Safety Pool™: The World’s Largest Database for Automated Driving Systems

The Automated Vehicle Bill has become law in the UK, signalling the potential arrival of a self-driving future in the coming years. However, this raises safety concerns among the public and road users. The question is, ‘How can we prove automated vehicles are safe?’

Research shows that self-driving vehicles would need to drive 11 billion miles to demonstrate that they are 20% safer than human drivers. This is the equivalent to making 46,119 trips from the Earth to the moon or having 100 vehicles driving 24 hours a day, 365 days a year, at a speed of 25 miles per hour for 500 years. Clearly, proving the safety of AVs through real-world testing alone is a time-consuming and costly process. On top of that, the opportunities for such extensive testing are limited in the UK and other parts of the world.

This is where scenario-based testing becomes essential for the safety assurance process of self-driving technologies. Engineers create virtual testing scenarios, which are the “what if” situations that an autonomous driving system (ADS) might face. They encompass all aspects such as the environment, weather, road conditions, and the behaviour of other road actors. Engineers can then test their systems virtually and assess the safety level of their products, such as whether they can respond to road incidents safely and promptly.

However, no technology developer can create every possible test scenario alone. As a global leader in safety research for automated transport systems, WMG’s Safe Autonomy Research Group is committed to developing innovative safety assurance methods for the safe deployment of these systems. They recognise that large-scale cross-industry collaboration is key to creating a more diverse set of test scenarios by leveraging each other’s experiences and knowledge across the ecosystem.

To facilitate this, WMG and Deepen AI, a leading US software developer, created the Safety Pool™ Scenario Database, which supports the development, verification, and validation of automated driving systems. Safety Pool™ enables organisations, regulators, and researchers to store and share their scenarios with others. As part of the CAM TestBedUK ecosystem, we assist organisations in testing their ADS on CAM TestBedUK and other international testbeds, as well as on simulation platforms.

Since its launch in 2021, the Safety Pool™ Scenario Database has become the world’s largest public scenario database for automated driving systems, housing more than 270,000 scenarios and attracting 200 active users across academia, policymakers, and technology developers worldwide, including the World Economic Forum, Transport Canada, and the UK’s Vehicle Certification Agency.

The pioneering DriveSafeAI project is also the user of the Safety Pool™ Scenario Database, using its scenarios to help Wayve and WMG develop scalable methodologies and mechanisms to prove that the use of AI is safe for self-driving vehicles.

To know more about the Safety Pool™ Scenario Database and potential collaboration opportunities, please visit https://www.safetypool.ai/.