I'm not sure what the story is here. I'm as skeptical of Waymo as anyone but a lot of people basically don't take Uber/Lyft/taxis on anything like a regular basis.
I realize a fair number of readers here probably live in cities and maybe don't have cars. But, in my case, I live outside a major city and, while in normal times, I do take reserved cars back and forth to the airport, I haven't taken a taxi/Uber/Lyft around where I live for many years. I do take them maybe a dozen times a year but that's when I'm traveling.
So the fact that there's relatively little use of Waymo in the Phoenix suburban sprawl where it operates doesn't seem especially noteworthy.
I'm pretty sure we'll see this on certain highways in certain weather conditions--which would actually be a killer feature for a lot of us. But it doesn't fundamentally change car ownership and use--and may or may not change trucking.
Yeah, the actual driving part is just part of the job. You would need personnel at every stop along the route to handle refueling or changing trailers. As well security concerns to an unmanned vehicle laden with goods. I mean an automated cruise control would allow night driving and such some savings there.
Eh, imagine two drivers for eight trucks... Takes care of the security problem, so long as the US doesn't devolve into road banditry, and the humans can handle refueling, and the trucks can drive without the usual restrictions on human fatigue.
Not owning a car seems hard for anyone with a life more complicated than single office drone. I often need to stash kids equipment in the car for later pickup; we keep medicine and books in the car used often, strollers, umbrellas, not to mention car seats. That’s a lot of stuff to lug between Uber to Uber. People who play sports after work, sales folks traveling with merchandise, etc. cars are a home away from home in our lifestyles.
IMO Society will adapt and people will find different solutions for these things. Take for instance two scenarios:
* London - I used to live there and commute on the tube. Not everyone uses a car on a regular basis, except maybe to go to the supermarket. I didn't find it particularly hard to bring my sportswear in a bag and put a kindle in my jacket pocket. I assume New York is the same!
* Amsterdam and The Hague - I used to travel to these cities often, and these societies exist almost entirely on bike.
I have a car, but also manage to walk into work and not bring everything with me everywhere. It just seems like the assumption is "it will only work if it can support me doing exactly what I do now", but this sort of technology has the potential to completely change the way people work and live.
Also cars equipped with boat racks, bike racks—often fairly specialized for specific equipment. Loading up cars the night before for a trip. Could definitely make a difference at the margins in cities but you already have Uber.
This might cut price in half? A lot of people seem to assume that this makes taxi rides almost too cheap to meter relative to owning a car or taking an Uber. And I'm pretty sure that's not true. You don't pay yourself to drive and the US IRS deduction for mileage is something like 50 cents per mile. That's probably a pretty good floor for a robo-taxi including the mileage it takes to get to you.
The combination of ride-share, Zipcar, and traditional rentals (plus scattered improvements in public transit and bike infrastructure in some places) has already done that to some degree. It's not about the person who needs to get in a car every time they go somewhere from their house. It's about replacing the car that maybe gets driven once or twice a week.
Lots of middle class folks from across the world have families and don't have cars. There are ways of life other than living in a SFH on a quarter acre an hour commute from work.
It is quite possible and even more convenient to live in a place with walkable areas and good public transit.
TBH I think there are all solutions to that when you have a big enough fleet. Rent a cargo van, rent a car for the day for storage, more storage options at places, rent a car with car seats for several years and sport equipment racks. When family is over, automatically scale with 3 or 4 cars, etc.
You could make specialized cars that would be way more convenient than the typical suv, but no person would keep as their permanent car configuration. I could see vans with roll in bike racks for example, which is more convenient than latching it on top of the typical suv sports rack for example. Groceries and other retail shopping I could see you just drop in the little cart drone and it gets to your house before you do, already in the pantry, etc. Grocery drones will be well insulated coolers, which is better than your typical suburban shopping trip freshness wise. No more melted ice cream anxiety when you remember something.
All these problems apply to anyone without a car anywhere. People in major cities with good transit systems figure it out, I don't see how this is any different. In fact, it's certainly preferable to a train/bike in terms of space to transport things.
"Once it works properly, you won't need to spend anything at all on marketing... people will flock to it in droves."
I agree.
However this really remains to be seen. If Waymo got everything working as right as they could it might still fail. This is a culture shift you're talking about, and even when past tech has required less to shift some has failed.
Trucking and highway driving are so much easier. And will probably be solved far sooner by other people, and Waymo will have missed an obvious stepping stone, business scaler, revenue stream.
Personally I'm not convinced that Waymo isn't continuing to flush a whole lot of money down the toilet. But this is a test/prototype in an area where I assume most households own cars and rarely take taxis and ride shares around the local area. So the fact that not a lot of people are taking a Waymo except maybe for the novelty is utterly unsurprising.
