The Insurance Institute for Highway Safety (IIHS) in the US has released more results for its new small overlap test. This severe test shows what happens when a car just clips another vehicle or a tree at 40 mph. In the US this type of frontal crash accounts for around a quarter of the serious or fatal injuries for front seat occupants. So why shouldn’t Europe also test cars in this way?

Dr Mervyn Edwards, head of structural crashworthiness at the Transport Research Laboratory, says Europe has a bigger problem that needs fixing first. Edwards is one of the researchers who advise regulators and Euro NCAP on how to improve crash safety.

I’ve got to be careful how I explain this. The small overlap situation is an issue in Europe, but Edwards says that currently “deceleration induced injuries related to the restraint systems” is more of an issue. This doesn’t mean people shouldn’t wear seatbelts: seatbelts are essential.

Edwards wants carmakers to improve the way that seatbelts and airbags work: “We need cars that sense the accident severity and then match the performance of the restraint systems to the deceleration and the different sizes and weights of the occupants,” he says. “In many crashes, the car’s structural integrity is OK: the passenger compartment isn’t crushed or ruptured. People are getting injured because they are experiencing a high deceleration as the restraint system restrains them.”

Basically, we need car companies to make the restraint system more intelligent and adaptive. To encourage that, regulators would need to introduce full-width tests carried out at two speeds and with a range of different sized dummies would be ideal, says Edwards. The only drawback is that these would cost too much to do.

Edwards is floating the idea of having one lower speed physical test and a higher speed test conducted using computer simulations. That would allow regulators to see how well people of different weights and sizes are protected. Promoting the adoption of more intelligent restraint systems is important and computer simulations are pretty powerful these days, but how much should we trust the technology when it comes to safety?