Relocalization under Substantial Appearance Changes using Hashing

Localization under appearance changes is essential for robots during long-term operation. This paper investigates the problem of place recognition in environments that undergo dramatic visual changes. Our approach builds upon previous work on graph-based image sequence matching and extends it by incorporating a hashing-based image retrieval strategy in case of localization failures or the kidnapped robot problem. We present a variant of hashing algorithm that allows for fast retrieval for high-dimensional CNN features. Our experiments suggest that our algorithm can reliably recover from localization errors by globally relocalizing the robot. At the same time, our hashing-based candidate selection is substantially faster than state-of-the-art locality sensitive hashing.

[1]  Gordon Wyeth,et al.  SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights , 2012, 2012 IEEE International Conference on Robotics and Automation.

[2]  Antonio Torralba,et al.  Spectral Hashing , 2008, NIPS.

[3]  Piotr Indyk,et al.  Similarity Search in High Dimensions via Hashing , 1999, VLDB.

[4]  Winston Churchill,et al.  Experience-based navigation for long-term localisation , 2013, Int. J. Robotics Res..

[5]  Gordon Wyeth,et al.  FAB-MAP + RatSLAM: Appearance-based SLAM for multiple times of day , 2010, 2010 IEEE International Conference on Robotics and Automation.

[6]  Wolfram Burgard,et al.  Robust Visual Robot Localization Across Seasons Using Network Flows , 2014, AAAI.

[8]  Xiang Zhang,et al.  OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks , 2013, ICLR.

[9]  Roberto Cipolla,et al.  PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[10]  Wolfram Burgard,et al.  Efficient and effective matching of image sequences under substantial appearance changes exploiting GPS priors , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[11]  Michael Milford,et al.  Convolutional Neural Network-based Place Recognition , 2014, ICRA 2014.

[12]  Paul Newman,et al.  Learning place-dependant features for long-term vision-based localisation , 2015, Auton. Robots.

[13]  Paul Newman,et al.  Work smart, not hard: Recalling relevant experiences for vast-scale but time-constrained localisation , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[14]  Zhe Wang,et al.  Multi-Probe LSH: Efficient Indexing for High-Dimensional Similarity Search , 2007, VLDB.

[15]  Niko Sünderhauf,et al.  Appearance change prediction for long-term navigation across seasons , 2013, 2013 European Conference on Mobile Robots.

[16]  Cyrill Stachniss,et al.  Lazy Data Association For Image Sequences Matching Under Substantial Appearance Changes , 2016, IEEE Robotics and Automation Letters.

[17]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[18]  Michael Milford,et al.  Vision-based place recognition: how low can you go? , 2013, Int. J. Robotics Res..

[19]  Simon Lacroix,et al.  Location graphs for visual place recognition , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[20]  Luis Miguel Bergasa,et al.  Fusion and binarization of CNN features for robust topological localization across seasons , 2016, 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[21]  Josef Sivic,et al.  NetVLAD: CNN Architecture for Weakly Supervised Place Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Paul Newman,et al.  Highly scalable appearance-only SLAM - FAB-MAP 2.0 , 2009, Robotics: Science and Systems.