A group of companies in Bristol are working on new ways to test the next generation of drones and driverless cars.
“Cyber physical systems will start to touch every part of our life from robotic vacuum cleaners to autonomous cars”
T&VS is working with the University of Bristol to develop new engineering methods to test software that controls cyber physical systems (CPS) such as driverless cars. They will be working with Thales, Dyson and SCISYS to better understand how to check and verify the software for these systems works as expected.
“Cyber physical systems will start to touch every part of our life from robotic vacuum cleaners and online orders delivered by drone to autonomous cars,” said Dr Mike Bartley, CEO and Founder of T&VS. “Ensuring such systems are fully tested and safe to use can be both expensive and time consuming and consequently potential barrier to market entry. The project will investigate if techniques successfully adopted in hardware design verification can be adapted to work with complex software.”
The technology for driverless cars is currently being tested in Bristol as part of the VENTURER and FLOURISH research projects.
Developers of these systems are recognising that the current software verification and validation techniques will not cope as cyber systems become increasingly complex.
The involvement of Thales, Dyson and SCISYS is key to the potential success of the project, which is backed by the Innovate UK government agency. All three are developing these cyber systems and have identified testing as a major technical challenge, so they will define what they need to test the hardware. T&VS will then develop the tools for them to test those systems to make sure they behave as expected.
Testing the tests
Hardware verification faced a similar issue 20 years ago and developed new verification techniques to overcome this. Those techniques are now widely adopted and hardware developers are able to verify increasingly complex designs so that they reach the market in good time and work as expected.
The University of Bristol has recently developed new model-based test generation techniques to target specific use cases or requirements far more effectively than conventional methods can. How to apply these techniques to complex real-world CPS will be investigated as part of this collaborative project.
“This project brings new understanding of practical requirements into the University to drive new research”
“Model Based Test Generation techniques can significantly improve verification productivity,” said Dr Kerstin Eder, Reader in Design Automation and Verification in the Department of Computer Science at the University of Bristol and leader of the verification and validation for safety in robots research theme at the Bristol Robotics Laboratory. “Achieving coverage closure faster and in an automated fashion saves engineering effort in practice. This project brings new understanding of practical requirements into the University to drive new research with access to real-world uses.”
You can see more on autonomous systems at the Bristol Robotics Lab.
