Fork me on GitHub

Reinforcement Learning 2D Humanoid Walkers

"Play that funky music, robot"

This project take a teach a 2d ragdoll to walk(ish) using reinforcement learning.

You can view the progress at differen't stages of training with the buttons below. You can also throw balls by clicking the animation.

The agent can move it's limbs in a realistic range of motion, it can feel the position of it's limbs and it's goal it to hold it's head upright and move to the right. The dark outlines are when the agent grips the floor, since I found walking was slippery otherwise. The balls are to provide obstacles.


How does it work?: This uses use reinforcement learning to teach the agent to walk. This is a branch of machine learning targeted at controlling systems over time such as systems of limbs or a self driving car. The agent is defined in 2d with a certain strength and range of limb movement. Training is done offline in tensorflow.js. The aglorithm is Deep Deterministic Policy Gradients with prioritized experince replay. The environment is in box2d for javascript and we use webpack to run the same code on the backend and frontend.


Source code https://github.com/wassname/rl_2d_walker.js

Credits:

Load brain:

Be mean: