NCAP won't test collision-avoidance systems on cyclists

Pedestrian-avoidance only criterion, says Volvo researcher

by John Stevenson   October 14, 2013  

Volvo Cyclist Detection System 01

Automotive collision-avoidance systems have been much in the news this year with developments from Ford and Volvo, among others. But the Euro New Car Assessment Programme (NCAP) standards will only test them for their ability to avoid pedestrians.

Speaking to Carlton Reid of and, Volvo software expert Dr Andrew Backhouse said: “Euro NCAP is planning to introduce ratings for collision avoidance systems. 2014 will see the first ratings. There is a lot of talk about testing the system on pedestrians. Nothing is planned for cyclists though.”

In 2012, 5,979 pedestrians were killed or seriously injured on UK roads, 420 of them fatally, so equipping cars with systems that detect and avoid them when the driver fails to  can only be a good idea.

But the situation isn’t much better for cyclists: 3,340 killed or seriously injured in 2012, and 118 fatalities. Manufacturers like Ford and Volvo are claiming that their systems can avoid cyclists - indeed Volvo’s system is called ‘Pedestrian and Cyclist Detection with full auto brake’ - so those claims should surely be tested by NCAP.

Volvo’s system includes adaptive cruise control, which slows a car down if it senses the car in front braking, and was proving “very popular” Dr Backhouse said even though it adds £2,400 to the price of a new car.

The logical extension of all this technology is driverless cars, and companies such as Mercedes and Google have been working toward that goal for a few years now.

Carlton’s article on how driverless cars might affect cycling and cyclists is well worth a read. The CTC’s Roger Geffen said: “It might lead to vast improvements in cyclists’ safety, eliminating the risks from those who drive aggressively, irresponsibly or just without paying attention.”

On the other hand, roads filled with phalanxes of driverless cars would have no place for cyclists.

Driverless cars are suggested as a way to increase road capacity but according to consultant Brad Templeton, “the only way this works is if you remove pedestrians and cyclists from the equation. As long as someone can still step out in traffic who is not controlled electronically, then you really can’t increase speeds and volume much beyond what they are now.”

7 user comments

Oldest firstNewest firstBest rated

On the other hand, roads filled with phalanxes of driverless cars would have no place for cyclists.

Wouldn't they? I'd imagine that you could just pull out into traffic, and the driverless cars would brake, then wait patiently behind you until they could overtake.

two wheels good; four wheels bad

posted by cat1commuter [1449 posts]
14th October 2013 - 16:24


I think the point was, if pedestrians and cyclists were prone to putting themselves in front of - and holding up driverless cars, then there would be demand from motorists to get them off the roads.

Training is like fighting a gorilla. I've never got around to that either.

posted by SteppenHerring [303 posts]
14th October 2013 - 16:50


Before driverless cars become widely available, most of the benefits will be accessible via "Driver Assist" systems, preventing accidents in the same way the control systems of a Boeing 747 stop the pilot from stalling the aeroplane.

Thats what these NCAP tests are aimed at.

Pedestrain detection and other system will mean cars are in lower insurance groups, which is quite important for young drivers.

Its perfectly possible to detect cyclists, but I douby there'll be much interest unless we campaign to make it part of NCAP testing so that a rigorous test and stamp of approval exists.

posted by ribena [169 posts]
14th October 2013 - 17:14


These systems are BS until the fully-fledged self-driving car turns up. I have a car with adaptive cruise control:

- Set the speed you wish to drive at and engage
- In the event of an obstacle being detected in front of you, the car slows to equal the speed of said object

Sounds great. Except the system can't anticipate. "Well that's obvious", I hear you say, but take this scenario:

Conventional driving without this gadget:
You're on the motorway (the only place I use it) and a car changes lanes to move in front of you. Typically, and unless the driver was erratic anyway, you have at least a few seconds warning of such an action - for example, their car was travelling quicker than the one in front, you saw them moving slightly to the right, you saw them indicate, then pull out. You have plenty of time to simply ease off the accelerator and adjust the stopping distance. No drama at all.

