Hierarchical Planning And Control Algorithms Of Personalized Automatic Lane Change System

Being an essential intermediate product of fully automation,commercialized ADAS(Advanced Driver Assistance Systems)such as ACC(Adaptive Cruise Control)and LKA(Lane Keeping Assistance)have validated their effectiveness to make highway driving safer and easier.The urgency of extending functionalities of current ADAS,and eventually progressing to highly automated highway driving necessitates the design of automatic lane change system.For trajectory planning,parameters of the traditional planner are fixed,thus unable to adapt to driver's lane change characteristics.For longitudinal and lateral control,nonlinearity and coupling effect have always been the difficulty.Supported by National Nature Science Foundation Project No.51575223(Research on control mechanisms and evaluation methods of the steer by wire system based on driver characteristics).This paper mainly focuses on trajectory planning and longitudinal and lateral control techniques of an automatic lane change system in highway driving scenario.A hierarchical trajectory planning algorithm,which combines parametric function and learning-based technique,is proposed to adapt to driver characteristics in different traffic situations.Both longitudinal controller and lateral controller are designed hierarchically to handle model nonlinearity.The upper controller computes the optimal solution based on linear model,while the lower layer extracts nonlinearity from the model and transforms the optimal solution into control signals using inverse models.Since high-speed vehicles must subject to multiple contraints,both upper controllers are designed based on MPC(Model Predictive Control)theory.Specifically,the longitudinal controller realizes coordinated tracking of two front vehicles during lane change,to avoid negative effects caused by the sudden change of tracking target.The lateral controller receives longitudinal tire force signals from longitudinal controller,and computes optimal lateral force of the front tire based on a force input model.Handling stability constraints and environment constraints are also designed to ensure stability and safety during lane change maneuver.The model nonlinearity and the coupling effect are considered in the design of the inverse tire model and handling stability constraints.In particular,this paper includes:(1)Reasearch on trajectory planning algorithm which adapts to different driver characteristics.Since the “human-vehicle-road” loop is nondeterministic by nature,traditional deterministic model fails to describe it completely.Therefore,a hierarchical planning algorithm is proposed to adapt to different driver characteristics.The upper GMM(Gaussian Mixture Model)receives traffic situation vector and outputs trajectory parameters,which are received by the lower trapezoidal acceleration model and outputs trajectory.Finally,based on the lane change data collected on the driving simulator,the model is trained and validated.(2)Treating the longitudinal control process during lane change as the coordinated tracking of two front vehicles,the functionality of ACC is extended.Hieararchical strategy is employed to handle nonlinearity.Based on the linear dynamics model between vehicles,the upper MPC controller computes optimal acceleration,while the lower layer transforms the optimal acceleration into throttle opening and braking pressure using invers longitudinal dynamics model.Fuzzy logic is employed to distribute weights on the front vehicles,and the weighted virtual vehicle is tracked.The tracking performance of the controller is validated based on the joint simulation of MATLAB/Simulink and Car Sim.(3)Research on lateral controller which considers tire nonlinearity and coupling effect,as well as guarantees stability and safety.In highway driving,control performance is affected by tire nonlinearity and coupling effect.Same with the longitudinal controller,a hierarchical algorithm is proposed.The linear force input model is employed as the prediction model of the upper MPC controller.Handling stability constraints and environment constraints are established to guarantee stability and safety.The lower inverse tire model transforms optimal tire force computed by the upper layer into front wheel angle.Coupling effect induced by the longitudinal tire force is accounted for in the inverse tire model and handling stability constraint.(4)Based on the driving simulator,typical lane change scenario and front vehicle emergency braking scenario are designed,and employed to validate the proposed planning and control algorithm.The results show that the automatic lane change system is able to incorporate comfort,safety and driver characteristics during lane change maneuver in different scenarios.