基于集员估计的室内移动机器人多传感器融合定位

本文针对室内移动机器人的长距离实时鲁棒定位问题进行了研究,考虑到单一定位手段存在的不足,以二维扫描激光和里程计作为主要的定位设备,采用多传感器数据融合技术实现了移动机器人的精确定位.论文首先通过引入基于点-直线特征匹配的改进迭代最近邻(iterative closest point,ICP)扫描匹配方法对激光采集的环境点云信息进行迭代匹配以得到相对位姿变换估计,并推导了其估计不确定性的保守包络矩阵形式,然后通过建立定位过程和观测模型,引入扩展非线性集员滤波器作为多传感器融合方法,利用扫描匹配结果校正由里程计滑移带来的定位误差,并获取定位自身的不确定性边界估计.实验结果表明了本文所提出的室内定位方法的精度、实时性和鲁棒性.

Multi-sensor fusion for mobile robot indoor localization based on a set-membership estimator

The robust long-distance localization problem of indoor mobile robots is studied in this paper. Taking into account the defects produced by using only a single localization means, a 2D laser scanner and an odometer are adopted as the main localization devices, with their data fused to achieve precise localization of the mobile robot. An improved iterative closest point (ICP) algorithm based on the point-line matching approach is proposed to estimate the relative pose transformation of the robot from point clouds collected by the laser scanner, and the underlying uncertainties for the pose estimation are also derived as a conservative envelope matrix. Through establishing the localization process and measurement models, the extended nonlinear set membership filtering (ESMF) is introduced as a multi-sensor fusion method to correct the cumulative errors of the odometer with the scan matching data, and the boundary estimations of pose uncertainties are also obtained. Experimental results show that the accuracy, the real-time property and the robustness of the proposed indoor localization system can be guaranteed.