The 10 Most Dangerous Roads In The World, And How Self-Driving Cars Would Fare
Let’s talk about dangerous roads.
In a moment, I’ll provide you with a recently published list of the presumed Top Ten most dangerous roads in the world.
Start an Emergency & Disaster Management degree at American Military University.
For some of you, the odds are that you’ll be happy that you’ve never had a cause to try and traverse these bad-to-the-bone roads, while others of you are probably going to put these alarming roads on your bucket list of places you have to go and give a whirl someday.
Do you prefer roads that are calm, easy to navigate, and present little or no qualms?
Or, do you relish roads that are crazy, a wild ride, and for which risk is on your shoulder for each inch driven?
But before we consider the topmost-list, it might be handy to ponder what makes a road dangerous and why we would be willing to agree that a cited road is indeed especially treacherous and worthy of keeping a steel wise grip on the steering wheel.
I canvassed some of my colleagues that are driving daredevils, the type that overtly seeks out these kinds of roads and I uncovered some of the key elements they give top priority to.
First, the road has to be a road and in one manner or another be passable.
This might seem obvious, but the point is that if a road is not really a road and merely a jumble of rocks or a bunch of sand dunes, it doesn’t quite count as a “road” and therefore should not be on a list of the most dangerous ones. You could perhaps place such instances on the most dangerous off-roads or made-up trails, but do not mix them up with actual intended-to-be roads.
Second, the road has to have vehicular traffic that goes on the road.
Once again, this has to do with the notion of whether the road is a road. If there aren’t any cars or trucks or other vehicles that go on the road, it does not seem to be an appropriate candidate for the list. Furthermore, the traffic is actually considered a factor in the dangerousness of the road, namely that it is not just the pavement or asphalt that gets your heart pounding, it is also the other drivers that add to the zany and perilous nature of the journey on the road.
Third, there should be some disquieting number of car crashes or roadway related deaths and injuries that occur on the road.
If the road is truly dangerous, the odds are that car drivers will misjudge and end up in a ditch, or worse, become a casualty of the hazardous road. Now, this can be somewhat misleading or misapplied in that say there is a freeway stretch in a congested city that gets a lot of fatalities, well, it is not necessarily the road per se and perhaps primarily due to the volume of traffic. As such, some suggest using a per-mile metric rather than a raw count of adverse outcomes or otherwise find a means to balance the quantitative numbers against the other factors warranting being considered a most dangerous road.
Fourth, the roadway design and its placement are likely a significant ingredient in the dangerousness.
Generally, roads that weave along a sheer cliff or that try to squeeze between two very tight canyon walls or otherwise present life-threatening pathways are likely considered inherently dangerous, quite obviously due to the apparent risks of driving even just slightly askew. Thus, the roadway design and where the roadway goes are bound to be a vital part of the danger. Just the littlest moment of taking your eyes off the road could lead to a really sour ending of a roadway run.
Fifth, the dangerous road must have stood the test of time.
Here’s where this precept goes. A road that temporarily has a fallen bridge or maybe a massive mudslide, though it’s not a good thing and you could assert that the road is dangerous, such a situation might only be temporary. In the proper spirit of being a persistent danger, the viewpoint is that the road must have been around for a long time and consistently presented itself as a danger. Sure, there are lots of one-time examples of roads that had a dangerous day, but the all-time list ought to be roads that proudly or imprudently have been enduringly dangerous.
Sixth, there must be speed involved.
One supposes that if you could drive a road at a snail’s pace of say 1 mile per hour, it would seem to knock down the dangerousness factor to some degree. Inching along would make things easier for the driver and allow for moment-to-moment re-calibration of the driving effort. On the other hand, if there is the speed involved, perhaps there is other traffic that is desirous of moving at a frenetic pace, this makes the danger come alive, given that you only have a fraction of a second to decide whether the road is curving to the left or the right.
Seventh and the last of this set of criteria or considerations is that opinion matters.
The notion underlying this condition is that the road ought to be one that people acknowledge as being dangerous. If a road is on the list and everyone balks at the inclusion, perhaps this implies that the road is not as dangerous as might be claimed. That being said, do not though be fooled by those smarmy drivers that will shake their head at any road on such as a list and out of the corner of their mouths say that the road is nothing of consequence and they could drive it blindfolded. There are always those sorts of braggarts or malcontents and are not to be taken at their word as puffery arbitrators of what is dangerous or not.
