The ability to associate the current location with previously visited places is an essential aspect of autonomous ground robots. Unstructured environments such as planetary surfaces pose a significant challenge for robots because their terrain is less distinctive. Meanwhile, traversability must be analyzed simultaneously for safe navigation. In the past, place recognition research has rarely considered traversability analysis despite its significance. This is because the structural information of terrains becomes quickly implicit during the encoding process. This paper provides a method that explicitly addresses both problems: place recognition and traversability analysis. It proposes a discrete Fourier transform (DFT) to represent the frequency components embedded in ground curvature, which underlies both concepts. Our place recognition function demonstrates excellent performance in extensive experiments using challenging planetary & urban datasets while estimating traversability.
The aggregation of frequencies can represent gentle ground curvature and urban scenes in a more informative way than approaches describing salient structures. The proposed max-sampling based approach enables robust place recognition against opposite revisitation.
The traversability analysis process aims to extract the regions of the point cloud forming each frequency that is regarded as a degree of traversability in this study. We highlight points constituting the target frequency by selecting smoothened surfaces reconstructed by low-pass filtering.
We demonstrate the high performance of the proposed method in unstructured and urban environments, where it exhibits superiority over baseline methods w. r. t. accuracy and robustness against noises. Furthermore, unlike existing descriptors, the proposed method simultaneously analyzes traversability based on the frequency components embedded in the ground.
MADMAX-E sequence
MADMAX-G sequence
LRNT-1 sequence
LRNT-2 sequence
Under different scales of uniform noise
Under different sparsities
The left shows the appearance of the target terrain. The middle shows overlaid masks, which indicate degrees of danger. The right shows an example of path planning that enables safe navigation based on segmented regions.
We appreciate the following works for releasing the event camera datasets used in our work:
MADMAX
L. Meyer, M. Smíšek, A. F. Villacampa, L. O. Maza, D. A. Medina, M. J. Schuster, F. Steidle, M. Vayugundla, M. G. M¨uller, B. Rebele, A. Wedler, and R. Triebel. The MADMAX data set for visual-inertial rover navigation on Mars. Journal of Field Robotics, 2021.
LRNT
M. Vayugundla, F. Steidle, M. Smisek, M. Schuster, K. Bussmann, and A. Wedler. Datasets of Long Range Navigation Experiments in a Moon Analogue Environent on Mount Etna. 50th International Symposium on Robotics, 2018.
S3LI
R. Giubilato, W. Stürzl, A. Wedler and R. Triebel. Challenges of SLAM in extremely unstructured environements: the DLR Planetary Stereo, Solid State LiDAR, Inertial Dataset. IEEE Robotics and Automation Letters, 2022.
This work was partially supported by JSPS KAKENHI under Grant No. 23KJ0170 and JST PRESTO Grant Number JPMJPR22C4.
@inproceedings{ayumi2024sdft,
author = {Ayumi, Umemura and Ken, Sakurada and Masaki, Onishi and Kazuya, Yoshida},
title = {SDFT: Structual Discrete Fourier Transform for Place Recognition and Traversability Analysis},
journal = {IROS},
year = {2024}
}