Robot Achieves One-Shot Imitation with Zero Fine-Tuning
Scobleizer · x · 2026-08-20
A researcher shared a breakthrough in robotics: while experimenting with few-gradient learning using GEN-1.5, they tested "physical prompting." The robot successfully imitated a task after seeing it demonstrated just once, requiring zero fine-tuning. Subsequent prompts for different tasks also resulted in successful rollouts. This achievement culminates extensive team work and years of community groundwork.
More from Embodied
- OpenRoboto Reaches SOTA in Three Weeks with Open Source Robot Brain — markjeffrey · 2026-08-20
- Robot Speeds: Tesla at 85mph, Racing Robot Hits 250mph — Scobleizer · 2026-08-20
- Waymo's new 'Ojai' vehicle launches in SF, Phoenix, and LA — SuzKP · 2026-08-20
- Peking U & MSRA: BCP optimizes robot replanning timing via lightweight policy — 机器之心 · 2026-08-20
- AI agent stacks blocks in browser physics sim as open embodied-AI arena debuts — NovaCoding · 2026-08-20
- Eclipse Ventures Robotics Portfolio Shines at Actuate Conference — Sethwinterroth · 2026-08-20