But their Arizona test is just that, a test. They are (or at least they state that they're) polishing the experience of using the service. What better place to do that than in a place where demand isn't going to be especially high, with ideal weather and lots of well-kept roads? Surely nobody thinks the Arizona service is Waymo's best stab at a productionalized offering.
They opened up a pork shop in a muslim area and we're not supposed to question that? These are the people supposed to make cars autonomous and they are utterly ignorant to all these things? Is it designed to fail or are they stuck in the suburban sprawl because that's the only place where the thing will not practically fail and kill someone?
these suburban sprawl places were also very gung ho about rolling out the regulatory equivalent of a red carpet. I'm not aware of any major city that has nearly as permissive autonomous regulations in their central area, probably because of concerns about reliability.
Surely, though, the cost of operation has little to do with the cost of rides? Google has poured billions into the project. A few dollars per ride isn't helping their accountants sleep any better.
Wut? Phoenix metro area is not the bay or Seattle or NY. Everyone out there has cars. It’s hot as hell and things are spread out. Spend any time there: there are only cars. Anyone on a bus is down on their luck. Same with bicyclists, either that or they are bike maniacs biking for political or health reasons. tldr anyone not using a car is already not in Waymos target audience because they can’t afford a taxi or are just really determined to ride a bike.
Downtown Phoenix, or ASU (students), or downtown Scottsdale (drunk night life) would have been the better test locations.
Senior citizens who aren't comfortable driving anymore? When other family members have taken all the cars but you still need to get somewhere? When your car's being repaired?
I think you’ve answered your own question. They don’t want drunk nightlife for these early tests. They want as predictable and car centric as possible.
They presumably mostly care about running a proof of concept at limited scale in as safe and boring an environment as they can without making it totally artificial. They also presumably don't really care about it being a generally useful taxi service as long as they can get some riders, which they seem to be doing.
It doesn’t service the major bar area of downtown Chandler, much less Old Town Scottsdale, Tempe
or Downtown Phoenix. Although it does service Chandler Fashion Center (one of the main malls in the south east valley), you can’t use it to send your 15 year old to the mall/theater, as you must be 18 to use the service alone. Intel’s Ocotillo campus appears to be in their service area, so if an Intel employee lived in the 40 square mile area, they might be able to commute with it. It doesn’t service either of the airports. The trial area effectively requires a vehicle for any practical lifestyle, and any of the use cases that I’d normally take a Lyft/Uber for, just aren’t serviced by Waymo.
The question isn't whether it's used much but whether it's at full capacity.
Waymo might as well use it in the suburbs where there are less corner case situations if it's at full capacity. The article suggests it is not used at full capacity, but isn't clear.
> you can’t use it to send your 15 year old to the mall/theater, as you must be 18 to use the service alone.
They could honestly relax that constrain.
> The trial area effectively requires a vehicle for any practical lifestyle, and any of the use cases that I’d normally take a Lyft/Uber for, just aren’t serviced by Waymo.
The trial area seems constrained because in case something goes wrong they still have to dispatch a human. That doesn't really scale.
I don't know what legal constraints/tradeoffs they're operating under but I was flying in planes by myself (as an adult) by the time I was 15. So it doesn't seem unreasonable.
Are you suggesting that stepping in an autonomous vehicle at home to be delivered to a museum and returning the same way is somehow more challenging than flying to a different state, taking public transportation, going to work at a McDonalds, or walking around a city--all of which are things that teens do? I assume these autonomous vehicles are monitored by humans at, at least, some level.
If I were waymo I'd keep that as long as i can because the downside of "child killed by freak accident in waymo car" probably is probably something I'd like to avoid.
It doesn’t service the major bar area of downtown Chandler, much less Old Town Scottsdale, Tempe or Downtown Phoenix
I live in Arizona and was going to say exactly this: Waymo doesn't go to many useful places right now. If it even expanded to cover Tempe (where ASU is located) and a good chunk of Scottsdale, it would be a lot more useful. Anyone on the margin of keeping a car or selling it, or not buying it in the first place, is unlikely to be heavily persuaded by Waymo availability because the area covered is so small.
> Anyone on the margin of keeping a car or selling it, or not buying it in the first place, is unlikely to be heavily persuaded by Waymo
I visit Arizona frequently and I don’t drive. Waymo’s current service area is useless. It’s in an area with a uniformly rectilinear road layout, low population density and lower incomes than its surrounding. It was picked for research ease. Not product-market fit.
It seems really unlikely that waymo wasn't aware of all these limitations when they selected the area they were going to trial the service in, and more like they selected the trial area to purposefully minimize demand.
Decided I’d try it out today. Drove into the service area, parked, and summoned a Waymo for a trip to the Chandler Mall. A short 10 minute wait until the vehicle arrived, fully autonomous. For the 4.4 mile drive I could have completed in 11 minutes, I was charged $11.30. Immediately, the Waymo failed to properly merge and replanned its route through a quiet neighborhood. It avoided the fastest route, the freeway, instead opting for surface streets. It ultimately arrived after 5.4 miles and 16 minutes. Overall, the braking at stoplights was better than I do, but it drives extremely cautiously, staying behind a bus for a good quarter mile, where a human driver would have passed.
