Abstract: In order to enable intelligent vehicles to obtain the current pavement adhesion coefficient in time during the trajectory tracking process for better trajectory tracking control, an intelligent vehicle trajectory tracking control method based on online real-time estimation of the adhesion coefficient was proposed. Based on the force situation of the current trajectory tracking of the intelligent vehicle, the Dugoff tire normalization model and unscented Kalman filter (UKF) algorithm were used to design the pavement adhesion coefficient estimator. Moreover, based on the estimated value of the current pavement adhesion coefficient, a linear quadratic regulator (LQR) with feedforward control was designed by the vehicle two-degree-of-freedom dynamics model and tracking error model, to realize the trajectory tracking control of intelligent vehicle. In addition, the joint simulation of Carsim and Matlab/Simulink was used to test the ability of trajectory tracking control and pavement adhesion coefficient estimation of intelligent vehicles. The simulation outcomes demonstrate that the proposed approach can precisely estimate the adhesion coefficient of each wheel when driving at various speeds on roads with different adhesion coefficients, and the trajectory tracking effect is good.

Key words: vehicle engineering; trajectory tracking; linear quadratic regulator (LQR); pavement adhesion coefficient estimation; unscented Kalman filtering

摘要： 为了让智能车在轨迹跟踪过程中及时获得当前路面附着系数以便更好地进行轨迹跟踪控制，提出了一种基于附着系数在线实时估计的智能车轨迹跟踪控制方法。该方法基于智能车当前轨迹跟踪的受力情况，采用Dugoff 轮胎归一化模型和无迹卡尔曼滤波（UKF）算法设计路面附着系数估计器，并基于当前路面附着系数估计值通过车辆二自由度动力学模型和跟踪误差模型，设计带有前馈控制的线性二次型调节器（LQR），实现智能车的轨迹跟踪控制。并通过Carsim和MATLAB/Simulink联合仿真，对智能车的轨迹跟踪控制及路面附着系数估计能力进行测试，仿真结果表明：在不同附着系数路面上不同车速行驶时该方法均能较为准确地估计各轮的附着系数，并且轨迹跟踪效果良好。

关键词: 车辆工程；轨迹跟踪；LQR；路面附着系数估计；无迹卡尔曼滤波