Waymo

[mikes_fsd]

They moved them to other areas. They still have the autonomous semi trucks that they're testing around the area though.

I am not sure what testing semi-trucks has to do with autonomous taxi service -- what am I missing here?
As for the "moved to other areas"... If they moved them to other areas, wouldn't we see reports on reddit or twitter of a bunch of Waymo vans in a new area?

 

[Doggydogworld]

Waymo has shown it has safe and reliable L4 autonomous driving. Their disengagement rate is the best of all the companies testing in CA. They even have FSD with no driver! They are clearly not incompetent when it comes to FSD. Their software clearly works. And as for their business model, Waymo has raised billions so clearly investors believe in their business model. And Waymo is the only company with commercial driverless ride-hailing robotaxis. And the service is receiving rave reviews. Waymo has proven that their ride-hailing model works. Now they can scale it up to more cities with the 5th gen FSD. So yeah, I think their business model is working just fine.

There's zero doubt about their technical competence.

They look more and more like Xerox PARC 2.0. It's 2021 and they're still running the 50 square mile faux service they launched in 2018. It's a technology showcase, not a business. Forget R&D costs, the service itself loses tons of money. And it seems the losses grow almost linearly with scale.

If the Robotaxi business model worked they'd be well into scale-up by now. They ordered 82k cars three years ago, so they clearly expected to do so. But they quickly found out that running a real service incurs all kinds of costs and inefficiencies that don't show up on Tony Seba's fantasy spreadsheets.

I half-think they've given up on Robotaxis and will use Chandler to get quicker regulatory approval for Waymo Via. Unlike Robotaxis, the trucking business model is spectacular. Almost as good as selling $10k Faux Self Driving!

 

[Dan D.]

There's zero doubt about their technical competence.

They look more and more like Xerox PARC 2.0. It's 2021 and they're still running the 50 square mile faux service they launched in 2018. It's a technology showcase, not a business. Forget R&D costs, the service itself loses tons of money. And it seems the losses grow almost linearly with scale.

If the Robotaxi business model worked they'd be well into scale-up by now. They ordered 82k cars three years ago, so they clearly expected to do so. But they quickly found out that running a real service incurs all kinds of costs and inefficiencies that don't show up on Tony Seba's fantasy spreadsheets.

I half-think they've given up on Robotaxis and will use Chandler to get quicker regulatory approval for Waymo Via. Unlike Robotaxis, the trucking business model is spectacular. Almost as good as selling $10k Faux Self Driving!

Good points. I would think the smaller local delivery van would involve much the same challenges as the Robotaxi model. Intercity trucking, though, is mostly highways, with the usual city interactions, should be easier? I presume they have to file a plan and report on disengagements too?

Wonder what their plan is for getting packages out of the vans and into the customers' store. I guess they could drive to the loading bay and get the customers' staff to unload it.

 

[diplomat33]

If the Robotaxi business model worked they'd be well into scale-up by now. They ordered 82k cars three years ago, so they clearly expected to do so. But they quickly found out that running a real service incurs all kinds of costs and inefficiencies that don't show up on Tony Seba's fantasy spreadsheets.

No, Waymo could not have scaled up with the 4th gen FSD. The 4th gen hardware was good for developing FSD but was always too expensive for scaling up a robotaxi service. But Waymo did learn a lot with the 4th gen both in terms of FSD and also in terms of the business side of ride-hailing. And the Waymo One service in Phoenix has proved to be quite popular with the public.

Waymo says that the 5th gen will unlock true scalability. So let's see if that happens. I think it is still too early to write off Waymo's ride-hailing service. Remind me in 5 years. If in 5 years, Waymo is still "stuck" in virtually the same ~100-200 sq mi area in Phoenix, then I will admit that Waymo's ride-hailing business model is a dead end. I am optimistic that once Waymo fully deploys the 5th gen hardware that we will see Waymo One expand to more cities. Having said that, I do agree that Waymo Via may prove to be very profitable for Waymo.

I would remind you that during autonomy day, Elon promised a Tesla robotaxi service would launch in 2020. It's early 2021, Tesla is still working on FSD Beta and has launched no ride-hailing service and no robotaxis yet. At least, Waymo has launched a popular true robotaxi ride-hailing service. A robotaxi ride-hailing service is harder than people think.

 

[diplomat33]

FYI, for anyone curious about how the new multiple stops feature will work, here is some info from the Waymo support page:

https://twitter.com/x/status/1374800277291757568

 

[Dan D.]

