Shuo Sun, Malcolm Mielle, Achim J. Lilienthal and Martin Magnusson
Proceedings of 2024 IEEE International Conference on Robotics and Automation (ICRA)
Abstract
Neural implicit surface representations are currently receiving a lot of interest as a means to achieve high-fidelity surface reconstruction at a low memory cost, compared to traditional explicit representations. However, state-of-the-art methods still struggle with excessive memory usage and non-smooth surfaces. This is particularly problematic in large-scale applications with sparse inputs, as is common in robotics use cases. To address these issues, we first introduce a sparse structure, tri-quadtrees, which represents the environment using learnable features stored in three planar quadtree projections. Secondly, we concatenate the learnable features with a Fourier feature positional encoding. The combined features are then decoded into signed distance values through a small multilayer perceptron. We demonstrate that this approach facilitates smoother reconstruction with a higher completion ratio with fewer holes. Compared to two recent baselines, one implicit and one explicit, our approach requires only 10%–50% as much memory, while achieving competitive quality. The code is released on https://github.com/ljjTYJR/3QFP.
@INPROCEEDINGS{10610338,
author={Sun, Shuo and Mielle, Malcolm and Lilienthal, Achim J. and Magnusson, Martin},
booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},
title={3QFP: Efficient neural implicit surface reconstruction using Tri-Quadtrees and Fourier feature Positional encoding},
year={2024},
volume={},
number={},
pages={4036-4044},
keywords={Surface reconstruction;Costs;Codes;Memory management;Multilayer perceptrons;Encoding;Robots},
doi={10.1109/ICRA57147.2024.10610338}
}