Ford- and Volkswagen-backed Argo AI is shutting down

[voyager]

Argo AI, an autonomous vehicle startup that burst on the scene in 2017 stacked with a $1 billion investment,
is shutting down — its parts being absorbed into its two main backers: Ford and VW, according to people
familiar with the matter. Ford, VW-backed Argo AI is shutting down

 

[diplomat33]

Argo AI, an autonomous vehicle startup that burst on the scene in 2017 stacked with a $1 billion investment,
is shutting down — its parts being absorbed into its two main backers: Ford and VW, according to people
familiar with the matter. Ford, VW-backed Argo AI is shutting down

It is being discussed in the autonomous driving progress thread.

Basically, Ford made the business decision to shut down Argo because they want to focus on L3 instead of L4. They feel L3 can be a more immediate and profitable product rather than waiting for robotaxis to scale. They will hire a lot of the Argo engineers to work on their BlueCruise ADAS system.

 

[voyager]

I think this news is big enough to justify its own thread.
Although the bad news regarding driverless have been HUGE this month.

There's an avalanche of bad news and doubt these past weeks about self-driving. Not good.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere. They were supposed to be the future. But prominent detractors, who pioneered…

t.co

 

[diplomat33]

I think this news is big enough to justify its own thread.
Although the bad news regarding driverless have been HUGE this month.

There's an avalanche of bad news and doubt these past weeks about self-driving. Not good.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere. They were supposed to be the future. But prominent detractors, who pioneered…

t.co

No. Many of those articles are FUD and clickbait. The bad news is being grossly exaggerated and they don't mention any of the very positive AV news. Self-driving cars are not "going nowhere" as one title says. We are seeing more and more robotaxi deployments. Baidu is deploying in China. Waymo announced LA as their next area, and Cruise is starting road testing of the Origin. And "$100 Billion and 10 Years of Development Later, and Self-Driving Cars Can Barely Turn Left" is a total nonsense clickbait title and totally false. AVs have made huge progress in 10 years and there are lots of autonomous cars that can turn left extremely reliably.

It should also be noted that Argo shutting down had nothing to do with their AV tech, which was quite good actually. It was purely a business decision by Ford and VW. So any attempt to use Argo shutting down as somehow proof that self-driving cars are failing or "going nowhere" is wrong.

 

[voyager]

Dream on, diplomat33...

 

[diplomat33]

Dream on, diplomat33...

It is not a fantasy. AVs are a reality now. They are not perfect of course but they will get better. I am not naive. I know there are still a lot of challenges left to solve in AVs. But I am optimistic about the progress. And I want a future where I can get in the back seat of a driverless car and it chauffeurs me around safely and comfortably. That will be a good future. The bottom line is that there is zero benefit to society if self-driving cars fail. There are a lot of benefits to society if self-driving cars succeed. So it makes sense to want them to succeed.

 

[Dewg]

"$100 Billion and 10 Years of Development Later, and Self-Driving Cars Can Barely Turn Left"

Thanks for that - I needed a good laugh today. Let's take Waymo out of the picture for now, who already have L4 running around in a few cities. Let's look at Tesla, which is working on L2. You are totally correct - Tesla couldn't turn left very well a year or two ago. That was a year or two ago. Today, FSD Beta is running around city streets like a champ. It's not complete, and still has issues that need to get ironed out. But just watch some of the 0 intervention drives that are more and more common today.

Chris is one of my favorites:

And want to see a more challenging drive?

Give those a watch and get back to us on "Self-Driving Cars Can Barely Turn Left"

 

[voyager]

Tesla Inc (TSLA.O) is under criminal investigation in the United States over claims that the company’s electric vehicles can drive themselves. The U.S. Department of Justice launched the previously undisclosed probe last year following more than a dozen crashes, some of them fatal. Exclusive: Tesla faces U.S. criminal probe over self-driving claims

 

[Dewg]

Tesla Inc (TSLA.O) is under criminal investigation in the United States over claims that the company’s electric vehicles can drive themselves. The U.S. Department of Justice launched the previously undisclosed probe last year following more than a dozen crashes, some of them fatal. Exclusive: Tesla faces U.S. criminal probe over self-driving claims

Yes, this is correct. It has to do with the naming of the systems.

A person gets into a Tesla with Autopilot. The name suggests that it can drive itself. The person enabled Autopilot in the menu, and a disclaimer comes up that warns them that the feature requires full attention and does NOT drive itself. The driver must accept this. Then the drive gets on the freeway and engages Autopilot, which brings up a warning on the screen that says the driver must pay full attention. Then the system monitors the driver to make sure they are paying attention with wheel torque requests every 30-60 seconds. With newer software and cars the system even watches the driver with the cabin camera to ensure the driver is paying attention to the road and not looking at their phone.

Where the issue is that people are getting into accidents and claiming they thought the car, with a feature called Autopilot, could drive itself, so they didn't pay attention and got into an accident. Or the driver was paying attention and saw a potential accident, but trusted the system would take care of it and let the car crash without intervening.

The question becomes: would a name change have prevented the accident? If the system was called "Tesla Driver Assist" and not "Autopilot", would the driver have still had the accident? Is public perception that Teslas are advanced and can drive themselves no matter what it's called?

"It's not my fault, officer - I had Autopilot enabled!"
"It's not my fault, officer - I had Driver Assist enabled"

 

[diplomat33]

Yes, this is correct. It has to do with the naming of the systems.

A person gets into a Tesla with Autopilot. The name suggests that it can drive itself. The person enabled Autopilot in the menu, and a disclaimer comes up that warns them that the feature requires full attention and does NOT drive itself. The driver must accept this. Then the drive gets on the freeway and engages Autopilot, which brings up a warning on the screen that says the driver must pay full attention. Then the system monitors the driver to make sure they are paying attention with wheel torque requests every 30-60 seconds. With newer software and cars the system even watches the driver with the cabin camera to ensure the driver is paying attention to the road and not looking at their phone.

Where the issue is that people are getting into accidents and claiming they thought the car, with a feature called Autopilot, could drive itself, so they didn't pay attention and got into an accident. Or the driver was paying attention and saw a potential accident, but trusted the system would take care of it and let the car crash without intervening.

The question becomes: would a name change have prevented the accident? If the system was called "Tesla Driver Assist" and not "Autopilot", would the driver have still had the accident? Is public perception that Teslas are advanced and can drive themselves no matter what it's called?

"It's not my fault, officer - I had Autopilot enabled!"
"It's not my fault, officer - I had Driver Assist enabled"

I don't think the issue is so much with the name "Autopilot". The issue is more to do with the name "Full Self-Driving". That name definitely suggests the car is driving itself.

Furthermore, back in 2017-2018, Tesla had a section on the order page when you would add FSD that described the car being able to do short and long trips with no action required from the driver. See below. That description certainly seemed to give the impression that the car would be able to drive itself:

Of course, there is also that disclaimer that states that functionality is dependent on validation and regulatory approval. So, one section says the car will drive you from A to B with no action on your part and the other section says "we can't promise when this will happen".

There is also this page before you order a Model S or X that claimed AP2 cars have the necessary hardware for full self-driving at safety greater than human:

 

[Dewg]

I don't think the issue is so much with the name "Autopilot". The issue is more to do with the name "Full Self-Driving". That name definitely suggests the car is driving itself.

Furthermore, back in 2017-2018, Tesla had a section on the order page when you would add FSD that described the car being able to do short and long trips with no action required from the driver. See below. That description certainly seemed to give the impression that the car would be able to drive itself:

Of course, there is also that disclaimer that states that functionality is dependent on validation and regulatory approval. So, one section says the car will drive you from A to B with no action on your part and the other section says "we can't promise when this will happen".

There is also this page before you order a Model S or X that claimed AP2 cars have the necessary hardware for full self-driving at safety greater than human:

I totally agree, but many of the investigations are about accidents, which are mostly on Autopilot.

 

[diplomat33]

Good thoughts from Aurora's CEO on the Argo news:

But I also want to be clear that this is not a signal that a future with self-driving technology isn’t real or imminent. In fact, it’s quite the opposite. Waymo recently announced the expansion of their robotaxi fleet to LA. Cruise is charging for their driverless robotaxi rides in San Francisco.

On seismic shifts and steady progress—a note from Chris Urmson

Sorry to interrupt your regularly scheduled programming, but major news was announced in the self-driving industry yesterday and I wanted to share my perspective. Argo is shutting down.

www.linkedin.com

 

[EVNow]

No. Many of those articles are FUD and clickbait

LOL. You don't think so about anti-Tesla news though !

Seriously, you guys should get your own forum.

 

[diplomat33]

LOL. You don't think so about anti-Tesla news though !

Of course I do. I call out anti-Tesla FUD.

Seriously, you guys should get your own forum.

No. This is a forum for Tesla owners, is it not? I am a Tesla owner. This forum is for me.

 

[ThomasD]

Will Tesla eventually install self cleaning cameras

 

[DrChaos]

It should also be noted that Argo shutting down had nothing to do with their AV tech, which was quite good actually. It was purely a business decision by Ford and VW. So any attempt to use Argo shutting down as somehow proof that self-driving cars are failing or "going nowhere" is wrong.

It does mean that the prospect of significant profitable robotaxi revenue was deemed to be remote, at least through Argo.

It's probably the right decision. It's a decision to prioritize acceptably low hardware costs that can be sold on an a customer's vehicle instead of paying for whatever hardware is necessary for L4 today, the reality being that high hardware costs are a certainty but L4 performance and robo revenue are uncertain.

