Challenges of SLAM in Extremely Unstructured Environments: The DLR Planetary Stereo, Solid-State LiDAR, Inertial Dataset

Abstract

We present the DLR Planetary Stereo, Solid-State LiDAR, Inertial (S3LI) dataset, recorded on Mt. Etna, Sicily, an environment analogous to the Moon and Mars, using a handheld sensor suite with attributes suitable for implementation on a space-like mobile rover. The environment is characterized by challenging conditions regarding both the visual and structural appearance: severe visual aliasing poses significant limitations to the ability of visual SLAM systems to perform place recognition, while the absence of outstanding structural details, joined with the limited Field-of-View of the utilized Solid-State LiDAR sensor, challenges traditional LiDAR SLAM for the task of pose estimation using point clouds alone. With this data, that covers more than 4 kilometers of travel on soft volcanic slopes, we aim to: 1) provide a tool to expose limitations of state-ofthe-art SLAM systems with respect to environments, which are not present in widely available datasets and 2) motivate the development of novel localization and mapping approaches, that rely efficiently on the complementary capabilities of the two sensors. The dataset is accessible at the following url: https://rmc.dlr.de/s3li_dataset

Precise pose estimation of the NASA Mars 2020 Perseverance rover through a stereo‐vision‐based approach

“…In the next years, novel highly movable rovers (speed >20 cm/s) will be launched, such as the lunar NASA VIPER rover (Utz & Fluckiger, 2021), and the advanced moving capabilities of these assets will enable the rovers to revisit the same areas multiple times. Simultaneous Localization and Mapping approaches could then take advantage of loop-closure detection to globally adjust the rover's trajectory while building a consistent map of the operational environment (Giubilato et al, 2022;Hidalgo-Carrió et al, 2018), paving the way for an effective human-robotic cooperative framework. To compare the localization performances of different VO-based localization methods, we processed the sequence of images acquired on sol 72 by also using a 3D-to-2D VO scheme (Supporting Information Section 3).…”

Localization for Autonomous Vehicles in Underground Mines

“…Feature detection methods are often designed for well-defined structures found in classical indoor environments. Moreover, the high presence of aliasing, with many tunnels and intersections that are locally very similar [45], are causing problems to get a globally consistent estimate and to discover correct loop-closures in the slam application. A loop-closure occur when a previously visited area is revisited and is of great importance for the performance of slam.…”

“…Position information can also be used for mine maintenance by e.g., logging where break-downs or failures often occur. While localization and mapping of an environment theoretically and conceptually often are considered solved problems, challenges remain in how these systems can autonomously operate long-term, over large areas where the environment is inevitably changing during the operation time [21,45]. This thesis focuses on how a highly accurate position estimate, with high availability in time, and robust towards changes in the surroundings, can be provided in the special environment of an underground mine.…”

A Superpoint Neural Network Implementation for Accurate Feature Extraction in Unstructured Environments

“…Moreover, although there is a plethora of datasets for evaluation and training of feature extraction algorithms such as HPatches (Balntas, 2017), Aachen (Sattler, 2008), COCO (Lin, 2014), Google Landmarks (Noh, 2017), etc, they focus on urban, indoor or vegetated environments while the datasets which represent unstructured scenes are quite few and they are designed mainly for SLAM (Simultaneous Localization and Mapping) evaluation (Meyer, 2021, Furgale, 2012, Giubilato, 2022, Hewitt, 2018 and not for training or keypoint detector or descriptor evaluation.…”