Open-source software for robots is becoming increasingly widespread in industry as well as academia. For some companies, the Robot Operation System (ROS) is already a competitive and innovative factor. The ROS-Industrial (ROS-I) Conference in Stuttgart, Germany, showed what developers and users are currently doing and why Amazon and Google are now using ROS for their robotics efforts.
Interest in ROS has increased significantly in recent years. Developed in 2007, ROS initially became the de facto standard, not only in research, but also for service robot technologies. With its range of basic but important functions for robots, it helped numerous start-ups to quickly develop their first products with just a few resources and to get them on the market.
ROS offers components for manipulation, navigation and image processing. It also includes drivers, algorithms, and diagnostic and development tools. These are freely available, manufacturer-independent, standardized, and created in participative work.
Once these modular components have been developed, they can be used several times, which makes the development and commissioning of robot systems more efficient and thus more economical. On the other hand, anyone can use the components as a basis for new or further development. Commercial use is already quite common and often includes both open-source and closed-source components.
The ROS-I initiative, which advocates the industrial use of ROS, has been in existence since 2013. It acts as an interface between the developer community and industrial partners. It performs an important networking function worldwide with its three consortia and 65 members.
Fraunhofer IPA manages the European branch. In this capacity, the institute hosted a major ROS event for the sixth time in December 2018. The conference included 27 presentations from research and industry to around 150 participants.
Presentations from the 2018 ROS-I Conference can be found here. The next training dates for ROS in Stuttgart, as part of the educational activities in the EU project ROSIN, are April 1-5 and June 24-28. You can register here.
The following six core messages can be derived from these presentations, reflecting trends, application highlights and challenges to be tackled with ROS.
Platforms, ecosystems, cloud technologies are the future
Henrik Christensen, Qualcomm Chancellor’s Chair of Robot Systems at the University of California, San Diego, and director of the Institute for Contextual Robotics, named three driving factors in his keynote about why robotic applications will no longer be isolated. Instead, they will operate and share data in larger, networked structures or ecosystems. Factors include:
- Increasing demand for flexibility in production
- The aging world population combined with the desire for more automated services at home
- Urbanization, which poses ever-growing challenges to logistics
Robots have to be cost-effective and robust in all contexts at the same time, which can only be achieved in the cloud. Integrating appropriate hardware into the system does not make sense for small household robots, for example, or the systems become too expensive, as with autonomous cars.
Numerous companies, primarily from the U.S. and Asia, operate within these new ecosystems. They have changed their business model to no longer sell technologies, but offer them as platform services. ROS can be a decisive factor in providing the necessary standardization for this development towards common platforms.
[Editor’s Note: Henrik Christensen is keynoting the Robotics Summit & Expo, which is produced by The Robot Report, from June 5-6 in Boston. Other keynotes include AWS Robotics, Microsoft, NVIDIA, the NSF, and Qualcomm.]
IT giants see added benefit in ROS
Prominent examples of these developments are Amazon and Google, which also presented at the ROS-I Conference. Each presented a new platform in 2018 to enable the development of robots using ROS in the cloud. This commitment by companies whose core business is far from robotics also caused a sensation beyond the industry as a whole.
The AWS Robomaker platform offers a browser-based development environment and numerous cloud-based services such as speech and image recognition for robots or tools for machine learning and analysis. A simulation environment and fleet management capabilities are part of Robomaker as well. Google’s Cloud Robotics platform providing a ROS interface also addresses the growing need for solutions that are scalable, offer collaborative capabilities and behaviour, and provide robust change management and monitoring.
[Editor’s Note: Roger Barga, General Manager, AWS Robotics and Autonomous Services, Amazon Web Services, will deliver a keynote at the Robotics Summit & Expo called “The Role of the Cloud in the Future of Robotics.” He will discuss how cloud services will allow developers to partition functionality between their physical robots and the cloud, in particular compute-intensive machine learning and artificial intelligence.]
Successful applications of ROS increase in industry
An entire conference session was dedicated to the topic of applications with ROS. For example, the FZI Research Center for Information Technology in Germany and Opel presented an application in which a robot attached a car door seal with around 35 metal pins. The application won the European robotics competition EuRoC. Although the application has not entered full production, Opel gave it a positive rating in terms of functional safety.
Other application examples included the cooperation between BMW and Microsoft and autonomous guided vehicles (AGV) from German AGV manufacturer E&K Automation.
Fraunhofer IPA presented two applications with its navigation software for mobile robots. On the one hand, the IPA software runs with ROS modules on AGV from system integrator Bär Automation. These AGVs are used in automobile production to flexibly transport the car bodies from assembly station to assembly station.
On the other hand, so-called “smart transportation robots” at the BMW Group plant in Regensburg navigate with the IPA software. Here, too, the aim was to enable very flexible navigation that also achieves an availability of 99 percent.
Easier system integration is necessary
Several speakers at the conference emphasized that system integration for ROS-based applications is still too complex and resource-intensive. Presentations by Acutronic Robotics, Denso, ABB, Bosch, Universal Robots, Fraunhofer IPA, and others presented components or development platforms that should enable simpler, faster, and more reliable integration.
Software must meet the highest requirements
In order to bring open-source software into industrial applications, it must be of high quality, especially for safety-critical applications. The European ROSIN research project, for example, is fostering software Quality Assurance measures among the ROS community. The project partners are also developing tooling for automatic code testing.
Another tool is the HAROS framework (High Assurance ROS), which offers a ROS-specific code test. The aim is to find possible errors in the code at an early stage and thus save development time and costs.
Functionality and IT security are indispensable
The discussions repeatedly focused on the need for reliable security in order to bring ROS-based applications into industrial use. One approach to this is the above-mentioned code testing. The Ubuntu Linux distribution, operated by Canonical, also offers security features important for ROS.
Pilz, a German security company, presented the topic of safety in an even larger context. Because service and industrial robots will work directly with humans in complex environments, they will require a completely new safety strategy. Pilz is intensively dealing with this change and presented its own robot arm based on ROS components last year.
Some speakers also pointed out that it is very difficult to get safety approval with open-source components. One cannot yet ensure individual software modules but only complete applications. That is why it is difficult to modify an application, since the whole assurance process would have to be redone. The consideration of open-source technologies in future security requirements is therefore desirable.
Public funding opportunities are numerous
Bringing ROS further into industrial applications is also the aim of several research projects. Companies can apply for support from the EU and soon at the national level. The EU project ROSIN offers financial support for ROS software components, as well as various further education measures and trainings for ROS.
Furthermore, the EU project RobMoSys makes it possible to receive funding for participation in a model-based ecosystem for robot software.
In the second half of 2019, the German research project SeRoNet (Service Robot Network) plans to make a public call for funded participation. The aim is to establish a platform that provides products and services for professional service robots and enables cooperation between hardware and software providers, system integrators, and end users.
In short: There is a lot going on in the ROS environment.
About the Author
Thilo Zimmermann is Project Manager at the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. Zimmermann can be reached at [email protected].