Reinforcement Learning on RC Car

I present before you how I implemented Reinforcement Learning on an RC Car. I walk you through the hardware setup, simulation and real world testing of the car.


This is a companion discussion topic for the original entry at https://community.robotshop.com/robots/show/reinforcement-learning-on-rc-car

Nice project! Very impressive!

Thank you Abram! :smiley:

1 Like

Wow, you demonstrate impressive unsupervised machine learning. (I hope I don’t wake up to find my Prius trying this out in the garage.)

Do I have this right, that the learning was on your laptop in the sim, and the RPi executes a resultant “learned situation-response”?

You rock!

Thank you Alan!

Yes you are pretty much right.

  1. Most of the learning was on simulation and little bit on the real car
  2. All the major computation, during training and testing, was on my laptop. RPi executes the low level motor commands commands that it receives from the laptop.

Hello,
An interesting project.
Would you like to share you software with us?

Thanks @bonzadog. I will try to organize the pieces of code and put it on github in a couple of weeks. Right now I am occupied with university work.