Tesla Model S can now park itself | Hacker News Reader

Tesla Model S can now park itself

techcrunch.com

236 pointsprostoalex169 comments

Open articleView on HN

It's about time. Volkswagen had that in 2008.[1] Audi in 2012. Chevrolet in 2014.[2]

[1] http://www.gizmag.com/volkswagen-demonstrates-fully-automati... [2] https://www.youtube.com/watch?v=3zh_5QGA1zw

BMW 2015 https://www.youtube.com/watch?v=UDQxcvYzAog

Honda 2013 https://www.youtube.com/watch?v=6MjmRDIVxws

Nissan 2014 https://www.youtube.com/watch?v=bQZ-UDY3lpQ

Audi 2013 https://www.youtube.com/watch?v=vt20UnkmkLI

Lexus 2009 https://www.youtube.com/watch?v=5ldLpPehy1k (TopGear)

It's a bit difficult to tell from the videos, but it seems like the only one of those that was a production vehicle was the Nissan, and there was quite a bit of manual intervention (shifting into forward and reverse).

To be fair, I don't really understand the point of the Tesla feature either... can it do anything but pull the car from right outside a garage to inside the garage and vice versa?

They have rolled out some of the autopark features. I think all the sold cars required the driver in the seat. The versions that I have tested myself have all been horrible. They were so slow, or required unnatural maneuvers to scan the space. One dealer told me the feature was only for "People who cannot look over their shoulder, you know ... old people." None of them would un-park the car. I still needed the basic parallel parking skills to get out of the spot.

We live in the centre of a city, Edinburgh, where there is only street side parking and I've been using autopark for years and it works really well - it is faster and more accurate than my parking and I thought I was pretty good at parallel parking!

Would you mind sharing what kind of vehicle you drive?

Nothing fancy - a ŠKODA Yeti (NB it's "park assist" rather than complete autopark).

I don't own one, but one possibility that springs to mind is, if you wanted, you could pack your garage with crap, leaving only a car-spaced hole. (You know, like most people do, except for the hole.) Then you use this feature to extract the car from the garage, without needing space to walk to it or open doors.

I'm only partially joking. Our two-car garage is basically large enough for two average size cars and.. that's about it. We prioritize parking over junk-collection, but I could see the use. Of course in the winter, in some climates at least, it removes one of the main benefits of the garage.

I guess one other benefit would be that you could always use your front door, whether walking or driving, so your shoes and stuff could all be in one place. Might be a minor advantage for some.

The Model S is a very large car, that makes it hard to fit in car parks in buildings (apartments, offices, shopping centres).

I've seen more than one or two stories of Model S owners having to climb in/out via the rear doors, because there's a pylon or wall in the way.

Having the car be able to pull out/park itself in those situations would likely be a major win for some people.

I rented a Ford Focus in France at the end of last year which could park itself either in a car park (i.e. cars parallel to each other) or parallel park at the side of the road.

I tried it parallel parking on a bend with a big drop next to it (so no walls to figure out where it should stop), and once I figured out which button to press it worked quite well.You still have to manually shift from forward to reverse (when it prompted you) and use the brake, but other than that it's hands off.

The best feature of this car wasn't this, but the technology that powers it - sensors all around the car to detect how close you are to something. It had a small indicator on the dashboard to let you know the distances.

The 2008 version still required manual braking (legal restriction), at least in Europe.

I doubt it was legally mandated, but it certainly isn't now: Model S brakes itself and IMHO it's much safer.

When I took a test drive of the Model S recently, one thing that struck me about the autopilot was that it's not (in my opinion) sufficiently clear what mode the car is operating in at the moment. Semiautomatic mode should be a glaringly obvious UI change, but instead it's just an icon. It reminds me of certain fatal plane crashes that led to improvements in cockpit layout and operation. Are we about to make the same mistakes with cars?

In particular, I think Aeroflot Flight 593 [0] is representative of this sort of problem. Autopilot control of the ailerons was accidentally disabled without clear feedback of the change in mode, this led to overcorrection by the autopilot using other control mechanisms as it attempted to restore the aircraft's heading and within a minute the plane entered an uncontrolled dive and crashed.

[0]: https://en.wikipedia.org/wiki/Aeroflot_Flight_593

This is an old video, but I love watching these airline pilot saftey presentations. This video in particular covers a lot of these automation issues.

https://www.youtube.com/watch?v=pN41LvuSz10

Your comment made me realize it would be a great idea to provide external signs as well, showing to people outside that the car is operating in autonomous mode. It could be lights for examples, a bit like when the car is going to move backward, only less powerful and with a different color.

This should also be standardized in a way across manufacturers.

Absolutely. I would suggest that cars have two other standardised lights in addition to turn signals:

a) Brake light on the front of the car

So that when you turn across traffic on a yellow light you can see that the oncoming car is braking.

b) Thank-you light on the back of a car. Maybe blue.

We can already express displeasure, but wouldn't it be nice to express pleasure with other road users.

> b) Thank-you light on the back of a car. Maybe blue.

I still wonder why those aren't popular:

http://www.thinkgeek.com/product/d138/

Is there a regulatory reason against placing something like it behind your rear windshield?

If not regulatory, then it's certainly a cultural thing. Many cars and buses in parts of the middle east (certainly Iran, Syria and Egypt, but my memory has blurred which countries in particular, it was some years ago) had some crazy extra lights in their rear windows.

As it says, it violates a few state laws. Mostly, it is seen as a distraction to the driving of others.

Blue in most jurisdictions is reserved for the police. Police are red-white-blue but often just red-blue, fire/ambulance are red/white and service vehicles are yellow.

I wonder if anyone here knows what purple lights are reserved for? A purple light on a car is thing all drivers are meant to know.

> I wonder if anyone here knows what purple lights are reserved for? A purple light on a car is thing all drivers are meant to know.

Apparently for anything, from funeral service to escorting oversized vehicles to HAZMAT response to whatever you feel like - depending on which country you live in :).

https://en.wikipedia.org/wiki/Emergency_vehicle_lighting

a) brake lights - at least of same color - in front will not happen - because you instinctively know - especially in poor visibility) to follow red lights and avoid the bright white ones - which is oncoming traffic...

b) in most of europe the thank you is done with a momentary press of the hazard button - thats why its generally so easy to access on most european-targeted cars.

I like the idea of a second subtle horn that sounds like clearing your throat. That would be better way to remind someone that the light turned green or that they're blocking you.

I always thought the brake lights should flash or get brighter during rapid braking. Sometimes it's hard to tell how fast cars are slowing down.

That's a good idea. Motorcyclists are taught to rapidly close and open the brake, in order to flash the rear brake light for extra visibility when stopping.

Thats already a feature in most modern cars. Seen it plenty of times on the Autobahn. It's really helpful/a life saver to know if the car in front of you just slows down or is doing an emergency brake.

>>> a) Brake light on the front of the car so that when you turn across traffic on a yellow light you can see that the oncoming car is braking.

Brake lights /= stopping, or even slowing down. They indicate that the brake pedal has been depressed 1/8th of an inch. They do not mean the other car sees you, is about to stop, or can stop. The lights you suggest would create a false sense of security, leading only to more danger. Do not ever pull in front of an oncoming car on the basis that you think it sees you and will stop. Wait the extra 2 seconds for it to actually stop.

Why? As a driver of another car, what exactly am I expected to do when I see a car in some kind of autonomous mode? It's not like reversing lights or turn indicators, which give a hint about what the driver is about to do.

If the 'autonomous mode lights' are meant to be some kind of warning, then that's just telling me that an autonomous mode isn't fit for real world use yet.

Currently manufacturers feel that they need to have their cards obey the law in every case, which is what people need to be warned about.

>It's not like reversing lights or turn indicators, which give a hint about what the driver is about to do.

On the contrary, such an indication would be precisely a hint to other drivers as to what sort of behavior one should expect from the auto-driving car.

I'd be quite a lot happier with the concept of autonomous automobiles if they'd just communicate better, personally .. like, autonomous cars should be publicly trackable with an open information network, imho .. so we, the public, can know when we are not dealing with a human but rather a machine, in life-critical situations.

>a hint to other drivers as to what sort of behavior one should expect

The point is--what behavior is that? If I saw such an indicator today, I'd have no idea what I should be expecting that's different from normal. Is it like a student driver sign (i.e. who knows what this car is about to do)? Then it probably shouldn't be allowed.

I'm sure this isn't what the OP had in mind but I think it would help adoption. At first you see one or two of the cars trickle onto the roadways and you see them behave. Then a few more. I think this would help the comfort level of people concerned about the safety and efficacy of AI driving.

Things like knowing that honking or flashing lights or screaming at the car will probably be pointless ?

Another reason is that we will definitely have bugs sometimes at the beginning, or non-human-like behaviors ( for which it's a bit early for me to give you real life examples).

