Running Containerized Unity Applications with RL Agents: GPU Acceleration

Introduction

Containerization is a popular method for deploying and managing applications, especially in the context of cloud computing and continuous integration and delivery. One of the most widely used containerization platforms is Docker. Unity, a popular game engine, can also be containerized to enable easier deployment and management. However, running containerized Unity applications, especially those utilizing machine learning (ML) or reinforcement learning (RL) agents, can be quite demanding and may require GPU acceleration. In this article, we will explore how to run containerized Unity applications with RL agents using GPU acceleration, excluding Xvfb as far as we know.

Prerequisites

Before we begin, make sure you have the following prerequisites installed:

Docker
NVIDIA Container Toolkit
Unity Hub

Setting Up the Environment

First, let's create a new Unity project that we will containerize. Open Unity Hub and create a new 3D project:


 Unity Hub > Projects > Create > 3D > New 3D Project

Next, we will create a Dockerfile to containerize our Unity project. Create a new file named Dockerfile in the root directory of your project:


 touch Dockerfile

Open the Dockerfile in your favorite text editor and paste the following content:


 FROM unity3d/2021.3.24f1
 WORKDIR /app
 ADD . /app
 RUN mkdir -p /app/Temp
 ENV UNITY_EDITOR_QUIT_ON_LAST_FRAME 0
 ENV UNITY_EDITOR_FORCE_MULTITHREADED_RENDERING 1
 ENV UNITY_EDITOR_DISABLE_DATABASE_AUTO_UPDATE 1
 ENV UNITY_EDITOR_DISABLE_MANAGED_ASSETS_AUTO_IMPORT 1
 ENV UNITY_EDITOR_ENABLE_DEBUG_LOG 1
 ENV UNITY_EDITOR_DISABLE_AUTO_SAVE 1
 ENV UNITY_EDITOR_DISABLE_AUTO_BACKUP 1
 ENV UNITY_EDITOR_DISABLE_AUTOSAVE_ERROR_MESSAGE 1
 ENV PATH="/usr/local/nvidia/lib:$PATH"
 RUN apt-get update && apt-get install -y unzip
 RUN mkdir -p /app/Plugins/x86_64
 ENV PATH="/app/Plugins/x86_64:$PATH"

This Dockerfile sets up a Unity 2021.3.24f1 container with GPU support and installs the necessary dependencies for running the project.

Building the Docker Image

Build the Docker image by running the following command in your terminal:


 docker build -t unity-project .

This command builds the Docker image with the tag "unity-project".

Running the Containerized Unity Application

To run the containerized Unity application, use the following command:


 docker run -it --rm --gpus all unity-project

This command runs the containerized Unity application with GPU acceleration.

Setting Up RL Agents

To set up RL agents, you can use popular ML frameworks such as TensorFlow or PyTorch. In this example, we will use TensorFlow.

In this article, we explored how to run containerized Unity applications with RL agents using GPU acceleration, excluding Xvfb. We covered the prerequisites, set up the environment, built the Docker image, and ran the containerized Unity application. For more information on running RL agents with TensorFlow, refer to the following resources: