Vision language models estimate uncertainty for safer robot navigation
Estimating Semantic Ambiguity via Gaussian Context Distributions for VLM-Driven Traversability Analysis
Robotics
Summary
Understanding complicated scenes is tough for robots, especially when the things they see can be interpreted in different ways. The authors of this paper developed a way to measure how unsure a robot is about what it sees by using math to represent possible meanings as a range rather than one answer. This helps the robot know which parts of the environment might be confusing or unsafe to cross. They tested their method on real outdoor scenes and showed it can highlight uncertain areas caused by tricky visuals or mixed surfaces. This approach could help make robot navigation more reliable and safer in the real world.
Autonomous navigationVision-Language Models (VLMs)Semantic ambiguityContextual uncertaintyGaussian Context DistributionTraversability estimationConceptual anchoringOpen-vocabulary predictionUncertainty quantificationScene understanding
Authors
Ramona Häuselmann, Mario A. V. Saucedo, Christoforos Kanellakis, George Nikolakopoulos
Abstract
Autonomous navigation in unstructured environments requires robust scene understanding, yet Vision-Language Models (VLMs) often suffer from semantic ambiguity, where conflicting predictions can lead to dangerous failures. To address this, we present a novel pipeline for vision-based traversability estimation that explicitly models contextual uncertainty. Our approach utilizes Conceptual Anchoring to ground open-vocabulary VLM predictions onto a continuous physical traversability scale. By formulating the model's responses as a Gaussian Context Distribution (GCD), we derive both a dense traversability map and a dense uncertainty map based on the statistical properties of the distribution. Experimental validation on the real-world GOOSE dataset demonstrates that our proposed uncertainty metric effectively correlates with sources of ambiguity, such as visual artifacts and mixed terrain overlap. The method exhibits competitive performance while offering the distinct advantage of providing statistical uncertainty estimates to address semantic ambiguity, enabling safer and more reliable autonomous behavior in complex outdoor settings.