Skip to main content

Lesson 1: ROS 2 Architecture and DDS

Learning Objectives​

By the end of this lesson, students will be able to:

  • Explain the DDS (Data Distribution Service) communication layer
  • Identify the key differences between ROS 1 and ROS 2
  • Understand the client library architecture (rclpy, rclcpp)
  • Describe the concept of ROS 2 domains and namespaces

Introduction to DDS​

The Data Distribution Service (DDS) is the underlying communication middleware that powers ROS 2. Unlike ROS 1's custom TCPROS/UDPROS protocols, ROS 2 uses DDS as its communication layer, which provides:

  • Decentralized communication: No master required
  • Language and platform independence: Standardized API
  • Quality of Service (QoS) policies: Configurable communication behavior
  • Security: Built-in authentication and encryption capabilities

DDS Concepts​

Topics and Publishers/Subscribers​

In DDS terminology:

  • Topic: A named data stream (e.g., /cmd_vel, /scan)
  • Publisher: Entity that sends data on a topic
  • Subscriber: Entity that receives data from a topic

Services and Clients​

  • Service: Request/reply communication pattern
  • Server: Responds to service requests
  • Client: Makes service requests

Actions​

  • Action: Long-running tasks with feedback and goal management
  • Action Server: Manages action goals and provides feedback
  • Action Client: Sends goals and receives feedback/results

Quality of Service (QoS) Policies​

ROS 2 introduces QoS policies that allow fine-tuning of communication behavior:

Reliability Policy​

  • Reliable: All messages are guaranteed to be delivered
  • Best Effort: Messages may be lost, but faster delivery

Durability Policy​

  • Transient Local: Late-joining subscribers receive last known value
  • Volatile: Only new messages are sent to subscribers

History Policy​

  • Keep Last: Store a specific number of most recent messages
  • Keep All: Store all messages (use with caution)

Practical Exercise: Creating a Simple Publisher​

import rclpy
from rclpy.node import Node
from std_msgs.msg import String

class MinimalPublisher(Node):

    def __init__(self):
        super().__init__('minimal_publisher')
        self.publisher_ = self.create_publisher(String, 'topic', 10)
        timer_period = 0.5  # seconds
        self.timer = self.create_timer(timer_period, self.timer_callback)
        self.i = 0

    def timer_callback(self):
        msg = String()
        msg.data = 'Hello World: %d' % self.i
        self.publisher_.publish(msg)
        self.get_logger().info('Publishing: "%s"' % msg.data)
        self.i += 1

def main(args=None):
    rclpy.init(args=args)
    minimal_publisher = MinimalPublisher()
    rclpy.spin(minimal_publisher)
    minimal_publisher.destroy_node()
    rclpy.shutdown()

if __name__ == '__main__':
    main()

Summary​

This lesson introduced the fundamental concepts of ROS 2 architecture and the DDS communication layer. Understanding these concepts is crucial for developing robust robotic applications that can handle real-world communication challenges.