Autonomous Car Progress

[diplomat33]

Has waymo addressed the question of whether they would ever be involved in self-driving cars that are designed for consumer purchase as opposed to robotaxi fleet usage?

I think the CEO of Waymo recently said that they are definitely interested but it would come "later". I think Waymo has said robotaxis come first then consumer self-driving cars later.

 

[diplomat33]

I would add that Waymo is only interested in autonomous driving that does not require the consumer to supervise the system. Remember back in the Google days, they experimented with hands-free L2 on the highway and found that the drivers became complacent. This is what motivated Waymo to only develop L4. So Waymo would never deploy consumer self-driving cars that require the driver to pay attention. But this means that Waymo will need to improve their FSD further to where it does need safety drivers anymore before they would deploy consumer self-driving cars. Robotaxis are a nice first step since they can be deployed in more controlled geofenced areas. But I think once Waymo is able to remove the safety driver everywhere, they would definitely do consumer self-driving cars.

 

[stopcrazypp]

I would add that Waymo is only interested in autonomous driving that does not require the consumer to supervise the system. Remember back in the Google days, they experimented with hands-free L2 on the highway and found that the drivers became complacent. This is what motivated Waymo to only develop L4. So Waymo would never deploy consumer self-driving cars that require the driver to pay attention. But this means that Waymo will need to improve their FSD further to where it does need safety drivers anymore before they would deploy consumer self-driving cars. Robotaxis are a nice first step since they can be deployed in more controlled geofenced areas. But I think once Waymo is able to remove the safety driver everywhere, they would definitely do consumer self-driving cars.

Given they had a crash that was a safety driver that fell asleep and pressed a pedal (thus putting it in manual mode), it's understandable that they want to eliminate safety drivers as soon as they can. They may not even want to have it in cars that still have direct physical controls (as the same thing can happen, "driver" falling asleep and hitting controls).
Waymo’s self-driving car crashed because its human driver fell asleep at the wheel

 

[diplomat33]

Here is the press release from Geely on their collaboration with Waymo with a few more details:

Waymo and Zeekr to Collaborate on All-electric, Fully Autonomous Ride-hailing Vehicle

December 28th, Hangzhou, China and California, USA. Geely Holding Group’s (Geely) premium electric brand, Zeekr, will collaborate with Waymo on the development of a new pure electric vehicle for deployment in the Waymo One autonomous ride-hailing fleet in the United States. The Program Explained

zgh.com

Here are the key points:

• The new purpose-built mobility vehicle is being designed and developed at Zeekr’s R&D facility, CEVT (China Europe Vehicle Technology Centre) in Gothenburg, Sweden.
• Zeekr will be designing and developing the future vehicle on a new proprietary and open-source mobility architecture.
• Waymo will take delivery of the vehicles in the United States and will then integrate its fully autonomous Waymo Driver into the vehicle platform.
• The new vehicle will be designed to be rider-centric from the outset.
• The new Zeekr vehicle has been designed for autonomous use-cases and will come with a fully configurable cabin, both with and without driver controls, that can be tailored towards rider requirements for the Waymo One unmanned ride-hailing fleet in the US.

 

[Doggydogworld]

Has waymo addressed the question of whether they would ever be involved in self-driving cars that are designed for consumer purchase as opposed to robotaxi fleet usage?

Over and over again, as documented in this forum. The post I linked has an excerpt from her most recent statement, at Code Conference. She's also written it in blog posts and other communications.

The new Zeekr vehicle has been designed for autonomous use-cases and will come with a fully configurable cabin, both with and without driver controls, that can be tailored towards rider requirements for the Waymo One unmanned ride-hailing fleet in the US.

It will have driver controls in the US unless/until FMVSS changes. GM asked NHTSA to change it years ago, as have Waymo and others. This article says they are looking at it, so we can probably expect a change sometime in the next few decades....

 

[diplomat33]

It will have driver controls in the US unless/until FMVSS changes. GM asked NHTSA to change it years ago, as have Waymo and others. This article says they are looking at it, so we can probably expect a change sometime in the next few decades....

Thanks for that info. The Waymo blog appears to make a reference to this, saying that the vehicle will conform to standards but "one day" will have no steering wheel or pedals:

While ensuring a level of safety consistent with U.S. federal vehicle standards - our Waymo One riders will one day experience an interior without steering wheel and pedals, and with plenty of headroom, leg room and reclining seats, screens and chargers within arm’s reach, and an easy to configure and comfortable vehicle cabin.

The Waymo blog also says that they will begin to introduce this robotaxi to the Waymo One service "in the years to come":

We’ll begin to introduce these all electric, rider-first, fully autonomous vehicles on U.S. roads within our Waymo One fleet in the years to come.

