Research On Emergency Collision Avoidance Motion Planning And Tracking Control Method Of Driverless Vehicle

Driverless technology can effectively reduce the traffic accident rate,save power and energy and alleviate traffic congestion.It is an important starting point and main foothold of the new round of scientific and technological revolution and industrial reform.It is also the main strategic arena for automobile power.Under the macro background that driverless engineering prototype can complete most working conditions,the ability of rapid identification and stable disposal of unusual working conditions with sudden and strong destructive force has become a key scientific and technical difficulty restricting the popularization and application of driverless technology.This project relies on the sub project "Research on vehicle vertical and horizontal decoupling robust control"(2017YFB0102603-3)of the key research and development plan of the Ministry of science and technology "decision-making and cooperative control technology of autonomous vehicles in complex driving environment",In view of the planning failure caused by the conservative design of general Collision Avoidance Trajectory Planning under emergency conditions and the insufficient tracking accuracy caused by the interactive interference of vertical and horizontal control of general tracking control method under composite input conditions,a systematic solution to collision avoidance regulation under emergency conditions is explored based on vehicle dynamics.The main research contents of this paper are as follows:(1)Explore the coupling characteristics and influence of vehicle longitudinal and transverse dynamics.Through the 3-DOF monorail vehicle model,wheel dynamics model and Uni Tire tire model,the generation mechanism of vehicle longitudinal and transverse coupling is discussed.Based on the control chart and Car Sim vehicle dynamics software,the influence of vehicle longitudinal and transverse dynamics coupling characteristics on vehicle lateral control,longitudinal control and tracking control is discussed from the perspective of theoretical analysis and data mining;Based on the data mining of vehicle tracking results,a quantitative evaluation method of vehicle vertical and horizontal coupling degree is proposed by analyzing the characterization characteristics and influence of vertical and horizontal coupling.(2)Facing the problem of improving the trajectory tracking accuracy due to the coupling of vehicle longitudinal and transverse dynamics,the inverse system decoupling method used in the past for the decoupling control of bearingless motor is applied to the trajectory tracking of driverless vehicle.Through the reversibility proof of vehicle system based on interactor algorithm,the vehicle inverse system model is established by BPNN,the vertical and horizontal decoupling of vehicle system is completed by series inverse positive system,and the vehicle system is transformed into a two-dimensional parallel and independent pseudo linear system with controllable longitudinal acceleration and lateral acceleration.In the application research of motion trajectory tracking,based on vehicle kinematics model and geometric model,the target motion trajectory is decomposed into target vehicle motion state,and the target vehicle motion state is tracked by pseudo linear system to realize the indirect tracking of target motion trajectory.In addition,in order to eliminate the cumulative error of vehicle state tracking,a trajectory correction module based on heading angle approximation is proposed to ensure that the vehicle position is controlled in the loop.(3)In order to comprehensively realize high-precision and high-efficiency tracking of target trajectory,the tracking controller scheduling strategy based on target trajectory coupling level is studied.Using the vehicle coupling state quantitative evaluation method,according to the vehicle longitudinal slip rate response,the vehicle motion state is divided into high coupling and low coupling,and the data samples are marked and the classified data set with vehicle coupling state label is established.In the classification effect through KNN,decision tree and its integrated classification algorithm,the bagged random forest model is selected as the vehicle coupling state classifier,combined with logical "or" operation to strengthen the conservation of motion trajectory coupling level classification.Finally,based on the coupling level of the predicted target trajectory,different tracking controllers are called for the coupling level of different motion trajectories,which makes comprehensive use of the advantages of high calculation efficiency of the pure tracking controller and high tracking accuracy of the coupled working conditions of the decoupled tracking controller.The effectiveness of the target trajectory coupling level scheduling strategy is verified by simulation.(4)Facing the strict requirements of collision avoidance under high-speed and short-distance emergency conditions,the vehicle trajectory planning method under emergency conditions is studied.In order to maximize the use of vehicle limit performance,a piecewise polynomial path planning method is proposed based on the idea of adaptive parking safety motion path planning,and the smoothest target path is selected according to the principle of "safe avoidance of the first path and stable pull back of the second path".In addition,using the dynamic programming algorithm with embedded vehicle stability predictor,the optimal target speed is planned according to the target path curvature constraint.In addition,aiming at the complex dynamic motion environment,the decision-making rule table of collision avoidance scheme under emergency conditions is designed,and the collision avoidance trajectory planning under different emergency conditions is realized through the integration of " decision rule +optimization planning".Through the co-simulation of Simulink and Car Sim,the proposed trajectory planning algorithm can guide the high-speed driverless vehicle not to collide with the emerging short-range obstacles.Through the integration of local path planning algorithm under general working conditions and collision avoidance trajectory planning algorithm under emergency working conditions,and the integration of trajectory planning algorithm and decoupling tracking control algorithm,the planning and control algorithm of driverless vehicle is constructed.Finally,the effectiveness of the integrated algorithm for emergency collision avoidance is simulated and verified based on the Car Sim model in the loop and the hardware in the loop based on Speedgoat mobile.The results show that the proposed integrated planning and control algorithm can effectively achieve vehicle safety collision avoidance and trajectory accessibility in case of emergency,and the time efficiency meets the hardware requirements.In addition,the collision avoidance results with or without decoupling tracking algorithm show that the proposed decoupling tracking algorithm can effectively improve the safety and stability of vehicles because it can improve the tracking accuracy in the face of emergency obstacles.