Research On Active Lane Change Motion Planning And Model Predictive Control Considering Real-time

Under the new round of technological revolution,automotive intelligence is in the stage of rapid development and accelerated industrialization.With the wide application of advanced driver assistance systems(ADAS),active lane change assistance system has become one of the current research hotspots.Based on the research of dynamic traffic scenarios,this paper considers the key issues of real-time active lane change planning and control.Aiming at the real-time requirements in the process of active lane change,based on the idea of path-speed decoupling,the three-dimensional planning problem including time,lateral position and longitudinal position is reduced to two two-dimensional planning problems,path planning and speed planning.In the design of lateral and longitudinal MPC controllers,the matrix blocking strategy is introduced to realize the dimensionality reduction of the solution variable,and a high-performance quadratic programming solver is selected to accelerate the MPC solution through design comparison experiments.The specific research content of this paper is as follows:(1)Research on active lane change motion planning: In order to reduce the computing power and improve the solution efficiency,the path and speed are planned separately.For the path planning problem,the reference line selection is first realized through the decision logic of active lane change behavior,and then the planning space is discretized under the Frenet coordinate system,and the static obstacle and virtual obstacle are mapped to obtain the SL diagram.In the SL graph,the path search is accelerated based on the A* heuristic search algorithm,and the convex space of the quadratic programming of the path is quickly opened up,and the path planning problem is transformed into a quadratic programming problem for solving.For the speed planning problem,the ST diagram is obtained by projecting the dynamic obstacle with the path planning result as the "reference line".In order to improve the computational efficiency,the ST graph is rasterized by non-uniform sampling,and the DP algorithm is accelerated by pruning to realize the fast search of the speed sequence.According to the search results,the velocity quadratic programming convex space is constructed,and the velocity planning problem is transformed into a quadratic programming problem for solving.In Simulink,the path planning and speed planning algorithm modules are developed separately,and the final complete lane change trajectory is generated by combining the trajectory rolling update algorithm,and the algorithm is verified under typical working conditions.(2)Research on active lane change model predictive control: Firstly,CVXGEN,a quadratic programming solver with fast solution speed and small tracking error,is selected by designing a typical double-line shift experiment,and the matrix blocking strategy of " front dense and back sparse " is introduced to reduce the dimensionality of the solved variables,so as to realize the real-time optimization of MPC tracking control under the premise of ensuring tracking accuracy.In terms of lateral tracking control,the linear two-degree-of-freedom lateral tracking error model is used as the prediction model,and the block matrix is introduced into the MPC controller design to realize the reconstruction of constraints and objective functions.In terms of longitudinal tracking control,a layered architecture is designed to establish an MPC upper controller based on first-order inertial system,and a lower controller based on drive/brake switching strategy and inverse longitudinal dynamic model.In the constraint and objective function construction of MPC upper-layer controller,the block matrix is also introduced to improve the solution efficiency.Finally,the active lane change lateral and longitudinal integrated MPC controller is designed with longitudinal speed as the coupling point,and the control algorithm is verified by simulation experiments.(3)Active lane change planning and control algorithm verification: First,an active lane change joint simulation verification environment is built based on Prescan/Carsim/Simulink,Prescan is responsible for the establishment of active lane change traffic scenarios,Carsim provides vehicle dynamics model for the main vehicle,and Simulink is used to develop planning and control algorithms.In order to verify the planning and control algorithm proposed in this paper,three typical working conditions are designed: conventional lane change scenario,lane change overtaking scenario and lane change conflict scenario.The simulation results show that the path and speed planning algorithms can plan feasible trajectories that meet the constraints in different scenarios,and the MPC tracking control algorithm has a fast calculation speed while ensuring small tracking errors,which considers the real-time requirements in the process of active lane change.