Nvidia Simulating Guide

This guide is for running the simulation in a Docker container, but with GPU passthrough, that means that Docker will be able to use your GPU, speeding up the simulation massively. This only works for Nvidia GPU’s because they are the only ones with this functionality. This guide also assumes you are running Linux, and was mainly designed for the Nvidia Jetson computers.

1. Setup the codebase

Clone the repository on the Jetson using git:

git clone https://github.com/TrickfireRobotics/gazebo-simulations.git
cd gazebo-simulations

2. Launch the simulation with GPU support

Use the sim CLI to build and launch the simulation with GPU passthrough:

cd gazebo-simulations
sim docker <robot-name>

The container detects the NVIDIA GPU and automatically enables GPU passthrough. You should see the container start and drop you into a shell.

3. Launch the simulation

You can now move onto the Running Simulations to launch your sim.

How to set up a Jetson

To set up a Jetson computer for Docker GPU support:

Install NVIDIA Container Toolkit:

curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
  && curl -fsSL https://nvidia.github.io/libnvidia-container/libnvidia-container.list | \
  sed 's/^deb /deb [signed-by=\/usr\/share\/keyrings\/nvidia-container-toolkit-keyring.gpg] /' | \
  sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list > /dev/null
sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

Configure Jetson for max performance (optional but recommended):
- Switch to maximum power mode (MAXN)
- Set fans to full speed
- Disable WiFi power management

You should now be ready to run the simulation container on the Jetson with GPU passthrough.