With all of those thoughts in mind, let’s next take a look at a recently reported list of the Top Ten alleged most dangerous roads in the world:
1. The “Street of Death” Road of North Yungas in Bolivia
2. The Road of Jalalabad-Kabul in Afghanistan
3. The Highway of James Dalton in Alaska USA
4. The Highway of Karakoram in Pakistan
5. The Guoliang Tunnel Road in China
6. The Pass of Zoji La in India
7. The Road of Skippers Canyon in New Zealand
8. The Pass of Los Caracoles in Chile
9. The Pass of Stelvio in Italy
10. The Highway of Sichuan
How do you feel about the list?
Also, if you’ve driven at least one of those roads, pat yourself on the back for having survived to tell the chilling tale.
And for those of you that have driven on all ten of the roads, one has to ask, do you have a death wish or are you just that kind of person that loves a good challenge?
Shifting gears, consider what the future will be like when we have AI-based true self-driving cars on our roadways.
Here’s today’s intriguing question: Will AI-based true self-driving cars be able to drive on dangerous roads, and if so, how will they fare in that treacherous endeavor?
Let’s unpack the matter and see.
The Levels Of 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-ons 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 in spite of 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 Dangerous Roads
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.
In discussing the handling of highly dangerous roads, keep in mind the earlier articulated criteria for what constitutes a dangerous road.
This is worthy of a reminder for several crucial reasons.
The most notable reason involves a quite significant matter that surprises many people about the nature of self-driving cars, including startling those that purport to know a lot about self-driving cars.
In the classification used to rate self-driving cars, the aspects of being able to have the AI drive off-road is considered off-the-table. This means that the levels of self-driving do not encompass off-road driving. The standard has nothing to say particularly about off-road driving and considers off-roading to be outside the purview of the existing standard (for my analysis on this off-road versus on-road issue for self-driving cars, see the link here).
That’s a shock to some.
Why wouldn’t the standard include off-road driving, many ask incredulously?
Generally, the thinking is that off-roading is so varied and open-ended that it made more sense to focus the standard toward on-road driving (and, some would assert that we need self-driving for on-road driving, but don’t necessarily “need” self-driving for going off-roading, though this is a debatable contention).
Keep in mind that there isn’t anything that precludes a standard that does focus on off-road, and nor does it preclude the existing standard from being later extended to add off-road driving aspects.
Anyway, in short, there is no requirement in the standard that an AI driving system has to drive off-road, at least in terms of meeting the standardized levels of semi-autonomous and autonomous driving.
Automakers and self-driving tech firms can decide if they want to encompass off-road driving or not do so. One small irony, some suggest, stems from the fact that the early days of self-driving were initially all about doing off-road kinds of driving, such as competitions in the desert, partially to ensure that no one would get hurt by using desolate areas for doing tryouts and experimentation.
This lack of an off-road stipulation does not seemingly factor into today’s question about the dangerous roads, since please recall that the suggested criteria emphasized that the road has to be a road and be somehow reasonably passable as a road.
Back to the matter at hand and the pondering of whether AI-based true self-driving cars could handle dangerous roads, including for example the reported Top Ten such roads.
The answer is somewhat amorphous because it comes down to the driver, namely, the AI system, and whether the AI has been appropriately readied for coping with the conditions and situations of a dangerous road.
Let’s delve into that facet.
For many of today’s roadway tryouts, the AI has been shaped to deal with normal and routine driving conditions. The AI is dealing with driving in quiet neighborhoods, or on conventional highways, or on well-kept freeways, etc. Dealing with a winding road that has severe potholes and makes its way along sheer cliffs, well, that’s not especially what the AI driving systems are yet crafted to do.
Furthermore, recall the point about speed.
If you were to have a self-driving car proceed at 1 mile per hour, the chances of successfully navigating a dangerous road are going to be a lot higher. Speed for AI is about as daunting as speed is for humans, in the sense that the faster the car is going, the harder the driving task becomes, simply due to the need to make split-second decisions and also be aware of the roadway status with little time to figure out what to do next.
Here’s an additional twist.
