Reinforcement Learning: Sim-to-Real Part I

​Hello YoungAI friends! We're running a two-part workshop on sim-to-real reinforcement learning. The goal: train agents in simulation, then deploy them on a real self-driving RC car. This is part 1, where we build the foundations.

​You'll implement PPO (Proximal Policy Optimization) from scratch and train agents in simulated environments. Everything you build here carries directly into part 2, where we deploy on a physical PiRacer Pro.

​No RL experience needed, just bring your Python skills and a laptop.

​Program:

• ​Mingling & snacks

• ​Intro to reinforcement learning

• ​Implement PPO (or use Stable-Baselines3 if you prefer)

• ​Train your agent on MountainCar and CartPole

• ​CarRacing challenge, who gets the fastest lap?

• ​Wrap-up & teaser for part 2

​You need:

• ​Laptop with Python

• ​An editor you like (VS Code is great)

​Setup instructions will be shared before the event so we can jump straight into coding.

​This is part 1 of 2. You can attend part 1 without committing to part 2, but we recommend both!

Big thanks to Visito for sponsoring the workshop and providing the PiRacer Pro we'll be deploying on in part 2.

​This event is mainly for our members, but if you're new, you're welcome to join one meetup for free. Interested in becoming a member? Check out our website: https://www.youngai.no.