What's next for EURO-NCAP 2029? BLUESKEYE AI’s chief scientist makes his predictions.
The January 2026 implementation of the Euro-NCAP safety rating is out. Naturally, everyone is now looking ahead at what will be required at the next implementation date, which is in 2029. That is harder to predict than you might think. As Niels Bohr said, “Prediction is very difficult, especially if it’s about the future!”. Nevertheless, I’m going to have a go at it here.
Euro-NCAP provides a safety rating scale for all cars sold in Europe. This allows customers to make informed choices; most customers want a safe car and will choose the safer of two otherwise equal choices. So the benefit for car makers is that they are rewarded for making cars safer than the legal minimum. This safety rating system indirectly improves road safety by encouraging car manufacturers to innovate on safety.
Encouraging safety innovation is core to improving road safety as far as Euro-NCAP is concerned. Ultimately with safety ratings it comes down to the art of the possible, acceptable and affordable. If there’s a cheap, easy, and effective way to measure that someone has been drinking, then sooner or later care manufacturers (OEMs) will be rewarded for implementing it. But innovation is expensive so Automotive OEMs and Tier 1s (the primary suppliers to OEMs) will delay it for as long as possible.
This is why Euro-NCAP generates some uncertainty in the safety ecosystem on purpose. If they spelled out in great detail what would be required in 2029, that would start a race to the bottom to find the cheapest, simplest solutions that only just met the requirements for their desired safety rating. And we all know that cramming just to pass the test is not in anyone’s best interest. That’s not exactly encouraging innovation. So, to start off only a direction of travel is provided, with details becoming clearer as the implementation date comes closer. For example, the final protocol for driver drowsiness and attention was released in March 2025, less than a year before the implementation date of January 2026, with a fairly well established outline of the requirements available about 18 months prior to the implementation date.
Given that the next implementation date is Jan 2029 and assuming a similar timeline as for the 2026 implementation, e should expect a final protocol for 2029 by March 2028, and stable provisionals around July 2027. Or in other words, there’s about two years during which true innovation will influence what will be included. That’s where the tea-leaf reading comes in. So, what do I think we can expect, going by the rather short 2030 vision paper and what we can glean from the activity of other innovators in the field?
Here’s my prediction of 7 things that I think may be included in the Euro-NCAP 2029 implementation in terms of driver monitoring:
Alcohol intoxication will become a serious safety item with detection needed both prior to setting off as well as during driving (to stop people drinking while driving as opposed to driving after drinking), and I’m confident it will reward options where breathalysers are only used to verify when a non-contact system has flagged a high chance of intoxication.
There will be more emphasis on interaction between existing safety features and quickly maturing human sensing capabilities e.g. lane assist and other self-driving capabilities relying on driver attention measurement, or airbags and automated braking systems performing better by making use of the position and body shape of the vehicle occupants.
Fatigue and drowsiness detection will reward prediction of dangerous fatigue levels ahead of time. Rather than detecting that someone is asleep, currently defined as continued eye closer for >= 3 seconds, evidence-based risk of falling asleep will be added. Also, being asleep will be redefined, if nothing else because some people apparently sleep with their eyes (partially) open.
It will become clearer what is meant by ‘relevant medical conditions impacting fitness to drive’, with some specific medical conditions defined, but I predict this will only be presented as only a small subset of possibly relevant medical conditions, leaving the door open for innovation in other conditions. Here are three reasons that this will remain a bit vague: firstly, there are very many conditions that could impact driving. For example, the UK’s DVLA has a list of 177 medical conditions that may affect your driving. Secondly, there is a real possibility that detecting such conditions renders a car a medical device, which requires a type of certification that the industry is new to. Thirdly, it raises ethical questions such as do you need to inform an occupant if a medical condition is detected that they don’t know about yet.
Interior sensing and exterior sensing will be combined so that attention estimation can move from the current ‘driver is looking through the windscreen’ to ‘driver is looking at that pedestrian’. All the sensor data is already available, it is ‘just’ a matter of combining the two. Good luck teams!
Not much progress will be made towards requiring cognitive workload itself to be measured, but attention will become more granular and detailed. For example, it may well become a requirement that a dead stare ahead has to be detected, where the driver is technically looking at the road but not processing the information of what they’re seeing.
Eye health and a person’s ability to properly see the road will become a focus. On the one hand that will mean detecting deteriorating eyesight due to an ageing body and/or health conditions. On the other hand that will mean detecting when visibility is poor due to weather conditions, which can be detected by noticing the driver is squinting, for example.
Time will tell how close these predictions come to the actual implementation. If you want to stay up to date with what’s new in the in-cabin monitoring space, follow me on LinkedIN where I will regularly post updates on this topic.