Understanding ROS: Framework and Ecosystem for Robotic Applications

First of all, ROS, short for Robot Operating System, is not an operating system (OS) like Linux or Mac. The simple and correct way to understand ROS is:

  1. Easy-to-use wrapper for communications
  2. Ecosystem (collection of packages and tools)

which are specifically designed for robotics applications.

1. Easy-to-use Wrapper for Communications

Regardless of ROS development, there already exists a well-established communication protocol known as DDS (opens in a new tab) (a kind of protocol) and its vendors (opens in a new tab). This has shown great technical credibility from various applications such as financial and military sectors.

ROS2 serves as a wrapper for this middleware, offering a straightforward implementation for robotic applications and convenient utilities and APIs. End-users like us only need to write codes with our familiar languages, C++ and Python with a few functions and classes provided by ROS

Communications provided by ROS

The communications available in ROS2 include the below:

  • Topic publish-subscribe: The publish-subscribe mechanism works like data-streaming to multiple receivers. Examples include the continuous reading of sensor inputs.

  • Service communication: This type of communication works as a one-to-one transaction between two parties.

Although there is another type of communication called action, this will not be covered at this moment.

Minimal Source code (No Need to Understand)

Using ROS allows you to perform the above two communications in your programs with a minimum amount of code lines, as in the below example.

#include <rclcpp/rclcpp.hpp>
#include <std_msgs/msg/string.hpp>
 
class MinimalPublisher : public rclcpp::Node
{
public:
  MinimalPublisher()
    : Node("minimal_publisher")
  {
    publisher_ = this->create_publisher<std_msgs::msg::String>("topic", 10);
    timer_ = this->create_wall_timer(
      std::chrono::milliseconds(500),
      std::bind(&MinimalPublisher::publish_message, this));
  }
 
private:
  void publish_message()
  {
    auto message = std_msgs::msg::String();
    message.data = "Hello, ROS2!";
    publisher_->publish(message);
  }
 
  rclcpp::Publisher<std_msgs::msg::String>::SharedPtr publisher_;
  rclcpp::TimerBase::SharedPtr timer_;
};
 
int main(int argc, char** argv)
{
  rclcpp::init(argc, argv);
 
  auto publisher = std::make_shared<MinimalPublisher>();
 
  rclcpp::spin(publisher);
  rclcpp::shutdown();
  return 0;
}

Once you provide the source codes, ROS will handle all subroutines for DDS communications, including message transport, callback scheduling, and interface generation for C++/Python.

2. Ecosystem

(Left) Physics and sensor simulation from gazebo (opens in a new tab). (Right) Realsense provides ROS wrapper (opens in a new tab). Most of widely used sensors and robots do the same.

Either indirectly or directly, ROS offers a vast ecosystem of packages that extend its capabilities, enabling developers to build sophisticated robotic applications. Some key components of this ecosystem include:

Project management

  • Package managing and workspace (colcon / ament_cmake): ROS offers a build system with workspace functionality and user-friendly dependency management to facilitate the seamless management of multiple packages in an organized manner.

  • Common Protocol Types and easy generation of custom types (rosidl): ROS provides a set of common message types, such as strings, integers, and custom-defined messages, which facilitate standardized data exchange.

Utilities

  • Visualization (rviz): The ROS Visualization (rviz) tool enables developers to visualize robot states, sensor data, and other important information in a 3D environment, aiding in debugging and monitoring.

  • Record and Play (rosbag): ROS provides tools for recording data during robot operation and playing it back for analysis and debugging, allowing developers to reproduce scenarios for testing purposes.

  • Coordinate Transform Management (tf): ROS includes a coordinate transform management packages (tf) to handle complex coordinate transformations, critical in multi-sensor and multi-robot systems.

Simulation & Hardware

  • Physics Simulation by Plugin (gazebo_ros): ROS integrates with physics simulation environments through plugins, allowing developers to test robot behavior in realistic virtual environments.

  • Hardware Support and ROS Wrappers: Many well-known hardware manufacturers offer ROS wrappers, allowing users to easily integrate and test their hardware with ROS. This compatibility promotes seamless hardware testing and development within the ROS framework.

And More

The ROS ecosystem continues to grow, with numerous additional packages available for various functionalities, such as perception, navigation, manipulation, and human-robot interaction. You can visit here (opens in a new tab) to look up the available repository in the ecosystem.