On the return trip, I compared pricing to Lyft and Uber. $9.75 for Waymo, while Uber attempted to extort me for $18.25, and Lyft offered a ride for $8.99 with a 3min wait, or $7.99 with a 10min wait. Again, Waymo avoided the freeway, snaking through residential areas instead.
If my Tesla Full-self driving (tried the subscription - I’d recommend against wasting money on even a single month of the subscription at this point) drove like this Waymo, and I could avoid liability for being intoxicated, I’d gladly fork over the $10k Musk is very proudly asking for his offering. I’m skeptical Tesla’s going to achieve Waymo levels in the next 5 years, or ever, without an improvement to the sensors.
The Waymo experience was much better than I expected, however, the car only fits 3 adults! There’s a child seat occupying one full seat, plus the middle in the rear seat. Both front seats were unavailable.
If I were offered an electric car with Waymo for $25k more than a Tesla, I’d gladly pay. It was impressive.
This is the 1st Waymo user review I've read. Thx for the post. Did you notice if the company has install road signs/hardware that help the cars complete their trips? I would think that would be possible since they have a small service area.
Changes to road hardware for autonomous cars have been happening on civil engineering timescales. For example, Caltrans commented in 2017 that it was increasing lane line thickness from 4" to 6" and phasing out the use of Botts' Dots. (Source: https://www.bizjournals.com/sanfrancisco/news/2017/07/18/cal...)
$0.50/mile is likely the low end of what’s possible for the base vehicle costs. $0.50/mile + autonomous hardware costs/expected vehicle lifetime would be the total costs. 200k miles is generous for a Pacifica. I’ve read the autonomous hardware is $250k (could the hardware outlast a vehicle and be transferred?)
$0.50 + $250k/200k mi = $1.75/mi
$1.75/miles * 5.4 miles actually driven = $9.45, very close to what I paid.
If Google can get the autonomous hardware package down to $50k, then this trip would have been $4.05. At $5k (the absolute lowest possible?), the trip would cost $2.85. At that point, this becomes exciting.
That would be much cheaper than a city bus, if you account for heavy subsidies the public buses get — the ticket revenues typically cover less than a half of operational cost of running buses, and if you include capex, it gets even worse.
I've been impressed with Waymo from watching JJRicks videos. In this video the Waymo vehicle gets cut-off by a driver changing lanes in front of them, and it handled it correctly.
The white 4runner? My read of that driver’s behavior is that lane change was predictable. The 4runner gives up a huge gap to let a space open up, generally indicative of a lane change, but the van wants to keep a constant distance from the car ahead of it. And the van seems to never leave the right-most lane, so I’d expect these events happen pretty often.
The nice thing about a "false consensus", of course, is that someone who believes in it can tell others they are victims of "pluralistic ignorance" and once enough people are brainwashed, it's a true consensus!
>> Waymo has also partnered with organizations—most recently, the Southern Christian Leadership Conference
>This is odd to me. I don’t see the connection. What am I missing?
The SCLC is a civil rights organization that advocates for social justice. Waymo partnering with them is a signal they want to make sure the Waymo cars don't avoid poor parts of town, are too cost prohibitive, etc.
The SCLC also has a strong "get voters to the polls" effort. I can see Waymo volunteering rides to people's polling stations.
Hah, the moment those Waymo cars go into poor areas they're going to get ransacked. I used to live in a poor area and got daily notifications of catalytic converters being stolen. There's tens of thousands of dollars of tech on that car and you can stop it by just standing in front of it. Cameras haven't exactly stopped thieves from doing what they do in the past.
> I used to live in a poor area and got daily notifications of catalytic converters being stolen
I mean, the catalytic converter is going to be like 800° and there's not exactly a big integrated lidar aftermarket yet, so maybe not a great imagined reason to preemptively redline anyone.
But there is a large market for server CPUs and presumably FPGAs on eBay. I doubt something like the CPU in a Waymo isn't a commodity off-the-shelf SKU, the brunt of the work is done by TPUs anyways.
Ultimately if robocars work, the little anecdotes you hear about people trusting them after the first ride seems to me to confirm we shouldn't expect human factors to kill them.
Singapore (a city I love, an awesome place) started a fairly large test of self driving taxis in 2016. I just looked for recent news about this service but couldn't find any. Anyone know if this is still a thing?
Another issue is I just never think about Waymo. I was without a car for maybe a couple of months? Didn't think about using Waymo once. Just grabbed an Uber.
Very true. Waymos are held in considerable contempt in Chandler. Local HATE them (we have a plant in Chandler).
But more broadly, the "learning territory" is bizarrely and truly inapplicable to 99% of the rest of roadways in the USA.
Chandler roads are:
* Unusually flat
* Unusually broad
* Unusually straight
* Unusually well-lit at high sun angles (no shadows)
* Unusually good weather most of the time (ignoring Haboob sand storms and monsoon downpours that happen for a few weeks in the summer)
* Unusually light traffic (compared to any urban center)
Honestly they are a MINIMUM of 20-50 year away from "any road in the USA" and that still presumes radical and unknown improvements in the technology arriving in a timely fashion!
I'm not holding my breath on this technology! I'm OK if someone wants to blow through some serious cash - that's their choice. But it's not going to pay off any time soon. Fools may believe otherwise but fools and their money are quickly parted and that's on the fools.
It's a PR stunt and an attempt to continue to show "progress" so they can keep getting funding. Their current design requires up-to-date ultra high definition 3d mapping, perfect weather conditions, and a minimal amount of traffic weirdness. Maybe 5g gives them a way to keep up to date 3d maps onboard, but even if that works out the data costs and cost to continually re-map the streets makes it so the unit economics don't work. It's not like they just need a few more miles of "learning", the foundation of their entire strategy is flawed and can't scale.
HD Maps - really aren’t as difficult as you’re making them out to be. Most AV companies can update their maps using sensor data from their own fleet. And the cars need to recharge/service at a facility every few hours anyway. So there certainly isn’t any need for 5g or expensive cell data to make HD maps scale.
That's encouraging. It's hard, and there's a market. If and when they can eliminate the "safety driver", it's real.
DMV has three levels of license for self-driving vehicles.
- "Testing with a driver". This has about the same restrictions as a learner's permit - must have a licensed driver onboard, no vehicles over 10,000 lbs, no carrying passengers for hire. Over 50 companies have that license. Uber flunked this and had their license to drive autonomously revoked in 2016. So Uber went to Arizona and killed a pedestrian. Then Arizona suspended their license and Uber stopped.
- "Testing without a driver". This is tougher to get. There are specific technical requirements. There are communications requirements. A "law enforcement interaction plan" ("How do we stop this thing?") SAE Level 4 or 5 is required; 3 is not good enough. Operation may be restricted to certain types of roads (freeway, campus) and limited in speed. Eight companies have this license now, which is encouraging.
- "Deployment". Only one company has qualified: Nuro, which makes slow-speed shuttle buses. This is the Commercial Drivers License level for self-driving vehicles.
DMV has a staged approach. It seems to work. Companies can test, and failures get reported and publicized. Complaints about the California DMV being too restrictive about testing have subsided. Most of the companies that couldn't meet the requirements are gone.
There's slow, steady progress. Alphabet/Waymo and GM/Cruise are getting close to deployable systems.
I live and work in and around the Waymo service area. Personally I have never heard anyone express annoyance with the vehicles, so I'm not sure where it's coming from. Driving I've never noticed them to affect me in any way.
Only one example I remember. I was going down a residential street to a midblock through street, usually I wouldn't expect any other cars. I pull up and find my self 6 cars back behind a Waymo to turn. And waited. The Waymo would just not turn whereas all the humans would have. The street has relatively constant traffic.
The only thing I can think of is that people are annoyed by them going the speed limit. Some streets in Chandler are 45 and I swear everyone is trying to go 60, so I suppose people that people that want to be going well over all the time get mad if they are stuck behind a Waymo. In Tempe the traffic tends to seem slower and more controlled so possibly they aren't as bothered by it.
> Local HATE them... truly inapplicable to 99% of the rest of roadways in the USA.
Have you considered they might be training the vehicles on you, not the roads? At some point roads become a solved problem and understanding/anticipating other drivers becomes thr hard problem.
Those conditions are probably why the area was chosen: because it's a good place to start safely, collect data. They are working out the tech probably less so the business model.
It's likely they can't deploy in some areas because snow/rain are just too hard to navigate safely at this point.
The trillion dollar question of course is are they 'almost there' or is this an 90/10 thing where the last 10 is really the 90.
What is the benefit of self driving taxis versus human taxi drivers with safety assist (collision mitigation braking, and forward-collision warning, etc.)?
Is the benefit of self-driving taxis purely cost reduction?
You can pay less and have an artificial intelligence drive you, or pay more and have a human (non-artificial intelligence) drive you?
I do think the initial question if phrased specifically around safety is an interesting one. This may be the optimist nerd in me but I believe that computers will eventually (if not already) be safer than humans when driving. I think it's too easy for us to become distracted and make mistakes, while a self driving car has 360 vision and can see through opaque objects. If one were to argue that a safety assist system should be capable of correcting all mistakes then I think the car is just basically self-driving at that point.
Apart from costs, I think removing the need for attention on the road and other drivers is a benefit for self driving cars, you can spend time engaging loved ones or whatever pleases you.
