Link Search Menu Expand Document
FogROS2 Docs

Launch Configuration

ROS 2 utilizes Python to create launch files which can specify what ROS 2 nodes should be launched. FogROS 2 utilizes these launch files to not only specify what nodes are launched, but where they are launched (i.e. on the robot or on the cloud machine). This page provides information about modifying existing launch files to work with FogROS 2 for deploying nodes to the cloud.

Robot Nodes

Robot nodes are ROS 2 nodes that are only running on the robot. There are no modifications necessary to the normal launch file to launch nodes that are running on the robot.

Cloud Nodes

Launching ROS 2 nodes in the cloud can be completed in 4 steps:

  1. Add the following import to the top of the launch file: import fogros2

  2. Replace the LaunchDescription in the launch file with fogros2.FogROSLaunchDescription

  3. Specify AWS EC2 machine parameters by creating a fogros2.AWSCloudInstance object

  4. Replace all Node objects that need to be executed in the cloud with fogros2.CloudNode

fogros2.AWSCloudInstance

Definition

fogros2.AWSCloudInstance(ami_image, region="us-west-1", ec2_instance_type="t2.micro", disk_size=30)

Description

An AWSCloudInstance is an object that represents an AWS EC2 instance that performs computation in the cloud. By specifying certain parameters for the machine, the AWS EC2 machine may be launched in different AWS regions or with different hardware. For help deciding some of the parameters, please visit the Getting Started page.

Parameters

ami-image: An AWS AMI is a template that is used to launch machines. The parameter itself is the Two recommended AMI images are ‘ami-00f25057ddc9b310b’ for Ubuntu 20.04 and ‘ami-0b6030c78f8b2f076’ for Ubuntu 22.04. Custom AMI images may be created to speed up deployment time, as any necessary dependencies can be installed in the AMI, so that every launch need not install all dependencies. For more information about AMIs, visit this link.

region: The AWS region the machine will be launched in. By default, this is us-west-1.

ec2_instance_type: The type of machine that will be launched by AWS. For help deciding what machine to launch, please visit Getting Started. By default, this is a t2.micro instance, which is a lower compute free instance type.

disk_size: The size of the disk of the EC2 instance in gigabytes. The default is 30GB.

fogros2.CloudNode

Definition

fogros2.CloudNode(machine, stream_topics=[], ...)

Description

A CloudNode is a ROS 2 node that is executing in the cloud.

Parameters

machine: machine is the fogros2.AWSCloudInstance created in step 3.

stream_topics: stream_topics is used to specify any topics that are used for video streaming so that appropriate compression can be deployed. For more details, visit Video Compression (H.264).

Make sure to leave in any normal arguments that are a part of a normal ROS 2 node. An example will be provided to make the transition more clear.

Talker-Listener Example

Consider the following talker-listener launch file, which launches two nodes, where the listener subscribes to a topic and talker publishes to a topic (more details here):

from launch import LaunchDescription
from launch_ros.actions import Node


def generate_launch_description():
    ld = LaunchDescription()

    listener_node = Node(
        package="fogros2_examples", executable="listener", output="screen"
    )
    talker_node = Node(
        package="fogros2_examples", executable="talker", output="screen"
    )
    ld.add_action(talker_node)
    ld.add_action(listener_node)
    return ld

This section will show how to deploy the talker node onto the cloud, while the listener node stays on the robot.

Step 1: Add the following import to the top of the launch file: import fogros2

After running through this step, the launch file should now look like this:

import fogros2
from launch import LaunchDescription
from launch_ros.actions import Node


def generate_launch_description():
    ld = LaunchDescription()

    listener_node = Node(
        package="fogros2_examples", executable="listener", output="screen"
    )
    talker_node = Node(
        package="fogros2_examples", executable="talker", output="screen"
    )
    ld.add_action(talker_node)
    ld.add_action(listener_node)
    return ld

Step 2: Replace the LaunchDescription in the launch file with fogros2.FogROSLaunchDescription

After running through this step, the launch file should now look like this:

import fogros2
from launch import LaunchDescription
from launch_ros.actions import Node


def generate_launch_description():
    ld = fogros2.FogROSLaunchDescription()

    listener_node = Node(
        package="fogros2_examples", executable="listener", output="screen"
    )
    talker_node = Node(
        package="fogros2_examples", executable="talker", output="screen"
    )
    ld.add_action(talker_node)
    ld.add_action(listener_node)
    return ld

Step 3: Specify AWS EC2 machine parameters by creating a fogros2.AWSCloudInstance object

The AWSCloudInstance that needs to be created will be called machine, as follows: aws_machine = fogros2.AWSCloudInstance(region="us-west-1", ec2_instance_type="t2.micro", ami_image=ami-0b6030c78f8b2f076)

The AWS region the machine will launch in is us-west-1 with a t2.micro instance, which is low compute and free, and with an AMI that will make the machine run Ubuntu 22.04 (the AMI is one of the recommended ones stated earlier).

Incorporating this into the launch file,

import fogros2
from launch import LaunchDescription
from launch_ros.actions import Node


def generate_launch_description():
    ld = fogros2.FogROSLaunchDescription()
    aws_machine = fogros2.AWSCloudInstance(
                    region="us-west-1", 
                    ec2_instance_type="t2.micro", 
                    ami_image=ami-0b6030c78f8b2f076
                )
    listener_node = Node(
        package="fogros2_examples", executable="listener", output="screen"
    )
    talker_node = Node(
        package="fogros2_examples", executable="talker", output="screen"
    )
    ld.add_action(talker_node)
    ld.add_action(listener_node)
    return ld

Step 4: Replace all Node objects that need to be executed in the cloud with fogros2.CloudNode

Since the talker node will be deployed on the cloud, the modification will be to the talker node only.

import fogros2
from launch import LaunchDescription
from launch_ros.actions import Node


def generate_launch_description():
    ld = fogros2.FogROSLaunchDescription()
    aws_machine = fogros2.AWSCloudInstance(
                    region="us-west-1", 
                    ec2_instance_type="t2.micro", 
                    ami_image=ami-0b6030c78f8b2f076
                )
    listener_node = Node(
        package="fogros2_examples", executable="listener", output="screen"
    )
    talker_node = fogros2.CloudNode(
        package="fogros2_examples", executable="talker", output="screen", machine=aws_machine
    )
    ld.add_action(talker_node)
    ld.add_action(listener_node)
    return ld

Notice that in this example, the machine was set to the aws_machine that was created earlier.