Something adorable could have been a heap of unpleasant trouble.
A kitten had decided to hide underneath a parked AI self-driving car.
I’m sure that you’ve seen or even personally witnessed having a cat or dog that opted to sneak under your conventional car and decided to hide out in this seemingly protective cove. Of course, we all know that once the vehicle starts up, things are going to get dicey for the cuddly pets if they don’t quickly scramble out from their hiding spot. Worse still, the likely now startled animal could shockingly freeze up in place and be ominously in the path of the wheels and be at risk of getting (heaven forbid) run over.
In my neighborhood, there seems to be a slew of meandering cats. Thus, I always do a quick peek underneath my outdoor parked conventional car whenever I am going on a driving journey. More than once I’ve caught a blissful feline that was catnapping and unaware of the potential dangers involved.
Any street-savvy cat has probably already figured out that it is time to rapidly move on whenever a car engine gets started. I usually assume that the cats undertaking the under-the-car gambit are seasoned prowlers that are constantly on alert and ready to skedaddle with tremendous grace and speed.
A kitten might not yet have learned this crucial school-of-hard-knocks lesson.
I dare say that few people seem to put much thought toward looking underneath their vehicle before they begin to go for a drive. You might remember in your teens taking a driver training class that repeatedly emphasized the importance of doing a careful and plodding walk around your vehicle before you get into the car itself.
This was a type of pre-flight check.
The idea is that you might notice a flat tire, or maybe a ding or dent in the exterior, etc.
In short, it makes abundant sense to ensure that your car is ready for a road trip. Plus, you want to know what your car looked like beforehand, in case something untoward happens during a driving adventure. Upon going to say a grocery store, you might believe that someone carelessly allowed a shopping cart to skirt and scrape across the side of your car and scratch it up, though it could be that the scratch was there earlier, and you had lackadaisically not glanced around to see it.
The most vital reason to do a walk around of your vehicle would be to ascertain that the car is fully functional and ready to go.
If you just jump into the driver’s seat first, you might not realize that your right rear tire is perilously low on air. Or maybe a nail or spike has somehow ended up on the gravel just where your right front tire sits, and for which the moment you pull forward there is going to be a bursting sound and a lot of angst and dismay on your part. Had you noticed the silently menacing piece of steel waiting to puncture your tire, you could have easily swept it out of the way beforehand.
Another reason to do a brief walk around your vehicle entails any otherwise unseen complications.
Here’s what I mean.
When my children were toddlers, it seemed like they used to regularly leave a toy behind my parked car. The car was usually seated on an outdoor pad in our front yard and was slightly on an incline. It was easy for the kids to inadvertently leave a toy laying here or there. It was equally easy for me to be in a hurry, rushing to work, and hastily back down the driveway without knowing that a toy was about to get summarily crushed.
To be clear about this, I would mindfully look to ensure that no humans and no animals were anywhere behind the vehicle. This was standard practice. That being said, a small toy could be laying flat on the ground, unmoving, and utterly blend into the pattern of the concrete pad. If that were the case, and assuming that the toy by a bad stroke of luck was in the path of the wheels, it was going to be a goner.
More than once I regrettably smashed such a toy. As a parent, your first thought seems to consist of chiding the kids for having left their toys in a vulnerable spot. One might even suggest that this is a valuable lesson for them, namely if you don’t take good care of your things they will be at risk of getting lost or damaged. On the other hand, there is also the rule of thumb that the parent backing down a driveway ought to be inspecting the path and ensure that there isn’t anything in the way.
I admit that I sometimes opted to get a replacement toy, explaining that the matter was unfortunate for all and yet we didn’t need to go quite overboard about lessons learned. Some others thought that was merely proving that two wrongs make a right, and my kids would lead a life of confusion and misunderstanding about how the world really works. No worries, things didn’t play out that way and those theories were disproven.
Anyway, back to the kitten and the looming danger of resting underneath a car.
