一种改进的几何约束分枝定界SLAM重定位算法

摘要

重定位技术是机器人在已有SLAM地图的环境中依靠自身传感器重新获得定位信息的关键技术.几何约束分枝定界重定位(GCBB)算法是一种有效的方法,但是其存在计算速度慢的缺点.针对GCBB算法的不足,从两个方面对其进行改进:一是采用分组方式进行数据关联;二是结合传感器探测范围在局部区域中选择特征进行数据关联.仿真结果表明,所提出的快速几何约束分枝定界重定位(FGCBB)算法能够正确实现重定位,且计算复杂度与观测数目两者之间服从线性关系,当处理观测数目较多的问题时,FGCBB的计算效率明显优于GCBB算法.%The relocation technology is a key technology for robot to recover the location information by its sensor in the existing SLAM(simultaneous localization and mapping)environment. The geometric constraints branch and bound(GCBB)algo-rithm is an effective method,but its computation speed is slow. To overcome the shortcoming of GCBB algorithm,the algorithm was improved in two aspects:the packet mode is selected for data association;the feature is selected in local region for data as-sociation according to the detection range of the sensor. The simulation results show that the proposed fast geometric constrains branch and bound(FGCBB)algorithm can relocate correctly,both the computational complexity and observation quantity are in accord with the linear relation,and the calculation efficiency of FGCBB algorithm is better than that of GCBB algorithm while processing much observation quantity.