Autonomous Vehicles

The Early Years

As of this writing, it'll be awhile before we hop into our car and say "Home, James". While Tesla might have something approaching driverless tech, they still aren't really there. The very first real fully autonomous vehicles will likely be in fleet service of some kind, like what Tesla and Waymo is working toward. Maybe that's a driverless taxi service or long-haul interstate-centric trucking.

What it almost certainly won't be is individually-owned vehicles at the retail level. At least not for most of us -- at least not anytime soon.

The Obvious Benefits

If it worked perfectly

Using a (truly developed, non-experimental, ready for prime time) self-driving vehicle could offer these benefits:

• You could nap or otherwise be occupied during your commute or other long drive.
• Stay out late and consume moderate alcohol with no worries about how you'll get home safely and no risk of a DUI.
• Reduce your likelihood of being in an at-fault accident.
• Self-parking; you could exit your car at a store front then it goes off to find parking on its own. When you're done, your car could come back and pick you up, all nice and cool (or warm) inside.
• Complete a lengthy road trip faster by extending driving hours or maybe not stopping overnight along the way by sleeping in the car as you might on a night train.
• Children, elderly, or those who don't/can't drive could still get around without a driver.

We'll probably see that sometime in the not terribly distant future. Ten years? I don't know. I hope that tech becomes widely available by the time I can no longer drive.

The Present State of Self Driving

Full Self Driving (FSD) is taking far longer than originally expected. But, of course, that's not surprising. Highly anticipated emerging technologies are often hyper-optimistic about their rate of developmental success -- and autonomous vehicles are a prime example of that.

Let's go over the various levels of autonomy.

Level 0: What we've always had. No self-driving whatsoever.

Level 1: Car offers at least one of these assists: lane-keeping, braking, or acceleration (adaptive cruise control), in limited situations. Most new cars today fit the level 1 definition.

Level 2: Introduces ADAS (Advanced Driver Assist System) This is where all the components of Level 1 can be used together on certain roads, allowing brief hands-off but nearby and ready to take over at any time.

>> This level is the state of most vehicles today that advertise "self driving" capability. But it's absolutely not FSD (Full Self Driving) even thought it could feel that way on an interstate.

Level 3: Introduces self-driving on select roads and under certain conditions. The car can steer, brake, accelerate, and manage certain driving challenges. At this level, the driver generally can keep hands off the wheel and eyes off the road so long as the self-driving system is happy. With level 3 driving, you might travel for many miles including navigating interchanges, entrance and exit aprons, without having to intervene. But the driver must be prepared to take over when the car requests intervention. No napping!

This is generally the level being tested when a development car has a "safety driver".

This is widely considered by industry experts to be the toughest level to reach. Once this plateau has been reliably achieved then the remaining levels (4 and 5) should be at least a little bit easier to achieve because most of the key elements needed will be in place.

None of today's advertised "self driving" cars for sale meet the requirements for this level as of this writing.

Level 4: This really is FSD but geofenced to certain roadways where highly detailed maps are available. While on such geofenced roadways driving under level 4, it's not expected that a human would ever need to take control. The software must be developed enough to handle any exigent circumstance it encounters. From a technical viewpoint, napping could be permitted here. Laws, of course, may vary.

This is the level where unmanned robotaxis become possible, along with cars for people who don’t drive but live within such geofenced operating areas.

Level 5: This is similar to Level 4 but vehicles could self-drive anywhere with little to no geofencing. Cars capable of level 5 might not even have driving controls at all.

The Best and the Worst

Waymo is widely considered the front-runner when it comes to corporate integrity and technical advancement in developing Level 4 FSD. They are farther ahead than anyone and by all accounts is a well organized and serious operation. If anyone can get this done, Waymo can.

Tesla, on the other hand, sits at the other extreme. Tesla's boss is prone to bold, boastful claims about FSD and where Tesla is on the development curve. Their inflexible camera-only* approach lags behind industry leaders who are more evidence-based in their sensor modeling.

