SSF teaches robots to reason about likely hiding spots when searching for hidden objects
_krishna_murthy · x · 2026-11-07
New work SSF tackles a concrete embodied-AI problem: asking a robot to find objects like car keys that could be anywhere, while the robot can only see part of the room — it must reason about likely hiding spots, actively go look, and open drawers and containers.
The authors share a thread demonstrating how SSF closes this search-reason-act loop, an exploration of embodied intelligence in real home settings.
More from Embodied
- 11,398 real-world robot evals across 124 sites confirm GPT-6 Astra is seriously strong — DJiafei · 2026-11-07
- Open-source robot Microduck production line first runs at Seeed's Shenzhen factory — huggingface · 2026-11-07
- FlexiTac hits CoRL 2026, releases FlexiTac-Gen for AI-assisted tactile sensor design — anand_bhattad · 2026-11-07
- JHU makes a Unitree G1 crawl like a spider with temperature-aware training, code open-sourced — _krishna_murthy · 2026-11-07
- BoopFace: MIT-licensed SVG robot avatar Web Component with an ESP32/STM32 embedded engine — Additional-Ad5527 · 2026-11-07
- Industrial robot installs hit 603k in 2025, China 59%; humanoids near 25k but mostly non-productive — ingliguori · 2026-11-07