不依赖里程计的机器人定位与地图构建

为了解决缺少里程计情况下的移动机器人同时定位与地图构建(SLAM)问题，提出一种机器人运动状态估计模型。通过将该模型与FastSLAM框架相结合，在SLAM过程中实现对机器人位置、姿态及其运动状态(如速度)的估计。该算法用估计的运动状态代替里程计，实现了在没有里程计情况下的SLAM。为验证算法性能，通过仿真和维多利亚数据库的实验将该算法与需要里程计信息的SLAM算法相对比。实验结果表明，该算法在大于30个粒子的情况下可以达到与需要里程计信息的SLAM算法相当的精度。

A model for estimating robot motion state was proposed to handle simultaneous localization and mapping (SLAM) without odometry. By combining this model with framework of FastSLAM, the proposed algorithm estimates the robot position, pose and motion state (such as speed) during SLAM. The proposed algorithm uses the estimated motion state instead of odometer, thus enables SLAM with noodometer. The performance of the algorithm was verified by comparison of the proposed algorithm with the SLAM algorithm including odometry information by simulation and Victoria database. Experimental results show that the proposed algorithm can achieve the same accuracy as that of the SLAM algorithm with odometer information in the case of the particles larger than 30.