Would the AI self-driving car have any human passengers in it?
You might be perplexed about why the aspect of having riders inside the self-driving car would be a consideration.
The reason is rather straightforward. The AI has presumably been programmed to keep the car within the allowable limits of what the human body can deal with. For the AI system, making a super-fast and sharp turn is no problem for the AI and nor the car, but the human passenger might get injured, even if wearing a seat belt (due to a whiplash effect).
You could say that AI is hampered by the inclusion of human passengers.
That is obviously the case if the vehicle was being driven by a human, the same limitations would exist.
Taking this idea to another realm, consider what the AI driving system could do if it didn’t need to worry about human passengers. The self-driving car can be completely empty and have no humans present at all, thus, in that use case, it can proceed to drive to the extreme limits allowed by the physics of the car (for more details about how AI driving systems can incorporate race car driver capabilities, see this link here).
Would an unencumbered AI driving system do better on a dangerous road than a human-driven car?
That’s hard to say, since the nature of the road, the speed of travel, and other factors all come to play, along with however the AI itself has been primed for the driving.
This brings up another important aspect.
For some of the existing public roadway tryouts, the roads being used have been pre-mapped and oftentimes have been pre-driven to allow the AI system to get up-to-speed about the roads. This can be handy for the use of Machine Learning (ML) and Deep Learning (DL), allowing the AI to figure out what the road is like and aim to do better on future travels of the road.
One nuance that might not be apparent is that if you have an AI self-driving car in a fleet and it traverses a particular road for the first time, in theory, the result can be shared with all other AI systems of the fleet, thus being able to get those other self-driving cars ready to drive the road too.
For humans, you cannot especially do the same trick, since having one person drive a road might be somewhat handy for others when the person explains what they did, but this is assuredly not the same as being able to transmit the moment-to-moment and detailed driving nuances involved.
In the use case of the dangerous roads, we ought to consider whether a human driver has had a chance to preview the road by driving it perhaps slowly one time and then increasing speed for later journeys. The same aspect can be considered in the case of AI self-driving cars.
Another factor to contemplate involves the risk threshold of the driver.
We all know of human drivers that are willing to take great risks while driving, weaving in and out of traffic and taking chancy moves that increase the odds of getting into a car crash or other adverse result.
For AI self-driving cars, there is an ongoing debate about the threshold of risks that the AI should be allowed to undertake (for my analysis of the risk thresholds of self-driving cars, see the link here).
While traveling on a dangerous road, what should the risk setting be for the AI system?
Presumably, if you dial down the acceptable risk, the AI is going to drive more slowly and cautiously. If you push up the risk meter higher in terms of risk tolerance, the AI will drive the car with greater speed and aim toward the brink of calamity.
As a final quick thought on this topic, consider what human passengers might do when regularly able to go for a ride in AI self-driving cars.
Suppose you are late for work. You urge the AI to push the pedal to the floor and get a move on. Essentially, the human rider is seeking to increase the risks of the driving act.
Should an AI driving system allow for the human riders to make such changes?
In theory, some believe that the AI will and should always drive in the same relatively low-risk way, regardless of the interests or desires of the passengers.
But, right away there are apparent exceptions, such as a passenger that is about to give birth and has to be rushed to the hospital or someone suffering from a gunshot wound or other emergencies that might require taking a riskier driving approach.
Let’s return back to the dangerous road topic.
Will we have AI self-driving cars that will allow us to take a wild ride on a dangerous road, doing so by telling the AI to maximally take risks on such roads, giving the humans quite a thrill (one presumes)?
For now, the automakers and self-driving tech firms have their hands full with getting self-driving cars to safely take people to the local grocery store, and thus this inquisitiveness about coping with especially dangerous roads is considered an edge or corner case (not something to be dealt with right now).
In the future, don’t be surprised if you start to see advertising for brands of AI self-driving cars that showcase they can readily drive on scandalously dangerous roads, which might become a marketing pitch to differentiate one AI driving system from another (some believe that self-driving cars will eventually become a business commodity, see my analysis at this link here).
You’d certainly seek out to take an AI self-driving car to get over to downtown for work if that AI driving system was known to have nimbly and safely handled death-defying roads of grand peril, enough so that you might even take a short catnap on the way to the office.