基于多维尺度法和卡尔曼滤波的机器人传感器网络跟踪定位

Research on tracking and locating of robot sensor networks based on multidimensional scaling coupled Kalman filtering

为解决当前工业机器人定位误差大,传统多维尺度(MDS)模型的精度受到环境噪声的影响,现有定位方法无法满足定位精度需求的现状,提出了一种将卡尔曼滤波和多目标跟踪相结合的方法来减小噪声对多源定位的影响.建立了高斯分布下加性噪声对距离测量噪声分布的预测模型,并据此建立了线性跟踪系统.实验数据表明,在不同噪声水平及不同传感器网络节点数量下,基于MDS法的传感器网络节点定位误差最高接近20%,基于EKL法的定位误差在7.46%～13.13%范围内变化,而本文提出的MDS和KL联合定位方法误差始终控制在8%以内.与现有常用的MDS和扩展卡尔曼滤波EKL定位方法相比,该算法有效地降低了实际环境中的噪声影响,具有很好解决移动传感器定位问题的潜力,满足了高精度定位的实际要求.同时,由于采用了线性滤波器进行简化,很好地适应了功率、内存和计算能力有限的小型嵌入式传感器,在机器人定位领域具有一定的参考价值.

The work aims to solve the problems that the positioning error of industrial robots is large,and the precision of the traditional multidimensional scale(MDS)model is affected by the environmental noise which cannot meet the requirement of positioning accuracy.A solution to attenuate noise effects to MDS by combining MDS with a Kalman filter was proposed.A model was built to predict the noise distribution with regard to additive noises to the distance measurements following the Gaussian distribution.From that,a linear tracking system was developed.The experimental data showed that,under the different noise level and sensor network node numbers,the location error based on MDS method is close to20%,and changes in the 7.46%~13.13%based on EKL,and always lower than8%based on multidimensional scaling coupled Kalman filtering.Compared with the current locating algorithms(MDS and EKL),the proposed algorithm had a higher positioning accuracy under different noise level and node number.It had a good potential to solve the positioning of mobile sensors.Besides,the linear filter is simplified,therefore it suits small and embedded sensors equipped with limited power,memory,and computational capacities well,which has certain reference value in the field of robot positioning.