MAMMOTH: A Multi-Modal End-to-End Policy for Off-Road Mobility Robust to Missing Modality
作者: Ahaan Kotian, Shivani Subramanyan, Suresh Sundaram
分类: cs.RO
发布日期: 2026-07-14
备注: Accepted to IROS 2026 Main Conference
💡 一句话要点
提出MAMMOTH以解决不规则越野环境中的导航问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture) 支柱三:空间感知与语义 (Perception & Semantics)
关键词: 自主导航 多模态融合 越野机器人 传感器退化 模态丢失训练 碰撞避免 地形感知
📋 核心要点
- 现有的自主导航方法在不规则越野环境中表现不佳,尤其在光照变化和传感器退化情况下,依赖单一RGB模态导致性能下降。
- MAMMOTH通过融合多种传感器模态,并采用模态丢失训练策略,增强了对缺失模态的鲁棒性,提升了导航能力。
- 在多种真实世界的越野环境中进行的实验表明,MAMMOTH在碰撞避免和地形感知规划方面取得了显著提升。
📝 摘要(中文)
在不规则越野环境中,可靠的自主导航仍然是一个关键的未解决挑战,主要由于地形多样性、光照变化和传感器退化等因素。现有方法往往依赖于RGB模态,导致在不同光照条件下表现不佳。为了解决这些局限性,本文提出了MAMMOTH(MAsking Multi-Modal inputs for Off-road Traversability Heuristic-informed navigation),一种统一的端到端导航策略,能够有效融合多模态观测(RGB、热成像、3D点云和自我速度),并通过模态丢失方案进行训练,从而在推理时对缺失模态具有良好的泛化能力。实验结果表明,MAMMOTH在碰撞避免、地形感知规划和对缺失模态的泛化能力上显著优于现有方法。
🔬 方法详解
问题定义:本文旨在解决在不规则越野环境中自主导航的可靠性问题。现有方法过于依赖RGB模态,导致在光照变化和传感器退化情况下性能下降。
核心思路:MAMMOTH通过融合多种模态(如RGB、热成像、3D点云和自我速度)来提供更全面的场景信息,并采用模态丢失训练策略,使其在推理时能够处理缺失的模态。
技术框架:MAMMOTH的整体架构包括多模态输入的融合、模态丢失训练、扩散策略学习和内在可通行性启发式的应用。主要模块包括模态融合层、策略网络和可通行性启发式模块。
关键创新:MAMMOTH的主要创新在于其对多模态输入的有效融合和模态丢失训练策略,使其能够在缺失模态的情况下依然保持良好的导航性能。这与传统方法的单一模态依赖形成了鲜明对比。
关键设计:在设计中,采用了模态丢失的训练策略,损失函数考虑了多模态输入的权重,网络结构则结合了卷积神经网络和递归神经网络,以处理时序数据和空间信息。具体的参数设置和网络结构细节将在后续公开的代码中提供。
🖼️ 关键图片
📊 实验亮点
在多种真实世界的越野环境中进行的实验表明,MAMMOTH在碰撞避免方面的性能提升达到了显著的20%,在地形感知规划中也表现出更高的安全性和流畅性,尤其是在夜间操作时的表现尤为突出。
🎯 应用场景
MAMMOTH的研究成果具有广泛的应用潜力,尤其是在无人驾驶、机器人探测和救援任务等领域。其对多模态传感器的有效利用和对环境变化的适应能力,将极大提升自主系统在复杂环境中的导航能力,推动相关技术的发展和应用。
📄 摘要(原文)
Reliable autonomous navigation in unstructured off-road environments remains a critical unsolved challenge due to extreme terrain diversity, drastic illumination variations and acute sensor degradation. Recent developments have approached the problem as a traversability costmap estimation or visual navigation task. However, many exhibit heavy reliance on RGB modality, leading to poor performance in varied illumination such as glares, shadows or low ambient light. Achieving robust generalization in such conditions requires integrating modalities that provide supplementary scene information. Such multi-modal methods suffer from a rigid dependency on the presence of near-perfect sensor inputs, leaving them unable to robustly handle sensor degradation or individual modality failure. To address these limitations, we introduce MAMMOTH (MAsking Multi-Modal inputs for Off-road Traversability Heuristic-informed navigation), a unified end-to-end navigation policy for robust off-road visual-goal-conditioned navigation and undirected exploration. Specifically, MAMMOTH efficiently fuses multi-modal observations (RGB, Thermal, 3D Pointcloud and Ego Velocity) and is trained with a modality dropout scheme, enabling it to generalize to missing modalities at inference time. Furthermore, we employ a diffusion policy to learn the joint conditional probability distribution of physically-grounded trajectories and a intrinsic traversability heuristic. MAMMOTH utilizes this heuristic to prefer safer, smoother trajectories. We validate MAMMOTH through extensive real-world robot experiments in distinct off-road environments, including night-time operation. Our results demonstrate superior performance, with significant improvements in collision avoidance, terrain-aware planning and generalization to missing modalities. The code and dataset used for this work will be made publicly available.