The trials and tribulations of training a physical robot with reinforcement learning

This is a follow up to my article “Training a real robot to play Puckworld with reinforcement learning”. In that one, to make it a little punchier, I showed the overview and end results of the project, but left out the insane number of little hurdles and decisions I had to figure out.

So this article will be about those details instead, partly for me to justify the pain, but maybe more charitably to show that for any project with a neat (hopefully?) presentation, there’s probably a harrowing saga of hair-pulling roadblocks behind it. Here, it’s roughly in the order I encountered things. There are lots I’m leaving out too. read more

Training a real robot to play Puckworld with reinforcement learning

After I trained an agent to play “puckworld” using Q-learning, I thought “hey, maybe I should make a real robot that learns this. It can’t be that hard, right?”

Hooooooooo boy. I did not appreciate how much harder problems in the physical world can be. Examples of amateurs doing Reinforcement Learning (RL) projects are all over the place on the internet, and robotics are certainly touted as one of the main applications for RL, but in my experience, I’ve only found a few examples of someone actually using RL to train a robot. Here’s a (very abridged!) overview of my adventure getting a robot to learn to play a game called puckworld. read more

Beating OpenAI games with neuroevolution agents: pretty NEAT!

Let’s start with a fun gif!

Something I’ve been thinking about recently is neuroevolution (NE). NE is changing aspects of a neural network (NN) using principles from evolutionary algorithms (EA), in which you try to find the best NN for a given problem by trying different solutions (“individuals”) and changing them slightly (and sometimes combining them), and taking the ones that have better scores. read more

Using Reinforcement Learning to solve the Egg drop puzzle

So last time, I solved the egg drop puzzle in a few ways. One of them was using a recent learn, Markov Decision Processes (MDP). It worked, which got me really stoked about them, because it was such a cool new method to me.

However, it’s kind of a baby process that’s mostly used as a basis to learn about more advanced techniques. In that solution to the problem, I defined the reward matrix and the transition probability matrix , and then used them explicitly to iteratively solve for the value function v and the policy p. This works, but isn’t very useful for the real world, because in practice you don’t know  and , you just get to try stuff and learn the best strategy through experience. So the real challenge would be letting my program try a bunch of actual egg drops, and have it learn the value function and policy from them.