![[2602.15875] Fly0: Decoupling Semantic Grounding from Geometric Planning for Zero-Shot Aerial Navigation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.15875] Fly0: Decoupling Semantic Grounding from Geometric Planning for Zero-Shot Aerial Navigation
Summary
The paper presents Fly0, a novel framework that separates semantic grounding from geometric planning to enhance zero-shot aerial navigation, improving performance in unstructured environments.
Why It Matters
This research addresses significant challenges in Visual-Language Navigation (VLN) by improving the efficiency and accuracy of aerial navigation systems. By decoupling semantic reasoning from geometric planning, Fly0 enhances the robustness of navigation in complex environments, which is crucial for applications in robotics and autonomous systems.
Key Takeaways
- Fly0 improves aerial navigation by decoupling semantic understanding from geometric planning.
- The framework operates through a three-stage pipeline, enhancing robustness and stability.
- Fly0 achieves over a 20% increase in Success Rate and a 50% reduction in Navigation Error in unstructured environments.
- The method reduces computational overhead by eliminating continuous inference.
- Extensive experiments validate Fly0's performance against state-of-the-art baselines.
Computer Science > Robotics arXiv:2602.15875 (cs) [Submitted on 2 Feb 2026] Title:Fly0: Decoupling Semantic Grounding from Geometric Planning for Zero-Shot Aerial Navigation Authors:Zhenxing Xu, Brikit Lu, Weidong Bao, Zhengqiu Zhu, Junsong Zhang, Hui Yan, Wenhao Lu, Ji Wang View a PDF of the paper titled Fly0: Decoupling Semantic Grounding from Geometric Planning for Zero-Shot Aerial Navigation, by Zhenxing Xu and 7 other authors View PDF HTML (experimental) Abstract:Current Visual-Language Navigation (VLN) methodologies face a trade-off between semantic understanding and control precision. While Multimodal Large Language Models (MLLMs) offer superior reasoning, deploying them as low-level controllers leads to high latency, trajectory oscillations, and poor generalization due to weak geometric grounding. To address these limitations, we propose Fly0, a framework that decouples semantic reasoning from geometric planning. The proposed method operates through a three-stage pipeline: (1) an MLLM-driven module for grounding natural language instructions into 2D pixel coordinates; (2) a geometric projection module that utilizes depth data to localize targets in 3D space; and (3) a geometric planner that generates collision-free trajectories. This mechanism enables robust navigation even when visual contact is lost. By eliminating the need for continuous inference, Fly0 reduces computational overhead and improves system stability. Extensive experiments in simulation and real-wor...
