基于低成本MEMS-INS的车辆变道判别方法

Identification method of vehicle lane changes based on low-cost MEMS-INS

针对传统车辆变道判别与预测方法的不足,提出一种基于低成本MEMS-INS的车辆变道判别与预测方法。利用车载低成本MEMS-INS和轮速传感器获取的纵向速度、加速度与横摆角速度信息,建立两级卡尔曼滤波器估计道路曲率,在此基础上,将所估计出的道路曲率信息作为观测量,利用交互多模型算法对车辆是否变道进行判别和预测,同时作为附带的效果,能够准确估计出车辆的航向角。实车试验结果表明,车辆变道识别率达到100%,航向角估计误差小于2%,延时小于1 s,具有成本低、精度高、延时短,环境适应力强的优点,克服了传统方法的不足,满足车辆碰撞预警系统的需要。

To overcome the shortcomings of traditional vehicle lane changes identification and prediction methods, a novel identification and prediction method of vehicle lane-change based on low-cost MEMS-INS is proposed. The speed, acceleration and yaw rate from MEMS-INS and wheel speed sensor are taken as the observed information of Kalman filter. By two-level Kalman filter recursion algorithm, the road curvature parameters are acquired. Then the road curvature parameters are considered as observed information, and the interactive multiple model method is employed for predicting lane changes. As an additional effect, the heading angle is estimated with high precision. The real vehicle tests demonstrate that the identification rate of lane changes is 100%, the estimation error of heading angle is less than 2%, and the latency times is less than 1 s. The proposed method has such advantages as low cost, high precision, environmental adaptability and very short latency times, which overcomes the shortcomings of traditional methods and meets the acquirements of vehicle collision warning system.