摘要
针对室内环境复杂,难以通过单一传感器对机器人精准定位的问题,以室内环境中的两轮差动移动机器人为研究对象,提出了一种自适应无迹卡尔曼室内定位算法;该方法以无迹卡尔曼滤波(UKF)算法为基础,融合里程计、超声波定位系统、电子罗盘等传感器数据,利用超声波定位低频特性好的特点,减轻里程计结合电子罗盘进行航迹推算的累积误差和打滑影响;鉴于实际中量测噪声往往难以确定,利用Sage-Husa自适应方法,并根据不同传感器的噪声特性设置不同的加权系数,在线更新量测噪声特性,以实现对量测噪声的自适应;通过仿真验证,该方法能在传感器噪声特性未知的情况下,有效适应传感器噪声的变化,从而能够在复杂室内环境下,实现较高精度和鲁棒性的位姿估计。
For the complexity of indoor environment and difficulty of accurate localization with a single sensor,an adaptive unscented Kalman indoor localization algorithm is proposed,using an indoor two wheel mobile differential robot as the research object.This algorithm,which is based on UKF(unscented Kalman filter),combines data of odometer,ultrasonic position system and electronic compass.The data of ultrasonic position system has great low-frequency character.It can be used to decrease the effect of slip and accumulated error on the dead reckoning using odometer and electronic compass.Usually,the actual measurement noise is difficult to determine.Based on the SageHusa adaptive method,different weighting coefficients are set according to the noise characteristics of different sensors.Measurement noise characteristics are updated online and the adaption of observation noise is realized.Simulation shows that AUKF(adaptive unscented kalman filter)algorithm can effectively adapt to the change of sensors' noise when the characteristic of noise is unknown.It can promise high precise and robust localization in indoor environment.
出处
《计算机测量与控制》
2018年第1期238-241,247,共5页
Computer Measurement &Control
基金
国家自然科学基金(61673214
61673217
61673219)
江苏省"六大人才高峰"项目(XNYQC-CXTD-001)
天津市科技重大专项与工程项目(15ZXZNGX00250)
