移动机器人同步定位的数据关联优化控制  

Data Association Optimal Control for Mobile Robot Simultaneous Localization

作　　者：石家兴 丁永生[1,2] 郝矿荣[1,2] 

机构地区：[1]数字化纺织服装技术教育部工程研究中心,上海201620 [2]东华大学信息科学与技术学院,上海201620

出　　处：《计算机仿真》2015年第9期377-381,共5页Computer Simulation

基　　金：国家自然科学基金重点项目(61134009);国家自然科学基金(61473078);上海市科学技术委员会重点基础研究项目(13JC1407500);上海市教育委员会科研创新项目(14ZZ067)

摘　　要：在移动机器人同步定位与构图(SLAM)问题中,在大规模复杂环境下,由于传统数据关联算法的速度和正确率随着地图规模的增长而降低,导致难以满足实时性和鲁棒性的要求。为提高定位性能,根据联合相容分支定界(JCBB)算法,提出了一种改进的IJCBB数据关联算法用于移动机器人同步定位优化控制。首先建立地图的KD树模型,生成优化候选路标集,以缩小关联搜索空间,提升关联速度;其次构造增补关联规则,对JCBB算法的初步关联结果进行增补再关联,提升关联正确率。仿真结果表明:IJCBB算法的关联速度和关联正确率均优于传统关联算法,具有较高的实时性和鲁棒性。In large-scale complex environment, the computing speed and accuracy of the traditional data associa- tion algorithm decrease with the growth of map scale, that causes the difficulty to meet the requirements of real-time and robustness of simultaneous localization and mapping （SLAM） for mobile robot. Based on the joint compatibility brunch and bound （JCBB） data association algorithm, an improved IJCBB data association algorithm is proposed. Firstly, a KD-Tree is built to optimize the candidate landmarks in order to shrink the association search space and speed up data association. Secondly, the primary result of the JCBB data association algorithm is processed by sup- plementary re-association using mutual and exclusion threshold principle to improve accuracy. Simulation results demonstrate that, compared with traditional data association algorithm, the IJCBB improves the computing speed and accuracy, and has better characteristics of real-time and robustness.

关 键 词：移动机器人 同步定位与构图 数据关联 联合相容分支定界 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]