Driver assistance: what the test proves, and what it leaves out

Every system here was signed off against a test, and a test is a scenario: stated speeds, a stated surface, stated light, a stated target. A car that passes has demonstrated that and nothing wider. Almost everything a driver wants to know lives in the space between the test and the road.

So rather than four descriptions in turn, four questions asked of all of them. What is each specified to detect, and to do about it? What does the published test measure? What degrades it, according to the maker and the instrument? What is left to the driver? The systems: automatic emergency braking, pedestrian and cyclist versions included; lane departure warning and lane keeping; adaptive cruise; blind-spot monitoring.

What each one is specified to do

Automatic braking is the only one of the four with a full performance standard behind it, and that standard is plain about its reason for existing: FMVSS No. 127, the United States standard for light vehicles, states its purpose as reducing the deaths and injuries from crashes in which drivers do not apply the brakes or fail to apply sufficient braking power. It requires the vehicle to warn and brake when travelling faster than 10 km/h and slower than 145 km/h with a lead vehicle ahead, and faster than 10 km/h and slower than 73 km/h with a pedestrian. The UN regulation used in Europe, No. 152, is written to a narrower band: active at least between 10 km/h and 60 km/h for vehicle targets, and at least between 20 km/h and 60 km/h for pedestrians and cyclists.

The lane systems split in two, and the split matters more than the marketing names. UN Regulation No. 130 defines a lane departure warning system as one that warns the driver of an unintentional drift out of the travel lane — a warning, not a steering input — and applies it to categories M2, N2, M3 and N3, buses and heavier goods vehicles rather than cars. The warning must arrive at the latest when the front tyre crosses a line 0.3 m beyond the marking. For cars in Europe the instrument is the emergency lane-keeping system, which is two things at once — a lane departure warning system active at least between 65 km/h and 130 km/h, and a corrective steering function that intervenes at least between 70 km/h and 130 km/h.

Adaptive cruise control and blind-spot monitoring have no equivalent. The European general safety regulation lists what every new motor vehicle must carry — intelligent speed assistance, alcohol interlock installation facilitation, driver drowsiness and attention warning, advanced driver distraction warning, emergency stop signal, reversing detection, an event data recorder — and adds advanced emergency braking and an emergency lane-keeping system for cars and light commercial vehicles. Neither list includes adaptive cruise control or blind-spot monitoring. What those two do is whatever the manufacturer built them to do, and the document that says so is the owner handbook.

What the published test actually measures

FMVSS No. 127 sets out its scenarios in the same detail as its requirements. The lead vehicle is stopped, or moving more slowly at 20 km/h, or decelerating at between 0.3 g and 0.5 g from a matched speed; the stopped one is approached at up to 80 km/h, or 100 km/h with a manual brake application. The pedestrian mannequin crosses the path, stands still in it, or moves along it, and the crossing tests run in daylight and again in darkness, on lower and upper beams.

The conditions are specified just as tightly. Temperature between 0 and 40 degrees Celsius. Wind no greater than 10 m/s in the lead-vehicle tests. A dry, uniform, solid-paved surface with a stated friction coefficient. Illumination at or above 2,000 lux for the daylight tests and at or below 0.2 lux for the dark ones. And then the sentence that does more work than any other in the standard: testing is not conducted during periods of precipitation or when visibility is affected by fog, smoke, ash or other particulate. The European braking regulation defines its conditions the same way, and the lane departure regulation adds one the braking tests have no equivalent for: lane markings in good condition.

None of that is a criticism: a test that allowed rain would not be repeatable. It is why the test cannot answer what a driver is actually asking, which is about a wet November evening on a road whose markings were last painted six years ago.

The independent test bodies push at the edges of the same box. The Insurance Institute for Highway Safety’s vehicle-to-vehicle evaluation runs at 31, 37 and 43 mph against a passenger car, a motorcycle and a dry van semitrailer, and against the semitrailer only the warning is tested. The choice of targets is itself a finding: you add a motorcycle when you suspect the systems are worse at them.

What the makers and the instruments say degrades them

The most candid sentence in any of these documents sits in the European braking regulation, which records that the performance it requires cannot be achieved in all conditions, and names vehicle condition, road adhesion, weather, deteriorated road infrastructure and traffic scenarios among the things that may affect it. A regulator writing down the limits of its own requirement is worth more than anybody’s reassurance.

