When you deal with safety-critical systems, you want to be sure that your software is really well tested. But testing a complex system by hand is not enough: there are too many combinations and possibilities to try out, and you can never be sure that you checked all of them. If your software is managing the braking system of a train or the temperatures in a chemical plant, forgetting to check just one of the possible use conditions can be very dangerous.
The University of Bremen and its spin-off Verified Systems International GmbH (founded in 1998) have developed the test automation System RT-Tester, and improved it in the context of the EU-funded COMPASS project, to address these problems in three different ways:
- Tests are generated from a model of the system in a fully automated way, guaranteeing that all the possible cases are covered (“completeness”).
- A health monitor observes the software while it is running and, just like a good doctor, understands immediately when there is a possible problem.
- Tests for large and complex “systems of systems”, where different cyber-physical systems are working together, can be automatically executed and monitored.
After the end of the COMPASS project, our company has further developed, optimised and integrated these features, which are now available into the commercial version of RT-Tester.
Meanwhile, at the University of Bremen, a team is working on challenges concerning the foundations of automated testing and safety-critical systems verification and validation. All research results are made publicly available in order to avoid patent controversies.
The biggest challenge Verified Systems is currently facing is the unwillingness of potential customers to change their test processes and the related supplier structure: many companies still consider testing as an activity of minor importance in the development process. A second problem consists in the fact that the development of test models requires skilled personnel and quite some effort. As a consequence, productivity of testing appears to be low at the beginning of a testing campaign, when test engineers are working on new test models. This is not always positively perceived by managers, even though the return on investment in the longer term is very good.
We are currently working to address these challenges both from a technical angle and from a marketing angle. We need to improve our marketing strategy in order to reach greater international visibility, and lead the company through a controlled growth phase.
Come to meet us at the Innovation Radar booth during the ICT 2015 to see a demo of RT-Tester (Wednesday 21st October 13:00-15:00). We are competing for the Innovation Radar Prize which will be awarded on Thursday 22nd October!