Deep learning for 2D scan matching and loop closure

Although 2D LiDAR based Simultaneous Localization and Mapping (SLAM) is a relatively mature topic nowadays, the loop closure problem remains challenging due to the lack of distinctive features in 2D LiDAR range scans. Existing research can be roughly divided into correlation based approaches e.g. scan-to-submap matching and feature based methods e.g. bag-of-words (BoW). In this paper, we solve loop closure detection and relative pose transformation using 2D LiDAR within an end-to-end Deep Learning framework. The algorithm is verified with simulation data and on an Unmanned Aerial Vehicle (UAV) flying in indoor environment. The loop detection ConvNet alone achieves an accuracy of 98.2% in loop closure detection. With a verification step using the scan matching ConvNet, the false positive rate drops to around 0.001%. The proposed approach processes 6000 pairs of raw LiDAR scans per second on a Nvidia GTX1080 GPU.

[1]  Kai Oliver Arras,et al.  FLIRT - Interest regions for 2D range data , 2010, 2010 IEEE International Conference on Robotics and Automation.

[2]  John J. Leonard,et al.  Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age , 2016, IEEE Transactions on Robotics.

[3]  Michael Bosse,et al.  Map Matching and Data Association for Large-Scale Two-dimensional Laser Scan-based SLAM , 2008, Int. J. Robotics Res..

[4]  José A. Castellanos,et al.  Linear time vehicle relocation in SLAM , 2003, 2003 IEEE International Conference on Robotics and Automation (Cat. No.03CH37422).

[5]  Stefan Kohlbrecher,et al.  A flexible and scalable SLAM system with full 3D motion estimation , 2011, 2011 IEEE International Symposium on Safety, Security, and Rescue Robotics.

[6]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  Wolfram Burgard,et al.  Geometrical FLIRT phrases for large scale place recognition in 2D range data , 2013, 2013 IEEE International Conference on Robotics and Automation.

[8]  Karl Granström,et al.  Learning to detect loop closure from range data , 2009, 2009 IEEE International Conference on Robotics and Automation.

[9]  Wolfgang Hess,et al.  Real-time loop closure in 2D LIDAR SLAM , 2016, 2016 IEEE International Conference on Robotics and Automation (ICRA).

[10]  Edwin Olson,et al.  M3RSM: Many-to-many multi-resolution scan matching , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[11]  Roland Siegwart,et al.  From perception to decision: A data-driven approach to end-to-end motion planning for autonomous ground robots , 2016, 2017 IEEE International Conference on Robotics and Automation (ICRA).

[12]  Geoffrey A. Hollinger,et al.  Deep Learning for Laser Based Odometry Estimation , 2016 .

[13]  Michael Bosse,et al.  Keypoint design and evaluation for place recognition in 2D lidar maps , 2009, Robotics Auton. Syst..

[14]  Wolfram Burgard,et al.  Improved Rao-Blackwellized Mapping by Adaptive Sampling and Active Loop-Closure , 2004 .

[15]  Sven Hellbach,et al.  Large scale place recognition in 2D LIDAR scans using Geometrical Landmark Relations , 2014, 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[16]  Sebastian Thrun,et al.  FastSLAM: a factored solution to the simultaneous localization and mapping problem , 2002, AAAI/IAAI.

[17]  Daniel Cremers,et al.  Robust odometry estimation for RGB-D cameras , 2013, 2013 IEEE International Conference on Robotics and Automation.