NeuralBlox: Real-Time Neural Representation Fusion for Robust Volumetric Mapping

被引:11
|
作者
Lionar, Stefan [1 ]
Schmid, Lukas [1 ]
Cadena, Cesar [1 ]
Siegwart, Roland [1 ]
Cramariuc, Andrei [1 ]
机构
[1] Swiss Fed Inst Technol, Autonomous Syst Lab, Zurich, Switzerland
关键词
D O I
10.1109/3DV53792.2021.00135
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present a novel 3D mapping method leveraging the recent progress in neural implicit representation for 3D reconstruction. Most existing state-of-the-art neural implicit representation methods are limited to object-level reconstructions and can not incrementally perform updates given new data. In this work, we propose a fusion strategy and training pipeline to incrementally build and update neural implicit representations that enable the reconstruction of large scenes from sequential partial observations. By representing an arbitrarily sized scene as a grid of latent codes and performing updates directly in latent space, we show that incrementally built occupancy maps can be obtained in real-time even on a CPU. Compared to traditional approaches such as Truncated Signed Distance Fields (TSDFs), our map representation is significantly more robust in yielding a better scene completeness given noisy inputs. We demonstrate the performance of our approach in thorough experimental validation on real-world datasets with varying degrees of added pose noise.
引用
收藏
页码:1279 / 1289
页数:11
相关论文
共 50 条
  • [1] Fasciculography: Robust Prior-Free Real-Time Normalized Volumetric Neural Tract Parcellation
    Ho, Hon Pong
    Wang, Fei
    Papademetris, Xenophon
    Blumberg, Hilary P.
    Staib, Lawrence H.
    IEEE TRANSACTIONS ON MEDICAL IMAGING, 2012, 31 (02) : 217 - 230
  • [2] Volumetric 3D Mapping in Real-Time on a CPU
    Steinbruecker, Frank
    Sturm, Juergen
    Cremers, Daniel
    2014 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA), 2014, : 2021 - 2028
  • [3] Robust Real-Time Volumetric Imaging Based On One Single Projection
    Xu, Y.
    Yan, H.
    Ouyang, L.
    Wang, J.
    Zhou, L.
    Jiang, S.
    Jia, X.
    MEDICAL PHYSICS, 2014, 41 (06) : 522 - 522
  • [4] RGB-D Fusion: Real-time Robust Tracking and Dense Mapping with RGB-D Data Fusion
    Lee, Seong-Oh
    Lim, Hwasup
    Kim, Hyoung-Gon
    Ahn, Sang Chul
    2014 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS 2014), 2014, : 2749 - 2754
  • [5] Robust Real-Time Tracking of Multiple Objects by Volumetric Mass Densities
    Possegger, Horst
    Sternig, Sabine
    Mauthner, Thomas
    Roth, Peter M.
    Bischof, Horst
    2013 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2013, : 2395 - 2402
  • [6] Saliency-aware Real-time Volumetric Fusion for Object Reconstruction
    Yang, Sheng
    Chen, Kang
    Liu, Minghua
    Fu, Hongbo
    Hu, Shi-Min
    COMPUTER GRAPHICS FORUM, 2017, 36 (07) : 167 - 174
  • [7] Motion2Fusion: Real-time Volumetric Performance Capture
    Dou, Mingsong
    Davidson, Philip
    Fanello, Sean Ryan
    Khamis, Sameh
    Kowdle, Adarsh
    Rhemann, Christoph
    Tankovich, Vladimir
    Izadi, Shahram
    ACM TRANSACTIONS ON GRAPHICS, 2017, 36 (06):
  • [8] Real-time Indoor Mapping by fusion of structured light sensors
    Kusari, Arpan
    Pan, Zhigang
    Glennie, Craig
    2014 UBIQUITOUS POSITIONING INDOOR NAVIGATION AND LOCATION BASED SERVICE (UPINLBS), 2014, : 213 - 219
  • [9] Real-Time Dense Visual SLAM with Neural Factor Representation
    Wei, Weifeng
    Wang, Jie
    Xie, Xiaolong
    Liu, Jie
    Su, Pengxiang
    ELECTRONICS, 2024, 13 (16)
  • [10] Deep Real-time Volumetric Rendering Using Multi-feature Fusion
    Hu, Jinkai
    Yu, Chengzhong
    Liu, Hongli
    Yan, Lingqi
    Wu, Yiqian
    Jin, Xiaogang
    PROCEEDINGS OF SIGGRAPH 2023 CONFERENCE PAPERS, SIGGRAPH 2023, 2023,