Traffic snarls are everywhere and seemingly endless.
Most major cities throughout the nation endure adverse congestion on their public roadways. Urban mobility has seemingly been getting worse and worse, despite the increasing options for getting around town (or, perhaps ostensibly worsening due to those added vehicular options). It takes forever to get to your destination. Even simple trips are exceedingly frustrating.
Could self-driving cars make a difference and potentially ease or even eliminate such congestion?
There are various pundits that keep touting the aspect that there will never be congested traffic after the advent of self-driving cars. Via self-driving cars, we’ll all be traveling in congestion-free heaven, and as a passenger in an autonomous car you’ll feel the exhilaration of moving at top speeds nearly all the time.
Imagine that you won’t suffer in stop-and-go traffic any more.
Though you won’t be driving the car, since true autonomous cars won’t need and won’t presumably have a provision for any human driving, especially the vaunted Level 5, you’ll certainly still be aware of the pace of the self-driving car and would likely be chagrined if the state-of-the-art vehicle was not smoothly proceeding on your journey.
Here’s the top news for the day: It’s quite unlikely that self-driving cars will bring us a congestion-free nirvana, certainly not in the foreseeable future.
Let’s unpack the matter.
Factors Impacting Driverless Cars On-The-Roads
There are various experiments that have been done on special closed tracks or proving grounds trying to gauge whether AI-driven autonomous cars can maneuver in a fashion that will avert traffic snarls.
If you had exclusively self-driving cars, they indeed would seem to be able to drive in a smoother manner than humans do, generally being able to maintain a proper distance between cars and adjusting quickly to the ebb and flow of the cars ahead of them.
In some cases, these experiments are done with the added benefit of V2V (vehicle-to-vehicle) electronic communications taking place among driverless cars. This might be akin to equipping human drivers with walkie-talkies or smartphones to group chat while driving, letting other nearby drivers know what they are doing at the wheel of the car.
For self-driving cars, V2V is likely even better utilized than if it was humans trying to converse with each other since humans might be delayed in verbalizing their actions, humans might mistakenly utter one thing and do something else, and the time required to orally describe your efforts is many times more time consuming than would be the bits-and-bytes communications among AI systems in the self-driving cars.
Though some put V2V onto a pedestal for driverless car efforts, I’ll just mention that it is not entirely as divine as might be assumed. With any kind of technology, there can be errors, glitches, transmission delays, and other snafus that occur with such real-time electronic communications.
There is also the concern that driverless cars might get swamped by V2V messages, zillions being broadcast all at once, and the AI might not readily be able to discern important ones from idle messages, plus there is the chance too of cybersecurity breeches messing with V2V aspects. Etc.
I’m not saying that V2V is not going to be a handy and vital element, it will be, and just want to make sure that we all realize that reality will still be in existence and the blemishes and qualms also need to be baked into what is possible versus what is likely.
In any case, the proving grounds experiments are often undertaken as though the world will be a different place when the advent of driverless cars arises. Let’s see if we can agree on one crucial proposition, namely that we’re going to have a mixing of human-driven cars and autonomous AI-driven cars, which will last for quite a while.
In the United States alone, there are around 250 million conventional cars. Economically it is not feasible to wave a magic wand and suddenly toss away those conventional cars and replace them with driverless cars, nor do I see much hope in trying to convert those conventional cars into autonomous cars.
Mixing Of Traffic With Human Drivers And AI-Drivers
I know many AI developers that are irked that society won’t just get rid of human drivers and let autonomous cars take over our roadways.
It would be a whole lot easier to develop the AI systems for driving if the AI didn’t need to contend with human driver foibles. Plus, humans are not likely to be able to adequately do V2V with AI-driven cars, at least not in the same way that the AI-to-AI of homogenous driverless cars would be able to do. Therefore, there are at least two strikes against human drivers, they tend to be gnarly and uncivil in their driving, and they wouldn’t tend to be very good V2V cooperators.
If you buy into my point that we are going to have a mixture of human-driven cars and of AI-driven cars, doing so for an extended period of time, many years to come, some suggest that maybe we can divide our roads into those that take the driverless cars and other roads that take human-driven cars (with the possibility too of allowing the driverless cars to go into the human-driven cars lanes but ban the human drivers from getting into the self-driving cars “only” lanes).
There are unfortunately stumbling blocks about trying to divide up the roads in this manner.
One is the logistics aspects of converting roads into those that permit only one type or the other. It is not physical and architecturally so easily undertaken. The cost too to do these conversions would be relatively high, and some might question whether it is a worthwhile cost for society to bear, just to allow driverless cars to proceed at a presumably faster pace.
