Research On Decision-making And Planning For Active Lane Change Of Autonomous Vehicles

L2+ level Advanced Driving Assistance System（ADAS）function is one of the hotspots of ADAS technology researches,and the Active Lane Change is the core component of that.The paper takes it as the research object to study the key problems such as decision-making,planning and control of Active Lane Change.Firstly,the Active Lane Change decision logic is designed based on state machine.The anthropomorphic autonomous lane change decision based on the decision link of "intention generation → feasibility judgment → generation of lane change behavior decision generation" is realized.There are three situations of lane change intention generation are proposed in the logic design.In the situation where "The speed of the vehicle ahead is lower",a judgment algorithm for the generation of lane change intention based on the idea of "changing lanes without deceleration" is proposed.Based on the idea of "trial calculation",the lane change motion planning is used to judge the feasibility of lane change.It means that if the lane change motion planning is feasible for the target space for the empty target lane,it has the feasibility of lane change.The reuse of motion planning algorithm can also be realized at the same time.In order to cooperate with the design of the above algorithm,the paper also studies the problems of traffic information preprocessing,traffic vehicle motion prediction,workshop safety time distance calculation and so on.Then based on the comparative analysis of various lane change trajectory planning methods,considering the characteristics of polynomial trajectory planning with smooth curvature and flexible planning route,the polynomial based trajectory planning method is adopted.Through the quintic polynomial,the longitudinal motion planning method of self propelled vehicle based on optimization method is designed;Considering the smoothness and comfort of longitudinal motion,acceleration and acceleration are introduced into the objective function.Based on the safe time distance model and the lateral constraint relationship between self vehicle and traffic vehicle,the time domain limit of self vehicle longitudinal motion is designed to realize the safe time distance constraint with traffic vehicle in the process of lane change.Through the seventh degree polynomial,the lateral motion planning method of self propelled vehicle based on optimization method is designed.Considering the requirements of lateral motion smoothness and comfort,acceleration and acceleration are introduced into the objective function,and the constraints on lateral displacement,velocity and other motion parameters are set.The parameters of the objective function are adjusted and optimized considering the length of lane changing time.Using the above methods,the motion planning problem of lane changing return vehicle is also studied.The horizontal and vertical planning of autonomous lane change and lane change return planning are simulated and verified.The tracking control problems of lateral motion and longitudinal motion of autonomous lane change are studied respectively.Firstly,a lateral motion tracking control strategy based on Feedforward and Feedback Control（FBC）architecture is designed,which has the following advantages: by taking the offset and heading angle at the preview point given by the camera as the feedback,the influence of vehicle yaw response lag is compensated,The feedforward steering wheel angle command is generated according to the road curvature at the preview point,which reduces the feedback error and improves the stability of tracking control.In the feedback link,the idea of weighting is used,and the influence of heading angle and offset is considered at the same time to realize the dual tracking of them.The simulation results verify the above advantages,but the lane line mutation simulation also shows that it is sensitive to the local mutation or distortion of lane line.The lateral motion control strategy based on model predictive control（MPC）is studied for the idea of "amortization" of the influence of local mutation or distortion of lane line over a distance.The constraint on the change rate of steering wheel angle is set In the MPC controller,and the requirements of minimum mean square of lateral motion deviation history and minimum mean square of control quantity history are included in the design of objective function.Therefore,the influence of local sudden change or distortion of lane line can be "amortized" in the whole prediction time domain to reduce the sudden change of steering wheel angle caused by meeting the requirements of lateral motion control.The simulation results verify the above analysis and show the advantages of MPC controller in this scenario.In the longitudinal control,a fuzzy self-tuning proportional integral（Fuzzy PI）controller is designed for speed tracking control.The autonomous lane changing trajectory generated by the trajectory planning algorithm is used to simulate the horizontal and vertical control algorithm,and the effects of the two horizontal controllers are compared.Finally,through the joint simulation of SCANe R and Simulink,the decision-making,planning and control algorithm of autonomous lane change proposed in this paper is verified.A real vehicle experimental platform is built（its ADAS controller takes Infineon’s tc297 chip as the core）.Limited by the computing power of the real vehicle platform and considering the feasibility of the algorithm,the off-line trajectory planning results are used to verify the vehicle lateral FBC control strategy.The experimental results verify the effectiveness of the algorithm.