An improved RGB-D SLAM algorithm

摘要：

　　This paper presents 2 improvements to traditional RGB-D SLAM algorithm.One is to build a minimum error function using graph optimization model to estimate camera pose,reduce the number of iterations; another is proposed a intelligent loop detection algorithm which according to camera trajectory curvature,to construct accurate loop closing.We also design the lost recovery system and use GPU in feature extraction.Compared with the tradition RGB-D SLAM,our system is 6.1 times faster,and the best precision increased 30%.

关键词： SLAM graph optimization GPU loop detection

作者: Wenchi Sun Shunyan Wang Jiancai Wu Xin Du

作者单位: School of Computer Science and Technology,Wuhan University of Technology Wuhan,China

会议类型: 国际会议

会议名称: 2017 IEEE 2nd Advanced Information Technology,Electronic and Automation Control Conference(IAEAC 2017)(2017 IEEE 第2届先进信息技术、电子与自动化控制国际会议)

会议地点: 重庆

会议语种:英文

页码: 1425-1428

在线出版日期: 2017-03-25（万方平台首次上网日期，不代表论文的发表时间）