Volunteering to be a test subject for Blindspin (project page) means that you may spend an hour or two of your life bicycling back and forth on a track and looking weird, probably picking up a heavy sweat, and potentially risking your personal safety. Why would anyone do that?
Short answer: because no one else has ever done this. Science means that you methodically pile bricks to create something useful. An individual brick is just a turd-colored clump of burned clay, but they all need to be in place. And in this particular niche of cycling safety, no one has really even picked up the first block.
What is the niche, and why is it interesting? The idea for this research spun off from work we’ve done in driver distraction. A large number of accidents are caused by the driver’s attention wandering to something irrelevant (these days, largely using a mobile phone). Yes, there are brute-force political ways to handle that particular problem, such as outright bans on mobile phone use (which don’t really work very well).
But if we actually want to approach the problems scientifically, we must ask simple-sounding questions that are measurable. In this particular case: how do we measure what driver distraction even means? How do we find numbers that allow us to compare how dangerous driver distraction would be in different scenarios?
We need those numbers before we can even think about how to think about answering more directly practical questions, such as: how do we design car interfaces so that they do not cause dangerous distraction; how do we intelligently attack the problem of driver mobile phone use; and even, how do we design roads (especially intersections) so that if and when distraction does occur, the impact will be as small as possible?
For cars, a lot of research has been done, and some of it has found its way into safety recommendations (very slowly, but steadily). With cars, there are well-established safeguards that can alleviate the effects of such distraction — lane markings, brakelights, and standards for traffic light design, for example. For cycling, such safeguards have been far less systematically studied.
The science of car road design. (Source: paulstaubin.ca)
The reality of cycleway design (Image source: madcyclelanesofmanchester)
In cyclist distraction, there have been a few mainly Dutch studies in the last few years, but nothing has been done in true field conditions. Right now, we don’t really even know how to think about measuring the numbers that are needed. That, in a nutshell, is what Blindspin is about. We want to get numbers that are needed to even kick off the field of cyclist distraction.
To make things measurable, we need to start from a concrete task, and so we ask how important vision is for safe cycling. What we are most interested in is cyclist distraction: what happens when a cyclist’s attention is drawn away from the road in front of him? What are the conditions in which this is particularly dangerous? Slow speeds? High speeds? Slightly curved cycleways? Narrow cycleways? Poor-quality cycleways?
We approach this question by the visual occlusion method, which works in a reverse way: rather than measuring the conditions in which say a one-second glance away from the path is dangerous, we want to measure the conditions under which such a glance is probably safe. To do that, test subjects need to wear a set of special goggles which can go dark or transparent by a key press (controlled by the subject). By studying how much darkness the test subject can tolerate, we get an estimate of the safe time lengths.
Without those numbers, there is no intelligent way to start asking more practical questions about cyclist distraction.
A set of professional occlusion goggles. We will be hacking ours from 3D glasses. Image source: CogLens
This is not going to change the world, but no one can change it if these numbers don’t exist. And it is extremely difficult to get these numbers while working within the confines of a university system. So, since we can do this outside university confines and with a budget that we are able to pay from our own pockets, we’ll just do it on our own.
The data will be made openly available after the project (though we need to anonymize them first), and if we are at all successful, we will theory and results as a real scientific paper (whether it will be published is another matter altogether). This is serious science, under all the craziness.
That, in a (long) nutshell, is why someone might want to volunteer. In a later posting, we will describe what the test subjects would actually have to do. But roughly speaking: one to two hours of cycling, wearing a set of goggles that sometimes go dark. If you are a moderately insane cyclist in the Turku region, and are potentially interested in volunteering, please contact Jakke.Makela{at}gmail.com.
The rest of this posting is more hardcore information for people who want to understand the science in more detail.
What is visual demand?
All research rests on one parameter: visual demand. Visual demand is an area which almost no one has tried to popularize — and visual demand is what we want to measure. We will now wave hands a lot, and try to popularize it.
In the way we use the term, visual demand refers to the amount of visual information that is needed to perform a given task — in this case, driving a car or a bicycle safely.
