Search papers, labs, and topics across Lattice.
This paper introduces AirForesight, a novel framework for Unmanned Aerial Vehicle Vision-Language Navigation (UAV-VLN) that enhances spatial reasoning by integrating current-map representation with future-trajectory prediction. By employing structured causal attention and a cross-space planning consistency loss, AirForesight effectively encodes both present scene structure and future motion intent, allowing for improved navigation in complex environments. Experimental results on OpenUAV and AerialVLN-S show significant performance gains, validating the framework's effectiveness in predicting 3D waypoints based on language instructions and multi-view observations.
AirForesight achieves superior UAV navigation by seamlessly integrating current spatial knowledge with future trajectory predictions, outperforming traditional methods that lack explicit scene grounding.
Unmanned Aerial Vehicle Vision-Language Navigation (UAV-VLN) requires agents to follow language instructions, infer spatial structure from sparse multi-view observations, and execute feasible 3D motion in complex outdoor environments. Despite recent progress with large language models, most existing methods still map vision-language inputs directly to actions, providing limited explicit scene grounding and future-aware spatial reasoning. We propose AirForesight, a current-to-future spatial map imagination framework for UAV-VLN. AirForesight first learns a structured current-map representation from multi-view observations. This representation is jointly supervised by current-map reconstruction and future-trajectory prediction, encouraging it to encode both present scene structure and future motion intent. Under structured causal attention, the current spatial knowledge is propagated to future-map reasoning, and the resulting current and future representations are aggregated to predict the next 3D waypoint. To make spatial imagination more relevant to navigation, we introduce a cross-space planning consistency loss that encourages directional agreement between the predicted map-space trajectory and the expert action direction derived from the ground-truth waypoint displacement. Experiments on OpenUAV and AerialVLN-S, together with extensive ablations, demonstrate strong performance and support the effectiveness and stability of the proposed framework.