Yes, if the "driver" unware of sounds and lights, the driver should not be driving.

Like the idea - maybe like a learner or probabtionary driver who has to display their L or P plates to indicate they are not fully competent.

I think it's just as clear as the typical cruise control indicator, and the failure to notice that has similarly serious consequences (If you think cruise control is off when it's on you're likely to hit something pretty quickly).

The major indicator is that the wheel becomes quite stiff and resists turning it. The next one is the sound it makes when engaging. The icon is relatively secondary.

And, crucially, disengaging - with a beep if you have hands on wheel and a panicky "crash imminent, we're all going to die, do something!" series of beeps if you don't. Being aware of the car's mode is not an issue.

This is a pretty misleading headline.

    s/drive/park/

That said, self-parking ability is still a big deal. Imagine when, say, 35% of the population gets them. You could have super-dense, extra-cheap parking lots for self-parking cars only. They could be packed extraordinarily tight, with no need for human valets.

Your car's boxed in all the way in the back? It'll just tell the ones in its way to please get out for a moment. One car breaks down? That's probably rare enough that you can move the ones around it and bring in a tow truck or human driver.

Of course it is. It's another TC article.

Packed extra dense? They could no doubt shave a few inches here and there, but the basics of parking remain.

(1) They would have to have lanes, else every car have to move every time someone wanted out. And these lanes would have to be wide enough that the car with the largest turning radius could still get out reasonably.

(2) There would still need to be some standard width for spaces, else a thin car leaving would leave a 3-4' wide space that nobody else could fit into. And a thin car moving into the gap left by a wide car wouldn't be very efficient.

(3) Accommodation would still be needed for humans/pedestrians to deal with issues or rescue stuck cars.

(4) The new nightmare of the loading area outside stores as autodrive cars line up like cabs at the airport pickup up passengers. At least in a parking lot everyone can load/unload without forcing everyone else to wait.

(5) The fire safety rules would still apply. Deep in building codes are provisions for exactly how much of various dangerous things are allowed inside parking lots. A bunch of tightly-packed, battery packed, teslas in a big concrete box is a firefighter's nightmare.

Imho far more space could be saved in north america if parking lots had better accommodation for motorcycles. Everyone who rides a bike rather than drive a car effectively saves 3/4 of a parking space if spots are made for them.

Not to mention you wouldn't be able to box cars in. You still need to be able to pull the car ou, with or without a driver.

This is just a personal valet service which is still exciting.

Why can't you box cars in? When I park in an attended lot in NYC and Philly, my car is often boxed in because space is so tight. I either tell them when I expect to be back or wait for them to jockey cars around. With self driving cars, it could be straightforward to program them to move around to extricate a boxed in vehicle.

And also win big at Rush Hour: http://www.amazon.com/Think-Fun-5000-ThinkFun-Rush/dp/B00000...

(1) They would have to have lanes, else every car have to move every time someone wanted out.

Which is no longer necessarily a problem.

Why? Are all of these cars going to automagically cooperate in some sort of parking ballet every time the guy at the back wants out?

Why not? Valets do this all the time.

Eventually, why not?

Because the human valet is effectively a common control system. To get all the cars to do this without humans they would need to be networked and agree some common communication standards. Cars aren't cellphones. The chances of a few hundred random vehicles, covering perhaps 20 different model years, all dancing together is a pipe dream.

They're not random vehicles. The special lots would only be for the cars with this feature. If 100% of Teslas, BMW's, and Apple Cars supported it, you'd quickly see a bunch of other automakers get on board (Why not? If the car can already self-park, this isn't much harder from that point.)

At first, it would just be a few lots in the densest parts of cities. Within, I don't know, seven years, you'd be seeing them all over the place. Every shopping mall would have a special area for them, right near the entrance.

Oh, and don't forget: Cars of the near future (and Teslas today) can have features like this added years after leaving the factory, as an over-the-air software update.

Ah. Teslas Apples and BMWs only. The rich man's lot. And this is going to be cheaper?

Software can be updated, but not the cameras and position sensors needed for cars to properly squeeze next to each other better than humans can. That tech will need hardware updates as the technology evolves. Unless of course those sensors will be integrated into the parking lot, as Honda did a couple years ago, but they weren't going for space efficiency.

Supercharger stations are Tesla-only today. Remind me, are they cheaper or more expensive than gas stations?

This is about parking, not powerplants.

