精确稀疏扩展信息滤波的多机器人SLAM研究

在机器人同时定位与地图创建(SLAM)问题中,多机器人SLAM成为目前机器人学中的研究热点.因此,基于精确稀疏扩展信息滤波算法(ESEIF)的多机器人SLAM问题,根据多机器人的运动模型和观测模型分别对多机器人位姿估计及环境特征点进行观测,并依据阈值划分观测特征点,以完成机器人的观测更新,同时边缘化机器人位姿并进行重定位.实验仿真数据表明:多机器人的位姿精度良好,观测更新阶段时间基本上恒定,与地图特征点数量无关,体现了ESEIF算法在研究多机器人SLAM问题的有效性.

Multi-robot SLAM with Exactly Sparse Extended Information Filtering

In the robot simultaneous localization and mapping（SLAM） problem, the Multi-robot SLAM has drawn much attention recently in robotics. This paper investigates the Multi-robot SLAM problem in using exactly sparse extended information filters algorithm（ESEIF）： Based on both the multi-robot motion model and observation model, the multi-robot pose is estimated and the environment features are observed. The threshold is set for partitioning and updating the observed features, and marginalizing the robot pose followed by relocating the robot. The simulation shows that the robots pose can be accurately estimated, and observation can be updated in a constant time fashion irrespective of the number of features in the map.