What does this “amount of visual information” mean? That is where things immediately get tricky. In older literature, the attempt was made to measure the number of “bits” that are needed to encode a given piece of road. In practice that is too simplistic.
Research is still fairly inconclusive, but we will use a simple definition: During a fixed 500 millisecond viewing time, a cyclist gets visual information about the track in front of him. Even if he looks elsewhere, he can use that stored information to continue cycling for a while — that is, he can genuinely drive blind. This skill is actually needed for safe driving; for example, it is necessary to check out other traffic at intersections. The area of focal vision is so tiny that such a glance outside causes, in effect, a moment when the driver’s focus is not on the road ahead.
It is known that for automobile driving, this occlusion time (”blind time”) is usually 500-1500 milliseconds, that is, not very long. A 500 millisecond glance is about the shortest glance that is possible to give any useful information — for example, a quick look at the speedometer takes approximately that long. Up to 1500 ms is not a problem on an empty and relatively straight road, although it can be enough to rear-end a braking car in traffic. There is a sharp cutoff above that, and for most situations, anything above 2 seconds is considered risky.
In a research paper we have submitted recently, we converted those occlusion times to occlusion distances, that is, the distance which can be driven blind. We found that the occlusion distance varies quite a lot from person to person and road type to road type, but on a straight road is around 10-15 meters.
No one knows what these limits are for cycling. If they are roughly the same, we can directly use lessons learned from car driver distraction. If they are very different, cyclist distraction will need to be studied from first principles.
It is also known from eye-tracking studies that experienced drivers do not move their eyes randomly. Typically, for the operational control of the vehicle, car drivers keep their eyes fixated at points that are located quite far away ahead, and use their peripheral vision to keep on the lane. We hypothesize that the situation will be much more complex for the operation control of cycling, and most likely will decrease occlusion times dramatically when there is even slight unevenness in the road.
We suggest several study hypotheses.
Visual input is needed for balancing
Unlike car driving, bicycling has two factors that cause uncertainty: the requirement to stay upright (equilibrioception), and the requirement to maintain direction and control of the bike. Car driving has only the latter.
We can suggest that at low speeds, the vestibular sense is not sufficiently sensitive to maintain balance on its own; visual information is needed besides it. The exact details are excruciatingly complex (see Wikipedia entry on bicycle dynamics).
There are two regimes of interest. At very slow speeds (near walking speed, about 5 kmh) it is difficult to maintain balance. This speed range is relevant when driving in traffic with pedestrians. The hypothesis is that near walking speeds, the visual demand will be anomalously high because of the balancing requirement.
At higher speeds, self-stability may (or may not) cause interesting effects. At some speed a typical bike will achieve self-stability, meaning that even if the driver were to disappear, the bike would continue moving. At this speed, there should be no need for visual input to keep balanced.
The self-stabilizing speed is usually between about 18-24 kmh (5-8 m/s) which is quite high for our purposes. The 85th percentile of urban speeds has been clocked at about 22 kmh, which would mean people rarely drive at the self-stabilization speed.
Curves raise visual demand dramatically
Turning in a curve requires a complex combination of counter-steering, steering, tilting, and moving the center of mass. We hypothesize that even a small curvature in the road would cause a rapid rise in the visual demand.
Eye fixations
We hypothesize that eye fixations needed in cycling are more complicated than for car driving. For an experienced car driver, the eyes normally just need to focus on distant points in the field of view (at least on a relatively straight and traffic-free road). For cycling, we suggest that the eyes will also need to scan the region right in front of the bike, as small-scale bumps can be dramatically unsafe.
This would need to be tested with an eye-tracking camera, but occlusion tests could give some clues. In particular, the 500-millisecond unoccluded time used in automotive research would not be sufficient. We should see this in the pre-tests already. More crucially, if this is correct, then any unevenness in the road should lead to much shorter occlusion times. Since the eyes have to scan constantly, they cannot afford to lose information.
The text was written by Jakke Mäkelä and Tuomo Kujala
See also Blindspin project page.
