When a tree falls onto a self-driving car.
getty
A popular riddle that has been asked since at least the 1800s is whether a tree that falls in a forest will make a sound if there isn’t anyone around to hear it.
This is a fun question to ask a child, ostensibly getting them to think stridently about something that appears simple and yet can actually entail great complexity. Adults too are often charmed by pondering how to answer this seemingly outside-the-box kind of curiosity.
Decidedly, there is a multitude of ways to answer the question.
One somewhat abstruse approach is notably conceptual and exceedingly philosophical, namely questioning the question itself. Perhaps a tree or otherwise anything else of any kind does not exist at all if there isn’t a human present to witness it. The existence of reality is only valid when human presence occurs. In short, there isn’t a forest and there isn’t a tree, until or when a human strolls along and experiences it.
Whoa, that’s pretty far gone and not the answer you would likely share with a child that is trying to merely ferret out this quandary about a sound in the forest as instigated by a falling tree. A more likely way to answer the question involves pointing out that a falling tree will disturb the surrounding air and create a series of airwaves. Those airwaves can in turn reach a human ear and cause a vibration that the sense of hearing deciphers as a sound.
Using this assuredly scientific form of reasoning, you can explain to a child that the question at its core has to do with what we all agree as to the definition of the word “sound” and thus gets them thinking seriously about words and their meanings. If you define a sound as only that which is heard by a human, well, in that case, since the question states there isn’t any human around, then obviously ergo there is no sound produced. You can enlarge the definition by stating that any living creature that can detect sound might be part of the definition, such that even if a human wasn’t there but some other animal that has hearing was nearby, you could reasonably conclude the falling tree did make a sound.
And so on.
Speaking of falling trees, sometimes a tree collapses or falls unexpectedly and catches a nearby human unawares.
The sound of the tree as it is falling might not provide sufficient notice to alert the person and therefore adversely undermines the needed reaction time to avoid getting struck. Or the sound is muffled and nothing more than a whisper, essentially bringing forth a silent falling of a tree. There is also the possibility that a human is wearing earplugs and thus cannot readily hear the falling tree, along with the instance too of a deaf person that would not be able to hear the falling tree at all.
There are news stories aplenty of falling trees that ended up harming a person that just so happened to be sitting under the tree or perchance was walking past the tree at the moment it fell. You don’t have to be in a forest for these kinds of deadly encounters. Our cities and suburbs have trees too and are usually placed in such a manner to invite people to come near them, by design. The idea is that humans relish being near trees, seeing trees, and otherwise welcome trees, most of the time. All told, in the United States, there are an estimated 250 billion trees (per numbers proffered by the United Nations FAO), though this staggeringly large number encompasses the vast forest lands too.
You don’t necessarily have to be standing or sitting outdoors to get struck by a falling tree. A variant of the falling tree scenario involves trees that manage to fall onto cars. If a person is inside the car, they are potentially going to get injured or in a worst-case possibly be killed.
We seem to readily accept the notion that a tree could fall onto a parked car, presumably because we know that a parked and unmoving car is a sitting duck. Though a human might be inside the vehicle and presumably could drive the car away before the tree strikes the car, this would generally require the person to be seated in the driver’s seat, they would need to start the engine, they would need to engage the car into gear, and they would need to drive away from the scene. That’s a lot of activity and seemingly too much to undertake unless the tree falls somehow in miraculously slow motion.
The instances when a tree falls onto an actively in-motion moving car and causes injuries or fatalities is perhaps the most startling of these tree endangering scenarios. It inexplicably seems like extraordinarily unlucky odds to be in a moving car that perchance goes past a tree that opts to fall at the same moment that the car was next to the tree. A few seconds earlier or later, possibly a few split seconds, and the car would have not been in the wrong place at the wrong moment in time.
A recent news report laments that a falling tree in Ohio incredibly clipped two vehicles as they both were in motion, whilst both were passing the tree at the same instant in time, leading to injuring one driver and killing the other driver. Unimaginable, and starkly tragic.
Shifting gears, this whole discussion seems to beg an unasked question, what causes a tree to fall over?
One supposes that if you asked a child whether a falling tree would make a sound, a smarmy kid might instantly ask why the tree fell. Perhaps the youngster is challenging the premise of the question. Maybe the child figures that if they knew what led the tree to fall, it might provide a clue to answering the overarching question about the sound-producing aspects. Of course, it could be that the youngster is stalling for time or might be seeking to avoid answering the question altogether.
Anyway, trees fall for a variety of reasons, plus we need to include the use case of a tree limb that falls versus the entirety of the tree per se. If the tree in totality falls, the odds are that the roots of the tree were probably in trouble or there was some huge gust of wind that shoved the tree from its foundational mooring. Limbs of a tree might break off and fall, due to age, or uneven growth, or perhaps a child had climbed on it and tried to get it to fall (hey, that’s another reason why a youngster might ask why a tree fell, potentially worried they would get accused of causing it to happen).
Returning to the situation of a tree falling on a car, we can contemplate such circumstances in the future and wonder whether the adoption of self-driving cars will come to play in this tree falling malady.
Here’s an intriguing point: Will the advent of AI-based true self-driving cars imply that we’ll never worry again about trees falling onto cars or is there still a chance of this undesirable problem occurring?
Let’s unpack the matter and see.
Understanding The Levels Of Self-Driving Cars
As a clarification, 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 (see my explanation at this link here), 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 contend, see my coverage at this link here).
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 Falling Trees
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.
We can begin our assessment of the tree falling aspects by pointing out that since the AI is the driver of a self-driving car, there would presumably be no one injured or killed if the self-driving car is absent of any passengers and meanwhile the vehicle happens to get struck by a falling tree. In a conventional car, there would always need to be a human driver, at least, and thus anytime a conventional car is in motion there is a possibility of a human getting injured or killed. Not so with a true self-driving car.
You might be perplexed as to why a self-driving car would be driving around and somehow not have any passengers. This is easily answered. First, a self-driving car might be roaming, akin to how ride-sharing cars do so today, hoping to be in the nearby area whenever someone needs a lift. Second, a self-driving car might be in the act of delivery, such as delivering groceries to someone homebound that ordered from their local store, and the self-driving car is being used essentially as a cargo mule. Third, the self-driving car might be driving to a designated location to pick-up a passenger. Etc. In short, you should expect that we’ll see a lot of self-driving cars wandering around on our streets and highways, empty of human passengers, likely for a lot of their driving time (see my analysis at this link here).
This situation of not having any passengers does not though preclude the possibility of humans getting hurt when a tree falls onto a self-driving car.
Suppose an empty self-driving car is minding its own business and driving down a busy city street that is jam-packed with pedestrians on the sidewalks. With the darned worst of chances, a tree falls onto the self-driving car. Fortunately, since there aren’t any people inside the vehicle, no one gets hurt.
But, wait for a second, we need to consider what happens to the in-motion self-driving car.
One possibility is that the self-driving car is struck by the tree and the vehicle goes careening onto the sidewalk, ramming into pedestrians. In a sense, you could abundantly state that the self-driving car was the perpetrator of those injuries, though the instigating factor was the fallen tree.
You might be tempted to insist that the AI would never allow the self-driving car to go into this ballistic mode of careening up onto the sidewalk. Certainly, you assume and rightfully expect, the AI driving system should never permit this to occur (hold onto that lofty thought, I’ll be shaking it apart in a moment).
This takes us into the realm of discussing what an AI driving system will potentially do when confronted with a falling tree situation. We can consider this admittedly rare scenario and also toss into the equation that there might be human passengers inside the self-driving car at the time of such a confrontation (i.e., there is the case of no human passengers present and the other case of human passengers being inside the self-driving car).
Let’s start with whether the AI will be able to detect a falling tree of this ilk.
Maybe, maybe not.
The sensors of today’s self-driving cars consist of video cameras, radar, LIDAR, ultrasonic, thermal imaging, and the like, all of which are devices used to detect the world surrounding the self-driving car. By-and-large, these sensors are focused on figuring out what is in front of the self-driving car. The AI driving system has to figure out whether another car is ahead of it, and the distance to the other car, along with whether a pedestrian is about to walk in front of the self-driving car, and so on.
Assume that a falling tree is most likely going to fall from the sidelines and not be directly in front of the self-driving car. This assumption seems reasonable since most of the time a falling tree is either in a sidewalk or front yard that abuts the street, or sometimes in the median, but rarely if ever somehow directly in front of a car (unless, of course, the car is parked and facing toward some trees).
Okay, the first difficulty for the self-driving car is that the falling tree is not likely to be within the primary focus of its sensory detection capacities. The tree is presumably going to be on the periphery. This means that the sensors are less likely to detect that the tree is in the midst of falling.
The odds are that the tree could land on top of the self-driving car and the AI would not have any particular forewarning or have detected that the tree was falling upon it. In that manner, you could reasonably argue that just as a human driver can be caught unawares, so can the AI driving system.
