Loopy-SLAM: Dense Neural SLAM with Loop Closures

Point-SLAM [43] ESLAM [26] GO-SLAM [74] Loopy-SLAM (Ours) ATE Depth L1 31.55 cm 7.94 cm ATE Depth L1 29.73 cm 42.89 cm ATE Depth L1 4.64 cm 8.79 cm ATE Depth L1 7.03 cm 3.81 cm Figure 1. Benefits of Loopy-SLAM. While Point-SLAM yields high-fidelity reconstructions it does not implement loop closure and may duplicate geometries due to drift. ESLAM is faced by the same problem due to the lack of loop closure. GO-SLAM implements loop closure, but computes rather low quality map geometry. In contrast to GO-SLAM which requires to save the entire history of input frames used for mapping to update the map after loop closures, our approach anchors the neural scene representation on points which can simply be shifted without recomputing the dense map from scratch. We show the ATE RMSE and the depth L1 re-rendering error on the mesh for the TUM-RGBD fr1 room scene.