The thing about a conventional car is that a human driver might notice beforehand that a kitten or other animal is in fact hiding out and ought to be dissuaded from staying where they are. The person would undoubtedly try to shoo out the hiding intruder. It is possible too that you might hear the cat meowing or perhaps making some form of noise while moving around. All in all, there is some viable chance that you might detect the hidden trespasser (well, I guess you could assert that the cat has every right to be resting in that position, presumably protected from predators and for which you’ve allowed the opportunity to exist).
What about an AI-based true self-driving car?
You see, there isn’t a human driver involved in a true self-driving car. Keep in mind that true self-driving cars are driven via an AI driving system. There isn’t a need for a human driver at the wheel, and nor is there a provision for a human to drive the vehicle. For my extensive and ongoing coverage of Autonomous Vehicles (AVs) and especially self-driving cars, see the link here.
Here’s an intriguing question that is worth pondering: What will an AI-based true self-driving car do or fail to do when getting underway and seek to ensure that no objects or living creatures might be in immediate harm’s way?
Before jumping into the details, I’d like to further clarify what is meant when referring to true self-driving cars.
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 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 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 Those Hiding Kittens
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.
One aspect to immediately discuss entails the fact that the AI involved in today’s AI driving systems is not sentient. In other words, the AI is altogether a collective of computer-based programming and algorithms, and most assuredly not able to reason in the same manner that humans can.
Why this added emphasis about the AI not being sentient?
Because I want to underscore that when discussing the role of the AI driving system, I am not ascribing human qualities to the AI. Please be aware that there is an ongoing and dangerous tendency these days to anthropomorphize AI. In essence, people are assigning human-like sentience to today’s AI, despite the undeniable and inarguable fact that no such AI exists as yet.
With that clarification, you can envision that the AI driving system won’t natively somehow “know” about the facets of driving. Driving and all that it entails will need to be programmed as part of the hardware and software of the self-driving car.
Let’s dive into the myriad of aspects that come to play on this topic.
We can certainly expect that AI-based self-driving cars will attempt to examine the surroundings when first powered on and getting ready to undertake a driving journey.
Self-driving cars are usually outfitted with a sensor suite consisting of video cameras, radar, LIDAR, ultrasonic units, thermal imaging devices, and the like. Not all brands or models will necessarily have the same set of sensors, such that some automakers and self-driving tech firms are aiming to use only certain kinds of sensing devices and somewhat eschew other such types (for my coverage on this at times controversial selection preference, see the link here).
In any case, the sensors are likely to right away begin capturing data about the immediate surroundings the moment that a driving task is going to be initiated. Some self-driving cars will even be making use of some of their sensors while parked and not at all underway, doing so as a precaution to try and detect and dissuade car thieves or carry on other desired observations.
Let’s assume that a driving mission has been given to the AI driving system and it is going to proceed with the assigned journey. Upon starting the engine and engaging the sensors, the programmed initial activity is assuredly going to be an assessment of where the self-driving car is, where it needs to go, and what might be directly within the starting path of the autonomous vehicle.
If the self-driving car is parked on a driveway and going to be backing down to the street, you can reasonably anticipate that the sensors on the rear of the vehicle will be trying to detect any obstructions that might be in the way. Perhaps a child has left a bicycle standing behind the self-driving car. Either the video cameras would pick up the image of the bicycle or possibly the radar would detect the bicycle. There could be a multitude of sensors that are able to simultaneously detect the presence of an object.
So far, so good.
Would a child’s toy that was on the ground, sandwiched between the driveway and the right front tire, be readily detected?
The odds are against this being detected.
The position of the toy is probably not within the scope of the video cameras. You could say that this particular location is a blind spot of sorts. None of the other sensors are likely to detect the toy, especially if it is lying flat and entirely motionless. The same could be said of a nail or spike that perchance was laying in that general locale.
Would a kitten or other small animal hiding underneath the self-driving car be detected?
Few self-driving cars have a video camera that is mounted or ostensibly positioned to see under the vehicle (likewise, you would not expect other externally attuned sensors to be there either).
