Study On Motion Planning And Control Of Automated Vehicles At Crossroads Based On Model Predictive Control

Vehicle networking technology and Automatic driving system has become a hot spot in the field of science and them play an extremely important role in Smart travel.Under the environment of the Internet of vehicles,the autonomous driving car can understand the surrounding traffic environment quickly such as people,cars and roads.And the autonomous driving car can also tell other its intention.Relatively,Autonomous driving uses its own perception system,motion planning and motion control system to understand of the surrounding environment.This way will cost more computing time and have potential risks.Therefore,this paper divides the dangerous scenes of intersections based on the premise of Internet of Vehicles,and aims at how to reduce the heavy computation burden ofcurrent MPC-based motion planning and control framework and algorithm,and how to deal with the vehicledynamics constraints duringmotion planning process.The main contents are as follows:(1)Established vehicle dynamics model A vehicle bicycle model.Based on reasonable assumptions use the linear tire model to make the nonlinear tire model approximately linearized,which can reduce the computation burden of model predictive control.And a bicycle dynamics model suitable for the motion planning and control problem of autonomous vehicle is established(2)The principles of two model predictions,nonlinear model predictive control(NMPC)and linear time-varying model predictive control(LTV-MPC),are studied.The prediction model of LTV-MPC algorithm and the linearized transformation equation of nonlinear model predictive control model is deduced in detail.Studiing the multi-constrained LTV-MPC optimization problem and transforming the problem into quadratic programming problem(3)A new framework and algorithm for motion planning and control of autonomous vehicles are proposed for the intersection scene.The time difference and threshold of conflicting vehicles from different driving directions arriving at the intersection conflict area are added to the control constraints of autonomous vehicles,and for the arrival of the vehicle,a method based on longitudinal safety priority is adopted to ensure the safe passage of vehicles.In terms of vehicle motion control,based on multi-constrained linear time-varying model predictive control and feedforward control methods,vehicle dynamics models and linear tire models,motion planning and motion control algorithms based on vertical and horizontal coupling are proposed.(4)Dangerous scenes at intersections were modeled based on the Full array from the number of traffic participants and the way of traffic(straight or turn).Simulate and verify all dangerous scenarios.The results show that the proposed autonomous vehicle motion planning and control framework and algorithm have good performance in planning and control.This method can be applied not only to low-speed signalless intersection scenarios,but also to large-scale high-speed signalized intersection scenarios.