Niantic's Object Interaction Makes Gaussian Splat Scenes Editable and Physics-Ready
lukas_m_ziegler · x · 2026-11-17
Niantic Spatial launched Object Interaction (beta), addressing Gaussian splatting's key limitation: photoreal captures are frozen, with no separate objects, collisions, or physics—useless for robot training.
The feature scans captures to identify whole movable objects, extracting each while preserving visual appearance and picking up collision information. After a quick review, objects become selectable, movable, or removable with simplified physical properties—turning a captured room from a fixed layout into an interactive sim environment where chairs slide and equipment shifts.
More from Embodied
- DeepMind's Raia Hadsell on transferring skills between robots at RAAIS — nathanbenaich · 2026-11-17
- Ropedia-Xperience-2 dataset: 1,500+ 3D scene scans and 131 hours of egocentric video — liuziwei7 · 2026-11-17
- A 4.5M-parameter TTS model runs fully offline on an ESP32-S3 — ngxson · 2026-11-17
- NVIDIA open-sources Alpamayo2-Edge, a VLA model for autonomous driving — nvidia · 2026-11-17
- AtomDog: a 3D-printed open-source quadruped robot you control from the browser — Stefania_druga · 2026-11-17
- Open-source LiDAR SLAM hits PyPI, no ROS 2 required: pip install lidarslam — rsasaki0109 · 2026-11-17