Research On Obstacle Avoidance Planning And Motion Control Of Intelligent Vehicle Based On Adaptive MPC

In recent years,autonomous driving has become a hot topic in academia and industry,and is considered as one of the solutions to improve driving safety and road traffic efficiency.With the rapid development of automobile intelligence,automatic driving is gradually entering the public’s vision.Autonomous vehicles mainly include environment perception,behavior decision,path planning and path tracking.The most important task of autonomous vehicles is to integrate the surrounding driving environment to make correct behavior decisions,and output an optimal obstacle avoidance path from the planning layer.Finally,the bottom control module completes the lateral and longitudinal tracking task.Autonomous driving motion control is one of the key technologies of obstacle avoidance,which is the key to determine whether the obstacle avoidance task can be completed successfully.At present,as the autonomous driving scene becomes more and more complex,the corresponding requirements for motion control layer are also higher and higher.Therefore,this paper studies the lack of planning and tracking accuracy and poor stability in the high-speed obstacle avoidance scenario of autonomous driving.The specific contents include:The first part is lateral path tracking control.The path tracking controller of vehicle dynamics is designed based on model predictive control(MPC)theory.Firstly,in order to improve the adaptability of the model predictive controller under different working conditions,the standard particle swarm optimization algorithm was used to optimize the predictive horizon and control horizon parameters offline to obtain the optimal horizon data set under different vehicle speeds and different adhesion coefficients,and the adaptive fuzzy neural network was used for online optimization.Secondly,the tire side stiffness estimation algorithm and the soft constraint of side slip angle were designed to prevent the model mismatch of the model prediction controller,so as to further improve the tracking accuracy and driving stability of the MPC controller.Finally,an improved particle swarm optimization algorithm with dynamic weight and penalty function is proposed to solve the problem of poor accuracy of quadratic programming.The second part is longitudinal velocity tracking control.Using the idea of hierarchical control,the upper controller can obtain the desired longitudinal acceleration according to the actual speed and reference speed of the vehicle.The lower controller obtains specific throttle opening and brake pedal opening through throttle/brake switching logic,inverse longitudinal dynamics model and engine calibration table.The third part is the lateral and longitudinal integrated control.The speed is taken as the comprehensive control point because the lateral and longitudinal control has a strong coupling degree.The expected speed planning is carried out combined with the curvature information of the expected path,and the principle of entering the curve deceleration and exiting the curve acceleration is followed,and the safe speed of sideslip and rollover cannot be exceeded.The fourth part is obstacle avoidance path planning and tracking control.Firstly,the risk field around the vehicle is modeled by the idea of artificial potential field method.Secondly,the potential field function is added to the objective function,so that a series of collision-free points in the future are planned by using the predictive function of model predictive control,and the curve is fitted by quintic polynomial.Finally,the path information is output from the obstacle avoidance planning layer to the motion control layer,and the speed and path tracking is completed by the lateral and longitudinal integrated controller.Finally,the effectiveness and real-time performance of obstacle avoidance planning and motion control algorithm are verified by dynamic and static obstacle avoidance simulation experiments.