The specifics below that belong to individual makers and must stay attached to them, because one maker’s stated limit is not a fact about every car. Kia’s handbook, for its forward collision-avoidance system, lists a camera lens contaminated by a tinted, filmed or coated windscreen or by damaged glass; sun glare obstructing the camera’s field of view; heavy rain, heavy snow or thick fog; a sudden change in brightness, such as entering or leaving a tunnel; a pedestrian or cyclist in clothing that blends into the background; and driving uphill or downhill. For lane keeping, the same handbook names markings covered with rain, snow, dirt or oil, markings that are indistinct or damaged, nearby markings that look like lane markings, and a vehicle ahead covering the marking. For blind-spot warning: a curved road, a sloped road, another vehicle passing so fast it is alongside only briefly, snow or rain over the rear corner radar. For adaptive cruise control: snow, rain or foreign matter on the front radar cover, reducing detection and temporarily limiting or disabling the system.

Read together, those four lists are one list: every system here is a sensor plus a decision, and every sensor degrades on dirt, water, glare, geometry, and anything that does not look like what it was trained and tested to recognise. The handbook supplies the sentence that ought to be on the dashboard — the system may not operate properly even though no warning message appears on the cluster.

What is still the driver’s job

The instruments answer this the same way, and none leaves it ambiguous. The federal standard defines forward collision warning as a signal designed to induce an immediate crash avoidance response by the vehicle operator — its specified purpose is to make the driver act. The European braking regulation requires that the driver be given the means to interrupt the warning and the braking, while making the system hard to abandon: it returns at each new engine start and cannot be switched off above 10 km/h. The general safety regulation states that the driver must be able to override these systems. Emergency lane keeping is permitted to suppress itself where markings are absent, unclear or in poor condition, or where another system is intervening — which the driver learns about, if at all, afterwards. And the mirror requirement is there too: the design must minimise warnings and braking where there is no risk of an imminent collision, because a system tuned to miss nothing would brake for bridges.

A fleet can know this, and usually does not

Both regimes reach into commercial vehicles: the federal standard covers trucks and buses at or below 4,536 kilograms, and the European requirements cover light commercial vehicles too. A company running vans is buying these systems whether or not anyone chose them.

Choosing them is the part that gets skipped. These systems are specified at purchase, on the order form: a fleet that decides once what it wants on every vehicle can brief drivers about it, and one that takes whatever each dealer had in stock cannot. The second costs nothing: knowing what you already have. Most fleets cannot say which of their vehicles has automatic braking and which has nothing, because nobody wrote it against the registration numbers — and that record is the precondition for driver briefings, for incident analysis, and for knowing whether the van that keeps getting rear-ended is the one without the warning.

What we cannot tell you

We cannot tell you how much any of these reduces your own risk. The published effectiveness work measures populations of fitted vehicles, not people, and the honest versions say so. What it gives is direction and size: the Insurance Institute for Highway Safety reports that forward collision warning with automatic braking cuts rear-end crashes in half, warning alone reduces them by 27 percent, automatic braking that recognises pedestrians cuts pedestrian crashes by 27 percent, and blind spot detection reduces lane-change crashes by 14 percent. Its analysis of police-reported crashes from 2009 to 2015 associated lane departure warning with 11 percent fewer single-vehicle, sideswipe and head-on crashes.

Those figures carry their own warnings, and the warnings are the interesting part. The same research notes that drivers frequently switch lane departure warning off, which drags the measured benefit down and says something about the system rather than the driver. Insurance claim rates have not moved for it the way police-reported crashes have. And pedestrian systems, the institute says plainly, can struggle to detect pedestrians in the dark, which is where pedestrians are hit.

We also cannot tell you what any particular insurer does about a car with these systems fitted. We read no published statement of it from any insurer, so we are not going to describe one.

Before the next long drive

Clean the windscreen in front of the camera, not just the swept area, and the bumpers where the radars sit. Read the handbook pages for the systems your car actually has — the only place the limits are written down, and pages most owners never open. Then, once, on a quiet road, find out what your car really does: whether the lane system buzzes or tugs, and whether the dashboard tells you when a system has quietly given up.

The specification is a promise about a test. Everything past the edge of the test is inference, and it is yours to make.