Many AI developers either haven’t considered putting a camera in that arrangement (the idea hasn’t arisen), or they don’t think it would be of much use. Plus, there is a cost associated with including yet another video camera in the self-driving car. All in all, the use of a costly video camera or other such sensors to be able to see what is under the car doesn’t seem to be in the cards, for now (I have predicted that this attitude will change once self-driving cars become prevalent and are being used in everyday real-world driving settings).
You could categorize this scenario or use case into the large bucket of so-called edge or corner cases.
Those are situations and circumstances that are considered rare and ranked as low priority to be considered. Most of the AI driving systems and the AI developers making them are struggling to just get self-driving cars to go safely from point A to point B. Worrying about what is perhaps underneath the self-driving car is something well beyond the focus of attention at this time.
Just don’t tell the kitten that they are ranked as low priority, you might get fervently cat scratched.
The bottom line is that most self-driving cars are generally blind (not set up to detect) whatever might be somewhere underneath the vehicle. Given that assertion, there is a bit of an escape hatch in that if the self-driving car were to detect an object before it ended up underneath the vehicle, this could aid the AI driving system in essentially calculating that the item has presumably gotten into that posture.
Consider how that might work.
Suppose that the sensors were active and detected a cat that was about ten feet in front of the vehicle and for which the feline was striding nonchalantly toward the front of the self-driving car. This is something that could be readily detected and tracked. Once the cat goes underneath the vehicle, the external sensors of the typical self-driving car are no longer able to detect the presence of the feline.
Meanwhile, the sensors that provide a 360-degree viewpoint around the vehicle would allow for detecting the cat once it comes out from underneath the self-driving car. If the cat does not appear on any of those sensory captures, it is reasonable to make the default assumption that the cat is still underneath the self-driving car. This is not a certainty, since perhaps there are some other means of the cat finding its way out, though we will not include that the cat has magical powers and nor that it is wearing an invisibility cloak.
Whenever a self-driving is underway, the AI driving system has the potential for figuring out what objects might have been driven over and perhaps be underneath the autonomous vehicle. This is due to detecting an object beforehand and then making calculated predictions about where the object might next be. For example, a discarded piece of splattered furniture on the freeway will at some point be underneath the self-driving car if the AI driving system opts to drive over the roadway object. Shortly thereafter, the object should be detectable as the self-driving car proceeds beyond the object and the strewn debris are now behind the vehicle and once again visible.
There are some dicey elements to this kind of computation.
Could the video camera image processing really account for all of the shattered remains of the furniture, or might some pieces still be underneath the vehicle and possibly stuck to the underframe in some fashion? The odds of being able to definitively make such a determination are rather slim.
Of course, this side tangent was based on the premise that the object of interest was detected before it went underneath the autonomous vehicle. If the self-driving car was parked on a driveway and entirely shut off, including all of the usual sensors, there would not be any “before” that could be relied upon to ascertain what might be underneath the vehicle.
Imagine two different variants of this scenario, whereby one instance consisted of the self-driving car going for a drive and doing so absent of having any passengers, while the other instance involves carrying one or more passengers.
In case you are wondering why a self-driving car would be getting underway without any passengers on board, that’s a straightforward and altogether commonly expected aspect regarding the use of self-driving cars. You see, if someone requests a ride, the self-driving car might need to first drive to wherever that person is. As such, the self-driving car will be entirely empty until it reaches and has the passenger embark on the desired ride. Many are worried that we are going to have a lot of self-driving cars that are roaming here and there, doing so absent of passengers, and ergo might inadvertently clog up traffic accordingly (see my columns for analyses of this qualm).
The second use case of having a passenger at the outset is especially interesting because it introduces a potential “solution” related to finding that kitten that is hiding underneath the parked self-driving car.
Here’s how that might happen.
A passenger comes up to a parked and inert self-driving car. The passenger rouses the AI driving system, either by using a smartphone to connect to it or possibly by speaking to the vehicle (a microphone is listening for a wake-up command, just like when you use Alexa or Siri).