Indeed, even after getting hit by a falling tree, the AI still might not detect what has occurred (whereas, we would expect nearly any human driver would realize such a catastrophic event). For more about the oddity of how self-driving cars might not realize they’ve been in a car accident or related incident, see my discussion at this link here.
Returning to the falling tree, we need to include time into this calculus.
As the self-driving car moves forward down a street, it is usually painting an internal picture, a virtual world model, consisting of the objects that are being detected. There is a strong possibility that the sensors detected the stationary tree that had not yet started to fall and included the tree in the virtual world model. If the tree begins to fall while the self-driving car is approaching a position adjacent to the tree, there is a chance that the AI might calculate that the tree is suddenly in-motion.
Now what?
Well, it depends upon what the AI developers have established within the AI driving system. The AI has undoubtedly been programmed to look for moving pedestrians, and for moving animals like dogs and cats, and for moving bicycles, and the rest. Whether it has been prepared to cope with a moving tree is a whole different enchilada, as they say.
The notion of a moving tree is pretty esoteric.
Given that most of the AI developers have their hands full with trying to get the self-driving car to handle the day-to-day normal stuff, the idea of preparing the AI to contend with a falling tree is a somewhat distant thought. In industry parlance, it is an edge or corner case or considered at the long-tail of what might arise, meaning those situations that are rarely expected and therefore can be placed on a lower priority list than everyday higher priority driving tasks.
Consider these permutations and combinations:
· It is possible that the self-driving car will not detect the falling tree and continue forward unabated, ultimately getting hit or crushed by the tree.
· The self-driving car might detect the tree and have posted into the virtual model that the tree exists, and yet not detect when the tree starts to fall, thus getting hit or crushed by the tree.
· The self-driving car might detect the tree, mark its position in the virtual model, and detect the tree as it starts to fall, but nonetheless take no evasive action because the AI has not been programmed to deal with this particular scenario. The self-driving car gets hit by the tree.
And, the preferred answer is that the self-driving car detects the tree, figures out that the tree is falling, and opts to do something about this unfolding calamity.
What would you have the AI do?
One obvious answer might be that the self-driving car should do whatever it can to avoid getting hit by the falling tree. I don’t think you’ll like that answer, as will be explained next.
Suppose the AI ascertains that it can avoid getting crushed by rushing up onto the sidewalk. Unfortunately, doing so endangers the pedestrians on the sidewalk. In that case, which has the greater importance, a passenger inside the self-driving car that is possibly going to get harmed when the tree lands on the roof of the vehicle, or those pedestrians on the sidewalk?
This is the essence of a famous thought experiment known as the Trolley Problem. Some industry insiders eschew the Trolley Problem as merely hypothetical and unworthy of attention. I sternly have exhorted that the Trolley Problem for self-driving cars is real and must be contended with, one way or another, including that ultimately there might be lawsuits and criminal cases against the automakers and self-driving tech firms that ignored the importance of dealing with the Trolley Problem (see my coverage at this link here).
Conclusion
I had mentioned that one possibility is that the self-driving car careens onto the sidewalk after getting hit by the falling tree. This action could be by the design of the AI and thus an overt act, or it could simply be that the AI is unable to control the car after the vehicle has gotten struck by the tree.
I point this out because some people seem to think that self-driving cars will be able to defy physics. They dreamily assert that self-driving cars will never get into car crashes or collisions. That is just crazy talk. A car is a car, subject to the law of physics, and if other forces are too strong, indubitably the AI is not going to be able to drive its way out of a messy situation (plus, keep in mind that the driving controls might get jammed or disrupted too). The same could be said of a human driver.
There are lots of other variations involved in this falling tree situation. For example, if a passenger is inside the self-driving car, consider whether their yelling to the AI and telling its Natural Language Processing (NLP) component that a tree is falling might be of aid to the AI driving system (hint: only if the AI NLP is ready to process such a proclamation).
As a final thought, would a self-driving car that is in a forest and without any passengers be able to hear the sound of a falling tree?
Some self-driving cars are being outfitted with external microphones to provide the capability of hearing ambulance and police sirens, along with detecting other various outdoor sounds. In that way, the AI might indeed detect the sound of a falling tree, but whether the AI has been prepared to make any semblance out of the sound, as to what it portends, well, that’s a different enchilada too.
And since the AI isn’t a human, albeit maybe someday sentient, we return to the beginning of this discussion about the meaning of words. When a machine such as AI “hears” a sound, does this count, or are we going to be recalcitrant that only living creatures can hear sounds?
Ask a child, they might have some clever answers.