Indeed, Tesla has been caught up in multiple lawsuits over their exceedingly optimistic claims about how ready their Full Self-Driving really is. For years, they’ve been selling "FSD" packages for thousands of dollars, as much as $12 to $15k, promising that full self-driving was just around the corner, but it hasn’t materialized yet and likely won't for some time. Customers are pretty PO'd at being misled, having spent upward $3.4b thus far on promised FSD capability that, so far, has not arrived.

* While most companies developing self-driving cars, including Waymo, use an array of different sensors, each with their strengths, Tesla’s CEO has said that since humans only need eyes to drive, then cameras alone should be enough for self-driving cars.

But that overlooks the fact that the human brain is insanely fast at reading complex situations and almost instantly and instinctively reacting -- and usually the correct way. Computers, not so much. They need the additional input that a full sensor suite can provide. A lot of industry experts have been pretty critical of Tesla’s camera-only approach because of that.

So, yeah, Level 4 or 5 FSD would be fabulous!

But...

A Compulsory Autonomous Future

Let us speculate and fantasize about what a future of only autonomous vehicles might look like. A future where no one manually piloted an automobile.

That is, no more human drivers. In addition to all the benefits described above, just imagine the new ideas and modes of operation that become possible once and only when humans are no longer piloting any vehicle on the road -- with the possible exception of certain industrial applications such as road construction and infrastructure maintenance vehicles and such.

Automobile Redesign

Autonomous-only cars and trucks don't need a driver position. No steering wheel, shifters, pedals, or other controls to operate the movement of the vehicle. You'd still have all the other stuff such as climate control, entertainment options, navigational, etc. Perhaps you could opt for darker tinted glass that is presently illegal. Seating arrangement could be different as well.

Infrastructure Redesign

This is where things really get interesting. With all vehicles being autonomous (computer operated), these elements of infrastructure and other trappings of a human-piloted landscape could be radically different.

Note that some of these feature would still be useful in a permissive autonomous environment to the extent that a dual mode car is presently in autonomous mode.

> I2V: This stands for "Infrastructure to Vehicle". This means fixed assets like roadways, intersections, draw-bridges, railroad crossings, crosswalks, etc. all communicate their status to approaching vehicles. That way, the cars know what to expect and can compensate -- all far faster and more accurately than a human can.

> V2V: This stands for "Vehicle to Vehicle". This means all the cars in the nearby area communicate with and know where each other are, their speed, direction of travel, how quickly they can accelerate or decelerate, size of vehicle, mission of vehicle (first responder, etc.), and myriad other metrics that give complete situational awareness of what's what in the vicinity.

> No Traffic Signals: Computers don't need traffic lights. Autonomous vehicles could transit an intersection using a reservation dispatch system that negotiates and schedules micro time slots. Each AV enters the contention zone with permission, seamlessly interleaving with other AVs.

AV tech doesn't need the spacing and buffers that humans do so more AVs could transit the contention zone without conflict, increasing the throughout by perhaps 100%.

> Pedestrians: Crosswalks and other examples of roadway / pedway conflict points could utilize "beg buttons" that give pedestrians a guaranteed safe crossing signal that no car could violate.

> Platooning: With coordinated autonomous control (platooning), vehicles can safely run at tighter longitudinal spacing and at somewhat higher speeds. Lane capacity gains of 100% or maybe more are plausible on controlled-access roads.

> Traffic Law Enforcement: Various law enforcement agencies could focus on other tasks. Speeding tickets and DUIs would be a thing of the past.

> Insurance Reform: Car insurance costs would drop significantly and there would be a new model for universal coverage. Risk is no longer indicated by the driver which has always been the wildcard. Indeed, most risk is eliminated. We'd still have coverage for force majeure and other wacky one-offs, but the majority of insurance costs would be eliminated because we're removing the single largest component of risk and unpredictability -- the fallible human driving the car.

Insurance would also have to cover damage from possible cyberattack or faulty software that could cause wrecks to occur. This isn't nothing, especially if a widespread cyberattack were to damage a large number of vehicles.

