This AI Must Learn to Escape (Reinforcement Learning)

Welcome back to Rollo’s Lab — my experiment where I train a small AI robot, Rollo, to escape puzzle rooms built in Unity. In this second part, the abandoned lab becomes even more dangerous. Rollo is no longer alone. After being released from containment, Hollow (Rollo’s predecessor) begins chasing him through the old chamber. While Rollo tries to solve new puzzles, activate buttons, cross moving platforms, and survive difficult jumps, Hollow is learning too… and getting better at stopping him. If you want to support more experiments you can join via    / @rolloslab   or   / rolloslab   Your name could appear as one of the Lab Members in upcoming experiments. Chapters: 00:00 - Intro 00:12 - Previously 00:43 - The Chase 08:02 - Outro Business inquiries: [email protected] Inspired by @aiwarehouse (AI Warehouse) Subscribers: 1071