基于线段特征匹配的EKF-SLAM算法  

EKF-SLAM Algorithm Based on Line Segment Matching

作　　者：张国良[1] 汤文俊[1] 敬斌[1] 程展欣 

机构地区：[1]第二炮兵工程大学301教研室,陕西西安710025

出　　处：《控制工程》2012年第6期1019-1024,1028,共7页Control Engineering of China

摘　　要：针对EKF-SLAM算法在机器人被"绑架"时失效的问题,提出一种新的基于线段特征匹配的EKF-SLAM算法——EKFLineSLAM算法。该算法在线段特征观测模型和改进的基于逐点搜索的线段提取算法的基础上,将线段特征匹配引入EKF-SLAM算法,并对线段长度和姿态角进行EKF更新,创建环境的线段特征地图。在未知室内结构化环境中,将该算法与弱匹配EKFLineSLAM算法进行比较,验证了EKFLineSLAM算法在结构化环境中克服机器人"绑架"问题的可行性和有效性。For the problem of EKF-SLAM algorithm being invalid when robot is kidnapped, a new EKF-SLAM algorithm called ＂EKF- LineSLAM Algorithm＂ based on line segment matching is presented. This algorithm is based on the line segment observation model, and the improved line segment extraction algorithm based on point by point search. It introduces the line segment match in the EKF- SLAM algorithm, and renews the line segment＇s length and posture angle by EKF to find the line segment characteristic map. Finally, the presented algorithm and the weak matching EKFLineSLAM algorithm are compared in an unknown structured indoor environment. The comparison results confirm the feasibility and the effectiveness of the EKFLineSLAM algorithm for the kidnapped robot problem in structured environments.

关 键 词：EKF-SLAM EKFLineSLAM 线段特征观测模型 线段提取 线段匹配 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]