> Lives Saved: Presently, over 30,000 people are killed each year in auto wrecks and well north of 100,000 are injured, many of them grievously. Imagine those numbers dropping by 90% or more. That is staggering to even contemplate. The daily gentle admonishment we give our loved ones to "drive carefully" and "be safe" are words we'd no longer have to say. Imagine that! No longer worry that your spouse won't make it home alive because some dumbass was playing with their phone, not paying attention.

> Radically Different Cityscape: Without human drivers, our car-centric cities can be designed very differently. For example, stores and office buildings would no longer need onsite parking for cars (deliveries and maintenance would still need loading docks or other staging areas). Instead, we'd have much smaller zones for passengers to enter and exit their vehicles.

There could be large dedicated parking lots scattered about a city every few miles instead of being located in front of every store and building. Homes would no longer need garages -- at least for cars. When you need your car, summon it from a nearby neighborhood parking lot and it could be there in a couple of minutes. With no human drivers, we no longer need to design infrastructure to include nearby parking.

> Vehicle Loaning: Some people who have no need to store personal belongings in a car while not in use could participate in a pool vehicle program. e.g. Two people owning the same car. Or belonging to a shared car club that owned a fleet of cars. Or if you owned your own car and you are home all day, your car could be out and about, hauling people around locally like an unmanned Uber of sorts. You could earn extra money while not needing your vehicle.

I'm sure you can imagine other ways that compulsory autonomy could change society. Some of those changes are good. A lot of what I'm saying sounds pie in the sky. Compulsory autonomy will almost certainly never happen and definitely not in my lifetime.

But it's cool to think about and ponder all possibilities. What the Germans would call a Gedankenexperiment.

Permissive full autonomy, on the other hand, is quite likely to happen in my lifetime. By the time I'm too old to drive (probably around 2045 or so), I expect autonomous vehicles to be widely available. Man, I can't wait.

Downside

The idea of compulsive autonomy isn't without its significant downsides. And to be sure I'm not saying I'm necessarily for it. This is a thought experiment. It's just a vision of what could be. Not necessarily what will be or even what I'd want it to be. There's certainly benefits -- and those benefits come with a price. But it has to be done right. That is, for society's benefit, and not the sole benefit of the corporations that are trying to bring this about.

A lot of who we are as Americans is tied up in our automobiles. America is a car culture unlike most other countries. It's hard to believe we'd give up driving our cars. A lot of why we buy the cars we do is for the identification, fun, and thrill of driving them. All that goes away with compulsive autonomy. I mean, if you can't actually drive the thing, then why have anything other than a generic people and thing mover? The individualism, the expression, the engine held tame under foot, your hand on the wheel, that call of the open road, only exists when we're in control. That would be a sad loss and a seismic shift in what it means to be American. I totally get it. Motorcycles would probably cease to exist, certainly not mixing with full compulsive autonomy. Which is sad because I've driven motorcycles my entire life.

But cities in the US today are mired in automobile-centric infrastructure. Most of our cities are already not walkable or safely bikeable. Some of the advantages of AVs, like higher density and faster speeds, can only come about with compulsory autonomous roadways with no at-grade interference from pedestrians and bicycles. Walking and biking becomes even more difficult than is already is. That is not good!

And more cars and at higher speeds, would be incredible noisy. Imagine how noisy it already is walking alongside a traditional stroad*. Now imagine there's 3x the cars moving 2x faster!

* A stroad is roadway that is a bastardized combination of a street and a road. A street is intended to be fairly low speed, with driveway curb cuts for accessing homes and businesses. A road is intended as a higher speed connector, bringing together different parts of a city. Ideally, a road has semi-controlled access with few or no driveway curb cuts and limited intersections. A stroad tries to do both of these mutually incompatible things and fails at doing either of them well.

But for you car-first enthusiasts, worry not. Compulsive autonomy, if it happens at all, will probably be no sooner than 50 to 100 years. It might be something your triple-great grandkids see.

Public Transit

Now you didn't think I was going to totally ignore that did you?

What we really need is more and better public transportation. I'll get into that in another article.