You are driving along on a long stretch of highway and minding your own business. So far, your driving journey has been relatively uneventful.
Upon reaching a curve in the highway, you suddenly spy a tow truck that is trying to hook-up a stranded car at the edge of the roadway. Both the tow truck and the incapacitated car are just barely out of the lane that you are in. Turns out that this portion of the highway has a very thin emergency lane and thus there are going to be just inches between your car and those two vehicles attempting to get connected and ultimately get off the highway.
What do you do?
What should you do?
The usual rule-of-thumb would be to try and get over into the next lane. This would enable you to pass those two vehicles and be plenty of distance from them as you do so. Not only won’t you skirt the sides of the tow truck and the standing car, but you’ll also be safer and so will the tow truck driver and any of the passengers in the stranded car if they perchance decide to walk around nearby their vehicles.
Though it seems unimaginable, there are sometimes people that get out of their stranded cars and stand adjacent to their vehicle, being within a whisker’s distance of cars zipping along at 65 miles per hour (or faster). Somehow these people do not realize the gravity of their situation. All it takes is for a car on the highway to veer a few inches or feet into the emergency lane and someone is going to be pushing up daisies, horrifically so.
For reasons that aren’t very clear cut, those motorists that stand or walk around on an active highway or freeway are playing a daring do-or-die kind of game. One supposes that it is easy to assume that no car is going to purposely divert from the normal lane and come into the emergency lane, though this does indeed happen with more regularity than is expected. Also, there are those drunk drivers that see a car in the emergency lane and seem to be drawn to it like a moth to light, as though the stranded car suggests that there is another viable lane available for use.
Another factor consists of daytime versus nighttime driving. At night, the vision for drivers tends to be obscured or otherwise diminished. It is possible to mistakenly believe that an emergency lane is perhaps an exit lane or has some other purpose such that it can be used for cars already in motion and that wish to use it. There are those outright scofflaws that use the emergency lane as a means of scooting around other cars, which is a definite no-no and once again a high-risk form of devious driving that is decidedly illegal.
Returning to the notion of moving over to avoid those that are sitting in an emergency lane, this is not necessarily as simple a maneuver as it might seem.
You might be caught off-guard about the brewing situation up ahead and not have enough time to switch lanes. At times, you need to weigh the challenge of making a risky lane change against the sour odds of staying in the lane and breezing right next to the vehicles on the side of the highway. There is also the possibility of other cars in the lane that you wish to switch into, and thus you might disturb those cars by making a seemingly rash switcheroo. It does little good to avoid the emergency lane calamity by instead starting a lane change calamity that could just as likely start a cavalcade of car crashes.
You’ve undoubtedly seen car drivers that are frantically trying to switch lanes and it is obvious they are doing so to get over and away from vehicles at the side of the road. If you are in the lane they want to barge into, presumably, you would politely let them in. Of course, if they are now going much slower than the prevailing traffic, you have to decide whether to hit the brakes to let them in or try to zip past them and assume they will enter into your lane once you’ve gone past. This can be a delicate dance and one that gets all parties frustrated.
The driver wanting to get into your lane is bound to be giving you the evil eye for not slowing down and giving them a chance to get in. You are going to be giving that driver your evil eye since they are unduly disrupting the flow of traffic. If the driver in the running lane doesn’t see why the other driver wants to feverishly change lanes, this ups the ante in terms of exasperation and irksomeness. It might appear as though there is no reason for that driver to make such a rash switch and therefore the driver zipping along is more determined to not let the interloper into the lane.
This brings up another facet about what to do when in the lane that is adjacent to the emergency lane, namely slow down.
The assumption is that by slowing down, you’ll have less chance of striking the vehicles or any wandering people, and be able to come to a stop quickly if absolutely necessary. The act of slowing down also means that if push does come to shove, heaven forbid, the resulting collision will be a lot less impactful at a lower speed. Slow speed also means more mental processing time is available and adds leeway somewhat to the decision making process about what driving action is best to be undertaken.
Many states have a “move over, slow down” law that is a mandatory requirement for driving on their highways and byways.
