基于自适应模型预测的自主车辆避障控制研究

Research on Obstacle Avoidance Control of Autonomous Vehicle Based on Adaptive Model Prediction

在无人驾驶车辆安全避障控制过程中,由于被控对象的非线性特性,线性模型预测控制器难以保证车辆避障系统的安全性及稳定性。以车辆二自由度运动学模型为基础,在已有感知障碍物信息的基础上,通过应用自适应模型预测控制理论,结合障碍物虚拟边界约束及车辆控制变量约束,设计了以节气门开度和方向盘转角为控制变量的线性时不变控制器,并在Simulink环境下与基于传统模型预测避障控制器进行比较。仿真结果表明:相较于模型预测控制,基于自适应模型预测控制理论设计的控制器能提高车辆避障安全性及舒适性。

In the process of unmanned vehicle safety obstacle avoidance control,it is difficult for the linear model prediction controller to guarantee the safety and stability of vehicle obstacle avoidance system due to the nonlinear characteristics of the controlled object. Based on the vehicle two DOF kinematics model and the premise of perceived obstacle information,an adaptive model predictive control theory was applied. The virtual boundary constraints and vehicle control variable constraints were combined. The linear time-invariant controller was designed with the throttle opening and steering wheel angle as the control variables. Finally,the obstacle controller based on the traditional model prediction theory was compared in the Simulink environment. The simulation results show that the controller based on the adaptive model predictive control theory can effectively improve the safety and comfort of the vehicle obstacle avoidance compared with the traditional model predictive control.