Search papers, labs, and topics across Lattice.
F3DGS is introduced as a federated learning approach for 3D Gaussian Splatting, enabling decentralized multi-agent 3D reconstruction without requiring centralized data access. It addresses geometric inconsistency and communication overhead by fixing Gaussian positions based on a shared geometric scaffold derived from registered LiDAR point clouds, while only appearance attributes are updated locally. Visibility-aware aggregation is used to resolve partial observability during federated optimization, achieving reconstruction quality comparable to centralized training in multi-agent settings.
Decentralized 3D reconstruction is now possible with comparable quality to centralized methods, thanks to a federated Gaussian Splatting approach that cleverly avoids geometric drift.
We present F3DGS, a federated 3D Gaussian Splatting framework for decentralized multi-agent 3D reconstruction. Existing 3DGS pipelines assume centralized access to all observations, which limits their applicability in distributed robotic settings where agents operate independently, and centralized data aggregation may be restricted. Directly extending centralized training to multi-agent systems introduces communication overhead and geometric inconsistency. F3DGS first constructs a shared geometric scaffold by registering locally merged LiDAR point clouds from multiple clients to initialize a global 3DGS model. During federated optimization, Gaussian positions are fixed to preserve geometric alignment, while each client updates only appearance-related attributes, including covariance, opacity, and spherical harmonic coefficients. The server aggregates these updates using visibility-aware aggregation, weighting each client's contribution by how frequently it observed each Gaussian, resolving the partial-observability challenge inherent to multi-agent exploration. To evaluate decentralized reconstruction, we collect a multi-sequence indoor dataset with synchronized LiDAR, RGB, and IMU measurements. Experiments show that F3DGS achieves reconstruction quality comparable to centralized training while enabling distributed optimization across agents. The dataset, development kit, and source code will be publicly released.
