ROS2 - The Robot Operating System

Philip Hörmandinger and Dr. Jürg M. Stettbacher
 
Due to the enormous complexities involved, the management of promising projects in the field of robotics today is often only possible for large companies that have the appropriate resources and financial means at their disposal. This could change drastically in the future through the use of ROS2.



Figure 1: ROS2-Logo

ROS2, the Robot Operating System, is an international open-source project born at Stanford University with the purpose of creating a universal and free operating system for all kinds of robots. The first version of ROS was still very university-oriented. The second edition, however, now claims to be "industry-ready."

Figure 2: Simulating a small robot with lidar in an office environment.

According to Stettbacher Signal Processing AG, ROS2 has enormous potential because its huge ecosystem gives it great clout, especially for small development teams. Instead of months or years, concept studies or functional samples can be realized within a few days or weeks. This is supported by the fact that ROS2 consists of various subprojects. For example, there is a subproject that deals with the navigation of mobile robots, be it on land, on water or in the air. Another subproject deals with the movement of robotic gripping arms. In addition, ROS2 offers a number of open simulators, such as Gazebo or Webots. With these, a system can be tested even before the hardware for it exists. Almost all common sensors can be simulated realistically, especially lidars. Thus, there are almost no limits to the possibilities.

At Stettbacher Signal Processing AG, a ROS2-based navigation system with automatic collision avoidance was ported to an existing robot within less than a year and advanced to production readiness. In the process, special requirements had to be met in terms of robustness with regard to various terrain and driving scenarios. These requirements could only be achieved by a specially designed chain of lidar filters (keyword computer vision) and a special MPC-based controller. Figures 2 and 3 show simulations for the robot vehicle and visualization of navigation around obstacles.

Figure 3: Visualization of the navigation. In red in the green circle is the vehicle. The white boxes represent the obstacles detected by sensors. Light blue is the safety distance to them and the red thin line is the planned path.

Like other successful open-source projects, ROS2 has the potential to become a de-facto standard for a wide variety of robotics applications in the medium term. Sure, it makes little sense to keep reinventing the wheel. Nevertheless, caution is advised when switching: It is true that the extensive documentation helps the beginner. But the subject matter is complex and mathematically demanding. At best, it is therefore worth working with a strong, reliable partner, such as Stettbacher Signal Processing AG.