,establish an objective function using the quantity to be optimized,and use the Fmincon function to transform the trajectory tracking problem into solving a multiobjective optimization problem with constraints.Apply the optimal control quantity in the predicted time domain to the system,achieving lane keeping control with active safety obstacle avoidance capability under rolling optimization.Finally,the simulation scenario was constructed and a joint simulation experiment was conducted in Prescan,Python,and Matlab&Simulink.The experimental results showed that the proposed lane keeping controller had good safety and rationality,and could achieve active safety obstacle avoidance during lane keeping control.