Reinforcement Learning · advanced · concept 190 of 311 · Updated Nov 4, 2026

🐒Imitation Learning

Imitation learning teaches a system by showing it examples of an expert doing the task rather than defining a reward. It is often the fastest route to a working robot skill. The classic weakness is drift: once the system strays from what it saw, it has no idea how to recover, so newer methods keep collecting corrections from the expert.