3 min read

Last Friday, a group of Spanish researchers have published a research paper titled ‘Towards a distributed and real-time framework for robots: evaluation of ROS 2.0 communications for real-time robotic applications’. This paper talks about an experimental setup exploring the suitability of ROS 2.0 for real-time robotic applications. In this paper, ROS 2.0 communications is evaluated in a robotic inter-component communication hardware case running on top of Linux. The researchers have benchmarked and studied the worst case latencies and characterized ROS 2.0 communications for real-time applications.

The results indicate that a proper real-time configuration of the ROS 2.0 framework reduces jitter making soft real-time communications possible but there were also some limitations that prevented hard real-time communications.

What is ROS?

ROS is a popular framework that provides services for the development of robotic applications. It has utilities like a communication infrastructure, drivers for a variety of software and hardware components, libraries for diagnostics, navigation, manipulation, and other things. ROS simplifies the process of creating complex and robust robot behavior across many robotic platforms.

ROS 2.0 is the new version which extends the concepts of the first version. Data Distribution Service (DDS) middleware is used in ROS 2.0 due to its characteristics and benefits as compared to other solutions.

Need for real-time applications in robotic systems

In all robotic systems, tasks need to be time responsive. While moving at a certain speed, robots must be able to detect an obstacle and stop to avoid collision. These robot systems often have timing requirements to execute tasks or exchange data. By not meeting the timing requirements, the system behavior will degrade or the system will fail.

With ROS being the standard software infrastructure for robotic applications development, demands rose in the ROS community to include real-time capabilities. Hence, ROS 2.0 was created for delivering real-time performance. But to deliver a complete, distributed and real-time solution for robots, ROS 2.0 needs to be surrounded with appropriate elements. These elements are described in the papers Time-sensitive networking for robotics and Real-time Linux communications: an evaluation of the Linux communication stack for real-time robotic applications.

ROS 2 uses DDS as its communication middleware. DDS contains Quality of Service (QoS) parameters which can be configured and tuned for real-time applications.

The results of the experiment

In the research paper, a setup was made to measure the real-time performance of ROS 2.0 communications over Ethernet in a PREEMPT-RT patched kernel. The end-to-end latencies between two ROS 2.0 nodes in different machines was measured. A Linux PC and an embedded device which could represent a robot controller (RC) and a robot component (C) were used for the setup. An overview of the setup can be seen as follows:

Source: LinkedIn

Some of the results are as follows:

Source: LinkedIn

The image describes the Impact of RT settings under different system load. They are

  • a) System without additional load without RT settings.
  • b) is system under load without RT settings.
  • c) is system without additional load and RT settings.
  • d) is system under load and RT settings.

The results from the experiment showed that a proper real-time configuration of the ROS 2.0 framework and DDS threads greatly reduces the jitter andworst-casee latencies. This mean a smooth and fast communication.

However, there were also some limitations when there is noncritical traffic in the Linux Network Stack is in picture. By configuring the network interrupt threads and using Linux traffic control QoS methods, some of the problems could be avoided.

The researchers conclude that it is possible to achieve soft real-time communications with mixed-critical traffic using the Linux Network stack. However hard real-time is not possible due to the aforementioned limitations.

For a more detailed understanding of the experiments and results, you can read the research paper.

Read next

Shadow Robot joins Avatar X program to bring real-world avatars into space

6 powerful microbots developed by researchers around the world

Boston Dynamics’ ‘Android of robots’ vision starts with launching 1000 robot dogs in 2019

Data science enthusiast. Cycling, music, food, movies. Likes FPS and strategy games.