Back

Adventures in Reinforcement Learning

completed

A couple of projects around reinforcement learning in simulated environment

Notes

I saw the deepmind documentary and read What is intellignece from Blaise Aguera y Arcas, and fell in a consummating rabbit hole about reinforcement learning.

alt text Park my car: training a reinforcement learning agent to park a car in a parking lot, in a simulated environment in Godot.

alt text The Bear: A banana eating bear in a random simulated environment with a few sensors needs to learn how to navigate around to stay alive. Still in godot.

alt text The hexapod: training several iterations of a 3D hexapod robot to walk in a simulated environment in Unity.

The three projects are using PPO.