FYI, for anyone curious about how the new multiple stops feature will work, here is some info from the Waymo support page:

https://twitter.com/x/status/1374800277291757568

So it's a customer-service convenience feature. A bit like having a time limited transfer on your bus ticket rather than having to pay for each separate trip. Or in Waymo's case it's a 5-stop transfer. They will try and keep the same car or another car "close by" for 20 minutes. Could get more people to use the service.

 

[MP3Mike]

FYI, for anyone curious about how the new multiple stops feature will work, here is some info from the Waymo support page:

https://twitter.com/x/status/1374800277291757568

So you still can't leave anything in the vehicle during a stop, which I thought would be one of the biggest advantages, because "your" car could get assigned to someone else.

 

[Dan D.]

So you still can't leave anything in the vehicle during a stop, which I thought would be one of the biggest advantages, because "your" car could get assigned to someone else.

That's why it's a taxi not a rental/long-hire, oh well. I suppose you can still keep the meter running if you want.

You can't leave your children apparently; because I always leave my kids in the taxi. Just gonna be in the casino for a few minutes I promise.

Can people stay in the car at a stop?
If under 5 minutes, it’s okay to stay in. But, past that, we ask you to exit. No matter the length of stop, please never leave children unattended in the vehicle.

 

[diplomat33]

So you still can't leave anything in the vehicle during a stop, which I thought would be one of the biggest advantages, because "your" car could get assigned to someone else.

Well, these cars are still taxis. They are not your car. But Krafcik has said that Waymo is planning a personal car ownership program where you could lease your own private Waymo robotaxi for a year. It would be your car for the duration of the lease. It would be in your driveway and you could use it like you would use your car. So there would be much more flexibility there, where you could leave your stuff in the car, and use it whenever you want.

 

[diplomat33]

I parsed the data from the CA DMV disengagement report. Waymo had 145 cars that drove autonomous miles in CA and thus reported to the CA DMV.

Here is the data for each car. I think it gives some perspective to the average of 1 disengagement per 30,000 miles. We can see that many cars did thousands of autonomous miles with ZERO disengagements.