I cannot think of any service that limits itself to an exclusive client list yet is cheaper than the same service that is open to a wider client base. Exclusivity always come at a premium. And I haven't seen any gas station charge anyone for parking. So yes, supercharge stations are not more expensive than gas stations. It's still a ridiculous comparison.

There already exist parking spaces that are for extra-small cars (like Smart cars) only. These will often be cheaper than alternate parking, or at least much more convenient. It's basically the exact same thing - a parking advantage offered because a feature of the car allows for it. In one case, size; in the other, automation.

Some local authorities in the UK provide free or heavily subsidised parking for electric cars.

https://www.westminster.gov.uk/electric-vehicles

Including the the cheap ones! ( http://www.goingreen.co.uk/product-category/g-wiz )

>I cannot think of any service that limits itself to an exclusive client list yet is cheaper than the same service that is open to a wider client base.

Exclusivity doesn't have to make a service more expensive. It may be possible to reduce operational costs by only allowing self-driving cars in a parking lot. The upfront costs might be higher (but not necessarily). If there is enough demand, it could be economically feasible.

Imagine a future where Level 4 self-driving cars are ubiquitous - the car would drop you off, choose the best parking lot (closest, cheapest), drive to it and park itself automatically.

> I cannot think of any service that limits itself to an exclusive client list yet is cheaper than the same service that is open to a wider client base

When approximately the same number of people had cars as had horses, how did the price of leaving the car somewhere overnight compare to the price of leaving the horse?

Awesome. "Can't afford a new car? You get to park at the back of the lot and walk."

I can afford a new car, but always park at the back of the log and walk, because I don't want anybody parked near my car. :)

I agree that standards are needed. However, a common control system is still an option. Just a robot with a radio instead of the human.

Cars aren't cellphones, but Teslas have both cellular and Wi-Fi transceivers in them.

Yes, that's exactly what they would do. If your car wasn't part of the 35% with that feature, you'd have to park in some other lot.

Now there are two parking lots? One by definition will always fill before the second, leaving people on the street while spots in the other lot sit empty. So much for any efficiency gains.

Incentivizing the people on the street to upgrade to self-parking cars. That's a giant feature, not a bug.

If you are selling a service, you want to be ahead of the demand. A parking lot, a business, isn't going to get far by lecturing to customers as to their car choice. That would be like me shutting my office to any client I though needed a better business plan. You have to take the customer as they are.

Sure, but not every business needs to accomodate all the needs of every consumer. Customers who are looking to purchase a tube of toothpaste probably wont be coming to your office in the first place.

No, you absolutely do not. Have you not heard of, "No shirt, no shoes, no service", or "You must be at least this tall to ride", or "This event is open to anyone 21 years of age or older" or "This software requires Windows 98 or later"?

It's very easy to imagine this happening, mainly because the economic incentives line up for everyone involved. If you are a garage owner, you could create a "networked self-parking car only lot" - benefit for you is that you can park a lot more cars into the same space, benefit for the driver is they have more options to park, and benefit for the manufacturer is they can upsell you.

The self-parking car is the hard part. The algorithm to get a bunch of cars to move out of the way seems quite easy by comparison.

And whose responsibility is it if something goes wrong ?

As a driver I am not going to invalidate my insurance just to save a few dollars in parking. Especially when the person managing this orchestra may not be capable of keeping everything in sync.

Those problems are bureaucratic problems that will be solved with the money at stake. You're already protected with valet parking, through a combination of your own insurance and the valets. This wouldn't be any different, you'd pay a few more bucks per year in insurance to save a large percentage in parking -- it would make economic sense in densely packed cities.

It is very inefficient, if not anything else.

What's inefficient is devoting large swatches of useful real estate to parking lots, when the same number of cars could be accommodated in much less space.

I'm thinking 3D. A cuboid parking area where the car rests/charges, stacked high, wide and deep of cars. No need for regular staff access. When a car is needed it disengages, connects with a ramp/lift loading/unloading platform, and drives itself to you.

That's getting much more complicated than the concrete slab that is a parking lot today. Cars on lifts/ramps hauled by robots has been done many times, but its far from cheap and has never caught on. It just requires too much maintenance. Valets are cheaper.

Isn't that a very common sight in Japan? The cost of land in metropolitan areas easily overshadows a bunch of hydraulics and elevators produced at scale.

Many companies already make similar products for parking, e.g. [1]. The autonomous, unoccupied drive to the parking structure and being connected to power without human intervention are the more pressing problems.

