Julia Gym DLL

What?

I have a Reinforcement Learning Gym that is a complex C++ model. To use it, I've created a .dll/.so with a C interface that makes calls into the true C++ Model; The original goal was to be able to use the C++ model from Python, as a RL Gym.

I like Julia. So as a side project, this is a Julia variant that pushes the interface up so it works with ReinforcementLearning.jl, from here: https://juliareinforcementlearning.org/

Usage

Setup

# To build the test environments
make

# To recreate the necessary environment+dependencies
import Pkg
Pkg.activate(".")
Pkg.instantiate()

Running

# To instatiate and run the Gym
using ReinforcementLearning

include("./DLLGymEnv.jl")

# To run it:
env = DLLGymEnv("../pendulum_gym.so")
A = action_space(env)
for i in 1:100
    env(rand(A), render=true)
    is_terminated(env) && break
end

License

This code is made available under the MIT license.

Gary chunky@icculus.org