Search papers, labs, and topics across Lattice.
The paper introduces BayesFusion-SDF, a CPU-centric probabilistic signed distance fusion framework that represents geometry as a sparse Gaussian random field to address the limitations of traditional TSDF and GPU-intensive neural implicit methods. It fuses depth observations using a heteroscedastic Bayesian formulation solved with sparse linear algebra and preconditioned conjugate gradients, enabling uncertainty estimation via randomized diagonal estimators. Experiments on a controlled scene and CO3D dataset demonstrate improved geometric accuracy compared to TSDF baselines and provide useful uncertainty estimates for next-best-view planning.
Ditch the GPU for probabilistic 3D reconstruction: BayesFusion-SDF achieves higher geometric accuracy than TSDF baselines on CPU while providing interpretable uncertainty estimates for active sensing.
Key part of robotics, augmented reality, and digital inspection is dense 3D reconstruction from depth observations. Traditional volumetric fusion techniques, including truncated signed distance functions (TSDF), enable efficient and deterministic geometry reconstruction; however, they depend on heuristic weighting and fail to transparently convey uncertainty in a systematic way. Recent neural implicit methods, on the other hand, get very high fidelity but usually need a lot of GPU power for optimization and aren't very easy to understand for making decisions later on. This work presents BayesFusion-SDF, a CPU-centric probabilistic signed distance fusion framework that conceptualizes geometry as a sparse Gaussian random field with a defined posterior distribution over voxel distances. First, a rough TSDF reconstruction is used to create an adaptive narrow-band domain. Then, depth observations are combined using a heteroscedastic Bayesian formulation that is solved using sparse linear algebra and preconditioned conjugate gradients. Randomized diagonal estimators are a quick way to get an idea of posterior uncertainty. This makes it possible to extract surfaces and plan the next best view while taking into account uncertainty. Tests on a controlled ablation scene and a CO3D object sequence show that the new method is more accurate geometrically than TSDF baselines and gives useful estimates of uncertainty for active sensing. The proposed formulation provides a clear and easy-to-use alternative to GPU-heavy neural reconstruction methods while still being able to be understood in a probabilistic way and acting in a predictable way. GitHub: https://mazumdarsoumya.github.io/BayesFusionSDF
