Optimal 3D Reconstruction of Caves Using Small Unmanned Aerial Systems and RGB-D Cameras

This paper discusses our work on building a SmartCaveDrone system [1] to function as “Co-Archaeologists” that can map large caves and enter dangerous or hard-to-reach spaces. We present an optimal 3D reconstruction framework that combines both sparse feature matching and dense surface alignment. Sparse feature matching is used for visual odometry and global camera pose optimization after loops are detected, while dense surface alignment is the way of closing large loops and solving surface mismatching problem. We collected data inside Maya caves in Belize, and evaluated different methods applied on this dataset. Our experiments demonstrate that sparse feature-based tracking is robust for steady tracking, while it is effective and critical to close surface loops. After our experiments, we discussed new emerging research problems that can be explored.

[1]  Juan D. Tardós,et al.  ORB-SLAM2: An Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras , 2016, IEEE Transactions on Robotics.

[2]  Yi Li,et al.  Fully Convolutional Instance-Aware Semantic Segmentation , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[3]  Vladlen Koltun,et al.  Robust reconstruction of indoor scenes , 2015, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[4]  Yangquan Chen,et al.  LoopSmart: Smart Visual SLAM Through Surface Loop Closure , 2018, ArXiv.

[5]  John J. Leonard,et al.  Kintinuous: Spatially Extended KinectFusion , 2012, AAAI 2012.

[6]  Michael Kaess,et al.  GravityFusion: Real-time dense mapping without pose graph using deformation and orientation , 2017, 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[7]  Paul H. J. Kelly,et al.  SLAM++: Simultaneous Localisation and Mapping at the Level of Objects , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[8]  Sebastian Thrun,et al.  Simultaneous localization and mapping with unknown data association using FastSLAM , 2003, 2003 IEEE International Conference on Robotics and Automation (Cat. No.03CH37422).

[9]  Andrew Owens,et al.  SUN3D: A Database of Big Spaces Reconstructed Using SfM and Object Labels , 2013, 2013 IEEE International Conference on Computer Vision.

[10]  Huimin Ma,et al.  3D Object Proposals for Accurate Object Class Detection , 2015, NIPS.

[11]  YangQuan Chen,et al.  SmartCaveDrone: 3D cave mapping using UAVs as robotic co-archaeologists , 2017, 2017 International Conference on Unmanned Aircraft Systems (ICUAS).

[12]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[13]  YangQuan Chen,et al.  Optimal Observation for Cyber-physical Systems: A Fisher-information-matrix-based Approach , 2009 .

[14]  G. Klein,et al.  Parallel Tracking and Mapping for Small AR Workspaces , 2007, 2007 6th IEEE and ACM International Symposium on Mixed and Augmented Reality.

[15]  Jana Kosecka,et al.  3D Bounding Box Estimation Using Deep Learning and Geometry , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[16]  Nicolas Pugeault,et al.  Taking the Scenic Route to 3D: Optimising Reconstruction from Moving Cameras , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[17]  Andrew W. Fitzgibbon,et al.  KinectFusion: Real-time dense surface mapping and tracking , 2011, 2011 10th IEEE International Symposium on Mixed and Augmented Reality.

[18]  Michael Gassner,et al.  SVO: Semidirect Visual Odometry for Monocular and Multicamera Systems , 2017, IEEE Transactions on Robotics.

[19]  Jörg Stückler,et al.  From monocular SLAM to autonomous drone exploration , 2016, 2017 European Conference on Mobile Robots (ECMR).

[20]  Nobuyuki Kita,et al.  3D simultaneous localisation and map-building using active vision for a robot moving on undulating terrain , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.

[21]  Matthias Nießner,et al.  BundleFusion , 2016, TOGS.

[22]  Stefan Leutenegger,et al.  ElasticFusion: Dense SLAM Without A Pose Graph , 2015, Robotics: Science and Systems.

[23]  Pat Hanrahan,et al.  Submodular Trajectory Optimization for Aerial 3D Scanning , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[24]  Tim Weyrich,et al.  Real-Time 3D Reconstruction in Dynamic Scenes Using Point-Based Fusion , 2013, 2013 International Conference on 3D Vision.

[25]  YangQuan Chen,et al.  Optimal Mobile Sensing and Actuation Policies in Cyber-physical Systems , 2011 .

[26]  Chieh-Chih Wang,et al.  Mapping and Localization in 3D Environments Using a 2D Laser Scanner and a Stereo Camera , 2012, J. Inf. Sci. Eng..

[27]  Juan D. Tardós,et al.  Visual-Inertial Monocular SLAM With Map Reuse , 2016, IEEE Robotics and Automation Letters.

[28]  Radu Bogdan Rusu,et al.  3D is here: Point Cloud Library (PCL) , 2011, 2011 IEEE International Conference on Robotics and Automation.