The largest public repository of scenarios for testing autonomous vehicles globally has been launched by WMG at the University of Warwick, and Deepen AI.
The database provides a diverse set of scenarios in different operational design domains (ODDs i.e. operating conditions) that governments, industry and academia can leverage. It will allow users to test and benchmark Automated Driving Systems (ADSs) and use insights to inform policy and regulatory guidelines.
Initial scenarios have been generated using a novel hybrid methodology developed by WMG, at the University of Warwick, using knowledge-based and data-based approaches. The Safety Pool Scenario Database will allow organisations to create scenarios in their libraries, collaborate with other organisations via shared and public libraries and enable the public to submit challenging real-world scenarios.
Enabling scenarios to be matched to specific environments and operating conditions means that trials and tests can be undertaken in the simulated environment, controlled test facilities, and public roads. Evidence from each environment can be used to inform understanding of safe behaviours, bringing Autonomous Vehicles closer to market at pace.
It is becoming ever more apparent that Autonomous Vehicles and the Connected and Automated Mobility (CAM) enable are one of today’s most exciting technological advances with industry, academia, and governments investing in the research development of safe and secure Autonomous Vehicles.
CAM will provide a once in a lifetime opportunity to have a global impact on societal issues around road safety, traffic efficiency and emissions.
However, to ensure that Autonomous Vehicles are road-ready and will be safer than the average human driver, it has been suggested that they must be tested on 11 billion miles of roads, an impossible goal in the real world. Therefore, the ability to test on virtual roads in simulation environments is paramount for manufacturers and government bodies to ensure safe behaviours and assure that Autonomous Vehicles positively influence road safety.
The true test of Autonomous Vehicles will not be in just the number of miles driven and the quality and complexity of those miles. Widespread industry adoption of a scenario-based testing approach will ensure that the Autonomous Vehicle’s behaviours and capabilities are ready for the real world.
Dr Siddartha Khastgir, from WMG, University of Warwick, holds a UKRI Future Leaders Fellowship enabling him to create methods to test autonomous vehicles over a seven-year programme. Dr Khastgir has already worked on the UK Government’s Centre for Connected and Autonomous Vehicles and Innovate UK funded Midlands Future Mobility. The project offers a real-world ecosystem for the development and trialling of Connected and Automated Technology as part of the Zenzic coordinated CAM Testbed UK capability and was fundamental in developing the scenario database that forms the core of Safety Pool initiative. Dr Khastgir stresses the importance of a global database of scenarios:
“Safety of automated driving systems is a hard research challenge and can only to solved by national and international collaboration and knowledge sharing. With the launch of Safety Pool Scenario Database, we are inching closer to seeing automated driving systems on the roads. Testing and validating automated driving systems transparently in an integrated simulation-based framework and in real-world scenarios will provide insights into the readiness of ADS and speed up the adoption globally. WMG and MFM are grateful for the support of CCAV and Innovate UK in developing the database and we are excited to be at the forefront of this revolution.”
“The Safety Pool Scenario Database lays a key foundation stone for autonomous vehicle safety”, said Mohammad Musa, CEO & Co-founder of Deepen AI. “We are working closely with governments across the world to create a framework for ADS certification that will bring vehicle manufacturers one giant step closer to deploying safe and secure autonomous vehicles on the roads.”
Scenarios in Safety Pool Database can be applied to a range of different autonomous vehicle systems, such as Automated Lane Keeping Systems (ALKS). ALKS see cars drive in an automated manner on motorways by adapting to speed and traffic around them, trucking, fully autonomous vehicles, and even pods that could be used in town centres and pedestrianised areas as a ‘last mile’ mode of transport.
Safety Pool Initiative invites stakeholders to share learnings in the form of scenarios to expedite validation, testing and certification for the entire community. Safety Pool Initiative is a global multi-stakeholder initiative with the mission of bringing transparent, certifiable safety to ADSs, uniting the autonomous vehicle community around standardised certification programs for ADSs worldwide.
“We are thrilled to work closely with Deepen AI & WMG, University of Warwick, to launch the Safety Pool Scenario Database. We believe Safety Pool Initiative is going to play a crucial role in standardising and bring transparency to ADS certification globally. We are already in advanced talks with many countries to adopt ADS certification frameworks based on Safety Pool database scenarios.”
Michelle Avary, Head of Automotive from World Economic Forum
“I am very pleased that the effort and hard work of producing this scenario database has been so successful and is now gaining the recognition it deserves. Scenario testing, both in simulation and physical tests, is widely recognised as the practical route to verifying the safety of ADS, and a comprehensive scenario database is crucial for that. We are proud to have supported this work.”
Richard Morris, Innovation Lead for CAV at Innovate UK
Safety Pool initiative says it welcomes government and industry stakeholders worldwide to join the initiative and take the front row in bringing safety standards and certifications to their country. Members of the autonomous vehicle industry can also join the Safety Pool community and access safety scenarios to test, validate and benchmark ADS transparently.
Comments