Consumer Automotive Safety Rating to Enhance CAV Uptake
Thatcham Research and Zenzic are developing a consumer safety rating, which they hope will be adopted by Euro NCAP.
The organizations have worked on safety testing for 25 years. During this period the focus was on passive and active safety. The latter was introduced a decade ago to examine the potential for radar and camera-based systems to prevent collisions – reveals Matthew Avery, director of research at Thatcham.
He explains: “The focus on crash prevention has underpinned a huge leap in safety for the UK’s passenger car fleet, leading to 65% reduction in passenger occupant injuries and deaths – comparing 1990 with 2010. That’s pre- and post- the inception of Euro NCAP, which is a big driver in setting the highest possible standards. As vehicles become increasingly automated the consumer will ask Euro NCAP ‘how safe are automated systems?’ So, we began testing the early generations of automation – assisted driving systems – in 2020. These Level 2 systems are a common feature of new vehicles. They support the driver with steering and braking, which can help to reduce fatigue on long journeys. However, the driver remains liable, and must keep their hands on the wheel and their eyes on the road.”
Automated vehicle regulation
The first automated vehicle regulation emerged in early 2021 at the United Nations. The UN signed it off to enable the driver to effectively sit back and relax – no hands on the wheel while also having no eyes upon the road to undertake other tasks in the vehicle. “Once the technology is available, it will be the first time we’ve had automated driving systems on the world’s roads for consumers to use,” says Avery.
However, at present there is car type approval for automated lane keeping systems (ALKS), and he believes there won’t be for a couple of years, when Level 3 functionality will be available. At Level 3 autonomy, drivers are required to be vigilant, to be ready to take back control immediately if the system encounters a situation it can’t cope with. Bad weather such as heavy snow or rain may require this to occur, or when the vehicle approaches unplanned roadworks. So, at Level 3 drivers won’t be able to take their eyes off the wheel nor off the road. They are required to monitor the situation.
He therefore comments: “This is the crux of the problem because drivers have an expectation of automation that the car is going to control everything, in all driving situations. There is a need for consumer education around how these systems work and they are also keen to understand how safe they are. So, the role we play at Euro NCAP and Thatcham, is to ensure that these systems are safe and to inform consumers why one system is better than another, while hammering home their obligations when driving.
“We assess current assisted driving systems on a test track with dummy target cars. However, we recognize that understanding outcomes for a driver who is not engaged or ‘in the loop’, would require more thorough testing. This is where virtual testing plays a pivotal part. It enables us to test a vehicle in many, many more situations than would be physically possible on a test track, because you can simulate scenarios with a computer.”
Integrating physical and digital testing
Zenzic’s technical and senior delivery leads on the consumer safety rating project reveal that they’ve found from the start of the project that any evaluation procedure of ALKS can’t rely solely upon physical testing. This is where a real vehicle is placed into a danger situation in a controlled test environment. They find there are too many variables and iterations to make this practical and so the majority of the ALKS testing will be conducted using simulations, involving a digital vehicle in a digital environment with potential safety hazards thrown at it to evaluate the “predicted results carefully”.
They add: “Following this, a handful of specific test cases can then be performed with physical testing. If the physical outcome matches the simulated prediction within reasonable error margins, we can gain confidence that the simulations are indeed representative.”
Designing the framework
Avery says a framework is being designed to consider how automated and autonomous systems will be assessed. He also confirms that virtual testing and physical validation are being integrated – after all digitally simulated assumptions need to be assessed in a real-life environment to ensure they are correct.
To achieve this, he says: “We are working in co-operation with Warwick Manufacturing Group (WMG), as part of a Zenzic-funded project that’s using the deep expertise available via the UK’s Connected and Automated Mobility (CAM) testbeds. The project brings all these elements together; for example, the virtual testing expertise of WMG, Thatcham Research’s relationship with Euro NCAP and consumer testing, and our expertise in developing active safety test protocols. We are also bringing in the expertise of Horiba MIRA via its test facility, which has been designed with automated vehicles in mind. It’s about multi-disciplinary UK expertise in automated driving and testing.” The outcome of all this will be an independent rating to help drivers to know how to use the automated systems, and to promote their safe adoption.
ALKS limited
So, why is the UK government limiting the use of ALKS? Zenzic’s project leads claim: “The UK government recognizes that ALKS technology has huge economic and societal benefits but also that its implementation must be managed very carefully to ensure safe rollout. Initially the systems are to be designed with a maximum operating speed of 60km/h (37mph), which is practical for most tedious stop-start traffic situations on motorways, whilst maintaining a sensible speed safety margin until consumer confidence is established.”
Avery says there is limited scope for the use of ALKS, which is at present contrary to consumer expectations of automation. For this reason, Thatcham and UK insurers have raised their safety concerns to ensure that the driver is required to take back control of the vehicle at short notice after long periods of letting the car do the driving.
“The fact that these vehicles will simply stop in lane should anything go wrong is also a concern,” he says before disclosing that there are a number of amendments being proposed by special interest groups within the UN process. He believes these should allow the speed cap to be increased, and the system will be permitted to change lanes rather than be static to one. “This will encourage adoption as drivers are more likely to see the benefit,” he suggests.
Automated systems inequality
Yet, despite the need to ensure safety, not all automated driving systems will be made equal. Zenzic explains that while two different automated driving systems may both meet “the relevant regulatory standards but one is easier to use or exceeds the standards”. In addition to this factor, Avery says significant financial pressures is placed on vehicle manufacturers to produce systems with different performance at different price points. Things are nevertheless changing as a decade ago, only premium cars have ADAS and so the same pressure will be applied when it comes to automation and ALKS specifically. Avery concludes: “By engendering trust. Consumers are more likely to believe the system is safe if it has been assessed by an independent body. Therefore, if you want consumer adoption, you need consumer ratings.”
With that aim in mind, Zenzic finds that some vehicle manufacturers are already announcing vehicles with Level 3 ALKS capability, such as Mercedes S Class. This leaves the need to build consumer confidence and to educate them on how best to use the technology, which an independent consumer safety rating for connected and autonomous vehicles (CAV) should be able to achieve. This ambition is being supported with a significant level of funding by the UK government because of its recognition of the value CAV technology. The funding is being provided to support innovation and the commercialization of such technologies via the CAM Testbed UK ecosystem.
