TOMD: A Trail-based Off-road Multimodal Dataset for Traversable Pathway Segmentation under Challenging Illumination Conditions

2025-06-24

AI Summary

Detecting traversable paths along narrow forest trails in unstructured野外 environments remains challenging due to low illumination, high clutter, and path narrowness—conditions poorly addressed by existing urban or wide-off-road datasets and models. Method: We introduce TOMD, the first multimodal野外 dataset specifically designed for narrow forest trails, featuring synchronized 128-line LiDAR, stereo images, GNSS/IMU, and ambient light sensor data. We further propose a lighting-aware dynamic multi-scale fusion semantic segmentation network, incorporating a novel three-stage feature fusion strategy (early, cross, hybrid) and illumination-adaptive scale modeling. Contribution/Results: Experiments demonstrate substantial improvements in trail segmentation under low-light conditions (mIoU ↑12.6%), validating the robustness of multimodal fusion under extreme illumination. TOMD is publicly released to establish a new benchmark for野外 autonomous navigation.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

Detecting traversable pathways in unstructured outdoor environments remains a significant challenge for autonomous robots, especially in critical applications such as wide-area search and rescue, as well as incident management scenarios like forest fires. Existing datasets and models primarily target urban settings or wide, vehicle-traversable off-road tracks, leaving a substantial gap in addressing the complexity of narrow, trail-like off-road scenarios. To address this, we introduce the Trail-based Off-road Multimodal Dataset (TOMD), a comprehensive dataset specifically designed for such environments. TOMD features high-fidelity multimodal sensor data -- including 128-channel LiDAR, stereo imagery, GNSS, IMU, and illumination measurements -- collected through repeated traversals under diverse conditions. We also propose a dynamic multiscale data fusion model for accurate traversable pathway prediction. The study analyzes the performance of early, cross, and mixed fusion strategies under varying illumination levels. Results demonstrate the effectiveness of our approach and the relevance of illumination in segmentation performance. We publicly release TOMD at https://github.com/yyyxs1125/TMOD to support future research in trail-based off-road navigation.

Problem

• Detecting traversable pathways in challenging off-road environments
• Addressing gaps in existing datasets for narrow trail scenarios
• Evaluating fusion strategies under varying illumination conditions

Innovation

• High-fidelity multimodal sensor data fusion
• Dynamic multiscale traversable pathway prediction
• Illumination-aware segmentation performance analysis