BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Painting On City Streets Can Impact Traffic Safety, Good Or Bad, For Human Drivers And For AI Self-Driving Cars

Following
This article is more than 3 years old.

If you follow the news, you most likely have seen the recent efforts of numerous cities that have been painting large block letters onto their streets.

In many cases, the streets are considered active in that once the painting has been completed, the roadway is reopened to everyday traffic. Thus, these are not seemingly specially set aside streets that are secured from vehicular traversal and instead are put back into their usual service after having been painted.

Customarily, paint that is used on an active traffic-going street is employed as a traffic control device, known in the roadway infrastructure realm as “road surface markings” and are used to officially depict navigational guidelines and directions.

When drivers proceed along a street, they are at times provided visual cues via painted asphalt surfaces that showcase where the median is, where crosswalks are, and generally is indicative of the curbs and other key roadway features. The colors of yellow and of white are particularly reserved for these purposes and drivers are accustomed to noting where those painted lines and areas are.

A driver that is familiar with a given street is apt to no longer overtly notice the painted surfaces, though they seemingly subconsciously still pay attention to the guidance and use it reflexively as they drive down a street so marked.

Drivers that have not previously entered onto a particular street and are approaching it for the first time are likely to be directly cognizant of the painted guidance, using it actively as they attempt to safely make their way along that street.

Federal standards for the use of painted surfaces as a traffic control device are included in a USA governmentally approved document referred to as the MUTCD (Manual on Uniform Traffic Control Devices for Streets and Highways, available at this link here) as published by the Federal Highway Administration (FHWA) and the Office of Transportation Operations (HOTO).

Additionally, the FHWA issued an additional official ruling on the various uses of painted surfaces as a supplemental memorandum (see the link here).

The supplement memorandum is often cited when examining what has been an already rising interest by cities in painting murals and other artistic renderings onto active streets and in-use roadway surfaces.

Why is painted art on the roadway worthy of added consideration and potential concern?

Some have expressed qualms that the artistry displayed could be distracting to those driving on such streets.

A driver might mistakenly interpret a portion of the artistic rendering to be a driving guidance directive and therefore drive improperly, either illegally driving or potentially driving in a means that could endanger themselves and other nearby drivers, perhaps also jeopardizing pedestrians.

Or, a driver might become distracted by the artistic presentation and thus fail to realize that a car ahead of them is braking suddenly, or that a pedestrian is jaywalking in front of the car. As such, the driver might plow into another vehicle or ram into a pedestrian as a result of being focused on the art and bereft of attention to the driving situation.

Another possibility is that the painted art has overlapped, obscured, or confounded the intended painted traffic control surfaces.

Suppose that an artistic mural extends over a crosswalk and as a result, the formal crosswalk-painted lines are less identifiable or possibly even no longer discernable at all.

Thus, the paint markings intended for traffic safety are no longer viably able to be seen and the drivers of the roadway cannot as readily gauge the nature of how to best navigate the street.

The use of painted surfaces as a traffic control mechanism is supposed to be a visual indicator for drivers and something that is observed and abided by on nearly an instinctive basis while driving. Indeed, the use of nationwide standards has provided a uniformity that enables drivers to minimize uncertainty about what the paint is trying to tell them, becoming second nature in detection, and decidedly is not supposed to distract drivers from the life-or-death nature of the driving task (i.e., the painted surfaces are considered a crucial aid to the driving task and the safety thereof).

So, when there is paint used on an active roadway surface and that paint does not have an ascertained traffic guidance purpose, one logical question arises as to whether the painted conveyance will undercut traffic safety or whether it will be neutral or, surprisingly to some, possibly even bolster traffic safety.

Let’s take a look at the potential outcomes.

Anticipated Driving Results Per Added Painting

Art murals are usually an aesthetic treatment, generally done for artistry purposes, and not as a goal of aiding traffic per se.

Likewise, it would seem that the use of large block letters is not usually being undertaken as a traffic control means.

Some argue that any such painting that is not intended for bearing on roadway traversal or street navigation is usurping safety and will ultimately generate added traffic injuries and fatalities.

Obviously, if so, this would be an adverse unintended consequence.

In that frame of reference, what trade-off is to be used when weighing the value of the painted surface that is not for roadway navigation versus not putting such painted artistry or messages onto the given street being otherwise used for designated driving purposes?

Even formally added painted conventions for roadway control tend to undergo scrutiny.

For example, recent extensions of the roadway painting approach for driving purposes have included the adoption of green painted lanes and street markings to designate bicycle riding zones. This is being done to encourage bike riding and hopefully guide car drivers to be cautious and aware of nearby bicyclists. Nonetheless, studies have been undertaken about the impacts on driving safety and whether the green painted markings are worthwhile or not.

When other non-driving painted efforts opt to use paint that is yellow, white, or green, do those choices possibly make differentiating the actual roadway surfaces even harder for drivers?

Do drivers potentially become numb to any painted markings if the use of paint on the roadway surfaces is undertaken for all sorts of purposes?

