Please wait a minute...
Front. Inform. Technol. Electron. Eng.  2015, Vol. 16 Issue (2): 152-165    DOI: 10.1631/FITEE.1400139
    
Design of an enhanced visual odometry by building and matching compressive panoramic landmarks online
Wei Lu, Zhi-yu Xiang, Ji-lin Liu
Institute of Information and Communication Engineering, Zhejiang University, Hangzhou 310027, China; Zhejiang Provincial Key Laboratory of Information Network Technology, Hangzhou 310027, China
Download:   PDF(0KB)
Export: BibTeX | EndNote (RIS)      

Abstract  Efficient and precise localization is a prerequisite for the intelligent navigation of mobile robots. Traditional visual localization systems, such as visual odometry (VO) and simultaneous localization and mapping (SLAM), suffer from two shortcomings: a drift problem caused by accumulated localization error, and erroneous motion estimation due to illumination variation and moving objects. In this paper, we propose an enhanced VO by introducing a panoramic camera into the traditional stereo-only VO system. Benefiting from the 360° field of view, the panoramic camera is responsible for three tasks: (1) detecting road junctions and building a landmark library online; (2) correcting the robot’s position when the landmarks are revisited with any orientation; (3) working as a panoramic compass when the stereo VO cannot provide reliable positioning results. To use the large-sized panoramic images efficiently, the concept of compressed sensing is introduced into the solution and an adaptive compressive feature is presented. Combined with our previous two-stage local binocular bundle adjustment (TLBBA) stereo VO, the new system can obtain reliable positioning results in quasi-real time. Experimental results of challenging long-range tests show that our enhanced VO is much more accurate and robust than the traditional VO, thanks to the compressive panoramic landmarks built online.

Key wordsVisual odometry      Panoramic landmark      Landmark matching      Compressed sensing      Adaptive compressive feature     
Received: 20 April 2014      Published: 29 January 2015
CLC:  TP391  
Cite this article:

Wei Lu, Zhi-yu Xiang, Ji-lin Liu. Design of an enhanced visual odometry by building and matching compressive panoramic landmarks online. Front. Inform. Technol. Electron. Eng., 2015, 16(2): 152-165.

URL:

http://www.zjujournals.com/xueshu/fitee/10.1631/FITEE.1400139     OR     http://www.zjujournals.com/xueshu/fitee/Y2015/V16/I2/152


基于在线建立与匹配压缩全景路标的增强型视觉里程计

目的:高效精确定位是移动机器人智能导航的先决条件。传统视觉定位系统,如视觉里程计(VO)和同时定位与三维重建(SLAM)算法,存在两点不足:一是由累积定位误差引起的漂移问题,二是由光照变化和移动物体导致的错误运动估计结果。
创新:通过引入全景相机到传统双目VO系统,提出一种增强型VO,高效利用全景相机360?视场角信息。(1)在线建立路口场景压缩全景路标库;(2)机器人以任意方向重新访问路标时,对定位结果进行全局校正;(3)当双目立体VO不能提供可靠定位信息时对航向角估计结果进行校正;(4)为高效利用信息量较多的全景图像,引入压缩感知概念并提出一种自适应压缩特征。
方法:首先,在压缩亮度特征基础上,增加压缩SURF特征提高其描述能力,通过分析特征区分度,使压缩特征可以根据具体图像特点自适应调节,最终构建自适应压缩特征(ACF,图2),该特征计算速度快(表3)、描述能力强(图6、7,表1),有效提高全景图像信息利用效率。然后,使用ACF对全景路标图像进行描述,提出一种任意方向的路标图像匹配算法,若当前全景图像与路标图像匹配成功,则对当前定位结果进行全局位姿校正(图4),抑制大范围环境中定位路径漂移问题(图10、11)。最后,介绍基于图像片匹配的航向角鲁棒估计方法,当双目视觉里程计因特征跟踪质量差而导致运动估计结果不稳定时,对局部运动估计结果进行校正,提高运动估计的精度(图9)。
结论:提出的增强型视觉里程计系统可以准实时提供可靠定位结果,极大抑制大范围挑战性环境中传统VO漂移问题和运动估计错误问题。实验结果显示,所提算法大幅度提高传统VO的准确性和鲁棒性。

关键词: 视觉里程计,  全景路标,  路标匹配,  压缩感知,  自适应压缩特征 
[1] Min Yuan, Bing-xin Yang, Yi-de Ma, Jiu-wen Zhang, Fu-xiang Lu, Tong-feng Zhang. Multi-scale UDCT dictionary learning based highly undersampled MR image reconstruction using patch-based constraint splitting augmented Lagrangian shrinkage algorithm[J]. Front. Inform. Technol. Electron. Eng., 2015, 16(12): 1069-1087.