1. http://www.parkmatic.com

> A bunch of tightly-packed, battery packed, teslas in a big concrete box is a firefighter's nightmare.

That's certainly moderately concerning, but I can assure you a firefighter's nightmares are far darker...

Pretty much. Confronted with a parking lot of burning electric cars, I might point some deck guns and aerials at it, but it'll be more "watch it burn". Risk a little for a little. It's insured property, not savable life(s).

Unless this is an underground lot and there is a worry that the heat of 300 burning teslas might do something to the building above. It would at least give off some nasty fumes. Someone will want that one car put out before it propagates.

And is water the correct thing to throw on a few hundred pounds of burning lithum batteries? I really don't know, but I suspect not.

  the heat of 300 burning teslas

"300 Burning Teslas" sounds like a band name.

Or a "Hollywood Undead" song...

Obligatory https://xkcd.com/1025/

We had a Tesla fire in Norway a couple of weeks ago, happened when fast-charging a fortunately empty car. It burned down completely. In the aftermath, there have been statements indicating that people won't put charging stations in underground car parks. But I also believe (lots of) water is the recommended way to put it out.

Lithium ion batteries don't actually contain that much lithium by mass. Lots of water is indeed the recommended approach to dealing with Li-ion battery fires.

The lithium in the batteries is far less of a concern than the magnesium used elsewhere in the car. If that gets hot enough to ignite, then you're going to have a very hard time putting it out (though the approach to burning magnesium in car fires (which is actually pretty common) is still water, water, and more water).

Water is the correct thing. There's not much lithium, and there's a lot of misconceptions around hybrid vehicles, for both firefighting and for extrication:

http://www.firerescue1.com/fire-products/Extrication-Tools-C...

There are YouTube videos of consulting companies throwing battery packs into fishtanks with nothing more than a fizzle.

It used to be "There are two voltage systems in a hybrid. You can cut the lines on one, but not the other". Not the case.

Really, the most dangerous aspect of a modern vehicle is HID headlights, and leaning in on a bumper where the arcing happens.

Our current system of packing litres of gasoline in a tight space can't be any safer, could it?

Those gas tanks are never very close together and are wrapped in about 100years of safety regulations. But it isn't the tank that catches fire. It normally starts in the interior (cigarettes) or with an electrical problem in the engine compartment. In that respect, electric car's aren't very different.

This happened just the other day:

http://www.abc.net.au/news/2016-01-06/100-cars-destroyed-in-...

Once that happens, it wouldn't be just tesla's that will be auto parking densely. Any other car is equally susceptible to cause fire as all cars have batteries and most have inflammable fuel. Its not yet here, but auto parking densely probably meant cars can be tightly packed as you don't have to leave enough space for you to open doors. As the car is self driving, it can go to the garage and park next to some other car. Its doable now, if those garages have some sort of beacon to notify unoccupied space and if all cars have beacons to notify/detect its edges. Its not hard to implement at all.

It's funny because dense packing is also a problem for Tesla batteries. They had to leave gaps for security.

It wouldn't be that much of a win using the feature as it currently stands. You could park them a bit closer together since you don't have to worry about getting the doors open, but that's it. A decent win, allowing maybe 20-30% denser packing, but not huge.

Packing tightly is efficient, but not always desirable. Sometimes you need a little more clearance in the rear for loading or unloading large cargo from the trunk or hatch. It's not going to work in a mall parking lot or a supermarket lot where you want to unload into the trunk from a shopping cart.

The ideal would be the car drops you off, parks itself somewhere nearby in a high density area, then returns itself for loading in an open area when you need it.

Ok we made the title say 'park'.

Only thing about all the self parking stuff is it's a bit too slow. I live in SF, and so far it's been difficult to use self parking in practice. If you take too long you get people pulling up so close to you in the back that you can't reverse, or endless honking. Especially with the Model S auto park since you actually have to pull up a lot farther than a normal person would before you can engage it

I like that idea, but that's basically how they park in many cities around the world already. It's not uncommon for drivers in for example Rome or Brussels to nudge cars with the bumper, to get an almost airtight fit when parallel parking. Underground carparks in Europe are also enormously tight, I don't think robot parking will make it that more packed. The bar is just lower here when it comes to using space efficiently. It's more of a mindset problem than a technological one.

But in those places, you still need to leave enough room for a driver to open the door and get in or out. That's a lot of wasted space.

Why even care about it being your car? How about just a car? At your bidding when you need it and no other time.

Because I got stuff I want to keep in my car. When I go shopping I sometimes hit more than one store and don't want to carry everything from Costco with me into Home Depo.

Also dogs.

In that case you'd indicate you need the same car on your return from Home Depot, and have to wait a while for it to show up from the parking lot, whilst people who don't need a specific car just grab the first one in the waiting area outside.

And no computer has ever failed to properly hold a reserved car. "See, you know how to take the reservation. You just don't know how to hold the reservation. And that's really the most important part."

https://www.youtube.com/watch?v=4T2GmGSNvaM

"When your reservation is over or you are finished with the vehicle, please place this key-like device into steering column where the traditional key would."

A manual lockout would be a smart thing here. There's all kinds of solutions to the "computers are dumb" problem.

I feel like a simpler solution would just be to store anything left in the boot when the car returns to the depot, and put it back in your next car. There could also be a safe for security and privacy. It's slightly less flexible, but for most cases it's much more efficient.

I also think it's slightly less error prone, as there's no user input required, and since the stuff gets removed from the car, it's much less likely that anything will end up being sent to another user. At worst, they forget to put your stuff back automatically, and it's relatively easy for them to send another car, or even send your stuff to your house. Indeed, that might even be a service people are interested in if they're doing a lot of shopping, or they take public transport for part of their journey.

That's like asking why do you even care about it being your home? How about just a home? When you come from work, you will pick one at random and only pay for the time you spend there.

You might look at a car as a transportation vehicle, I look at it as an extension of my private space. It's my car, so I can customize it, I can make it comfortable for me, I can keep my stuff in it, I know it's always available, I know what state it is in, etc.

Because of gun-free zones, sometimes it's necessary to secure one's firearm in one's car.

Regardless of where I live, I keep a toolkit, trauma kit, and emergency supplies in the trunk… just in case. I also will have region specific stuff:

- When I lived in the great white north, I used to keep a snow shovel, cat litter, etc. in my car during the winter months.

- While I live in an arid desert, I keep extra gallons of water and a solar blanket in the car.

I really like knowing exactly where my car (and by extension) stuff is. Also, I hate having to readjust my mirrors and seat.

High density automated parking is a solved problem.

https://en.wikipedia.org/wiki/Automated_parking_system

In a few years, you will tell your smart fridge, "buy a car". The fridge will make the appropriate selection (a Volvo if you drink Organic milk etc.), make the payment from the appropriate account (keep the Swiss account for another day). In about 10 minutes, the car that now belongs to you drives itself to your home, while you go out for a walk.

Your scenario is reminiscent of Homer Simpson's recent quote: "Siri, tell Amazon to drone me a beer." A drone carrying a beer immediately flies through the window and drops it in his hand.

I could see "tell X to Y" becoming a central UI pattern for future voice interfaces, where the system's voice agent (Siri, Cortana...) forwards arbitrary command Y to app/service X which is responsible for interpreting Y.

Not so exited about this, much more exited by the car driving itself to be serviced if possible (and back)

Or just "your car is your own Uber driver"

This should be interesting. If I were in a Walmart parking lot facing an empty car and it was trying to take a space, I would probably accelerate at it assuming it would back off, since there's no one who cares. If everyone did this, would it end with some parade of never-parking-Teslas that just drive around the parking lot until their owner gets done shopping? That's not necessarily a bad thing...but it would be a weird new aspect to parking.

Nobody seems to have considered the possibility of actually driverless cars. Would not be remotely legal, but interesting to consider them just aimlessly driving city streets or looking for parking that it will never find.

OK, say you're in Manhattan. Assuming a private car, instead of paying for a garage, should it be acceptable to just leave an automated car driving around, or instead, driving outward until it finds a public space? Would it result in cheaper garages, or some other crazy problem?

what if the cost of having my empty car wander the city is cheaper than paying for parking?

will parking costs go down? will there be some laws forbidding this behavior?

This is a very interesting hypothetical from a public policy standpoint. Perhaps the city would find a need to subsidize parking for the sake of pollution, traffic, and energy consumption concerns.

edit: maybe subsidize is the wrong word, but rather that they will incentivize affordable parking and make more available.

Aimless driving would be difficult to regulate.

Instead, if you know roughly how long you'll be gone, they could temporarily Lyft while you shop.

It seems to me that a completely empty car wandering the city would make a more enticing target for theft.

More than an empty parked car? I'd bet a car doing 80 on the highway is harder to rob than one parked in a lot.

Problematic questions would be raised about cars that redlined certain city neighborhoods.

There is illegal robbery and legal robbery... once neighborhood redlining became popular, you can guarantee cities would start raising revenue by ticketing self driving cars caught cruising residential zones, entertainment districts, etc. We have plenty of existing laws and case law against "cruising". You can imagine the bug reports already "Your service failed to inform my car that the sportsball team was having a game last night, placing the stadium district under no-cruise law, so I got ticketed, towed, impounded, and had to pay for a taxi home"

You have the car do taxi runs - lyft/uber/whatever - or deliveries until you need it again.

In Manhattan this is already happening. Not-so-rich people drive around for an hour looking for "free" or cheap street parking. Rich people pay others to sit in their cars while they are illegally parked.

Having a robot do it instead of a person would be an improvement. Better would be to make the behavior stop regardless of driverlessness.

This is irritating

It's fine to want to pay less, but in the end you might have just parked at a train/subway station far away and taken the train downtown

Faster, quicker and probably cheaper

I still have yet to see what driverless cars will do in light of the trolley problem. [1]

[1] https://en.wikipedia.org/wiki/Trolley_problem

Just search online and you'll find lots of pointless discussion about it.

Surprised to get downvoted as I was not trying to be snarky.

I've seen the pointless discussions and I probably did not word my previous comment properly to indicate I am still trying to find compelling arguments one way or another.

Also, is there a new kind of 'road rage' where one's automated car is defeated in some way by someone else's automated car? Or, if one is blocked out by superior automation? I'm a bit new to thinking about this, but I'm starting to imagine the confrontations. The bitterness will be amplified and at the same time limited by the fact that cameras are recording everything.

Yes. Unless you pay a monthly premium. Then your tesla will always win the fight. The cars of the poor will politely give up their spots.

Except my spot. Which will have a tesla-proof orange cone forever protecting it from skynet.

The car already has to be lined up with the space to park without the driver in it, and the driver still has to be nearby. Your strategy won't work.

A lot of people seem to be commenting on this without understanding how it works.

>When Autosteer is engaged on a restricted road, Model S’s speed will be limited to the speed limit of the road, plus an additional 5 mph.

This means the car is programed to break the law. Does this have liability implications or is 5mph not enough to matter?

From the notes it looks like they're just setting the upper limit on TACC to speed limit +5pmh on residental roads. You can still set TACC to exactly the speed limit without any issues.

Right, so in other words, the car allows you to break the law - just like any other car - but unlike most, it prevents you from breaking it by much.

There is going to be some friction between true self-driving cars and unrealistic speed limits in the near future though.

I wouldn't say it prevents you from doing much of anything. The car will still speed however you want it to, just not with the machine in control.

Yes, sorry, I was referring to the operation with automation engaged. As opposed to a standard cruise control, which can be set to any speed you want. Obviously in manual mode it will let you speed.

It means the car can be commanded to break the law by the driver. Which is no different from any other car on the road. It's not just spontaneously deciding the exceed the speed limit, it does so when the driver tells it to.

Interesting. I suppose the questions everyone is asking are: Can we extrapolate that liability to the 'driver' of a driverless car? By pressing the 'on' button, are you commanding the vehicle to break the law should it do so?

Just had a very low visibility landing into BWI. Feeling thankful for autoland, but also for the unimaginative FAA bureaucrats regulating it...

It's entirely possible let your pilot landed manually. All major airports have detailed instructions on how to land there even in the case of zero visibility at ground level. Basically you fly to a specific point at a specific heading, altitude, and speed. Once there you follow the instructions stating you should fly n seconds, then adjust, and repeat until you touch down on the runway.

He announced it was required to be an autopilot landing.

Pilots do not manually land without visibility. The procedure you describe would result is a lot of airliner parts suddenly occupying the runway.

I expected a bit more from the much hyped Auto-Park feature. It seems that all that the new "Summon" feature does is move up to 13m forward or backward in a straight line.

I can imagine that there are some cases where this is useful, but in general you'll still need to do that annoying 3-point turn yourself...

It does steer a bit to avoid obstacles. If you have it go forward into a garage, for example, it will center itself within as it goes.

The options for automatic parking while you're in the car are broader. It can parallel park or reverse into a perpendicular space for you in that situation.

This is actually very clever. Your car could drive itself in a sandbox before stable technology has been evolved or even legislations on self-driven cars were passed.

