ECCV26 Oral: Flow Matching Enables Single-Stage Multi-View Point Cloud Registration
ducha_aiki · x · 2026-09-11
A Stanford and University of Bonn team released Register Any Point (RAP), an ECCV26 oral-candidate paper that recasts multi-view point cloud registration as conditional generation.
- Method: A learned continuous point-wise velocity field transports noisy points into a registered scene, from which each view's pose is recovered. Unlike prior pipelines that do pairwise correspondence matching then pose-graph optimization, RAP directly generates the registered cloud, gaining efficiency and point-level global consistency.
- Extension: It extends the team's NeurIPS25 Rectified Point Flow work to single-stage, multi-scale multi-view registration.
- Results: Scaling training data plus test-time rigidity enforcement yields superior zero-shot performance and the shortest runtime on a cross-domain multi-view registration benchmark, generalizing across view counts, scales and sensor modalities.
- Paper, code, benchmark and demo are open-sourced.
More from Embodied
- cixliv: robot fighting may be the first realized market where bots beat average humans — cixliv · 2026-10-28
- Can the Hiwonder MentorPi T1 run fully offline as a robot pet? Pi 5 questioned — oulu2006 · 2026-10-28
- Testing Mint Modeler to Build the Menlo Research Robot — freelerobot · 2026-10-28
- Khosla-backed startup builds a 'mind-reading' cap with 100,000 tiny chips to write AI prompts — Distinct-Question-16 · 2026-10-28
- Racer turns Ray-Ban Meta glasses into an AI track coach with Opus 5.5, gains 5.3 seconds per lap — Vjeux · 2026-10-28
- Fine-tuning VLAs on TAMP data hits distribution mismatch; DATAFARM gets JAX port — tomssilver · 2026-10-28