Team Rented a PC Bang and Recruited 1,000+ People to Train Game AI
Kangwook_Lee · x · 2026-09-29
- Kangwook Lee shares how his team collected data for the game AI "Ally": earlier versions had to actually interact and play with human players to improve quality.
- The team rented a Korean PC bang and recruited over 1,000 people to play with the AI and collect on-policy rollout data.
- Following the DAgger approach, they corrected the collected trajectories, kept training, then deployed the new model for the next round of data collection—iterating repeatedly.
More from Embodied
- Idle time in demos teaches robots to do nothing — dev open-sources a LeRobot dataset editor — lhoestq · 2026-09-29
- Microagi hires Georgia Tech professor Animesh Garg plus 7 researchers via stealth acquisition — lukas_m_ziegler · 2026-09-29
- NOOP open-source app makes WHOOP band work locally without subscription — mhdfaran · 2026-09-29
- Recursive Harness Distillation lifts robot manipulation success from 37.3% to 64.0% — SeoulNatlUniv · 2026-09-29
- Tsinghua humanoid robot plays badminton with one policy from just 30 min of human motion data — ChongZzZhang · 2026-09-29
- Bose's $99 USB-C wired ANC earbuds ditch the battery brick by drawing power from your phone — aakashgupta · 2026-09-29