移动机器人SLAM问题的研究

Research on Problem in SLAM for Mobile Robots

摘要:

针对移动机器人同时定位与地图创建(SLAM)的问题,就扩展卡尔曼(EKF)算法所存在的缺陷即不适合大范围环境及密集环境等特征数量较大的场合,提出了一种改进的EKF-SLAM算法;它在扩展卡尔曼(EKF)算法上采用Rao-Blackwellise的分解思想-分解估计构架,将SLAM问题分解为路径估计和地图估计两个问题从而进行预测步骤,观测步骤,更新步骤和向量增广步骤4个步骤；仿真结果显示改进的EKF-SLAM算法比EKF-SLAM算法在特征数量较大的场合更具有优异性;它大大降低了计算复杂度,提高了准确性,为在比较复杂环境下实时解决移动机器人同时定位与地图创建(SLAM)的问题提供了一种有效方法。

Abstract:

Aiming at the problem of simultaneous localization and mapping(SLAM) for mobile robots,an improved EKF-SLAM algorithm was proposed while considering the existing defects of extended Kalman (EKF) algorithm-not suitable for a large number of characters,occasions such as a wide range of environments ,an intensive environment and so on.It put the decomposition thought of Rao-Blackwellise on the extended Kalman (EKF) algorithm-Decomposition Estimate framework ,which broke down the SLAM problem into the two issues of the path estimation and map estimation so that predicting,observating,updating and increasing vectors .The simulation results show that improved EKF-SLAM algorithm is more superior than EKF-SLAM algorithm in a number of features,occasions. The improved algorithm greatly reduces the computational complexity and improves the accuracy so it provides an effective method for the problem of simultaneous localization and mapping (SLAM) for mobile robots in the more complex environment.