AirAnchor improves drone navigation by combining local and global spatial data
AirAnchor: Bridging Local and Global Spatial Information for Zero-Shot Aerial Vision-and-Language Navigation
Computer Vision and Pattern RecognitionArtificial IntelligenceRobotics
Summary
Drones need to understand both nearby surroundings and big-picture maps to follow spoken directions in cities accurately. The researchers created AirAnchor, a new system that connects close-up visuals and long-term landmarks, helping drones navigate better without training on new tasks. It uses three parts: one picks out important local spots, another remembers key objects over time, and the last uses both to make smart navigation choices. Tests show AirAnchor works better than previous methods that only focus on one type of information.
What this means in practice
- •For drone navigation engineers: Enhance drone autonomy for following natural language routes through complex city environments by using combined local and global spatial cues.
- •For robotics platform developers: Incorporate spatial anchor-based memory systems to improve zero-shot navigation capabilities in aerial robots operating in urban areas.
Authors
Shanwei Fan, Bin Zhang, Zhiwei Xu, Yingxuan Teng, Siqi Dai, Lin Cheng, Guoliang Fan
Abstract
Aerial Vision-and-Language Navigation requires drones to follow natural-language instructions and navigate through complex urban environments. Accurate navigation relies on both local and global spatial information, which support immediate action grounding and long-horizon path planning, respectively. However, existing zero-shot methods typically operate at a single spatial scale, relying either on local representations constructed online from current observations or on global memories built offline from historical experience. To address this limitation, we propose AirAnchor, a new paradigm that bridges local and global spatial information through spatial anchors and integrates both into a shared navigation framework, enabling comprehensive spatial grounding for decision-making. AirAnchor consists of three core components: (1) Query-Driven Spatial Anchor Grounding, which identifies decision-relevant anchors from visual observations and organizes them into local spatial representations; (2) Persistent Object Spatial Memory, which incrementally maintains an object knowledge base as persistent global spatial memory and retrieves landmark-related spatial priors; and (3) a Spatially-Informed Navigation Agent, which explicitly integrates both local and global spatial information into an agentic framework for decision-making. Extensive experiments on AerialVLN demonstrate that AirAnchor substantially outperforms existing zero-shot baselines, validating the effectiveness and efficiency of the proposed paradigm.