Another perspective is that perhaps human-driven cars could only be on our roadways at certain times of the day or days of the week. Maybe you can drive your car on Tuesday and Thursdays during midday, but not the rest of the week, which is reserved solely for self-driving cars. Upon giving this notion some thought, I believe you’d reject it nearly outright as not especially practical and the pushback by the public would be substantial, one would think.
Yet another idea involves charging a fee, similar to what today is being called congestion pricing, meaning that if you are using a car on certain roads at certain points in time, you’ll need to pay a special fee to do so. This would presumably discourage traffic at those times and get people to somehow spread out when they need to travel. Such a fee could be applied to say human-driven cars and not driverless cars, acting as a discouragement of human drivers and an encouragement of using autonomous cars.
In the case of driverless cars, some are referring to this as a kind of zombie-car tax, wherein the zombie is a driverless car that has no human driver (I’m guessing the AI-driving systems might get offended at being referred to as a brainless zombie).
Some critics worry that any kind of division or subdividing might cause unintended adverse consequences leading to other strife or societal repercussions. Suppose only the wealthy could afford driverless cars and an entire set aside of the roads was now being used more so by the well-to-do and essentially denied to those that aren’t rich. It could become a fairness issue.
City planners have been discussing whether to make downtown areas become autonomous car zones, meaning that any car entering into a boundary area of downtown could only use a driverless car within those downtown streets.
Again, not an easy way to solve the problem.
If you work downtown and are driving your own car to work, does this mean that you now need to find a place outside of downtown to park your car, and then take a ridesharing driverless car to your office? That doesn’t’ seem to make much sense in terms of the use of cars and getting around.
Okay, so we’re seemingly back to the conundrum that we’ll have a mixing of human-driven cars and AI-driven autonomous cars, and that the mix likely suggests that we’re not going to have streamlined traffic.
As an aside, I realize and acknowledge that those supportive of mass transit are irked by this entire discussion and debate since they would say stop using cars as the dominant mode of travel.
Get out of those darned cars, whether human-driven or AI-driven and use mass transit instead.
Thus, problem solved.
Well, it’s certainly a worthy idea, but for the moment let’s continue the everyday usage-of-cars deliberation and revisit mass transit in a future discussion.
Mixing And Congestion
Some hope that human drivers might become more tolerant and civil drivers if they were surrounded by driverless cars.
If an autonomous car next to them is providing the proper driving distance of some number of car lengths to the car ahead of it, maybe human drivers will do the same. I’m not holding my breath on that one. So far, human drivers are more likely to exploit the civility of a driverless car, jumping into the gap that was intended by the AI as a safer driving measure, rather than respecting the purity of such driving approaches.
The other side of the coin is that some suggest we need to make the AI be a tougher, meaner, and more realistic driver, mimicking the antics of human drivers.
Well, there is some merit to the idea, but it certainly would not seem to solve the traffic congestion problem. Presumably, you’d worsen traffic congestion by having more and more cars all playing the same kinds of tricks on each other, regardless of human-driven or AI-driven. In a twist on this notion, it could be suggested that the AI self-driving cars might so intimidate human drivers that either human drivers will shape up to be more civil at driving or they will toss in the towel and switch over to driverless cars (a bit of controversy there).
You might try to argue that the advent of autonomous cars will give rise to expanded use of ridesharing, which seems to make sense. Maybe that will dispense with congestion.
Curiously, some studies of ridesharing in downtown areas tend to show that ridesharing can actually make congestion worse, rather than better. This seems somewhat to be due to the aspect that the ridesharing cars need to drop off and pick up passengers, often doing so at places on the streets that causes blockage or traffic snarls. Also, ridesharing doesn’t necessarily seem to be getting people out of their own cars, plus it seems that whereas people might have walked or taken a bus for short downtown hops they instead are using ridesharing.
Daydreaming about the day when you’ll be able to go anywhere in a car and not be burdened by traffic congestion seems to be of some folly. The emergence of autonomous cars is going to be relatively gradual, along with the getting rid of or retiring conventional cars. For an extended period of time, we’ll have a mixture of human-driven cars and AI-driven cars, which will tend to defy the attempts to optimize traffic flows.
I know that some believe the semi-autonomous cars are going to aid in reducing traffic congestion, but the co-sharing of driving by a human and the AI is not necessarily a boon toward alleviating traffic congestion, since it once again allows for human intervention in the driving task.
There’s another viewpoint that maybe society can legislate the driving behavior of humans by setting in place rules about when they can take over from an AI system when co-sharing the driving task. All I can say is that trying to adjust human behavior is harder than it might seem on paper.
The dance goes on as we seek to rid ourselves of too much traffic. Don’t give up hope. If we can get to the moon, some would say that we can certainly cure ourselves of traffic jams.