More on the interactions between humans and self-driving vehicles

Up from Central Square towards Harvard Square is a stretch of Massachusetts Avenue that is mixed residential and commercial, with metered parking. A few weeks ago I needed to stop at the UPS store there and ship a heavy package. There were no free parking spots so I soon found myself cruising up and down along about a 100 meter stretch, waiting for one to open up. The thought occurred to me that if I had had a level 4 or 5 self driving car I could have left it to do that circling, while I dropped into the store.

Such is the root of anti-social behavior.

And more:

(1) People will jump out of their car at a Starbucks to run in and pick up their order knowingly leaving it not in a legal parking spot, perhaps blocking others, but knowing that it will take care of getting out of the way if some other car needs to move or get by. That will be fine in the case there is no such need, but in the case of need it will slow everything down just a little. And perhaps the owner will be able to set the tolerance on how uncomfortable things have to get before the car moves. Expect to see lots of annoyed people. And before long grocery store parking lots, especially in a storm, will just be a sea of cars improperly parked waiting for their owners.

(2) This is one for the two (autonomous) car family. Suppose someone is going to an event in the evening and there is not much parking nearby. And suppose autonomous cars are now always prowling neighborhoods waiting for their owners to summon them, so it takes a while for any particular car to get through the traffic to the pick up location. Then the two car family may resort to a new trick so that they don’t have to wait quite so long as others for their cars to get to the front door pick up at the conclusion of the big social event. They send one of their cars earlier in the day to find the closest parking spot that it can, and it settles in for a long wait. They use their second car to drop them at the event and send it home immediately. When the event is over their first autonomous car is right there waiting for them–the cost to the commons was a parking spot occupied all day by one of their cars.

In sum:

They are seeing the technical possibilities and not seeing the resistance that will come with autonomous agents invading human spaces, be they too rude or overly polite.

That is by Rodney Brooks, the piece has other points of interest, via Tim Harford.


So, a self-driving car is basically identical to the way that a taxi can be used?

Sure, except that the self-driving car doesn't charge $72 an hour waiting time, and is clean, and has your stuff in it.

Price matters.

It does (charge $72/hr or maybe day or half day) when the autonomous tow truck hooks it and drags it off to hang suspended from another hook in the impound lot.

This is why we need pricing on road usage as much as we need pricing on parking.

If you want to circle the block, sure, but you're gonna pay $5 a minute to do it. This would be unthinkably complex 20 years ago, but rather easy with modern technology plus self-driving cars.

The biggest, er, roadblock will be that people hate paying to drive just like they hate paying to park.

I love this luddism, this angle is almost as good as the thought experiment of who the car will pick to kill, the driver or the child in the middle of the street?

How about this instead: we will continue to develop autonomous cars regardless of what all of the Luddite arts majors type into their iPhones?

I could not be more wrong in this case!

Hey Hey. If you're going to talk about "majors" we always need to remember that engineers are losers who are the lowest paid and lowest status profession on the totem pole.

Maybe not the lowest status, but point taken!

+1. This reminds me of a recent post where Tyler talked about the need to raise the status of scientists. This is a common theme among smart people. Yet no one ever talks about the need to raise the status of engineers (that's how low status they are).

There are, by my count, at least 2 bajillion movies where scientists are portrayed as heroes. In how many movies are engineers given a heroic role? They are few and far between. The only one that I can think of is Apollo 13.

A great engineer can make just as big a difference in the world as a great scientist. And I am willing to bet that the median engineer is making a much bigger impact than the median scientist. Given the widely publicised problems with academic research, the median scientist might not even be doing any good.

(I say all of this as an engineer by training, who now works in academia as a scientist, and thinks science is really important).

I'd say "The Martian" was the most engineering-centric movie of recent times. Though the hero was a botanist IIRC.

Wouldn't you consider Iron Man an engineer? Still agree with your point though that scientists are favored much more in movies

I think "comic book inventor" is a somewhat different thing and not really a celebration of real-world engineering.

"If you’re going to talk about “majors” we always need to remember that engineers are losers who are the lowest paid and lowest status profession on the totem pole."

Yours will be the first autonomous car to drive off a cliff for no apparent reason. Remember that every time you get into an autonomous car...for the rest of your life.

The preceding was merely friendly advice from someone well-aware of the generally anti-social tendencies of engineers.

In this day and age of invigorated political innovation there are people still obsessing trivially about self driving cars? I fear that with Trump dominant Tyler will lose his compass altogether, and cohort ever more with an out of touch mass media, but with Alex gone to India he will be irredeemably lost (almost as much as Alex will be lost in India!). What a dreadful shock. The lesson of the age is ... self-drive. The most intelligent people hope Trump may introduce a new domestic paradigm — individual responsibility, taking the wheel, getting yourself from A to B without fancy assistance.

Yes, the days when Americans would, instead of taking the wheel, simply flap their arms and take flight have already taken on a tinge of nostalgia in my memory.

What would Trump know about individual responsibility?

#1 seems easily solvable. Just give unattended cars in non-parking spaces tickets (as happens now). Enforcement methods, and perhaps regulations, may need to be tweaked to deal specifically with self-driving cars, but I see no intractable problem.

#2 is a bit farfetched. And some edge cases aside, clearly the scenario "people get dropped off by their autonomous vehicle" results in lower overall parking usage than "people drive to events and park".

I think you misread #2. In either case, beware the autonomous tow trucks.

I understand that #2 involves a car blocking a parking space all day. But my point is, that specific scenario seems rare, and it implies a general circumstance where people are often getting dropped off at events by autonomous cars, which would almost certainly reduce parking space usage overall.


Result: more driving, less parking.

Or, gosh, we could actually consult a traffic engineer (, I love you) or maybe just go look at how these hypotheticals actually work in the real world (like Manhattan, with its Yellow, Green, Black, and Uber cars, busses, subways, and even helicopters.)

I foresee a need for realistic driverbots, so an uninhabited driverless car can orbit with an apparent driver. Oh sure, there have been many amateur attempts to make dummy passengers for using carpool lanes. I'm talking about something far more advanced. It would move its head, look both ways before entering an intersection, mouth some obscenity when cut off. Only extreme scrutiny would reveal it's a dummy occupying the seat of an orbiting car -- scrutiny which would not be cost-effective for traffic law enforcement.

Would it be illegal for a car to drive unoccupied? I think cities would want to allow the cars to leave and park somewhere else to solve parking issues.

Of course that would make it hard to police circling cars.

If the way we define a car as "parked" right now (that is, only when they are standing still without an adult in the driving seat) were to become obsolete, together with the concomitant way we charge for parking, some form of congestion pricing seems almost inevitable.

The fact that this post has even appeared shows the incredible cultural dominance of the automobile in the US. When a corporation names an employee of the month, will her autonomous car be parked beside the autonomous car of the boss? Will the average distance walked from the car to the front door of the super market shrink by what percent when autonomous cars take over? Won't that lead to higher average weights? Why even go to the market when the autonomous car can go pick up the muesli and arugula by itself? Why actually go to an event when it will be available instantly on a youtube channel? UPS has facilities with lots of parking, no problem getting a package on one of their brown trucks. Soon nobody will really need to go anywhere.

I disagree. Autonomous cars will make services like taxis or Uber so cheap that almost no one will have a car, or two. In the future, you will go to a venue and autonomous taxis will be waiting and coordinating automatically among themselves to get you in and out as quick as possible, better than today.

However, I do not believe the hype that they are around the corner. I think that it will be in 20 years or so.


I don't think it will get down to "almost no one". There are benefits to owning a car, such as being able to carry your own junk around in it and having instant availability. Plus cars are entertainment and/or status symbols for many.

Still, this is a very good point. Self-driving car services should greatly reduce the demand for car ownership.

Ubers basically can't carry children. Also, Uber works a lot better in dense areas so dispatch is quick and involves a nearby vehicle. Outside the city core, ownership makes more sense.

If driverless cars never crash, do you still need the child car seat laws?

If they really never crash, getting rid of child seats would be awesome. They have the minor benefit of keeping the child sitting in one location, but otherwise they are a total pain.

I didn't think of that possibility, but it seems like it might work in the long term. Cool!

You still have to worry about other drivers, plus many accidents are not caused by driver error.

"Outside the city core, ownership makes more sense."

Insanity is owning a car 24 hours a day that's sitting parked 22-23+ hours a day.

Most Americans live in metro areas, but in the suburbs and exurbs -- not in dense central cities. The Chicago metro area, for example, has about 9 1/2 million people and more than 2/3rds live outside the city. When/if autonomous vehicles become viable, most Americans will still have their own cars and they'll use them to go shopping, where they will find lots with free parking.

In fact, more people will own cars, because old people and children can viably use them in this future. And urban dwellers for whom parking was more expensive than the car payment can now economically own them.

The Uber effect will in some respects be an offset, but Uber will also be stealing customers from traditional public transit, so it's not an obvious negative for car ownership. It will make little sense to have subways when autonomous Ubers are cheaper.

It makes little sense to have subways _now_ in many cities. In NYC, they might make sense for poor people since you have an existing dense network with lots of rolling stock. But in places like Boston, where uberPOOL is already within a factor of two of subway costs, stops are few and far between with major inaccessible zones, and trains may be 10 minutes apart, subways make little sense.

Everyone uses the subway in NYC, not just poor people. I use the subway. The partners making $3m+ a year at my firm use the subway. Door to door to work with the subway for me is about 20 mins if you presume maximum wait time for trains (which are 6 mins apart). If I took an uber to work it would take me about 30 mins. If I took an uberpool it would probably take 45 minutes.

Your point may be true in Boston now, but if you took away the subways, the price of uberpool goes up and more traffic increases your commute time too.

To each his own, I guess. Your firm might be unusual.

I'm in NYC a lot for business and I never touch the subway. If you Google "uber vs subway in nyc", you'll see some controversy, but it by and large comes out in favor of Uber for speed.

Which isn't that surprising - NYC is pretty easy to drive around. Or at least it was before the election.

The subway is well used, but often a 20 min ride is to where a person parked, and another 30 min drive is still needed.

NYC is hardly a typical American city, as its high density and extensive subway network make it atypical.

Nonetheless, I'm not surprised that Uber (or any car service) is often faster than the subway. For the problem with the subway is the problem with all transit: it can seldom offer a single-seat ride.

I don't doubt that the subway is faster if your trip involves boarding at an express stop, riding a long distance, and then exiting at another express stop on that subway line; it's just that actual subway journeys usually involve more than one train plus a substantial amount of walking (which may be good for you, but, it does take time and it's inconvenient if you're carrying anything too big to put in a backpack).

As others have remarked, it's (usually) easy to drive in NYC, with some notable exceptions (four of the City's five boroughs are on islands, which can produce intense congestion at bridges and tunnels). BUT the reason why NYC is so driveable is because so few people drive in it, due to the high cost and limited availability of parking. BUT if many subway riders were to switch to Uber, those streets would be far more crowded.

wait's firm is typical of NYC (which is, of course, an atypical city.) My boss, at $10M+/year, took it to get to and from work. So did I, so did the guy who shines our shoes in the office.

Remember that police forces will have autonomous cars, too. These cars will drive around giving tickets to autonomous cars.

Of course, with adequate public transit the potential problems with autonomous vehicles would be avoided. But don't expect public transit, or public anything. As for autonomous vehicles, I will remind readers that, according to the candor (or gaffe) of Google's engineers, the maximum speed for autonomous vehicles will be about 30 mph as long as they must share the road with non-autonomous vehicles. Will those suburbanites venture into the city in their autonomous vehicle if the darn thing can't go more than 30 mph? That means one of two things must happen: non-autonomous vehicles must be abandoned or a separate right of way must be built for the autonomous vehicles. Are suburbanites really going to give up their massive SUVs and Porsches? I think not. That leaves building a separate right of way for autonomous vehicles. In other words, transit but with a twist: it's public in the sense that the public pays for it, private in the sense that private parties (e.g., Uber) profit from it. Yes, the dreaded public-private partnership!

You can't go more than 30 mph in Central Square now in a non-autonomous car. Suburbanites still visit, despite the traffic and parking hassles. I assume alleviating both hassles with fully autonomous cars would make this more likely, not less.

The frustration component of traffic would be a less. Currently dense traffic is frustrating because you have to pay a lot of attention, notice that that guy two cars up could have made the light but didn't. If the car were autonomous, you would't have to pay attention. Watch tv, post on MR, talk to your companions face to face. The extra five minutes won't even be noticed.

I do agree people will have large vehicles. They will be more oriented inside to activities other than driving. For instance, turn the seat around. The delays will be even less noticed.

Driving slowly in dense city traffic actually isn't that hard, it just feels hard because of the attention humans must pay. Dense cities would also have the advantage of being mapped to the micrometer so that auto cars could know in advance where every bump and pothole is located.

I do think we will see separate lanes on interstate highways where only auto cars are allowed to go, and do things that humans would surely break. Like caravaning with minimal following distance using coordinated braking.

Most challenging will be the suburban/rural non-limited-access roads, the 45mph arteries where more human games are played, like nosing your way out of the driveway to get someone to let you in.

Well I'm glad it's going up. All those times last year you claimed the max speed is 25 mph made me worry. Do your friends at google predict another 5 mph increase this year? Soon we'll all be doing 100.

I'm guessing that when the automobile first came on the scene, people had trouble envisioning how it would mimic the way that horses were used.

If self-driving cars become a reality, I'm pretty sure that different norms and environments will emerge, if allowed. I can only imagine that the layout of our grocery store parking lots will not stay the same. For that matter, perhaps the nature of the stores themselves will change. Maybe big parking lots disappear. Who knows?

After all, the only way these quaint nostalgic town squares from the 1800s survive today, with their coffee shops, kitsch stores, niche restaurants, and frustratingly limited parking, is through the efforts of the political sector. Otherwise, these useless, yet somewhat charming, old town-centers designed for horses would have disappeared long ago.

I used to hear stories from old men in the rural area where I grew up about how they missed taking a horse and carriage home because you could fall asleep and the horse still knew where to take you. That feature was fundamentally different on the car, and has only been re-realized via autonomous technology.


Self driving cars:
1. Reduce the cost use a 'drop off zone' (increase demand)
2. Reduce demand for parking (car picks someone else up or drives around the block)
3. Have no short term impact on the supply 'drop off zones ' or 'Parking spaces'

We can conclude that self driving cars will:
1. cause either congestion or fees when using 'drop off zones'
2. cause parking prices to fall, or create empty spaces.

1. It doesn't matter who owns the cars.
2. Drop off zones are just a special case of 'roads' which will also face increased demand, with similar supply.

More congestion, perhaps especially around pick

Rod Brooks is brilliant. In many ways he's the opposite of Robin Hanson - he's the chief proponent embodied intelligence.

His examples are the tip of the iceberg. The first big effect of real autonomous cars, assuming they work out technically, is a reduction in the need for parking relative to cars. Autonomous driving will bring lots more cars into the city because parking them will no longer be a $500/month proposition. People differ on predictions for the number of cars we'll need, but I think a big reduction in the cost of ownership will mean more cars all around. It will definitely mean higher utilization of existing cars and more driving miles irrespective of the number of cars.

It may be economic to have the family car drive parent a to work and then drive home to be available to parent b during the day, only to return to pick up parent a at the end of the day. Half the cars, almost double the driving (parent a's driving x 2 + parent b's driving) but parking goes to zero.

Cars will also get bigger. If I'm not driving, there's no reason to be hunched over the controls like a fighter pilot. Cars will start to look more like mobile living rooms or offices. Think about the backseat of a Maybach.

Longer-term, cities will accelerate the trend of spreading out rather than going up. There will be much greater need for drop-off points to access buildings, which is going to be tough because that's not something we do well.

Maybe, but people don't do this now, even if they can afford them (and a driver). The Maybach was basically discontinued and just used to brand larger S-class Mercedes. This is why I don't think autonomous cars will take off. It will be a niche market.

(2) is intriguing and I have actually done this (manually) in Boston; dropped a car off in a parking garage to circumvent huge parking increases that occur before an event. Free parking in Boston is a rarity.

Post author Rodney is not thinking of the other effects - he's considering that autonomous cars will follow the same patterns as traditional cars.

In the UPS/Starbucks case, idling out front, there is no reason the self-driving car has to wait outside. It could check google maps and find a 10-minute route that it would circle once and stay out of the way. So there wouldn't be a bunch of cars piling up. There is also the case that the owner would not have to actually go to UPS, but send the car to drop off the package by itself.

In the second case, the longer events are actually easier. Once dropping the passengers off, the car could easily drive a couple miles to a remote parking lot and wait to be summoned, or know when the event ends and move closer to the venue at the designated time. Thinking about Chicago: The car drops the family off at the Chicago theater at State and Randolf, then heads over to Lower Wacker and drives south or west for ten minutes, finding plenty of parking. The cars could also coordinate and park more densely. If they knew and communicate which car needed to leave when, they could box in the later cars. If an exception arose, the boxed-in car could request that the others let it out early.

More fanciful schemes include letting the car go make "Uber runs" while you're in the show and earn money to defray the cost of ownership, but I am sure many people will want "their car" and see it as a form of private living space.

I'm getting old enough to anticipate that I might live long enough to be unable physically or mentally to drive. Auto-mobiles will be wonderful for those who can't drive. As airlines pack us into seats that are barely large enough for the average adult, I'd much rather use my car and take a couple of books I've been meaning to read. Road trip will have a whole different meaning. A problem with remote parking is that if everyone does it, then there could be a LOT of traffic trying to get to the same place at the same time. And what if "Autonomous Vehicles Prohibited" signs/(laws) (for parking, and perhaps certain curb lanes) becomes common? Not to mention pirates. I don't think enough thought has been put into thinking about how people will abuse it, and how people will screw it up either because they have an agenda, or just because.

There will certainly be more sex in cars.

I should think a seating arrangement more conducive to doggie style, for example, will be offered.

Show rooms are suddenly going to be more intersting places.

Maybe wait until real self driving cars are a reality. there are still a lot of issues.

Outside of specific problems like the one in the OP, the larger point is that self-driving cars have to thread their way into our very complex transportation system. This means there will be unknown unknowns and unintended consequences. We simply do not know how this will shake out. What happens when the first autonomous car malfunctions and kills people? What happens if a car doesn't malfunction but still kills someone because its AI told it that was the best option? What happens when people start gaming the system by exploiting the rules autonomous vehicles follow? We are just starting to think about the second-order and third-order interactions resulting from such a massive change to one of the core mechanisms of society,

The truth is we simply have no idea how this will shake out, and probably will not be able to predict or plan for these changes - we are treading into areas where we will need new social norms, new legal rules, and major changes to a very complex system. A social media storm could result in laws that completely change how we interact with autonomous vehicles.

The Segway was supposed to revolutionize transportation by solving the 'last mile' problem of public transit. Some people were predicting that it was so revolutionary it would change the way we build cities. Technically, the Segway did what it promised. What couldn't be predicted was that society generally rejected the idea of sharing sidewalks and elevators with them. Laws were passed, and the Segway was relegated to niche uses.

Likewise, Google glass was a technical success that had some very valuable uses. It failed not because of the technology, but because wearing it made you look like a douchebag. That's a social interaction failure that was utterly unforseen by Google engineers, economists and futurists.

Cities should charge for street parking and congestion with or without autonomous vehicles.

I wrote a post [1] about Rodeny Brooks' (and others') "self-driving-cars-and-humans" issues, and especially about how manufacturers of self-driving cars should verify their solutions for these issues. Here is the summary:

The interactions between Autonomous Vehicles and people can be complex, which complicates AV deployment. This post summarizes some recent related publications, and then tries to predict the verification implications of all that. For instance, it suggests that verification teams will try to track total-accidents, AV-specific-accidents and AV-specific-annoyances.


Instead of leaving the carbot in the street why not have it circle around until the rider flags it down? The rider can wait for the carbot to cicle back without blocking traffic. As a former city driver in Boston and Cambridge I noticed it was the double parked vehicles, incompetent parkers, and people stopped and waiting for someone to leave a parking space that caused the most traffic delays. Carbots are going to make city and suburban life more pleasant.

The first problem could be solved technologically: the car's sensors (video and infrared) could detect the presence of humans and interpret possible gestures. Then, the car's actuators could use visual and auditory outputs to communicate what it was planning to do. For example, the car could play the "screeching tires" sound, project a crosswalk, or play the "crosswalk almost over" beep to indicate that it was stopping/stopped. It could then flash colored lights or play the starting/revving engine sound to communicate that it was about to move from stopped position.

The second problem is a negative externality, and should could be solved by taxation per mile, minute, or gallon of fuel. Hopefully, human citizens would support taxation of the robots sharing their roads. I think people an asking for an autonomous car is like Henry Ford's customers asking for a faster horse.

I hope that autonomous cars in suburban/urban areas do not drive faster than 30 MPH. As a neurologist, I am concerned by the 10x mortality increase among pedestrians hit at 20 vs 30 MPH. I also hope that AI will have a response prepared for when it inevitably injures or kills a human (which will certainly happen after enough miles driven).

This degenerate scenario presumes that culture will remain static in the face of innovation. Why would you bother bringing your cars when autonomous taxis are ubiquitous and operate just above the marginal cost of fuel? Why would you bother owning a car at all?

Who will buy music after Amazon Music Unlimited is less than $5 a month? (Maybe lovers of the atonal music that wasn't on Amazon when you posted about it.)

How easy is parking going to be for autonomous cars? Existing parking lots full of pedestrians seem pretty nightmarish to program a robot car to deal with.

I don't think so. It's much easier to make the technology work at 5mph.

Self-driving car bombs...

So it will be like owning a donkey.

"And before long grocery store parking lots, especially in a storm, will just be a sea of cars improperly parked waiting for their owners."

He's apparently unaware of the near certainty that groceries will not be picked up a stores, but will be delivered by autonomous delivery vans.

In 30 years, less than 30 percent of people in the U.S. will own a car. And more than 70 percent of present brick-and-mortar stores will be shuttered.

Comments for this post are closed