VIN	Disengagements	Miles
2C4RC1K70HR797365	0​	0.1​
2C4RC1K77HR833889	0​	3.1​
2C4RC1K7XHR833885	0​	3.1​
2C4RC1K79HR833778	0​	3.2​
2C4RC1K70HR805481	0​	6.8​
2C4RC1K73HR786120	0​	14​
SADHW2S17M1602661	0​	19.9​
SADHW2S15M1X06468	0​	26.4​
2C4RC1K71HR833838	0​	38​
2C4RC1K71HR822192	0​	41.3​
2C4RC1K75HR786121	0​	71.8​
2C4RC1K7XHR822207	0​	77.9​
2C4RC1K71HR805523	0​	95.7​
2C4RC1K78HR833870	0​	105.3​
2C4RC1K73HR805488	0​	108.3​
2C4RC1K75HR805489	0​	114​
SADHW2S15M1X06484	0​	121.5​
2C4RC1K75HR805508	0​	122​
2C4RC1K79HR786123	0​	137.6​
2C4RC1K74HR797367	0​	144.5​
2C4RC1K75HR534658	0​	148.8​
2C4RC1K71HR822208	0​	149.2​
2C4RC1K78HR805485	0​	185.5​
2C4RC1K7XHR833756	0​	193.1​
2C4RC1K77HR534676	0​	202.1​
2C4RC1K76HR805467	0​	204.7​
2C4RC1K76HR824259	0​	213.6​
2C4RC1K75HR833857	0​	224.4​
SADHW2S12M1X06474	0​	228.5​
2C4RC1K77HR786122	0​	279.9​
2C4RC1K75HR833759	0​	284.5​
2C4RC1K74HR827242	0​	358.6​
2C4RC1K75HR797295	0​	370.1​
2C4RC1K78HR534699	0​	388.1​
2C4RC1K71HR833810	0​	404.3​
2C4RC1K78HR534637	0​	409​
2C4RC1K78HR797386	0​	429.5​
2C4RC1K76HR797399	0​	435.5​
2C4RC1K73HR833856	0​	533.5​
2C4RC1K73HR822193	0​	547.2​
2C4RC1K75HR822194	0​	607.9​
2C4RC1K73HR827264	0​	619​
2C4RC1K70HR797415	0​	650.2​
2C4RC1K79HR797378	0​	727.2​
2C4RC1K71HR833841	0​	767.7​
2C4RC1K72HR833816	0​	779.6​
2C4RC1K73HR797375	0​	825​
2C4RC1K74HR534666	0​	839.1​
2C4RC1K74HR797322	0​	841.2​
2C4RC1K72HR824243	0​	844.7​
2C4RC1K7XHR534641	0​	945.4​
2C4RC1K73HR534710	0​	971.6​
2C4RC1K70HR797317	0​	978​
2C4RC1K74HR797353	0​	1072.6​
2C4RC1K77HR833858	0​	1111.8​
2C4RC1K7XHR534672	0​	1145.4​
2C4RC1K73HR833873	0​	1168.6​
2C4RC1K70HR534695	0​	1266.6​
2C4RC1K70HR534650	0​	1299.1​
2C4RC1K78HR695697	0​	1344.6​
2C4RC1K7XHR833854	0​	1385.1​
2C4RC1K73HR833808	0​	1396.2​
2C4RC1K75HR827234	0​	1420.3​
2C4RC1K75HR827265	0​	1426.8​
2C4RC1K74HR534652	0​	1531.1​
2C4RC1K75HR805461	0​	1573.1​
2C4RC1K79HR534677	0​	1589.5​
2C4RC1K76HR805453	0​	1634.2​
2C4RC1K77HR534693	0​	1677.6​
2C4RC1K72HR805465	0​	1719.9​
2C4RC1K76HR833785	0​	2013.2​
2C4RC1K7XHR695698	0​	2653.6​
2C4RC1K70HR534664	0​	2658.6​
2C4RC1K77HR534659	0​	2679​
2C4RC1K76HR797354	0​	2778.7​
2C4RC1K78HR695702	0​	2833.8​
2C4RC1K71HR534673	0​	2887.6​
2C4RC1K76HR534698	1​	2944.6​
2C4RC1K77HR534662	0​	2945​
2C4RC1K75HR695706	1​	3033.3​
2C4RC1K73HR534674	0​	3076​
2C4RC1K77HR534645	0​	3131.5​
2C4RC1K75HR695690	0​	3189.1​
2C4RC1K72HR534701	0​	3275.7​
2C4RC1K76HR534670	0​	3356.2​
2C4RC1K71HR534687	0​	3461.2​
2C4RC1K76HR534703	0​	3702.5​
2C4RC1K74HR695700	0​	3774.3​
2C4RC1K73HR797361	0​	3974.3​
2C4RC1K70HR833877	0​	3980.3​
2C4RC1K73HR695705	0​	4026.9​
2C4RC1K7XHR534705	1​	4238.5​
2C4RC1K72HR534696	0​	4438.5​
2C4RC1K74HR805516	3​	5064.2​
2C4RC1K76HR695696	0​	5088.5​
2C4RC1K76HR833835	0​	5167​
2C4RC1K76HR534653	0​	5201.4​
2C4RC1K7XHR805486	0​	5388.3​
2C4RC1K73HR534691	0​	5388.4​
2C4RC1K70HR534681	0​	5598.4​
2C4RC1K71HR695699	1​	6100.3​
2C4RC1K70HR805478	0​	6277​
2C4RC1K72HR836179	0​	6300.8​
2C4RC1K75HR797345	1​	6463.7​
2C4RC1K72HR797318	0​	6488.5​
2C4RC1K72HR534665	0​	6509.4​
2C4RC1K7XHR833787	0​	6690.2​
2C4RC1K79HR824255	0​	6754​
2C4RC1K7XHR827231	0​	7010.2​
2C4RC1K72HR827238	0​	7432.3​
2C4RC1K77HR805512	2​	7609.7​
2C4RC1K78HR833769	0​	7796.2​
2C4RC1K78HR833867	0​	8384.3​
2C4RC1K78HR805518	0​	8490.2​
2C4RC1K7XHR797308	0​	8589.6​
2C4RC1K73HR534660	0​	8918.9​
2C4RC1K73HR833758	0​	9163.7​
2C4RC1K79HR833859	1​	9243.3​
2C4RC1K74HR805466	0​	9477.8​
2C4RC1K70HR833796	1​	9941​
2C4RC1K71HR797343	0​	9965​
2C4RC1K73HR797330	0​	10014.5​
2C4RC1K78HR797405	0​	10142.5​
2C4RC1K7XHR827262	0​	10302.5​
2C4RC1K73HR797389	0​	10376.9​
2C4RC1K79HR822215	1​	10584.3​
2C4RC1K7XHR805455	2​	10591.8​
2C4RC1K77HR833827	0​	10997.1​
2C4RC1K75HR812331	0​	11702.2​
2C4RC1K73HR797358	0​	11796.6​
2C4RC1K73HR833842	1​	11822.8​
2C4RC1K78HR797355	0​	12102.7​
2C4RC1K7XHR797325	1​	12252.2​
2C4RC1K76HR797404	2​	12928.2​
2C4RC1K70HR827271	1​	13079.3​
2C4RC1K70HR797379	0​	13227.5​
2C4RC1K74HR805497	0​	13717.8​
2C4RC1K78HR833772	0​	13932​
2C4RC1K76HR822205	0​	14522.1​
2C4RC1K7XHR822188	0​	14947.6​
2C4RC1K74HR833879	0​	15439.3​
2C4RC1K75HR797362	0​	16047.2​
2C4RC1K74HR827239	0​	16757​
2C4RC1K78HR797324	1​	17126.1​
2C4RC1K79HR797381	0​	17269.9​