There's a limited number of drivers and they are competing for the most profitable trips at busy times. A self driving fleet can cover a larger area 24/7 because they don't mind parking for hours until the next fare.
I really don't understand the pervasive idea that we are going to hold robot cars to so much of a higher standard than human drivers. I'll take 1.2x . Human drivers kill a lot of people every year in what are largely preventable accidents. Lets use robot cars in places that it makes sense to do so and save some lives and they'll get better over time.
I think the issue is the kind of accidents will be very different from humans, uncomfortable and very fatal. You also have to compare against sober and awake drivers, not 'the typical driver', because you have a choice to be sober and awake if you're more safety minded, while you don't really have those choices with AI driving.
Like tesla cars going straight into concrete barriers on highway exits instead of down the 2 lane options on autopilot mode.
It's usually the optics/media that will torpedo the effort. "Big scary robot car was in an accident".
Proof? 13 American soldiers died in Afghanistan in a combat zone and people are up in arms. 1800 Americans died yesterday from covid while going about their daily lives and it's pretty mum.
Self-driving cars are nowhere as safe as humans per mile. People touting self-driving cars like to talk about human accidents but people are remarkably safe per mile. Self-driving cars don't have that many accidents attributed to them but they haven't driven very much compared to humans.
Wikipedia gives a summary of the accident rate in the US [1], humans drive millions of miles without accidents. Filip Piekniewski has an ongoing series on self-driving car problems. [2] (I'd look up the earlier articles in the series if I had time). This article I googled [3] give some statistics: ") At the moment, self-driving cars have a higher rate of accidents compared to human-driven cars, but the injuries are less serious. On average, there are 9.1 self-driving car accidents per million miles driven, while the same rate is 4.1 crashes per million miles for regular vehicles." However, nearly all self-driving cars are driven in highly controlled conditions, the safest driving areas available and/or an operator ready to take over - yet they still have a higher rate of accidents than human drivers.
There's a reason self-driving cars aren't being deployed yet.
Self-driving aint happening until we get AGI drivers.
Reading these comments trivialising the driving task makes me wonder if such commenters have ever actually driven or if they are just prone to magical thinking.
As of 2021 computers can not reliably and flexibly out reason people about open ended real world problems. Safe driving in the real world requires such reasoning. It is edge cases all the way down. Ergo self driving is not happening anytime soon.
When I see HN commenters say the robocar sees or thinks I wonder if these people are for real.
It would probably be easier to make a robo physician who gives more reliable diagnoses than a human physician than it would be to make a self driving car that out performs humans.
Reading these comments trivialising the driving task makes me wonder if such commenters have ever actually driven or if they are just prone to magical thinking.
I would say it's a testament to the human ability to reduce very complex tasks to very simple descriptions that can be kept out of rational awareness. The impressive thing is that even the wrong beliefs of a given person work for them. Remember, the people effectively arguing that driving is "easy" still understand that humans fail at it regularly. So it's mismatch in their understanding of how smart humans are, something that probably doesn't impact their daily lives.
I think waymo's best customers in suburban america would be the old and young people who cannot drive themselves and cannot walk to their destinations. Waymo to shuttle your 10 year old unsupervised to school and back, your 70 year old grandma who a cannot drive anymore and disabled people who cannot drive would be amazing in those communities, but in both cases, they are either not allowed, or are too old to use a smartphone effectively too. I wonder if waymo has human phone dispatchers, which is something I could see the old people being more into, if it would result in way more usage by that segment.
"FHP Orlando
@FHPOrlando
·
10h
Happening now: Orange County. Trooper stopped to help a disabled motorist on I-4. When Tesla driving on “auto” mode struck the patrol car. Trooper was outside of car and extremely lucky to have not been struck. #moveover. WB lanes of I-4 remain block as scene is being cleared."
I've got a question on Waymo. Say you're an old person who takes quite a bit of time to exit the vehicle. Is there a button outside to indicate you're clear of the vehicle? Or are you 3/4 out when it gets another ride, slams the van door on you and takes off at 45 mph with you hanging out.
I have seen many descriptions of the tech, some of it at a pretty high level on how it navigates but never anything how it ensures that the passenger has had a successful exit.
I don't have any 1st person experience with this (or work at Waymo) but there are probably several easy ways they avoid this.
The car will probably never drive with the door open and all modern cars have sensors to detect this.
Second, their cars (and any competent self-driving car) needs to have a very accurate awareness of its surroundings. True it might be difficult to get a clear view of a person just barely outside the car but I'd bet that they have cameras/radar/lidar positioned to see if at least something is there.
Lastly, they have internal passenger cameras that can detect whether a person is inside the car.
They have specifically strengthened the motors in the sliding door to cut the average adult person in half if they take longer than 30 seconds to exit. (lol I like your example scenario)
> “I am very aware of [Waymos] when I'm running on the road or riding my bike,” Stacy, the Chandler resident, told us. “When I was on my bike, I had one creep up on me. I chose to stop and wait for it to pass because it would not go around me as I was moving.”