California recently amended its version of the “move over, slow down” provision, which encompasses all manner of vehicles that might be standing alongside the roadway, including tow trucks, fire trucks, ambulances, police cars, and so on. The existing law as per the DMV says that “a person driving a vehicle on a freeway approaching, among other things, a stationary authorized emergency vehicle that is displaying emergency lights to approach with due caution and, before passing in a lane immediately adjacent to one of those specified vehicles, absent other direction by a peace officer, either proceed to make a lane change into an available lane not immediately adjacent to one of those specified vehicles, or slow to a reasonable and prudent speed that is safe for existing weather, road, and vehicular or pedestrian traffic conditions, as specified.” Part of the expanded scope includes that not only are the state highways part of the rule, so too are local ones.
Hopefully, drivers in California and other states are mindful about trying to avoid anyone or anything that is standing alongside the roadway. Prudence dictates that it is wise to get over and avoid the standing particulars if you can safely navigate away and that it is advisable to slow down as you pass such situations. We would not seem to need a law to instruct us as such, but then again, not everyone drives cautiously and mindfully, thus the law adds teeth and a reminder of the civil thing to be done.
Shifting gears, let’s ponder what the future of driving might be like. There is going to gradually be the appearance of self-driving cars on our highways and byways.
This brings up an interesting question: Should AI-based true self-driving cars also observe the “move over, slow down” provisions?
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 Move Over
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.
It is widely anticipated that self-driving cars will strictly abide by the laws of the road (see my coverage about the controversy underlying that assumption, at this link here). In that case, we would certainly expect that the AI driving system would obediently do a “move over, slow down” whenever it detected a suitable circumstance.
That’s part of the rub about this matter.
The question arises as to whether the AI driving system will be able to detect a situation that entails invoking the “move over, slow down” protocol.
Notice that the California law states that if there is a “stationary authorized emergency vehicle that is displaying emergency lights” the provision so applies. This makes things pretty clear cut that there needs to be a stationary vehicle, and for which it needs to be an authorized emergency vehicle. Furthermore, this said vehicle, if detectable as such, would also need to display emergency lights.
You might think that this is a rather obvious and abundantly detectable facet.
For humans, perhaps that is the case, though even humans admittedly can mess-up on doing so. In any case, most AI driving systems make use of a special sensory suite consisting of video cameras, radar, LIDAR, ultrasonic units, thermal imaging, and the like. These devices are mounted on the self-driving car and are used to collect data about the driving environment. They are essentially the eyes and ears, as it were, for the AI driving system (for more details, see my discussion at this link here).
Typically, AI driving systems are pre-trained on what different kinds of vehicles look like. This is usually done by presenting thousands upon thousands of pictures or videos to a Machine Learning or Deep Learning (ML/DL) algorithm which uses computational pattern matching to try and figure out what those objects consist of. You might provide an ML/DL with a large set of images that showcase say an ambulance, and the computational pattern matching attempts to ferret out the visual characteristics that seem to pertain to ambulances.
When a self-driving car is cruising along on a highway, it is capturing the video streams and applying the ML/DL to try and identify what else is around the self-driving car. Coming upon a scene that has an ambulance at the side of the road and standing still, the video is examined by the ML/DL and a probability associated with what the vehicle is will then be provided to the rest of the AI driving system. Once the rest of the AI driving system has a semblance of what is around the self-driving car, there is an internal virtual world database that is marked with what objects are nearby and what they signify.
It is readily conceivable that the emergency vehicle might not be recognized via the pattern matching and therefore not necessarily be classified as an emergency vehicle. There is also the chance that the object will be classified as an emergency vehicle though with an assigned probability, such that the pattern matching only perhaps got a partial impression and has stipulated that there is say a 60% chance that the vehicle is an ambulance.
This has to do with the kind of “facial recognition” per se associated with spotting an emergency vehicle. You might be thinking that it doesn’t matter much since there is also the stated requirement of the emergency lights. It would seem trivial to be able to spot the use of emergency lights. As such, put aside whether the vehicle is an authorized emergency vehicle and focus solely on those emergency lights.
Well, once again, this can be somewhat tricky to figure out. Suppose the emergency lights are hard to see because they are obscured by a stranded vehicle that is behind the emergency truck or ambulance. Perhaps the emergency lights are on, but they are low to the ground, or maybe the sunlight is adding glare and it is not readily feasible to see the emergency lights. And so on.
The crux of this emphasis is that you cannot summarily assume that a self-driving car can readily detect an emergency vehicle and nor detect that emergency lights are being displayed. This is a harder problem than it might seem.