Source: Disengagement Reports - California DMV

 

[Doggydogworld]

No, Waymo could not have scaled up with the 4th gen FSD. The 4th gen hardware was good for developing FSD but was always too expensive for scaling up a robotaxi service. But Waymo did learn a lot with the 4th gen both in terms of FSD and also in terms of the business side of ride-hailing. And the Waymo One service in Phoenix has proved to be quite popular with the public.

Waymo says that the 5th gen will unlock true scalability. So let's see if that happens. I think it is still too early to write off Waymo's ride-hailing service. Remind me in 5 years. If in 5 years, Waymo is still "stuck" in virtually the same ~100-200 sq mi area in Phoenix, then I will admit that Waymo's ride-hailing business model is a dead end. I am optimistic that once Waymo fully deploys the 5th gen hardware that we will see Waymo One expand to more cities. Having said that, I do agree that Waymo Via may prove to be very profitable for Waymo.

I would remind you that during autonomy day, Elon promised a Tesla robotaxi service would launch in 2020. It's early 2021, Tesla is still working on FSD Beta and has launched no ride-hailing service and no robotaxis yet. At least, Waymo has launched a popular true robotaxi ride-hailing service. A robotaxi ride-hailing service is harder than people think.

Waymo said Gen 4 was for scaling. At the 2017 Detroit Auto Show Krafcik said 90% lidar cost reductions enabled them to scale. That was more than FOUR. YEARS. AGO.

They "ordered" 20,000 Jaguar i-Paces, good enough for "a million rides per day", THREE years ago yesterday. They "ordered" 62,000 Pacificas before that. Krafcik said they'd launch a commercial service without drivers during 2018. There was never any thought given to waiting for Gen 5 sensors. They always planned to scale with Gen 4 then layer in Gen 5, 6, etc. as they continued to scale.

I fully realize Elon has failed to achieve any of his ridiculous claims. That doesn't change the fact that Waymo planned to start scaling in 2018 and then...... didn't. And still hasn't. And it's not because Gen 4 cost too much. They already factored that cost into their plans back in 2017-18.

 

[diplomat33]

Waymo said Gen 4 was for scaling. At the 2017 Detroit Auto Show Krafcik said 90% lidar cost reductions enabled them to scale. That was more than FOUR. YEARS. AGO.

They "ordered" 20,000 Jaguar i-Paces, good enough for "a million rides per day", THREE years ago yesterday. They "ordered" 62,000 Pacificas before that. Krafcik said they'd launch a commercial service without drivers during 2018. There was never any thought given to waiting for Gen 5 sensors. They always planned to scale with Gen 4 then layer in Gen 5, 6, etc. as they continued to scale.

I fully realize Elon has failed to achieve any of his ridiculous claims. That doesn't change the fact that Waymo planned to start scaling in 2018 and then...... didn't. And still hasn't. And it's not because Gen 4 cost too much. They already factored that cost into their plans back in 2017-18.

Yes, Waymo made big progress with reducing cost of the lidar on the 4th generation. I'm just going on Waymo statements that the 5th generation is built for scaling up.

Maybe they thought the 4th gen would be good enough for scaling and it wasn't, but now the 5th generation is even cheaper and simpler to manufacture, so they think the 5th generation will be good enough for scaling.

Bottom line: everybody misses deadlines when it comes to something difficult like FSD. Waymo has L4 but failed to scale up as quickly as maybe they had planned. Tesla is working on getting FSD.

 

[Doggydogworld]

Phoenix New Times article about police reports involving Waymo. A couple of interest:

- January 2020: the department issued a citation to one of its own officers for rear-ending a Waymo vehicle. ....When a streetlight turned green, the Waymo vehicle "began to move forward and then stopped prior to entering the intersection per the programming of the vehicle." Both vehicles had slight damage.

- 10/8/20: Waymo minivan in the middle of three westbound lanes on Chandler Boulevard.... unexpectedly braked for no reason. A red Chevrolet Silverado pickup behind the vehicle swerved to the right but clipped its back panel, causing minor damage. No citation issued.

