cs.RONov 2, 2026

MS-DOTT: A Multi-Modal Dataset for Off-Road Traversability Estimation

Authors: Shivani Subramanyan, Saksham Sharma, Samahith SA, Ahaan Kotian, Suresh Sundaram

Organizations: Dept. of Aerospace Engineering, Indian Institute of Science, Bengaluru, India

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Autonomous robots are gaining traction in off-road environments, driven by their rising demand in agriculture, defense, and search-and-rescue operations. However, these environments are inherently unpredictable and navigation remains a formidable challenge given the complex environmental conditions including challenging terrain and varying illumination. Off-road autonomy requires robust perception across multiple spatial scales and imaging modalities, yet most sensing setups struggle to capture both far-field environmental context and near-field terrain detail simultaneously across various illumination conditions. To facilitate development of robust off-road perception and traversability estimation systems, we propose MS-DOTT (Multi-Sensor Dataset for Off-road Terrain Traversability) a comprehensive multi-modal off-road dataset with long-range 3D sensing and short-range, fine-grained perception to facilitate high-fidelity terrain understanding by combining proprioceptive signals for tracking vehicle dynamics with exteroceptive imaging and pointcloud data. The dataset includes trajectories recorded across nominal lighting, low-light, and no-light conditions with diverse 2D and 3D sensing for both long-range large-scale and short-range high-resolution sensing, as well as 4D RADAR measurements that provide velocity estimates for detected pointclouds. We demonstrate the utility of the dataset by adapting a SOTA multi-sensor fusion method for the purpose of traversability estimation. This dataset will aid the robotics community in advancing research and development of methods for traversability estimation, off-road perception and multi-sensor fusion in unstructured environments.

Figures & tables

Fig. 1 : MS-DOTT dataset overview. MS-DOTT consists of various complementary sensor data which allows for comprehensive terrain traversability estimation in both day and night conditions.
DatasetThermalRGBToFLiDARRadarIMUStereoPlatformIllumination
RUGD [ 13 ]✗✓✗✓✗✓✗UGVDay
RELLIS-3D [ 14 ]✗✓✗✓✗✓✓UGVDay
CaT [ 22 ]✗✓✗✓✗✓✗ATVDay & Night
TartanDrive 2.0 [ 21 ]✗✓✗✓✗✓✓ATVDay
ORFD [ 19 ]✗✓✗✓✗✓✗CarDay & Night
GOOSE [ 15 ]✓✓✗✓✓✓✗CarDay
TABLE I : Comparison between Off-Road Datasets
Fig. 2 : MS-DOTT Data Collection Bench.
Sensor NameOrientationHzRangeFOV (HxV)Mean points per scanPurpose
Livox HAPForward10 Hz150m120° ×\times 25°44.91 ×103\times 10^{3}Long-range environmental awareness, predictive path planning, and place recognition.
Livox MID360Downward10 Hz40m360° ×\times 52°19.83 ×103\times 10^{3}Mid-range mapping, traditional terrain traversability baseline.
ToF: Lucid Helios2+Downward15 Hz5m (8.3m max)67° ×\times 51°233.16 ×103\times 10^{3}Short-range, fine-grained terrain understanding and micro-obstacle detection.
ARS548 RDI RADARForward10 Hz300m120° ×\times 30°3 ×102\times 10^{2}Weather agnostic perception
TABLE II : 3D Sensor Specifications and Functional Roles within the MS-DOTT Dataset Platform
Sensor NameModalityHzResolutionFOVPurpose
Trimer TRA_L6D_ZDNThermal30 Hz640x48030° ×\times 23°Short-range terrain understanding using thermal perception in low/no light.
FLIR Blackfly-SRGB30 Hz1920x108048° ×\times 36°Short-range, baseline RGB data for terrain understanding in daytime.
Lucid Vision Labs TDR054SHDR RGB10 Hz1920x108070° ×\times 50°Short-range, fine-grained terrain understanding robust to low-light and glares.
ZED2 Stereo CameraRGB + Stereo depth10 Hz1280x960110° ×\times 70°Maximum 3D detection range of 20m, auxiliary sensor for short-range terrain understanding.
TABLE III : Vision Sensor Specifications and Functional Roles within the MS-DOTT Dataset Platform
Sensor NameModalityHzSpecsPurpose
Xsens MTi-680GINS400 Hz9-axis INS, L1/L2 RTK GNSS receiver, 0.2° roll/pitch accuracy, 0.5° heading accuracyLocalization and vehicle inertial data
TABLE IV : Proprioceptive Sensor Specifications and Functional Roles within the MS-DOTT Dataset Platform
Varying TerrainIlluminationTrajectoriesDistanceRuntimeCamera FramesPointclouds
Muddy terrain, grassy fields, loose gravel paths, and undulating landscapesDay and Night113000m90 mins432,000243,000
TABLE V : Data Collection
Fig. 3 : Performance comparison of different visual sensors across diverse lighting environments. All costmaps (Bottom) are generated by fusion of Mid360 LiDAR and corresponding camera inputs (Up). (a) shows the effect of glares and shadows while (b) shows performance in nighttime conditions. Red arrow depicts pose of ego robot.
Fig. 4 : Performance comparison of different LiDAR sensors for multi-scale traversability estimation. The image on the left displays the reference FLIR camera frame. The images on the right illustrate the pointclouds and corresponding costmaps arranged vertically from top to bottom for the Helios, MID360, and HAP sensors, respectively. Red arrow depicts pose of ego robot.
More in this topic

Multimodal Dataset

See all 64 papers

Explore similar work

CardsList

  1. STONE Dataset: A Scalable Multi-Modal Surround-View 3D Traversability Dataset for Off-Road Robot Navigation

    Mar 10, 2026Konyul Park, Daehun Kim, Jiyong Oh +7Off-Road Robot NavigationRobot Traversability Estimation

    Interdisciplinary Program in Artificial Intelligence, Seoul National University, Seoul, 08826, Korea. · Department of Electrical and Computer Engineering, Seoul National University, Seoul, 08826, Korea.

  2. GO: The Great Outdoors Multimodal Dataset

    Jan 31, 2025Peng Jiang, Kasi Viswanath, Akhil Nagariya +8Mobile RoboticsRobotic Perception

    Mike Walker ’66 Department of Mechanical Engineering, Texas A&M University, College Station, TX, USA. · DEVCOM Army Research Laboratory, Adelphi, MD, USA. · University of Texas at Austin Center for Autonomy, Austin, TX, USA.

  3. Learning to Drive on Mars: Visual Multimodal Traversability Estimation for Off-World Navigation

    Sep 21, 2026Darren Chiu, Cole Wilson, Andrei Tumbar +2Off-Road Robot NavigationSpace Robotics

    University of Southern California, Los Angeles, CA 90089, USA. · Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA. · Washington State University, Pullman, WA 99164, USA.