The Natural Language Processing (NLP) component of the AI driving system might engage the human passenger in a brief dialogue. Besides asking where the person wants to go, the AI driving system might also ask the passenger to please glance around and underneath the autonomous vehicle. If the passenger notices anything out of the ordinary, the person could verbally alert the AI driving system accordingly. The AI driving system would have been programmed with some provision related to not getting underway until the matter is somehow properly resolved (that’s an entirely different story, which will be covered in a later column).
Just as a human driver might normally be expected to look around their vehicle to do a pre-flight kind of check, an AI driving system could ask a human passenger to take on that role. It would be easy for passengers to do this, taking just a quick walk around and bending down to look underneath the self-driving car.
That seems to close this matter and we can go home now.
First, we’ve already noted that this is just the use case when a passenger is involved. Therefore, we still have the gaping loophole of the use case for which there isn’t a passenger at the get-go.
Second, mull over this notion of asking passengers to look around, and I think you’ll quickly realize the folly of such an approach.
Some passengers will give only a cursory look-see. Their effort to examine what is underneath the vehicle is especially likely to be skimpy, at best (would you be willing to get down on the ground onto likely grimy asphalt, plus what if it is raining or some other adverse factor comes to play). The chances that a passenger will reliably spy on something is highly questionable.
I would bet too those passengers would feel as though they are being put upon by having to do “work” for the self-driving car. Why should a passenger have to be enlisted to do something that the AI driving system ought to be taking care of? When you get into a human-driven car, you don’t usually expect the human driver to ask you to walk around and check out things. The response by passengers would probably be something like getting your lazy posterior out of the car and check around for yourself.
All told, asking passengers to look around seems misguided. It could be a temporary and half-baked way to momentarily deal with such issues but ultimately will need a more systemic solution.
If you are wondering what else could be done, other than mounting sensors that are oriented to inspect underneath the autonomous vehicle, there is in fact a potential additional solution.
Yes, there are various efforts underway of pairing up drones with self-driving cars. A drone could be used for a variety of handy purposes. As mentioned in my columns, there are tests underway of using a drone to rapidly fly a fast-food meal to a moving self-driving car that has a hungry passenger inside (see my discussion at this link here).
Rather than simply providing some hamburgers and fries, let’s consider how the drone helps to find that adorable kitten.
When a self-driving car begins a driving journey, it might directly launch its paired drone (which was sitting in a rooftop casing, the passageway of the casing is automatically opened and the drone is sent airborne). The drone briefly circles around the self-driving car. A video camera on the drone is beaming electronic signals to the AI driving system and showcasing the images being captured. The AI driving system mathematically examines the images and tries to determine if there are any objects in the way of the autonomous vehicle.
In addition, the drone is flown very close to the ground and immediately adjacent to the self-driving car. This allows the video camera of the drone to peer underneath the vehicle. If needed, the drone can be flown underneath the self-driving car to get a closer look at what might be there.
That is the future and not something that you would find in today’s self-driving cars.
You might be wondering what happened to the kitten that was snoozing underneath a contemporary (budding) self-driving car.
Turns out that the self-driving car had a human backup driver, sometimes referred to as a safety monitor, that was taking the autonomous vehicle for a test drive. This is frequently the case these days, namely that automakers and self-driving tech firms hire people to go along for a ride, watching to see what the AI driving system does. If there is something that the AI driving system appears to be amiss on, the human driver can disengage the AI driving system and take over the driving controls (for details about the nature of this role, see my column discussions).
Well, it turns out that upon approaching the self-driving car, the human backup driver startled the kitten.
Out darted the lively kitten!
I suppose you could say that this was all in a day’s work for the human backup driver. They are intended to keep an AI self-driving car from going awry, and apparently able to save kittens from dangerous situations.
I wonder if the kitten used up any of its nine lives by surviving unscathed. This might be a necessity until self-driving cars are made kitten-proof.