Trajectory Tracking And Human-machine Cooperative Control Of Intelligent Vehicles

Under the L3-L4 level of automatic driving,this paper focuses on the control layer of automatic driving technology.Based on the self-driving intelligent vehicle,the trajectory tracking and human-machine cooperative control are mainly studied considering the uncertainties in the vehicle-road model of the intelligent vehicle.And a reasonable subjective and objective evaluation system is established for this research.The main research work of this paper is as follows:(1)To compensate the state deviation caused by the changing model parameters and adhesion coefficient and realize the real-time performance of the controller during the trajectory tracking of intelligent vehicles,a fast trajectory tracking co-driving control strategy considering predictive state deviation is proposed.Firstly,with the uncertain influence between the road surface and the vehicle-road model,the Dugoff tire model is introduced to obtain the relationship between vehicle-road model and road adhesion coefficient,which can enhance the adaptability of the vehicle-road model on the low adhesion road surface.Secondly,aiming at the problem of state deviation mainly caused by time-varying model parameters and the varying adhesion coefficient,the predictive error feedback item of historical state is introduced to the predictive model according to the idea of error feedback control,which can compensate the actual state and enhance the robustness of the vehicle trajectory tracking.Thirdly,aiming at the computational complexity of the controller caused by the online optimization of predictive control,the introduced Dugoff model and error feedback item,the arithmetic-segmented aggregation strategy is introduced at different operating points to reduce the computational dimensions of the optimized control outputs,so as to reduce the solving complexity of the predictive controller and improve the real-time performance of the system.And the stability analysis of the controller with the arithmetic-segmented aggregation strategy is given in theory.(2)Aiming at the human-machine cooperative control strategy and the flexible allocation of driving weight under human-machine cooperative control framework,a human-machine cooperative control strategy for intelligent vehicles considering the driver load and trajectory tracking accuracy is proposed.Firstly,the cooperative control model is built by the trajectory tracking model and the allocated control output considering the driver load and trajectory tracking accuracy.Based on the receding horizon control theory,the predictive equation and optimization problem is derived to get the co-driving controller based on the cooperative control model.The co-driving controller can response to the driver’s output in real time and a parallel indirect cooperative control structure is formed with the driver model.Aiming at the flexible allocation of driving weight between the driver and co-driving controller,the basic linear cooperative control structure is improved at first.And based on fuzzy inference theory,the fuzzy inference systems are respectively designed based on the drivers’ output-lateral deviation and the lateral velocity-lateral deviation to achieve the flexible allocation of driving weight coefficients.This allocation mechanism can improve the accuracy of vehicle trajectory tracking,and reduce the driving load of drivers and the conflict of human-machine cooperative driving.(3)Aiming at the problems of time-varying parameters of the system and measurable error interference of the driver input in human-machine cooperative control,a robust cooperative control strategy considering the measurable error interference of the driver input is proposed.Firstly,the LPV vehicle-road model is built considering the time-varying parameters and uncertain interference.Secondly,the weight transfer functions affecting trajectory tracking accuracy,outputs of the driver and co-driving controller and external interference are designed respectively.Combined with the LPV vehicle-road model,the generalized control system is constructed.Then the design and solution of LPV/H_∞ codriving controller are derived.Thirdly,aiming at the weight allocation problem of humanmachine cooperative driving,on the one hand,the LPV model is built considering the timevarying characteristics of the driving weight coefficient;on the other hand,the fuzzy driving weight allocation mechanism is designed based on longitudinal velocity,lateral velocity,lateral deviation and the driver output,so as to achieve a reasonable weight allocation of human-machine driving and reduce human-machine control conflict.And a generalized driving weight allocation structure is formed,which provides a generalized scheme for human-machine cooperative driving weight allocation(4)Aiming at the testing and evaluation of human-machine cooperative control under on-road experimental conditions,the on-road experimental platform and evaluation criterion of human-machine cooperative control are constructed,and a double-closed-loop human-machine cooperative control strategy is proposed.Firstly,aiming at the control interface of human-machine co-driving,the steering system is modified to steer-by-wire system generating three driving patterns: the driver drives alone,the co-driving controller drives alone and human-machine co-driving.Secondly,aiming at the problem of humanmachine co-driving control under real vehicle conditions,the longitudinal ACC control is designed based on the PD theory.The lateral control divides into two modes.The outer loop angle control is designed based on the fuzzy PID theory,and the inner loop torque control is designed based on the active disturbance rejection theory.The longitudinal control can achieve stable cruising at a constant speed.And the lateral control can achieve the humanmachine cooperative control experiment under the angle(single closed loop)and torque(double closed loop)modes.Thirdly,for the subjective and objective evaluation and analysis of human-machine co-driving,the subjective evaluation takes four indicators of the driving safety,driving accuracy,driving comfort and overall driving experience into account in the human-machine cooperative driving process.And the objective evaluation takes three indicators of the trajectory tracking performance,drivers’ driving load and human-machine conflict into account.This research provides a feasible scheme for the test and evaluation of human-machine co-driving under real vehicle conditions.