TAIL: A Terrain-Aware Multi-Modal SLAM Dataset for Robot Locomotion in Deformable Granular Environments

TL;DR

A Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains is proposed and several state-of-the-art SLAM methods are benchmarked against ground truth and provide performance validations.

Abstract

Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported. All associated resources are accessible upon request at \url{https://tailrobot.github.io/}.

Figures (8)

• Figure 1: (a) Terrain hazards encountered by NASA Spirit rover smith2009new. To study terrain-related hazards, we build a versatile sensor suite to record different data sequences in challenging soft, granular terrains. Both (b) wheeled robot and (c) quadruped robot platforms are used for carrying the designed sensor suite during terrain traversals of diverse sandy terrains, where robot locomotion can become quite challenging in certain scenarios.
• Figure 2: The specification for the multi-modal sensor suite. (a) CAD model and hardware components of the setup. (b) The mounted device's size and relevant coordinate frames of the sensor system. High-resolution 3D LiDAR is located at the center and an RGBD camera is installed below. Two frames and RGBD cameras are mounted on the left and right sides, complemented by an internally placed high-precision IMU and a rear-installed RTK-GPS. All these sensors are mounted on the same rigid aluminum frame and can be mounted on various robot platforms, e.g., see Fig. \ref{['fig:intro_picture']}. Thus, their spatial relations have no deviation, making it an effective test tool.
• Figure 3: The description of sensor hardware time synchronization process. (a) Triggering implementation using different marked signals. (b) Visualization of the triggering pulses on the oscilloscope. (c) Validating the accuracy of hardware triggering using the Light Emitting Diode (LED) board.
• Figure 4: Sampled challenging scenarios shows the in-sequence diversity. We capture a variety of terrain characteristics with stereo frame cameras, three RGB-D cameras, and LiDAR.
• Figure 5: Overview of our field experiments for the TAIL data collection. (a) Acquisition locations are depicted on the satellite map. (b) Coarse, steep beach from shanhaili. (c) Fine, smooth beach from wanke.