And of course now you could really show off with that Summon feature before all those losers.

This is what we call "clever".

I'm curious what happens when there are two dozens of these in an active parking lot (or near each other on a highway) and they get false sensor echos.

Then one day there will be half a dozen brands with slightly different sensors messing with each other via accidental interference and echos.

Public has no idea how this stuff works and starts putting a little too much faith into it can lead to some horrible accidents.

Not exactly the batmobile / car on command I was hoping for. The 'driver' is still responsible for the car's behavior. My standard for "without you" remains either no person in sight, or the driver in the back seat sleeping off a night at the bar.

If they're releasing this to the public then the truth of the matter is that the technology is "there" for driverless parking but the regulatory bits are not. "There" in this case meaning the AI is less likely to get into an accident than a human.

Have they said that? I don't see anything from tesla saying that these cars are allowed to move without the driver watching. In fact Musk has made more than a few statement denouncing the behavior of tesla owners who have let their cars loose as unsafe.

The technology is there. The car doesn't require a human watching it while it moves. The government does.

More specifically, even if the AI gets in only 50% of the accidents a human driver would get into Musk simply cannot recommend relying on it until the regulatory bodies determine he (or Tesla) is not financially responsible for those accidents. We're still a ways off from that.

Aren't there a number of cars that can already do this? I recall watching an episode of Top Gear where the car parked itself while the driver was not in it.

I only wondered what happens when it can't find parking?

The door will automatically open and a soothing, motherly voice will tell you to please get back in the car and find an appropriate spot to park.

I'm surprised there isn't any news from Tesla about this change. This is pretty cool. Yeah, technically it is driving without you but that's a bit misleading. Still really cool!

The pull into garage feature seems targeted towards people who have tiny carports like you see everywhere in SF.

Tiny? Watch this: https://m.youtube.com/watch?v=QksqWRqEfy0

Haha - wonderful :-)

More than 3 cm on each side is a waste of space :-)

I am actually happy it doesn't drive without the driver.

Not that I have a Tesla, but I would never want to own a performance vehicle that drives without me. The whole point is to enjoy the experience.

They already took away engine noise and manual transmission. Now might as well just have someone else drive you.

Then you can go to work and send back the car to your significant other to take the kids to school and then send the car back to you when you need to come back home.

I wasn't arguing that it's not practical.

One of the troubles with 150mph performance vehicles is you spend 90% of your time going slow stuck behind traffic.

Even if it did, so what? Nobody is forcing you to use these features. Don't like them? Don't activate them.

It's not unthinkable that computer-driven cars will be so much safer on the public road that it'll become illegal to have human drivers, as they are much more dangerous.

Maybe not in the immediate future, but it'll probably happen at some point.

Could be, but I see little relationship between adding features like this to a normal, driver-piloted car, and a mysterious future where autonomous cars are the only ones allowed on the road.

If you've never driven a Tesla then you really have no idea what you're talking about.

You know how you get the best 0-60 time on the P90DL? You take your foot and kick the accelerator. Anything less and you lose 1-2 tenths of a second because the car is already moving by the time your foot is halfway down.

Take a test-drive, I guarantee you'll be singing a different tune afterwards.

Telsas don't have launch control?

They do. As of last month. It helps a bit, but it's almost entirely unnecessary.

From the way the procedure works, it sounds like it's just a way to shave off that delay from getting the pedal to the floor.

Launch control in a more typical car gets the engine to the best RPM and sets everything up to transfer that energy to the wheels in the best way possible without breaking traction. In a electric car with no clutch or transmission where it's just a matter of sending the proper massive amounts of current to the motors on command, there isn't much else needed.

From a standing start, an electric often needs to limit the power to the motors else they start snapping parts. That certainly happened in the tesla from day one. Then once the wheels are turning, launch control manages the traction control settings to maintain maximum acceleration, perhaps allowing for more/less slip than would be otherwise. It isn't as simple as just powering the motors.

Power limits apply to all driving, and that's why I said "proper" rather than "maximum." For slip, there's a separate "slip start" setting that gives better acceleration. There really isn't much else for launch control to do that you don't get already.

I would love to do a test drive.

However, the fun part with launching is not stomping on the accelerator or going fast, but matching the RPMs perfectly at the exact time of the race start. Then letting go of the clutch smoothly but rapidly. Followed by heel-toe downshifting when turning. When you do all these things perfectly, there is this exhilarating rush. Like you are one with the machine.

Reply on news.ycombinator.com