That's something underdiscussed. A lot of driving is predicting other driver's behaviours. After a while you get a sense of what people will do, and in which broad category they fall under.
With self-driving cars, now we have new actors on the road that don't act naturally. And each software update might change these behaviours. This has some interesting feedback implications. As we get more of those on the streets, we might see more accidents after software upgrades. Not because of the software, but because people got used to the old behaviour and didn't anticipate the new one.
117 comments
[ 5.2 ms ] story [ 192 ms ] threadI realize a fair number of readers here probably live in cities and maybe don't have cars. But, in my case, I live outside a major city and, while in normal times, I do take reserved cars back and forth to the airport, I haven't taken a taxi/Uber/Lyft around where I live for many years. I do take them maybe a dozen times a year but that's when I'm traveling.
So the fact that there's relatively little use of Waymo in the Phoenix suburban sprawl where it operates doesn't seem especially noteworthy.
Trucking will be up-ended. Freight will change. People will in a short period of time stop owning cars and call them as needed.
I'm just not sure we'll ever get there.
* London - I used to live there and commute on the tube. Not everyone uses a car on a regular basis, except maybe to go to the supermarket. I didn't find it particularly hard to bring my sportswear in a bag and put a kindle in my jacket pocket. I assume New York is the same!
* Amsterdam and The Hague - I used to travel to these cities often, and these societies exist almost entirely on bike.
I have a car, but also manage to walk into work and not bring everything with me everywhere. It just seems like the assumption is "it will only work if it can support me doing exactly what I do now", but this sort of technology has the potential to completely change the way people work and live.
This might cut price in half? A lot of people seem to assume that this makes taxi rides almost too cheap to meter relative to owning a car or taking an Uber. And I'm pretty sure that's not true. You don't pay yourself to drive and the US IRS deduction for mileage is something like 50 cents per mile. That's probably a pretty good floor for a robo-taxi including the mileage it takes to get to you.
It is quite possible and even more convenient to live in a place with walkable areas and good public transit.
You could make specialized cars that would be way more convenient than the typical suv, but no person would keep as their permanent car configuration. I could see vans with roll in bike racks for example, which is more convenient than latching it on top of the typical suv sports rack for example. Groceries and other retail shopping I could see you just drop in the little cart drone and it gets to your house before you do, already in the pantry, etc. Grocery drones will be well insulated coolers, which is better than your typical suburban shopping trip freshness wise. No more melted ice cream anxiety when you remember something.
I agree.
However this really remains to be seen. If Waymo got everything working as right as they could it might still fail. This is a culture shift you're talking about, and even when past tech has required less to shift some has failed.
then the question becomes chicken and egg.
[1] https://twitter.com/Ryandoofy/status/1410253369420226560
Downtown Phoenix, or ASU (students), or downtown Scottsdale (drunk night life) would have been the better test locations.
Chandler? Wow
Senior citizens who aren't comfortable driving anymore? When other family members have taken all the cars but you still need to get somewhere? When your car's being repaired?
I think the real answer here is they clearly don't really want people to use it yet. They just want to say they are the first robotaxi.
Waymo might as well use it in the suburbs where there are less corner case situations if it's at full capacity. The article suggests it is not used at full capacity, but isn't clear.
They could honestly relax that constrain.
> The trial area effectively requires a vehicle for any practical lifestyle, and any of the use cases that I’d normally take a Lyft/Uber for, just aren’t serviced by Waymo.
The trial area seems constrained because in case something goes wrong they still have to dispatch a human. That doesn't really scale.
I don't know what legal constraints/tradeoffs they're operating under but I was flying in planes by myself (as an adult) by the time I was 15. So it doesn't seem unreasonable.
If I were waymo I'd keep that as long as i can because the downside of "child killed by freak accident in waymo car" probably is probably something I'd like to avoid.
I live in Arizona and was going to say exactly this: Waymo doesn't go to many useful places right now. If it even expanded to cover Tempe (where ASU is located) and a good chunk of Scottsdale, it would be a lot more useful. Anyone on the margin of keeping a car or selling it, or not buying it in the first place, is unlikely to be heavily persuaded by Waymo availability because the area covered is so small.
I visit Arizona frequently and I don’t drive. Waymo’s current service area is useless. It’s in an area with a uniformly rectilinear road layout, low population density and lower incomes than its surrounding. It was picked for research ease. Not product-market fit.
On the return trip, I compared pricing to Lyft and Uber. $9.75 for Waymo, while Uber attempted to extort me for $18.25, and Lyft offered a ride for $8.99 with a 3min wait, or $7.99 with a 10min wait. Again, Waymo avoided the freeway, snaking through residential areas instead.
