Skip to content

Latest commit

 

History

History
103 lines (77 loc) · 3.68 KB

README.md

File metadata and controls

103 lines (77 loc) · 3.68 KB

ROS Launchable FogROS2-SGC

Dependencies

pip install icmplib

Configuartion file

FogROS2-SGC requires a separate configuration file to expose the topics. While it supports automatic topic scanning, we want the users to expose the topics globally only if they want to, to enhance the isolation and protect the privacy of the robots and services. An example of the configuration file can be found as following:

identifiers: 
  # crypto for the task (same for all robots/services on the same task)
  task: test_cert 
  # crypto unique for the robot 
  # note #0 : 
  # for now, it's not bound to an actual certificate yet, name whatever you want
  # but make sure to match this with the assignment 
  # note #1 : this value is overriden by rosparam's whoami
  # note #2 : feel free to define it here or define it at rosparam
  whoami: machine_talker 
  
topics:
  - topic_name: /chatter
    topic_type: std_msgs/msg/String

# declare possible states 
# pub: publish to the swarm
# sub: subscribe to the swarm
# note this is reversed from prior version of SGC config file
state_machine: 
  talker: 
    topics:
      - /chatter: pub 
  listener: 
    topics:
      - /chatter: sub 

# name: state
# name need to match the identitifer's whoami
# state should be declared in possible states 
# Phase 1: only allow changing the assignment at runtime 
assignment:
  machine_talker: talker
  machine_listener: listener

We observe that the cloud services are usually the reversed of the robots' configuration file (e.g. cloud is the publisher then robot is the subscriber). We want the user only need to write and distribute one copy of the configuration file, and mark themselves as different only at whoami in identifiers. This value is overriden by the ros2 launch file's parameter.

In the example, the configuration configures whoami as machine_talker, which is assigned with the talker's state (at aassignment). The talker state is defined to pub (publish) to the topic chatter. The chatter is configured to be of type std_msgs/msgs/String. Vise versa, the listener only need to set whoami as machine_listener, and the states (how the topics are handled) are automated.

Examples

Profiling Example

Remember to colcon build and source install/setup.bash the workspace.

On local robot, run robot sensors by

ros2 launch sgc_launch profiler.robot.launch.py

then we run the local service modules by

ROS_DOMAIN_ID=2 ros2 launch sgc_launch profiler.service.local.launch.py

We use different domain ID, and use FogROS-SGC to bridge them.

On the cloud, run

ROS_DOMAIN_ID=1 ros2 launch sgc_launch profiler.service.cloud.launch.py

by default, the cloud will be in standby state and only active when reconfigured.

For reconfiguration, run

ROS_DOMAIN_ID=2 ros2 service call /sgc_assignment_service sgc_msgs/srv/SgcAssignment '{machine: {data: "machine_local"}, state: {data: "standby"} }'

to change the local service to be the standby mode (doesn't do anything). To resume, run

ROS_DOMAIN_ID=2 ros2 service call /sgc_assignment_service sgc_msgs/srv/SgcAssignment '{machine: {data: "machine_local"}, state: {data: "service"} }'

TODO: later wrap this in a more user friendly CLI

H264 Example

cd sgc_ws/src
git clone https://github.com/KeplerC/h264_image_transport.git

The SGC will automatically incoperate H264 streaming ROS2 message type in its build system.

Automated version without configuration file

SGC supports autoamtic topic discovery by simply not setting the configuration file. This works with heristics that if the topic misses the publisher (publisher count = 0), then it exposes as a remote subscriber, and vise versa.