Note that this kind of painting is also a potential complication for pedestrians.

Pedestrians might assume that a painted area that seems to have a message or artistry is reserved for access by pedestrians and thusly wander into the street to more closely examine the painted surface. Those pedestrians might become distracted as they perhaps take pictures or look directly at the ground, meanwhile failing to be observant of cars that might be coming down the street.

Furthermore, pedestrians coming upon such a street are perhaps apt to enter into the street and disrupt traffic, causing driver delays, which some assert can foster road rage or other untoward driving acts.

Those with a penchant for eschewing the use of paint for anything but roadway control are prone to arguing that the streets should first get a makeover of the potholes and cracks, for which the added paint can at times hide. This hiding factor means too that drivers might fail to avoid those roadway imperfections, damaging their cars or causing them to drive awry, and too that the now concealed nature of the potholes and cracks will diminish the odds of infrastructure improvements to remove or fix those roadway blemishes.

It might seem at first glance that the stance of painting non-driving related exhibitory displays is fraught with so many downsides that it ought not to be undertaken altogether.

There is another side to this coin.

Some point out that there are potentially traffic-calming effects that can bolster safety on those streets.

Presumably, drivers will slow down due to the extra painted surfaces, either due to lack of familiarity and wanting to figure out how to navigate the street or due to being interested in whatever the added painted aspects are seeking to convey.

Coupled with this slowing down is the possibility that the drivers will be triggered to be more alert than normal.

Whereas they might have mentally zoned-out and not given any direct attention to the street and any nearby pedestrians, the special nature of the additional markings has jolted them out of their usual driving stupor.

Research studies that have examined these types of matters are often difficult to compare as to conclusive findings due to the variability in such facets as the extent of the painted surfaces, the pre and post traffic patterns in existence on any given street, the temporary time-frame versus permanent application of the painted surfaces, and so on.

Lawsuits brought to contend with the added painted efforts can at times get mired in the jurisdictional boundaries involved, such as whether a particular street is under the guise of the local authorities or the state or federal jurisdiction.

One thing that can be seemingly inarguably said is that the painting of streets beyond the conventional roadway markings will undoubtedly continue for some time to come.

This brings up an interesting question: Will AI-based true self-driving cars be able to cope with painted roadway surfaces that are beyond the scope of conventional roadway markings?

Let’s unpack the matter and see.

The Role of AI-Based Self-Driving Cars

True self-driving cars are ones that the AI drives the car entirely on its own and there isn’t any human assistance during the driving task.

These driverless vehicles are considered a Level 4 and Level 5, while a car that requires a human driver to co-share the driving effort is usually considered at a Level 2 or Level 3. The cars that co-share the driving task are described as being semi-autonomous, and typically contain a variety of automated add-on’s that are referred to as ADAS (Advanced Driver-Assistance Systems).

There is not yet a true self-driving car at Level 5, which we don’t yet even know if this will be possible to achieve, and nor how long it will take to get there.

Meanwhile, the Level 4 efforts are gradually trying to get some traction by undergoing very narrow and selective public roadway trials, though there is controversy over whether this testing should be allowed per se (we are all life-or-death guinea pigs in an experiment taking place on our highways and byways, some point out).

Since semi-autonomous cars require a human driver, the adoption of those types of cars won’t be markedly different than driving conventional vehicles, so there’s not much new per se to cover about them on this topic (though, as you’ll see in a moment, the points next made are generally applicable).

For semi-autonomous cars, it is important that the public needs to be forewarned about a disturbing aspect that’s been arising lately, namely that despite those human drivers that keep posting videos of themselves falling asleep at the wheel of a Level 2 or Level 3 car, we all need to avoid being misled into believing that the driver can take away their attention from the driving task while driving a semi-autonomous car.

You are the responsible party for the driving actions of the vehicle, regardless of how much automation might be tossed into a Level 2 or Level 3.

Self-Driving Cars And Painted Roadways

For Level 4 and Level 5 true self-driving vehicles, there won’t be a human driver involved in the driving task.

All occupants will be passengers.

The AI is doing the driving.

To drive a car, the AI relies upon a slew of special sensors that are added to a vehicle, including various cameras, radar devices, LIDAR, ultrasonic, thermal imaging, etc. The sensors are used to collect an indication of the surroundings and then synthesized together in an approach called sensor fusion or sometimes referred to as MSDF (Multi-Sensor Data Fusion).

The automakers and self-driving tech firms are each opting to construct sensor fusion algorithms of their choosing, and likewise selecting hardware sensors of their choosing. This in turn means that it is problematic to make overly broad statements about what any specific brand or model of self-driving car might be able to do or not do.

In any case, generally, most of the self-driving cars are programmed to detect the painted surfaces of the roadway.

In fact, especially during the earlier days of self-driving car development, the painted lines and other markings were particularly crucial, and some might argue they were overly relied upon. A frequently employed “trick” or technique has involved a “follow the line” algorithm by the AI system and can be a dangerously simplistic means of driving (see my analysis at this link here).

