基于Kalman滤波的车位侧方距离修正方法

Vehicle Side Distance Correction Method Based on Kalman filter

摘要:

通常车位识别技术通过超声波传感器获取侧方障碍物位置信息来判断车位边缘,由于测量时超声波传感器与障碍物形成波束角的跳变,及其本身的固有特性会带来随机噪声,导致不能直接得到状态变量的真实精确值。通过建立合适的距离修正系统状态方程和观测方程,采用Kalman滤波算法,由 k-1 时刻的超声波传感器测量值与随机噪声获得该时刻的协方差,并与测量噪声计算获得 k 时刻的Kalman增益,再结合k 时刻的超声波传感器测量值与观测方程得到k+1时刻的距离修正值。仿真结果表明,经过150次迭代计算后的绝对误差为1.575cm,平均修正时间仅需0.028s。该方法可有效降低了随机噪声干扰,具有良好的准确性和实时性,滤波测量距离修正值更加逼近真实值。

Abstract:

Generally, the parking space recognition technology uses the ultrasonic sensor to obtain the position information of the side obstacle to determine the edge of the parking space. Due to the jump of the beam angle formed by the ultrasonic sensor and the obstacle during the measurement, and its inherent characteristics will bring random noise, which may not be directly obtained. The true exact value of the state variable. By establishing a suitable distance correction system state equation and observation equation, Kalman filtering algorithm is used to obtain the covariance of the moment from the ultrasonic sensor measurement value and random noise at time k-1 , and obtain the Kalman gain at time k from the measurement noise calculation. Then, the distance correction value of k+1 time is obtained by combining the measured value of the ultrasonic sensor at time k with the observation equation. The simulation results show that the absolute error after 150 iterations is 1.575cm, and the average correction time is only 0.028s, which is much smaller than the average correction error of the dual probe average method, the dual probe fusion method and the sensor error compensation method. It effectively reduces random noise interference, has good accuracy and real-time performance, and the filter measurement distance correction value is closer to the true value.