SnowballTarget is the environment we created at Hugging Face, and the assets are from Kay Lousberg. We have an optional section at the end of this Unit if you want to learn to use Unity and create your environments.
The first agent you’re going to train is Julien the bear 🐻 (the name is based after our CTO Julien Chaumond) to hit targets with snowballs.
The goal in this environment is that Julien hits as many targets as possible in the limited time (1000 timesteps). To do that, it will need to place itself correctly from the target and shoot.
In addition, to avoid “snowball spamming” (aka shooting a snowball every timestep), Julien has a “cool off” system (it needs to wait 0.5 seconds after a shoot to be able to shoot again).
The reward function is simple. The environment gives a +1 reward every time the agent’s snowball hits a target and because the agent’s goal is to maximize the expected cumulative reward, it will try to hit as many targets as possible.
We could have a more complex reward function (with a penalty to push the agent to go faster, etc.). But when you design an environment, you need to avoid the reward engineering problem, which is having a too complex reward function to force your agent to behave as you want it to do. Why? Because by doing that, you might miss interesting strategies that the agent will find with a simpler reward function.
In terms of code it looks like this:
Regarding observations, we don’t use normal vision (frame), but we use raycasts.
Think of raycasts as lasers that will detect if it passes through an object.
In this environment our agent have multiple set of raycasts:
In addition to raycasts, the agent gets a “can I shoot” bool as observation.
The action space is discrete: