【摘要】 为了得到较高的估计精度,基于扩展卡尔曼滤波的同时定位与地图生成算法(EKF-SLAM)需要完成多次路径闭合。这不仅消耗大量的时间与能量,而且增大了机器人发生故障的概率。本文提出一种添加约束的EKF-SLAM算法。该算法通过分析协方差矩阵确定目标路标对,用测量信息与全局先验方向对原估计结果进行约束,能够极大改善估计效果,兼顾高效率与高精度。实验结果及其分析充分表明了算法的有效性。更多还原

【Abstract】 In order to achieve good result,multi-loop closure is necessary with the extended kalman filter approach to simultaneous localization and mapping (EKF-SLAM). This will not only waste lots of time and energy but also increase the possibility of robot fault. This paper presents a novel method which applies constraints to EKF-SLAM to improve estimate accuracy. The method determines the pair of landmarks by analyzing the covariance matrix and applies the constraints constructed with distance measurement and prior global heading information to improve the estimate result of EKF-SLAM. The experiment results and corresponding analysis demonstrate the approach could achieve good mapping efficiency and accuracy simultaneously.更多还原