analiza aiqbench
UniNav advances embodied AI by unifying visual foresight and action generation into a single diffusion process, eliminating the need for costly planning rollouts common in current world models. The architecture's ability to jointly denoise observations and trajectories while incorporating geometry-aware tokens demonstrates significant efficiency gains without sacrificing spatial accuracy on standard benchmarks. This work represents a notable step forward in creating more capable and computationally efficient navigation agents for robotics applications.
UniNav: A Unified World-Action Diffusion Model for Visual Navigation
arXiv:2608.03244v1 Announce Type: new Abstract: Image-goal visual navigation is a fundamental capability for embodied agents. Existing navigation policies efficiently predict waypoint trajectories but lack visual foresight, while navigation world models can anticipate future observations but often require costly planning rollouts. We present UniNav, a unified world-action model that generates future visual observations and continuous waypoint trajectories through a single diffusion process. Given history frames and a goal image, UniNav jointly denoises visual and waypoint tokens within a single transformer, unifying future prediction and action generation in a shared framework. To improve spatial grounding, we incorporate geometry-aware camera tokens. We also train on both trajectory-labeled navigation data and video-only data, enabling the model to benefit from diverse videos without waypoint annotations. Based on this unified framework, we introduce two variants: UniNav-Full jointly predicts interpretable future observations and their corresponding trajectories, while UniNav-Fast removes future-image tokens at inference for efficient trajectory prediction. Experiments on navigation benchmarks show that UniNav outperforms the strongest baseline in ATE across all datasets. With one-step inference, UniNav-Fast achieves a latency of 0.1s without a substantial accuracy drop. Code will be released.
Abstract:Image-goal visual navigation is a fundamental capability for embodied agents. Existing navigation policies efficiently predict waypoint trajectories but lack visual foresight, while navigation world models can anticipate future observations but often require costly planning rollouts. We present UniNav, a unified world-action model that generates future visual observations and continuous waypoint trajectories through a single diffusion process. Given history frames and a goal image, UniNav jointly denoises visual and waypoint tokens within a single transformer, unifying future prediction and action generation in a shared framework. To improve spatial grounding, we incorporate geometry-aware camera tokens. We also train on both trajectory-labeled navigation data and video-only data, enabling the model to benefit from diverse videos without waypoint annotations. Based on this unified framework, we introduce two variants: UniNav-Full jointly predicts interpretable future observations and their corresponding trajectories, while UniNav-Fast removes future-image tokens at inference for efficient trajectory prediction. Experiments on navigation benchmarks show that UniNav outperforms the strongest baseline in ATE across all datasets. With one-step inference, UniNav-Fast achieves a latency of 0.1s without a substantial accuracy drop. Code will be released.
Artificial Intelligence (cs.AI)
arXiv:2608.03244 [cs.AI]
(or arXiv:2608.03244v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2608.03244 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 273 cuvinte
Subiecte conexe
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · 25 sept. 2026
WAM-Nav: Asymmetric Latent World-Action Modeling for Unified Visual Navigation
arXiv · 16 iun. 2026
WAM-Nav: Asymmetric Latent World-Action Modeling for Unified Visual Navigation
arXiv · 4 iun. 2026
Developing Healthcare Robotics with GPU-Native Medical Physics Simulation
NVIDIA · 28 iul. 2026