- In the other two Chandler incidents involving minor rear-end collisions, the Waymo vehicles were in manual mode, and only one of the drivers received a citation.

- Waymo vehicles were involved in four other collisions, including two hit-and-runs. No injuries were reported. In one of the hit-and-run collisions, police noted that Waymo provided video of the Tesla that hit its minivan, but blurred out the Tesla's license plate.

- One man in Chandler was found passed out in a Waymo vehicle from suspected narcotics use.

- A police officer saw a car reverse rapidly toward a Waymo vehicle, barely avoiding a collision. The driver was arrested for driving on a revoked license.

- Someone threw an ice-cream cone into the open window of a Waymo vehicle.

- A man in a black car threw eggs at multiple Waymo vehicles. The Waymo backup drivers didn't want to be contacted by police.

 

[diplomat33]

Phoenix New Times article about police reports involving Waymo. A couple of interest:

- January 2020: the department issued a citation to one of its own officers for rear-ending a Waymo vehicle. ....When a streetlight turned green, the Waymo vehicle "began to move forward and then stopped prior to entering the intersection per the programming of the vehicle." Both vehicles had slight damage.

- 10/8/20: Waymo minivan in the middle of three westbound lanes on Chandler Boulevard.... unexpectedly braked for no reason. A red Chevrolet Silverado pickup behind the vehicle swerved to the right but clipped its back panel, causing minor damage. No citation issued.

- In the other two Chandler incidents involving minor rear-end collisions, the Waymo vehicles were in manual mode, and only one of the drivers received a citation.

- Waymo vehicles were involved in four other collisions, including two hit-and-runs. No injuries were reported. In one of the hit-and-run collisions, police noted that Waymo provided video of the Tesla that hit its minivan, but blurred out the Tesla's license plate.

- One man in Chandler was found passed out in a Waymo vehicle from suspected narcotics use.

- A police officer saw a car reverse rapidly toward a Waymo vehicle, barely avoiding a collision. The driver was arrested for driving on a revoked license.

- Someone threw an ice-cream cone into the open window of a Waymo vehicle.

- A man in a black car threw eggs at multiple Waymo vehicles. The Waymo backup drivers didn't want to be contacted by police.

Shame on that Tesla that did the hit-and-run!

 

[diplomat33]

00:00 Multistop trip demo
01:20 Car arriving
01:39 Boarding
01:51 Ride start and fixed the lighting!
02:45 Sees pedestrian inside a building
02:58 What's up with the turn signals
03:20 Unprotected right
04:46 Reading two traffic lights at once
05:10 Bit of heavy braking
07:40 Comment shoutout of the day
08:58 Right turn with a pedestrian in the crosswalk
09:32 Forty seconds of serious needle threading (cars parked on either side of the street)
10:22 "Close, but don't touch" unprotected right
11:15 Where that silly voiceover sound came from
11:50 Work/construction zone
15:20 Multistop trip shenanigans begin
15:59 Pedestrian walking in the road
17:21 Rider support call

 

[diplomat33]

Krafcik is leaving Waymo to pursue other ventures. The current COO and CTO will be taking over as co-CEO.

Twitter announcement:

https://twitter.com/x/status/1378029941602848772

My thoughts: I like Tekedra and Dmitri. As COO, Tekedra knows the operational side and Dmitri as the CTO, definitely knows the technical side. So I think they are a good match to lead the company. I am really hoping that we see a faster deployment of the 5th generation and a bigger push to expand to more areas.

 

[DanCar]

The axe has fallen.

 

[diplomat33]

The axe has fallen.

Could you be more specific? Do you think Waymo pushed Krafcik out?

 

[DanCar]

Could you be more specific? Do you think Waymo pushed Krafcik out?

Perhaps more cordial than that. Larry and Sergey said, we think it is time for new leadership. Krafcik: o.k. Likely there were some milestones that needed to be met, and they were not met.

 

[diplomat33]

Perhaps more cordial than that. Larry and Sergey said, we think it is time for new leadership. Krafcik: o.k. Likely there were some milestones that needed to be met, and they were not met.

Thanks for the reply. The announcement does sound like the departure was on good terms. Your explanation makes sense to me. It is possible that the company had goals like expanding faster and those goals were not met obviously. I think a lot of people had a feeling that Waymo was not expanding fast enough. So maybe, they thanked Krafcik for his leadership in bringing Waymo this far but decided that new leadership was needed to take Waymo to the next level.

Also, while "spending time with family" has become a cliche, there might be some truth to it. Being a CEO is a stressful job so it is entirely possible that after 5 years, Krafcik was looking for a change in scenery. So the truth might be a little of both, the company wanted different leadership and Krafcik wanted a break so the timing worked out.