Expanding our Waymo One fleet with Geely’s all-electric vehicle designed for riders first

Over a decade into our journey building the World’s Most Experienced Driver, we’ve partnered with some of the world’s leading automakers to deploy our fully autonomous technology as we help people and things get where they’re going. So, we’re pleased to announce our latest OEM collaboration with...

blog.waymo.com

 

[marketisopen]

I actually think that this is just a separate partnership and is not representative of Geely’s own autonomy plans.

Yes, it's just a design win by Geely for Google. It has nothing to do with Geely (from my reading of the press release)

 

[EVNow]

Has waymo addressed the question of whether they would ever be involved in self-driving cars that are designed for consumer purchase as opposed to robotaxi fleet usage?

They say “eventually” they will license the technology to consumer car OEMs. But they have not shown any kind of timeline or how the HD maps will be maintained in places they don’t operate or even how the sensors will look. At this point it’s just a zero cost feel good statement, not likely to happen this decade.

 

[diplomat33]

TuSimple completes 80 miles autonomous drive in their class 8 semi truck with no human in the truck and no interventions. Here is unedited video:

TuSimple successfully operates 80-mile, autonomous semi-truck ride on public roads without a human present

TuSimple announced it has completed the world's first fully-autonomous semi-truck ride on open public roads without a human present.

electrek.co

 

[daktari]

TuSimple completes 80 miles autonomous drive in their class 8 semi truck with no human in the truck and no interventions. Here is unedited video:

TuSimple successfully operates 80-mile, autonomous semi-truck ride on public roads without a human present

TuSimple announced it has completed the world's first fully-autonomous semi-truck ride on open public roads without a human present.

electrek.co

Impressive.
But why does the steering wheel and the vehicle wobble so much? @12 minutes for example. Shock wave from the passing truck, giving the trailer a wobble, and instead of speeding up a bit to get the trailer sorted, it just keep on counter steering?

 

[diplomat33]

Impressive.
But why does the steering wheel and the vehicle wobble so much? @12 minutes for example. Shock wave from the passing truck, giving the trailer a wobble, and instead of speeding up a bit to get the trailer sorted, it just keep on counter steering?

That sounds like a good explanation. I am guessing the autonomous truck does not know it should speed up to fix that issue. It just senses the truck not staying in the lane so it keeps counter steering to keep the truck centered in the lane.

 

[Terminator857]

"Why self-driving cars are a lie", from a UK guy. Click bait title, click bait video but good for entertainment purposes. Has a good section of history of lies by the car industry on self driving, starting around 3:10.

 

[ThomasD]

So TruSimple has achieved level 4 or 5 in their Semi? With the Snow and Rain now would be a perfect time to get these self driving vehicles out on the road.

 

[diplomat33]

So TruSimple has achieved level 4 or 5 in their Semi? With the Snow and Rain now would be a perfect time to get these self driving vehicles out on the road.

TruSimple autonomous trucks are L4.

 

[ThomasD]

I was watching the show Mud Mountain Haulers and was wondering if a self driving semi could be programed to handle the austere conditions better than these guys operating the trucks.

 

[diplomat33]

Brand new video from Mobileye. It is a very informative deep dive into their SuperVision driving policy with lots of examples. The video also explains challenges of driving policy and the pros and cons of different driving policy approaches such as end-to-end and monte carlo. The video is 1 hr but it is well worth a watch.

 

[diplomat33]

FYI, I know there has been some lengthy debates on this forum about Mobileye's "true redundancy" approach. During the Q&A part of the video, CTO Prof. Shai Shalev-Shwartz says that the radar+lidar system does use 1 camera but only for traffic lights. So no, it is not 100% radar+lidar-only. But the radar+lidar system does do all of the FSD except for that one thing, traffic lights. So "true redundancy" is really "vision-only" and "radar+lidar + 1 traffic light camera". He adds that both systems can drive autonomously "for hours". So the "radar+lidar + 1 traffic light camera" can do FSD. Also, he says that the idea of multiplying the probability of each system failing to get the probability of the whole system failing is not a perfect formula but it is close enough. He says that while the two systems are not perfectly independent, they will still provide significant redundancy. I hope that clarifies things.

 

[ThomasD]

So if 10 semis drove 100,000 miles with zero disengagements. Does that mean Waymo one is safer than the average truck driver?

 

[Daniel in SD]

So if 10 semis drove 100,000 miles with zero disengagements. Does that mean Waymo one is safer than the average truck driver?

I think you'd want to look at the rate of severe collisions. There is a fatality in a large truck crash about every 100 million miles (mostly in passenger cars).

Fatality Facts 2021: Large trucks

A summary of fatality statistics about large trucks compiled by IIHS from 2021 Fatality Analysis Reporting System (FARS) data.

www.iihs.org

Here's a great interview with head of trucking at Waymo. What makes trucks harder than passenger cars is that they operate much closer the the performance envelope of the vehicle. Presumably what makes them easier is that they can be economically viable while driving a relatively small number of routes.

 

[ThomasD]

Autonomous trucking rodeo. Pitting Autonomous trucks against each other. Like they have for Truck drivers