Splatblox: Traversability-Aware Gaussian Splatting for Outdoor Robot Navigation

2025-11-23

AI Summary

Outdoor robots face significant challenges in real-time differentiating traversable vegetation (e.g., tall grass) from rigid obstacles (e.g., tree trunks) amid dense foliage, irregular obstacles, and complex terrain—hindering robust long-range navigation. To address this, we propose a semantic-geometric fusion approach for online Euclidean Signed Distance Field (ESDF) mapping with traversability awareness. This work introduces Gaussian splatting to robotics navigation for the first time, jointly leveraging RGB semantic segmentation and LiDAR point clouds to construct an ESDF that simultaneously ensures geometric fidelity and semantic discriminability. The system enables real-time map updates and tight multi-sensor fusion. Experimental validation on quadrupedal and wheeled platforms demonstrates: >50% improvement in task success rate, 40% reduction in freezing behaviors, 5% shorter path length, up to 13% faster arrival time, and reliable autonomous navigation over 100-meter distances.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Natural Language Processing: Safety and Robustness

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Location- and context-aware Web and WoT applications and services
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

We present Splatblox, a real-time system for autonomous navigation in outdoor environments with dense vegetation, irregular obstacles, and complex terrain. Our method fuses segmented RGB images and LiDAR point clouds using Gaussian Splatting to construct a traversability-aware Euclidean Signed Distance Field (ESDF) that jointly encodes geometry and semantics. Updated online, this field enables semantic reasoning to distinguish traversable vegetation (e.g., tall grass) from rigid obstacles (e.g., trees), while LiDAR ensures 360-degree geometric coverage for extended planning horizons. We validate Splatblox on a quadruped robot and demonstrate transfer to a wheeled platform. In field trials across vegetation-rich scenarios, it outperforms state-of-the-art methods with over 50% higher success rate, 40% fewer freezing incidents, 5% shorter paths, and up to 13% faster time to goal, while supporting long-range missions up to 100 meters. Experiment videos and more details can be found on our project page: https://splatblox.github.io

Problem

• Real-time autonomous navigation in complex outdoor environments
• Distinguishing traversable vegetation from rigid obstacles
• Constructing traversability-aware maps using fused sensor data

Innovation

• Fuses RGB and LiDAR via Gaussian Splatting
• Builds traversability-aware ESDF with semantics
• Enables real-time semantic reasoning for navigation