CoCam4D: Geometry-Aware Cooperative 4D Perception for Camera-Only Autonomous Driving

This research proposes CoCam4, a Bayesian framework for collaborative perception in autonomous driving that explicitly models geometric uncertainty. It uses a VGGT-based network to generate 3D Gaussian scene representations with uncertainty estimates, enabling multiple agents to combine observations without LiDAR. Dynamic Object Primitives (DOPs) are introduced for efficient communication, and experiments show improvements of 11.48% on OPV2V+ and 10.62% on DAIR-V2X-C over vision-only methods.