Skip to main content

Cloud Setup Guide

This guide provides instructions for establishing cloud-based development environments to run tutorials and projects in this course. This is an alternative if you lack access to recommended local hardware.

Supported Cloud Platforms​

(This section will be populated with specific cloud platform recommendations, instance types, and setup instructions as content evolves.)

General Cloud Setup Steps​

  1. Choose Cloud Provider: Select provider offering GPU-accelerated instances (e.g., AWS, Google Cloud, Azure).
  2. Select Instance Type: Choose instance type with adequate CPU, RAM, and GPU resources. Refer to Hardware Guide for minimum recommendations.
  3. Operating System: Launch instance with Ubuntu 22.04 LTS.
  4. SSH Access: Configure SSH access to your instance.
  5. Install NVIDIA Drivers: Install appropriate NVIDIA GPU drivers and CUDA toolkit for your instance.
  6. Clone Repository: Clone course's companion code repository.
  7. Install Dependencies: Follow environment setup instructions in Module 1 installing ROS 2, Gazebo, and other required software.

Running Docker Containers in Cloud​

Docker containers provide consistent and isolated environments, making them ideal for cloud-based development. The course provides Dockerfiles setting up ROS 2 and Isaac ROS environments.

1. Build Docker Image (Example: ROS 2 Humble)​

First, build Docker image for desired environment. Navigate to docker/ directory of course's code examples on cloud instance.

cd /path/to/course-code/docker
docker build -t ros2_humble_dev -f Dockerfile.ros2_humble .

This command builds image named ros2_humble_dev using Dockerfile.ros2_humble.

2. Run Docker Container​

Once image is built, run container from it. This command launches interactive container with necessary privileges and mounts for graphical applications if needed.

docker run -it --rm --privileged \
    --network host \
    --runtime nvidia \
    -e DISPLAY=$DISPLAY \
    -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
    -v /path/to/your/ros2_ws:/ros2_ws:rw \
    --name ros2_dev_container \
    ros2_humble_dev bash

Parameter explanations:

  • -it: Interactive and pseudo-TTY.
  • --rm: Remove container after exit.
  • --privileged: Gives extended privileges (often needed for robotics simulations).
  • --network host: Allows container using host's network stack (important for ROS 2 discovery).
  • --runtime nvidia: Enables GPU access within container (requires NVIDIA Container Toolkit on host).
  • -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw: Enables graphical forwarding for GUI applications.
  • -v /path/to/your/ros2_ws:/ros2_ws:rw: Mounts local ROS 2 workspace into container.
  • --name ros2_dev_container: Assigns name to container.
  • ros2_humble_dev: Docker image name to use.
  • bash: Runs bash shell inside container.

3. Develop Inside Container​

Once inside container, ROS 2 environment will be sourced, and mounted workspace (/ros2_ws) will be available. You can then proceed with building and running ROS 2 packages as if on native Ubuntu installation.

4. Isaac ROS Specific Containers​

For Isaac ROS development, typically use NVIDIA's pre-built Isaac ROS Docker images. Process is similar: pull image and run with GPU and network configurations.

docker pull nvcr.io/nvidia/ros:humble-desktop-cuda-jammy
docker run -it --rm --network host --privileged \
    --runtime nvidia \
    -e DISPLAY=$DISPLAY \
    -v /tmp/.X11-unix/:/tmp/.X11-unix:rw \
    -v /path/to/your/ros2_ws:/ros2_ws:rw \
    --name isaac_ros_dev_container \
    nvcr.io/nvidia/ros:humble-desktop-cuda-jammy bash