Geo-VLA: Geometry-Aware Vision-Language-Action Planning via Internalization of Map Semantics

2026-08-18

AI Summary

本文提出Geo-VLA框架，通过学习几何感知视觉表示来增强VLA模型，以解决复杂驾驶环境中图像表示不足的问题。

Abstract

Vision-language-action (VLA) models have advanced end-to-end autonomous driving by leveraging foundation models for semantic reasoning and long-tail generalization. However, their planning performance remains limited in complex driving environments because image-only representations inadequately capture planning-relevant road geometry and topology. In this paper, we propose Geo-VLA, a plug-and-play framework that enhances VLA models by learning geometry-aware visual representations. During training, Geo-VLA internalizes geometric map semantics to strengthen road-structure representations, while requiring no HD maps or additional lane information during inference. To support this approach, we introduce Geo-QA, a geometry-focused question-answering dataset that injects road geometry into vision-language representations through contrastive learning and instruction tuning. Experiments on NAVSIM v1 demonstrate that Geo-VLA consistently improves VLA planners with distinct action-generation architectures, achieving 92.1 PDMS and establishing a new state-of-the-art among single-camera VLA planners.

Problem

• Vision-language-action
• planning performance
• complex driving environments
• road geometry
• topology

Innovation

• geometry-aware visual representations
• internalization of geometric map semantics
• Geo-QA dataset