How policies trained in simulation transfer to real hardware. The reality gap is the central challenge: domain randomization, system identification, and progressive fine-tuning are the key strategies for bridging it. Builds on the Sim Loop & Reward Shaping lesson.