ai-robotics

AI Robotics tutorials for hobbyists

Summary

Learn how to train a virtual humanoid/robot to do football tricks with Reinforcement Learning

From -> To

Learn how to:

• tune training hyperparameters for better model performance
• optimise reward functions for faster learning
• get results in a couple of hours of training, and for free (i also don't own any GPUs)

Tutorials

1 - Foot tricks

2 - Head tricks

3 - Penalty taking and stopping

4 - Robot tricks

5 - Box handling [Coming soon]

Notes

• Most of these results can be achieved in a few hours of GPU training. Sometimes less. And for free on Kaggle (30h per week usage limit)
• Notebooks originally adapted from Mujoco's MJX tutorial