This approach is more likely to lead to cheaper hardware (consumer level scale) and clever machine learning that eventually makes L4 robotaxis more cost effective and a feasible business when these have been further developed.

 

[ninefiveone]

It does mean that the prospect of significant profitable robotaxi revenue was deemed to be remote, at least through Argo.

It's probably the right decision. It's a decision to prioritize acceptably low hardware costs that can be sold on an a customer's vehicle instead of paying for whatever hardware is necessary for L4 today, the reality being that high hardware costs are a certainty but L4 performance and robo revenue are uncertain.

This approach is more likely to lead to cheaper hardware (consumer level scale) and clever machine learning that eventually makes L4 robotaxis more cost effective and a feasible business when these have been further developed.

The prospect of significant robotaxi revenue soon would probably have changed the decision, but I think it's really about Ford and VW as investors.

They're not ideal long-term investors right now. Both need to focus capital and attention on the transition to EV's. Self-driving for either of them won't matter if they don't exist by the time self-driving is solved, and they haven't secured their EV futures yet. Self driving could be solved a lot sooner than anyone thinks, but as investors they need to be ready for a 10 year timeframe.

 

[diplomat33]

It does mean that the prospect of significant profitable robotaxi revenue was deemed to be remote, at least through Argo.

It's probably the right decision. It's a decision to prioritize acceptably low hardware costs that can be sold on an a customer's vehicle instead of paying for whatever hardware is necessary for L4 today, the reality being that high hardware costs are a certainty but L4 performance and robo revenue are uncertain.

This approach is more likely to lead to cheaper hardware (consumer level scale) and clever machine learning that eventually makes L4 robotaxis more cost effective and a feasible business when these have been further developed.

I understand why Ford made the decision. But IMO, L3 is not a viable long term strategy. The future is L4. If Ford develops L3 with a plan to quickly transition to L4, that will work. But if Ford becomes content with just L3, they might "win" short term but they will lose long term. Right now, L3 is very limited. L3 only works in stop and go traffic at low speeds. It is not very useful. Perhaps, Ford/Argo engineers will develop better L3? But if they release L3 that is just stop and go traffic at low speeds when other companies are deploying L4, they will fall behind.

I just think Ford is making the typical mistake, that a lot of the legacy automakers make, of thinking too small. They will settle for L3 and miss the boat when L4 starts to really scale in a few years.

The prospect of significant robotaxi revenue soon would probably have changed the decision, but I think it's really about Ford and VW as investors.

They're not ideal long-term investors right now. Both need to focus capital and attention on the transition to EV's. Self-driving for either of them won't matter if they don't exist by the time self-driving is solved, and they haven't secured their EV futures yet. Self driving could be solved a lot sooner than anyone thinks, but as investors they need to be ready for a 10 year timeframe.

Sadly for Argo, having investors like Ford and VW was a major disadvantage because it caused them to get shut down, not because their tech was bad or because they had no prospects, but simply because they had "bad" investors that could not stick it out.

 

[DrChaos]

I understand why Ford made the decision. But IMO, L3 is not a viable long term strategy. The future is L4. If Ford develops L3 with a plan to quickly transition to L4, that will work. But if Ford becomes content with just L3, they might "win" short term but they will lose long term. Right now, L3 is very limited. L3 only works in stop and go traffic at low speeds. It is not very useful. Perhaps, Ford/Argo engineers will develop better L3? But if they release L3 that is just stop and go traffic at low speeds when other companies are deploying L4, they will fall behind.

I just think Ford is making the typical mistake, that a lot of the legacy automakers make, of thinking too small. They will settle for L3 and miss the boat when L4 starts to really scale in a few years.

I think the bet is the idea that something that scales up and advances will most likely do so from a low cost, high volume base. The discipline needed to integrate electronics and sensors sufficiently inexpensively and train fleets at scale from crowdsourced data will help develop L4 eventually as sensors get better---that's the bet.

The iPhone now has much better technology than any low-volume military platform that might have once started as more advanced than consumer electronics. Microsoft grew out of DOS and Windows 3.11, not IBM downsizing MVS/370.

The ADAS will help sell cars now. Mobileye is going down this path as well. Their revenue is L2+/L3 ADAS.

Lots of investment has been made on the assumption that the L4 was already going to scaleup by now, but it didn't. Ford literally can't afford to spend lots more cash while it makes the transition to EVs in the rest of the business and winding down LICE costs.

And it may also be that the economics don't work for a $100K robo taxi, with a $30K car and $70K sensor and computer suite, but they do work for a $35 K robotaxi, with a $5K sensor and computer suite. There's going to be lots of vandalism when there's no human aboard.

I think it's a reasonable approach, and obviously entirely opposite from Waymo. Maybe not a guaranteed success but much less of a financial risk. Despite Elon Musk's big mouth, it's clear where his priorities lie as actions speak louder than words: excellent hardware profit margins are the core goal, then whatever ADAS can be hacked onto that is a free bonus.