The concern about this kind of overreliance stems from the possibility that painted lines and markings can readily fade over time, thus, being excessively dependent on such painted aspects alone is worrisome and could lead to adverse results.

Also, often there are painted lines and markings that are allowed to be discontinued in use and new ones painted elsewhere instead, yet the older and no longer used painted indications are still visible and can be (undesirably) detected.

Human drivers can get tripped up on such matters too, though they tend to be experienced in realizing that faded lines are not to be strictly observed and that the stronger or newer painted surfaces are likely to be the ones of attention (though, this is not necessarily an easy differentiation for humans to discern either).

Okay, so the AI of self-driving cars usually is programmed to use the camera images to look for painted surfaces.

So what?

Well, based on those analyzed images, the AI then directs the driving controls of the car, choosing to steer, hit the gas, or use the brakes, partially as a result of attempting to interpret the painted surfaces.

Some advocates of self-driving cars have vigorously been pushing to have the national roadway infrastructure get newly painted, freshening up and correcting potentially improperly painted areas, along with including specialized paints that are more modern and readily detectable.

Roadway infrastructure bills and propositions are being asked to include such matters (see the link here).

The point is that by making those changes, it could bolster the advent of self-driving cars, which in of themselves are proffered to provide numerous benefits for society, including reducing the annual driving injuries and fatalities, along with becoming a mobility-for-all capability.

There are some in the self-driving car arena that fervently contends that the paint is the paint, meaning that the AI ought to be able to cope with the same faded and muddled painted surfaces that human drivers do, thus, there is not a need per se to make such changes, though they would tend to energetically support those changes as something that would enhance both human driver safety and AI self-driving car usage at the same time.

Of course, the AI is generally crafted to assume that the painted surfaces are existent for purposes of traffic control and roadway navigation, just as humans are likewise trained and become familiar with the same uses of painted markings.

How well does the AI do when painted roadway surfaces diverge from the norm?

It depends.

First, please know that today’s AI does not have any semblance of common-sense reasoning (for more on this, see this link here), and nor is the AI of today anywhere near being sentient or having reached the so-called singularity (a time at which AI supposedly will cross over into becoming a sentient being of some kind, for more about this aspect see my discussion at this link here).

This is important to keep in mind since the AI is not going to magically or miraculously divine that an artistic mural is indeed an artistic mural, and nor that large block letters are a message beyond the scope of traffic navigation.

Generally, the AI would likely attempt the usual visual processing analyses and if it could not discern the true traffic-intended painted surfaces, it would likely fall back on its other sensory devices as an added aid (not necessarily for figuring out the paint, but for figuring out the roadway structure and its shape and form).

One would hope that the AI has been crafted to assume that the cameras might be non-functioning or faltering, therefore the AI should have fallback postures already instilled. We all know that when in snowy conditions, for example, the street markings are likely to be obscured, and thus the AI if well-rounded would be prepared for circumstances of something like added paint that obscures the official roadway painted markings.

What seems to be a bit more challenging consists of the painted markings potentially being confounded with the official ones, in which case the AI could be misled into assuming it “knows” the street guidance and yet it is misinterpreting the painted surfaces (which, could happen to human drivers too).

Many of the self-driving cars are being pre-loaded with detailed maps of the cities that they are driving in, and for which this then can potentially help to contend with situations involving roadway over-painting and the like.

By-and-large the automakers and self-driving car firms are including in their initial public tryouts the use of a human driver onboard, also known as a safety or back-up driver (for more about this role, see my discussion here), doing so as the self-driving car is repeatedly perusing back-and-forth in a city area, attempting to train the AI on the driving particulars on those streets and yet having a human driver for any just-in-case circumstances.

Suppose that after being so trained, the AI is allowed to drive a self-driving car and there is no human back-up driver included.

Any subsequent painted changes might be potentially internally realized by the AI, being able to rely upon the prior “learned” image processing of those same streets and thus not become disturbed or distracted by the latest art mural or messaging that has been painted on a street (via the use of Machine Learning or Deep Learning, see more at this link here)

Naturally, this type of fallback has to be performed with caution since the AI cannot immediately necessarily attribute the newly painted surfaces as being done for artistry or messaging in contrast to a city having painted new traffic control markings.

Conclusion

We might reasonably assume that the AI if properly devised will be able to generally contend with the added painted markings on roadway surfaces.

There is an added twist though.

Since we are going to have a mixture of both human-driven cars and AI self-driving cars, likely for decades to come, if the human drivers are kilter over the painted surfaces and opt to drive recklessly or without proper attention, there is a heightened chance that the self-driving car could ram into such a vehicle or that a human-driven car might ram into a self-driving car.

In any case, one does wonder that if AI progresses sufficiently, will it one day be driving along and upon detecting a painted art mural or a painted block letters message be able to robustly “understand” what it is, and maybe engage the passengers in a thought-provoking dialogue about the meaning and substance of the painted roadway aspects?

Time will tell.

Follow me on Twitter