一种基于UKF的井下机器人超声网络定位方法

UKF-based ultrasonic network localization for a mine robot

机器人将在未来煤矿少人或无人化开采中发挥重要作用。根据井下工作光线较暗,空间封闭的特点,提出了一种基于无损卡尔曼滤波(unscented Kalman filter,UKF)的井下机器人超声网络定位方法。方法的核心是通过对光电码盘和电子罗盘定位以及超声网络定位输出的井下机器人位置坐标和航向角度进行UKF滤波,来对井下机器人进行位置更新和预测。由于对机器人进行位置更新和预测是复杂的非线性函数,采用UKF能有效提高滤波精度,降低定位误差。由于越接近当前时刻的转弯半径误差对转弯半径影响越大,故采用Sigmoid函数作为转弯半径误差系数来计算M个转弯半径误差和。以误差和为权重调节机器人左右驱动轮的转弯半径,可以减小转弯半径误差。仿真结果表明,采用所提出的基于UKF的井下机器人超声网络定位方法实现了井下机器人更稳定和更精确的定位。

Robot will play a significant role on unmanned coal mining in the future. According to the dark and confined characteristics in underground coal mine,an ultrasonic network localization technique based on unscented Kalman filter（ UKF） is proposed. The core of the method is to update and predict the robot position through UKF which combines the robot position output of light code disc and electronic compass localization with the robot position output of ultrasonic networks localization. Since the location update and prediction for a robot is a complex nonlinear function,the filtering accuracy is improved and the error of positioning is reduced effectively by the UKF. Since the nearer error of turning radius affects the turning radius more,the Sigmoid function is applied as the error coefficient to calculate the error sum of M turning radius. The robot turning radius error is reduced by adjusting the robot turning radius according to the error sum. The simulation results show that the more accurate and reliable location of robot is achieved with the proposed ultrasonic network localization based on UKF.