Managing Editor of Future Vehicle magazine, Steve Welch, interviewed Mike Dempsey, Managing Director of Claytex and two other D-RISK partners prior to the LCV2021 exhibition.
Safe and Sound
Mike Dempsey explains how creating ultra-realistic test scenarios can form the basis of a virtual driving test for autonomous vehicles.
What happens if you’re driving in your EV down a country road and a family of ducks wanders into view? “A human will take half a second to respond to that situation,” says Dempsey. “The autonomous driving systems being developed will process the information in milliseconds. It’s much faster than a normal driver.”
Take another example of a broken-down car at the side of the road. Should the car break the law and cross double white lines to avoid a crash? A human would know the answer, but a machine won’t – unless you tell it. “There are times where it would be safer to allow the vehicle to push the boundaries,” says Dempsey. “But it all has to be assessed and become part of your safety case.”
Using the rFpro platform, Dempsey and his colleagues create ultra-realistic test scenarios that can form the basis of a virtual driving test for autonomous vehicles.
Claytex is part of D-RISK, a co-innovation project aimed at improving the safety of AVs to make them a commercial reality in the UK and beyond. The project involves four partners: dRISK.ai, DG Cities, Imperial College London and Claytex and is part-funded by the Centre for Connected and Autonomous Vehicles and Innovate UK.
Together, D-RISK project partners are creating a standardised testing approach that will help establish a trusted model for bringing AVs to the roads. Our tests prepare AVs for every possible scenario they might face on the road – including those that have never occurred before.
They describe these as “edge cases.” Edge cases are those tricky, unplanned and unexpected events that happen while driving – such as a child appearing from behind a car or a plane landing on a motorway.
Read the full article here: Safe and Sound
Making self-driving vehicles safer
The D-RISK project is a consortium made up of dRISK.ai, DG Cities, Claytex and Imperial College London. We spoke to two of these partners to find out more.
Listening and learning
DG Cities describe themselves as “urban innovators”, dedicated to using new technology and data to transform towns and cities. “We’ve been engaging the public in exploring the future for autonomous vehicles (AVs),”
says Ed Houghton, Head of Research & Service Design at DG Cities. “We want to know how the public thinks and feels about AVs, as well as generating edge cases required for simulation scenarios. “There’s a lot of important data that are coded in traditional ways, such as traffic cameras, but we’ve worked on finding non-traditional sources of information, including crowdsourcing data.
On the edge
From road trips to Scotland, traffic cameras in London to jam-cams at a Bangalore roundabout, there’s almost nowhere dRISK.ai won’t search for edge cases. Identifying the scenarios and codifying them is critical to developing a driving test for AVs, says Chess Stetson, CEO of dRISK.ai. “When a box falls out the back of a truck and disgorges a bunch of electronics on the road, do you stop or keep going?” he asks. “These are the kinds of things that autonomous vehicles have very, very little intuition about and don’t do well at.”
Chess has been involved in the development of AVs for almost two decades, tracing his history back to the DARPA funded projects in the States. Frustrated at the lack of progress, an Innovate UK bid brought Chess to Britain, where dRISK.ai was created in 2019. Today, the company is part of the D-RISK consortium, working together to develop a new driving test for AVs.
“You don’t need to just stimulate better,” says Chess. “The most important thing is knowing what to simulate.”
Read the full articles here: Making self-driving vehicles safer
If you have any questions or for more information, please do not hesitate to contact us.
Telephone: +44 1926 885900 — Email: firstname.lastname@example.org