Driving with Adaptive Cruise Control:
Their car (or more accurately, the driver) was travelling quicker than the one in front, then moves slightly to the right, indicates, then pulls out. Adaptive Cruise knows nothing about any of this, not even the blatant flashing of an orange light warning of the manouvre. Now the car is in front. ACC makes a calculation of the vehicle's speed, taking nearly a second. It's travelling 10mph slower than you. The radar only works to approx 100 metres so that car is fairly close for you to have your feet nowhere near the pedals. Now it brakes, harder than you would have had you been making the decisions. This startles the drivers behind you, who assume that you're seriously dim not to realise earlier that you needed to be travelling at less speed than you were.

It maintains good stopping distances, but ironically that means that many people simply think that it's a sufficient gap to pull into, making the above happen more often.

This is just an example, but any competent road user, regardless of vehicle, knows that being able to anticipate - even something as obvious as being TOLD via flashing lights that someone else is about to pull out or whatever - reduces the drama no end.

These half-based attempts at self-driving cars are just a combination of Adaptive Cruise, the other now-common gadget that warns you of straying outside your lane (with a mechanism to auto-correct the steering), and a collision-avoidance aspect for pedestrians etc. All it's doing is lulling the driver into a false sense of security about what their car can do.

Proper self-driving cars will be able to, possible in combination with external technology measures, take into account:
A policeman directing traffic
Funny roadworks and temporary traffic lights
A cat/dog running into the road (no idea what the minimum size threshold is for these radars)
Erratic driving from idiot-driven cars
An endless list of other scenarios I can't think of right now.

I do wonder how they will handle cyclists. If they wait until there really is a good view down the road, it's wide enough, there isn't a double white line etc etc, will drivers simply not buy them?

Assuming they can be made to work properly, can detect cyclists, drive safely etc, I say bring it on. Who wouldn't fancy being able to do something else while on the journey? Buy a van and stick the turbo in the back! Sufferfest on the commute.

In theory, there is no difference between theory and practice, but in practice...

posted by notfastenough [3723 posts]
14th October 2013 - 18:08


There's no substitute for careful, sensible driving skills, where the driver respects other peoples right to use the public right of way even if they choose not sit in a motorised tin can!

I agree that more "safety gadgets" do nothing for the real world safety of anyone not sitting in the car. It's just a bunch of nerds playing with their electronics kit with little thought for "if it's actually a good idea in the real world". Makes me feel sad, like selling cars that can do 150 mph for use on roads with a MAXIMUM speed limit of 70 mph. All that is for is making fat old men feel sporty again without having to make any effort.

It's all wrong when someone driving licence is considered more important than public safety by the courts. Until the government start taking this subject seriously we'll have draft "robot car" nonsense to deal with.

Endorphines going up and adrenaline going down, who needs drugs?

posted by banzicyclist2 [285 posts]
14th October 2013 - 20:32


Spot on ribena - automated systems rely on getting humans to set them up with the right algorithms and you can spot systems that fail spectacularly because a detail was overlooked (automated share trading anyone?)

Automated systems without human intervention often go unstable as the feedback loop (error-correct>overcorrect=error-correct>overcorrect) and can often go into a violent and increasing (in amplitude) oscillation. We see this even with human computers in moving motor traffic when vehicles travelling too close with poor driver response suddenly go from travelling at speed to an emergency stop as a wave of deceleration moves through the line after the leading vehicle brakes slightly, and each following vehicle has to brake harder to avoid a collision (untill ultimately a collision occurs)

47 years of breaking bikes and still they offer me a 10 year frame warranty!

A V Lowe's picture

posted by A V Lowe [560 posts]
15th October 2013 - 9:23


I am skeptical about all this sci-fi stuff.

For example, on the bike, on a main road, white van came up behind, overtook terrifyingly close at high speed despite the fact it was a stupid moment to overtake as there was a car coming the other way also at high speed, and as a bonus white-van-man screamed abuse out the window at me as he did so.
Or the day before that, cycling on a minor side road, a BMW on the wrong side of the road suddenly accelerates head on straight at me (presumably just for the laffs) forcing me to swerve almost into the back of a parked car to get out of its way.

How would this technological stuff help when drivers (overwhelmingly young male drivers, it has to be said) are too often quite deliberately threatening and aggressive? Would the BMW have been automatically forced to violently brake (probably throwing the driver into the windscreen with the force of it?)

posted by FluffyKittenofT... [1031 posts]
16th October 2013 - 13:03