A decentralized, platform-agnostic visual spatial foundation model that learns spatial priors from data to accurately predict relative poses and local Bird's Eye Views without requiring camera overlap or existing network infrastructure.