If my Tesla Full-self driving (tried the subscription - I’d recommend against wasting money on even a single month of the subscription at this point) drove like this Waymo, and I could avoid liability for being intoxicated, I’d gladly fork over the $10k Musk is very proudly asking for his offering. I’m skeptical Tesla’s going to achieve Waymo levels in the next 5 years, or ever, without an improvement to the sensors.
The Waymo experience was much better than I expected, however, the car only fits 3 adults! There’s a child seat occupying one full seat, plus the middle in the rear seat. Both front seats were unavailable.
If I were offered an electric car with Waymo for $25k more than a Tesla, I’d gladly pay. It was impressive.
Changes to road hardware for autonomous cars have been happening on civil engineering timescales. For example, Caltrans commented in 2017 that it was increasing lane line thickness from 4" to 6" and phasing out the use of Botts' Dots. (Source: https://www.bizjournals.com/sanfrancisco/news/2017/07/18/cal...)
I would have hoped a company called "Caltrans" to provide 100 train routes and 1000 bus routes around California.
Shit, that's expensive. I would have imagined autonomous cars to be 1/4 the price of a human taxi or less.
$0.50 + $250k/200k mi = $1.75/mi
$1.75/miles * 5.4 miles actually driven = $9.45, very close to what I paid.
If Google can get the autonomous hardware package down to $50k, then this trip would have been $4.05. At $5k (the absolute lowest possible?), the trip would cost $2.85. At that point, this becomes exciting.
This is interesting though...
"Waymo says it provides hundreds of rides a week"
Which almost certainly means less than 100 rides per day. Typical Uber drivers do 4-6 trips in a 4 hour shift.
https://www.youtube.com/watch?v=LO3DxyTxdUU
I don't see why more people aren't taking advantage of it.
This commenter covers it: https://news.ycombinator.com/item?id=28338580
It's when you have a reason for doing or not doing something, and you assume you're the only one, but really, everybody else has the same reason.
As a result, everybody is passive and doesn't act like they would if they knew everyone agreed.
It's the opposite of https://en.wikipedia.org/wiki/False_consensus_effect
Ah, this is it: https://en.wikipedia.org/wiki/Pluralistic_ignorance
The nice thing about a "false consensus", of course, is that someone who believes in it can tell others they are victims of "pluralistic ignorance" and once enough people are brainwashed, it's a true consensus!
> Waymo has also partnered with organizations—most recently, the Southern Christian Leadership Conference
This is odd to me. I don’t see the connection. What am I missing?
> Waymo walked away from the term “self-driving” this January in favor of a more precise, multisyllabic “fully autonomous driving tech.”
I assume this is because Tesla marketing had polluted the term?
>This is odd to me. I don’t see the connection. What am I missing?
The SCLC is a civil rights organization that advocates for social justice. Waymo partnering with them is a signal they want to make sure the Waymo cars don't avoid poor parts of town, are too cost prohibitive, etc.
The SCLC also has a strong "get voters to the polls" effort. I can see Waymo volunteering rides to people's polling stations.
I mean, the catalytic converter is going to be like 800° and there's not exactly a big integrated lidar aftermarket yet, so maybe not a great imagined reason to preemptively redline anyone.
But more broadly, the "learning territory" is bizarrely and truly inapplicable to 99% of the rest of roadways in the USA.
Chandler roads are:
* Unusually flat * Unusually broad * Unusually straight * Unusually well-lit at high sun angles (no shadows) * Unusually good weather most of the time (ignoring Haboob sand storms and monsoon downpours that happen for a few weeks in the summer) * Unusually light traffic (compared to any urban center)
Honestly they are a MINIMUM of 20-50 year away from "any road in the USA" and that still presumes radical and unknown improvements in the technology arriving in a timely fashion!
I'm not holding my breath on this technology! I'm OK if someone wants to blow through some serious cash - that's their choice. But it's not going to pay off any time soon. Fools may believe otherwise but fools and their money are quickly parted and that's on the fools.
https://arstechnica.com/cars/2021/05/real-robotaxi-service-g...
HD Maps - really aren’t as difficult as you’re making them out to be. Most AV companies can update their maps using sensor data from their own fleet. And the cars need to recharge/service at a facility every few hours anyway. So there certainly isn’t any need for 5g or expensive cell data to make HD maps scale.
DMV has three levels of license for self-driving vehicles.
- "Testing with a driver". This has about the same restrictions as a learner's permit - must have a licensed driver onboard, no vehicles over 10,000 lbs, no carrying passengers for hire. Over 50 companies have that license. Uber flunked this and had their license to drive autonomously revoked in 2016. So Uber went to Arizona and killed a pedestrian. Then Arizona suspended their license and Uber stopped.
- "Testing without a driver". This is tougher to get. There are specific technical requirements. There are communications requirements. A "law enforcement interaction plan" ("How do we stop this thing?") SAE Level 4 or 5 is required; 3 is not good enough. Operation may be restricted to certain types of roads (freeway, campus) and limited in speed. Eight companies have this license now, which is encouraging.
