WMG and partners complete a three-year programme of research that lays the foundations for proving the safety of autonomous vehicles

Challenge

To prove automated vehicle systems are safe by driving them in the real world would require an estimated 11 billion miles of test driving. Given the infeasibility of this method of safety assurance, innovation is required.

The work of WMG’s Verification and Validation team is centred around how a regulator or developer can confirm an automated system is safe in a commercially viable time frame.

In the OmniCAV collaborative research project, the challenge for WMG was to help build a simulation platform capable of proving Automated Driving Systems (ADS) are safe. The project used virtual testing, to vastly reduce the time required for real world testing, enabling commercial opportunity for using ADS and improving safety in the UK.

The simulation platform had to reflect the environment and scenarios faced by vehicles on a test loop of real roads in both urban, and rural settings in Oxfordshire.

Solution

OmniCAV, led by transport simulation solution providers, Aimsun, has laid the foundation for a modular and validated simulation platform to accelerate the safe deployment of automated vehicles.

The OmniCAV project brought together a diverse set of stakeholders, enabling a 360-degree perspective on what it would take to deploy automated vehicles safely. The consortium includes: Aimsun, Admiral, Ordnance Survey, Arcadis, Oxfordshire Country Council, Arrival, RACE, Alchera Technologies, and Thatcham Research, and is supported by the UK Government’s Centre for Connected and Automated Vehicles (CCAV) and Innovate UK.

Verifying that an ADS is safe requires putting the system through many scenarios. The diverse set of stakeholders in the OmniCAV project meant the test scenarios generated from the research reflected a wide range of real-world situations .

Dr Siddartha Khastgir, Head of Verification & Validation at WMG explains, “WMG generated a large number of scenarios for OmniCAV using two methodologies: knowledge based, using expert knowledge based safety analysis, and data-based, using insurance claim records and accident databases”.

In one example, WMG data scientists analysed accident claim data from insurance provider, Admiral to derive likely accident causes. WMG computer scientists analysed information from a diverse set of sources and transcribed into a common description language by creating the Scenario Description Language, SDL and creating a structured catalogue of over 250,000 test scenarios.

These scenarios have been stored in the world’s largest public library of test cases for connected and automated vehicles, the Safety Pool™ Scenario Database. The enhanced database enables an informed and specific approach (known as an Operational Design Domain [ODD] approach) to testing which significantly cuts development time.

To prove this, the project partners undertook research and development to map, digitalise and simulate real-world routes of Oxfordshire. Within this virtual representation of the real-world, WMG virtually-tested an automated vehicle on a selection of the OmniCAV test scenarios. The simulated environment allows a variety of unusual and complicated situations to be run which is much faster and more effective than driving a real vehicle.

Impact

OmniCAV resulted in the development of a novel safety assurance framework by the WMG team. The research, published in several world-leading academic journals, has created the framework for future technology development.

As a result, WMG’s research team have taken a leading role in the creation of global standards at the International Organization for Standardization (ISO) and inputting into the upcoming regulations and guidance at the United Nations Economic Commission for Europe (UNECE).

The project was also very valuable to the project partners. Oxfordshire County Council, the first local authority in the UK to include driverless vehicles in their local transport plan, have achieved many benefits from the OmniCAV project, as Ruth Anderson iHub Future of Mobility Team Leader, Oxfordshire County Council explains:

“Working with WMG and other partners has given us a better understanding of the questions we should be asking potential trialling organisations”

Project lead Aimsun, whose role included the key project task of building a microscopic traffic simulation model of the Oxfordshire test loop, also experienced great results from the project. Dr Mark Brackstone, Head of Research Projects at Aimsun, explains:

“Working on OmniCAV was an exciting project that not only allowed Aimsun to explore and perfect new ways of building and calibrating models, but also added substantially to our understanding of the role of simulation in the Verification and Validation of autonomous vehicles.”

The OmniCAV project has enabled WMG to complete a key step on the journey to realising the safe introduction of automated vehicles.

Find out more about WMG’s Verification and Validation research here: Verification and Validation (warwick.ac.uk)

Discover more about the OmniCAV project story.