Industrial Applications of Reinforcement Learning
summary
Five industrial case studies show concrete results: simulated-to-real robotics copies dog motions and coordinates 50 motors synchronized to perform a 90° turn; a robot learns valve manipulation from vision-only input (single camera) and transfers to varied valves; humanoid agents produce adversarial soccer behaviors; a Berlin pick‑and‑place system generalizes to unseen objects; and a web UI agent maps raw HTML states and CSS-selector actions to automate multi‑step tasks. Each uses MDP framing, learned rewards, and simulation-to-deployment pipelines.
AI-written summary · about
chapters (8)
- 0:00Intro
- 2:36Reinforcement learning
- 8:06How can RL help?
- 11:50What do I need?
- 17:23Manufacturing (i.e. Robotics)
- 28:32Tech
- 32:24Transport
- 33:44Outro
related talks
13mReinforcement learning and it's growing role in AI
16mReal-World Robotics as the Next Frontier for AI?
37mA practical perspective on training reinforcement learning bots.
14mHuman-in-the-Loop Reinforcement Learning
01h 35mDeep Robotic Learning
31mDesigning game AI with reinforcement learning
31mFrom Single AI to Multi AI Systems
23mAlphaDow: Reinforcement Learning for Industrial Production Scheduling