非结构化道路的无人车MFAC大曲率横向控制方法研究

An unmanned vehicle MFAC large curvature lateral control method based on GPS positioning system

针对无人车大曲率转向控制时考虑道路曲率的视觉预瞄策略在非结构化道路难以应用,基于GPS定位导航系统,综合考虑车速、道路弯曲度与预瞄距离之间的关系,提出一种无人车可变序列预瞄策略,并设计了PID型无模型自适应横向控制方法,进而采用改进的粒子群优化算法实现对控制器几个重要参数的智能调参,最后通过仿真实验验证了控制方法的有效性。结果表明:采用可变序列预瞄策略的无人车能够根据前方道路弯曲度与车速实时调节预瞄距离,且智能调参后的控制器性能优于人工调参,能够确保车辆以较高精度跟踪大曲率路径。

Due to the lack of consistent features on unstructured roads,it is difficult for visual sensors to accurately understand and recognize this environment.Therefore,environmental perception and expected trajectory detection of unstructured roads usually need to combine pre-collected reference trajectories with high-precision global positioning capabilities.At this time,the vision-based preview strategy is obviously no longer applicable.Aiming at the problem that it is difficult to apply the visual preview strategy that considers the road curvature to unstructured roads in the steering control of unmanned vehicles with large curvature,firstly a variable sequence preview strategy for unmanned vehicles based on GPS positioning and navigation system is proposed,which comprehensively considers the relationship among the speed of the vehicle,road curvature and preview distance.Secondly,through the full-format dynamic linearization technique,the unknown lateral control discrete model of unmanned vehicle is transformed into an FFDL model with 3-dimensional pseudo-gradient time-varying vector,on the basis of which a PID-type model-free adaptive lateral controller is designed.Through this data-driven PID model-free adaptive control method,the modeling of complex mechanisms of unmanned vehicles under large curvature control can be avoided.The model-free adaptive controller can use the online I/O data of the controlled system to adjust parameters online.Compared with the PID controller,it is a low cost controller which has better control performance and adaptability in nonlinear complex systems and is simple in calculation.However,in order to design a model-free adaptive controller,some key parameters must be manually adjusted,which is a very complex problem,and the control parameters of the model-free adaptive lateral controller will affect the control performance,but during the practical application of particle swarm optimization,it is found that it is easy to fall into local extreme points because of the influence of the global group experience,which means that the performance of the conventional particle swarm optimization algorithm is not enough to automatically adjust the parameters of the controller.To further improve the convergence accuracy,local group experience is introduced into the standard PSO algorithm.The improved particle swarm optimization algorithm is used to intelligently adjust several important parameters of the controller.Finally,based on the MATLAB environment,and referring to the GB/T 6323—2014 standard,the simulation path of the double-shift line is designed,and the linear two-degree-of-freedom model is used as the controlled object on this path.The effectiveness of the control method is verified by simulation experiments.The results show that:(1)The unmanned vehicle adopting the variable sequence preview strategy can adjust the preview distance in real time according to the curve of the road ahead and the speed of the vehicle.(2)The controller effectively adjusts the front wheel rotation angle by obtaining the preview deviation angle to ensure that the vehicle can track the target path with high accuracy,and the PID model-free adaptive lateral controller still has a good control effect when the vehicle speed changes,which give full play to the robustness of the controller.(3)The designed improved particle swarm algorithm intelligent parameter adjustment method The parameter adjustment cost can be greatly saved and better controller performance can be obtained compared with the manual parameter adjustment method.