If the AI driving system does not detect the presence of the emergency vehicle and nor the emergency use of lights, what then?
One obvious answer is that if there is anything at the side of the road, regardless of flashing lights and even not an authorized emergency vehicle, by gosh the right thing to do is get over and slow down.
A counter-argument to that rule-of-thumb would be that the self-driving car might be continually switching lanes to avoid whatever might perchance be alongside the highway. Yes, this could be the safest idea when construed in a narrow viewpoint, but then again it could also mean that the self-driving car is making a lot of lane changes. Each lane change carries its own associated risks.
Okay, you say, then just have the self-driving car stay out of the lane that is adjacent to the side of the highway. In that case, there is presumably no need to generally be switching lanes, since the self-driving car is already in a “safer” lane to begin with. This has some merit, on the other hand, inevitably it would seem that a self-driving car has to get into the lane adjacent to the side of the highway if nothing else to prepare to exit or when coming onto the highway. And, notably, in either of those settings, there is always the chance that an emergency vehicle just so happens to be sitting on the side of the highway and needs to be dealt with.
As a measured stopgap, most of the automakers and self-driving tech makers are programming their self-driving cars to always attempt to get away from anything stranded on the side of the highway. The assumption is that this is the most prudent action.
That takes us into the other dicey part of this equation.
The act of moving over into another lane is fraught with potential problems. As mentioned earlier, human drivers are notorious for not letting other cars into their lane. The same kind of reluctance can assuredly be done to self-driving cars. This means that a self-driving car that wants to move over can be potentially stymied by the human drivers in the desired lane.
Another question comes up about the slowing down aspect. How slow should a self-driving car be programmed to go in these circumstances? In theory, a human driver might go just a tad slower than prevailing traffic or might drop down their speed to a crawl. It all depends. There isn’t a magic number that can be relied upon. As per the California DMV, the notion of slowing down is to “slow to a reasonable and prudent speed that is safe for existing weather, road, and vehicular or pedestrian traffic conditions.”
Thus, we are expecting an AI driving system to essentially exercise a kind of human judgment about what speed is the proper one to use in such settings. You can bet that this will inevitably be a contested issue in some future court cases, whereby a self-driving car slowed down, but arguably not enough, and somehow that led to a car crash or other injury. The other side of that coin is rife for attack too, namely that the self-driving car went too slow and therefore caused or generated a car collision or injury thereof.
If you are AI, you can lose either way.
Some pundits are quick to point out that we will be using V2V (vehicle-to-vehicle) electronic communications that allow self-driving cars to message each other. This implies that whenever a vehicle is stopped alongside the highway if a self-driving car spots this occurrence, it could rapidly use its V2V to message other nearby self-driving cars and give those AI driving systems a greater opportunity to move into a further lane.
Also, those vehicles stranded could be emitting a V2V message, letting any self-driving cars that are nearby be alerted to their presence, potentially doing so long before the self-driving car manages to reach that part of the highway. An AI driving system might decide to go a different route entirely, or at least make sure to be in a further lane to avoid the stranded vehicle.
Yes, those are all handy possibilities. The issue is that it will be a while before V2V is widely adopted, and meanwhile, there will be lots of conventional cars that do not have V2V. Overall, the V2V usage has a more pronounced long-term advantage, while in the nearer term there is still going to be an expectation that a self-driving car can figure out what to do on its own and without any V2V forewarning being provided.
You can even argue that despite an eventual widespread use of V2V, you would still want a self-driving car to be able to undertake its own actions without V2V since it is always possible that the V2V might not be functioning or otherwise have some form of electronic communications difficulties.
A final twist.
Believe it or not, some argue whether it is “move over, slow down” or ought to be “slow down, move over” in terms of the appropriate phrasing. You can get yourself tied into a knot by trying to be finicky about that wording.
Perhaps we can all agree that moving over is a usually good idea, and slowing down is a usually good idea, for which you can mix or match those two driving rules, as long as they make sense and aim to avoid any disastrous roadway incidents.
We can assuredly get the AI to observe the rule in that manner, and so the question remains, what about those dogged human drivers that think they rule the road.
You might say that new tricks are hard to learn, but this is an old trick that ought to already be part and parcel of anyone behind the wheel.