SJTU AutoLab releases Drive3R, a streaming geometric foundation model for long-horizon driving 3D reconstruction
AutoLab-SJTU · hf · 2026-10-16
AutoLab at SJTU introduced Drive3R, a block-wise streaming geometric foundation model for scalable, long-horizon multi-view 3D reconstruction in autonomous driving.
- Uses block-wise causal attention plus confidence-aware re-anchoring to reduce drift and maintain global consistency over long sequences.
- Benchmark charts show SOTA results over baselines like VGGT, π³, DA3, StreamVGGT and DVGT-2 across point cloud, depth and ego-motion evaluation on Waymo, nuScenes and KITTI, with visibly cleaner long-horizon reconstructions.
More from Research
- Logit lens on a Beat Saber finetune reveals it reuses language grammar circuits — ctjlewis · 2026-10-16
- Handwired compiles plain JavaScript into an exact, fully differentiable neural net — ctjlewis · 2026-10-16
- 128×128 record stands: how intermediate-work experiments foreshadowed reasoning models — ctjlewis · 2026-10-16
- TeamFusion: preference-based AI proxies debate to preserve minority views in team decisions — burkov · 2026-10-16
- MIT's C-GMT gives humanoid robots whole-body compliance without sacrificing motion tracking — MIT_CSAIL · 2026-10-16
- SUNY Albany to offer Topological Deep Learning course, from Perceptrons to WL test — burny_tech · 2026-10-16