WorldFly: A World-Model-Based Vision-Language-Action Model for UAV Navigation
This work addresses the challenge of robust UAV navigation under partial observability in dense urban environments, where existing VLA models struggle due to drastic viewpoint transitions.
WorldFly introduces a world-model-based VLA framework for UAV navigation that jointly generates future video predictions and actions via dual-branch coupled flow matching, outperforming baselines in dense urban environments with severe occlusions and sharp turns.
End-to-end Vision-Language-Action (VLA) models have shown promise in UAV navigation. However, existing approaches typically rely on historical observations to directly predict actions, often struggling in dense urban environments where severe occlusions and sharp turns result in drastic viewpoint transitions. We argue that the ability to "imagine" future states -- inherent in World Models -- is critical for robust decision-making under such partial observability. To address this, we construct a challenging Urban Canyon Traversal Benchmark, specifically designed to evaluate spatial understanding in scenarios characterized by severe occlusions and drastic viewpoint transitions. To this end, we propose WorldFly, a novel world-model-based VLA framework that employs a dual-branch coupled flow matching mechanism to jointly generate future video predictions and navigation actions, thereby explicitly guiding the agent's policy via spatial imagination. Extensive evaluations on our benchmark demonstrate that WorldFly outperforms other baselines, particularly in unseen environments, validating the effectiveness of integrating world models into embodied aerial agents.