- "Deployment". Only one company has qualified: Nuro, which makes slow-speed shuttle buses. This is the Commercial Drivers License level for self-driving vehicles.
DMV has a staged approach. It seems to work. Companies can test, and failures get reported and publicized. Complaints about the California DMV being too restrictive about testing have subsided. Most of the companies that couldn't meet the requirements are gone.
There's slow, steady progress. Alphabet/Waymo and GM/Cruise are getting close to deployable systems.
Only one example I remember. I was going down a residential street to a midblock through street, usually I wouldn't expect any other cars. I pull up and find my self 6 cars back behind a Waymo to turn. And waited. The Waymo would just not turn whereas all the humans would have. The street has relatively constant traffic.
The only thing I can think of is that people are annoyed by them going the speed limit. Some streets in Chandler are 45 and I swear everyone is trying to go 60, so I suppose people that people that want to be going well over all the time get mad if they are stuck behind a Waymo. In Tempe the traffic tends to seem slower and more controlled so possibly they aren't as bothered by it.
Have you considered they might be training the vehicles on you, not the roads? At some point roads become a solved problem and understanding/anticipating other drivers becomes thr hard problem.
It's likely they can't deploy in some areas because snow/rain are just too hard to navigate safely at this point.
The trillion dollar question of course is are they 'almost there' or is this an 90/10 thing where the last 10 is really the 90.
It’s basically Uber, but the driver is your grandma that is fine waiting several minutes to make a left turn “just to be safe.”
Is the benefit of self-driving taxis purely cost reduction?
You can pay less and have an artificial intelligence drive you, or pay more and have a human (non-artificial intelligence) drive you?
How much cheaper will the self-driving taxis be?
If you can't answer my question, maybe the Emperor isn't wearing any clothes.
Apart from costs, I think removing the need for attention on the road and other drivers is a benefit for self driving cars, you can spend time engaging loved ones or whatever pleases you.
Like tesla cars going straight into concrete barriers on highway exits instead of down the 2 lane options on autopilot mode.
Proof? 13 American soldiers died in Afghanistan in a combat zone and people are up in arms. 1800 Americans died yesterday from covid while going about their daily lives and it's pretty mum.
There's a reason self-driving cars aren't being deployed yet.
[1] https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in... [2] https://blog.piekniewski.info/2021/05/12/ai-mid-2021/ [3] https://carsurance.net/blog/self-driving-car-statistics/
The real world.
See also: https://blog.piekniewski.info/2021/05/12/ai-mid-2021/
Self-driving aint happening until we get AGI drivers.
Reading these comments trivialising the driving task makes me wonder if such commenters have ever actually driven or if they are just prone to magical thinking.
As of 2021 computers can not reliably and flexibly out reason people about open ended real world problems. Safe driving in the real world requires such reasoning. It is edge cases all the way down. Ergo self driving is not happening anytime soon.
When I see HN commenters say the robocar sees or thinks I wonder if these people are for real.
It would probably be easier to make a robo physician who gives more reliable diagnoses than a human physician than it would be to make a self driving car that out performs humans.
I would say it's a testament to the human ability to reduce very complex tasks to very simple descriptions that can be kept out of rational awareness. The impressive thing is that even the wrong beliefs of a given person work for them. Remember, the people effectively arguing that driving is "easy" still understand that humans fail at it regularly. So it's mismatch in their understanding of how smart humans are, something that probably doesn't impact their daily lives.
"FHP Orlando @FHPOrlando · 10h Happening now: Orange County. Trooper stopped to help a disabled motorist on I-4. When Tesla driving on “auto” mode struck the patrol car. Trooper was outside of car and extremely lucky to have not been struck. #moveover. WB lanes of I-4 remain block as scene is being cleared."
[1] https://twitter.com/FHPOrlando/status/1431565185899171840
I have seen many descriptions of the tech, some of it at a pretty high level on how it navigates but never anything how it ensures that the passenger has had a successful exit.
Uber showed us what we can do with an app, now the job is to adapt our mass transit systems to these new ways of reasoning about traffic.
The car will probably never drive with the door open and all modern cars have sensors to detect this.
Second, their cars (and any competent self-driving car) needs to have a very accurate awareness of its surroundings. True it might be difficult to get a clear view of a person just barely outside the car but I'd bet that they have cameras/radar/lidar positioned to see if at least something is there.
Lastly, they have internal passenger cameras that can detect whether a person is inside the car.
That's something underdiscussed. A lot of driving is predicting other driver's behaviours. After a while you get a sense of what people will do, and in which broad category they fall under.
With self-driving cars, now we have new actors on the road that don't act naturally. And each software update might change these behaviours. This has some interesting feedback implications. As we get more of those on the streets, we might see more accidents after software upgrades. Not because of the software, but because people got used to the old behaviour and didn't anticipate the new one.