Autonomous Car Progress

[Dewg]

How many people would pay 400 dollars for a round trip ticket from California to Las Vegas on High Speed Rail?

LA-Vegas high-speed rail tickets could be $400 round-trip

(The Center Square) - A round trip ticket from Los Angeles to Las Vegas on a planned high-speed rail line could eventually cost up to $400, the project’s developer says.

www.thecentersquare.com

Sounds like FUD

 

[DoctorWTF]

However these companies will have a tough time coming to US (probably won’t) and will have a tough time entering some other markets. And it may be hard for many US based companies to acquire Chinese made tech.

I understand BYD is planning on building a factory in Mexico to sidestep import regulations. Though there's still the question of how much interest Americans would have in cars from a Chinese company, given their track record.

 

[WyoDude]

I understand BYD is planning on building a factory in Mexico to sidestep import regulations. Though there's still the question of how much interest Americans would have in cars from a Chinese company, given their track record.

I’ve heard this also but also think American customers would be skeptical and I can see the US making changes to also make this harder for them to do.

 

[diplomat33]

Wayve gets a new president.

Erez brings to Wayve a remarkable career spanning 20 years at Mobileye, where he contributed significantly to developing advanced driver assistance systems (ADAS) and automated driving (AD) technologies. With expertise in Computer Vision, Erez ascended to key leadership positions, including Executive Vice President of Product and Strategy and Vice President of Intel Global. During his tenure, he led the company’s product portfolio and strategy, driving the deployment of ADAS in over 120 million vehicles, pioneering L2+ solutions with automotive partners and scaling this architecture to lay the groundwork for its L4 mobility-as-a-service business unit.

https://twitter.com/x/status/1768200665312993438

You can read Erez' blog on why he thinks the new e2e approach is best suited to solve the long tail problem.

Solving the long-tail with e2e AI: “The revolution will not be supervised”

Discover the power of Wayve's e2e Embodied AI in solving the long-tail problem for automated driving and ADAS systems. Wayve's President shares insights on how this AV2.0 technology can reshape the automotive industry.

wayve.ai

I am most curious what actual products Wayve is able to deploy. Does Wayve license their Driver to OEMs as an "eyes on" self-driving system? Do they deploy driverless? I suspect since Erez comes from an ADAS background, that Wayve will look to license the Wayve Driver as an "eyes on" self-driving system on consumer cars. They could provide a "FSD beta" like self-driving system to carmakers who don't have anything like that yet.

That will be interesting to watch.

 

[mspohr]

Wayve gets a new president.

https://twitter.com/x/status/1768200665312993438

You can read Erez' blog on why he thinks the new e2e approach is best suited to solve the long tail problem.

Solving the long-tail with e2e AI: “The revolution will not be supervised”

Discover the power of Wayve's e2e Embodied AI in solving the long-tail problem for automated driving and ADAS systems. Wayve's President shares insights on how this AV2.0 technology can reshape the automotive industry.

wayve.ai

I am most curious what actual products Wayve is able to deploy. Does Wayve license their Driver to OEMs as an "eyes on" self-driving system? Do they deploy driverless? I suspect since Erez comes from an ADAS background, that Wayve will look to license the Wayve Driver as an "eyes on" self-driving system on consumer cars. They could provide a "FSD beta" like self-driving system to carmakers who don't have anything like that yet.

That will be interesting to watch.

Interesting. Looks like they are adopting Teslas approach. But, they have far less training data.

Unlike traditional methods, e2e Embodied AI is built on the premise of learning how to drive from rich and extensive exposure to recordings of driving in various environments end-to-end—from sensing, as an input on one end, to driving actions, as an output on the other. Importantly—it does so in a self-supervised manner.

 

[diplomat33]

Interesting. Looks like they are adopting Teslas approach. But, they have far less training data.

Unlike traditional methods, e2e Embodied AI is built on the premise of learning how to drive from rich and extensive exposure to recordings of driving in various environments end-to-end—from sensing, as an input on one end, to driving actions, as an output on the other. Importantly—it does so in a self-supervised manner.

Wayve was doing e2e before Tesla. So really Tesla is adopting Wayve approach. But yes, Tesla has more data. That is why Wayve is using synthetic data to make up the difference.

 

[diplomat33]

Zoox update.

They are expanding their ODD:

First off, we’ve expanded our Las Vegas geofence: the virtual boundary we operate our purpose-built vehicles in. This geofence spans approximately five miles from our Las Vegas headquarters to the south end of the Strip along multiple routes. It’s larger and more complicated, with three-lane roads, required lane changes, unprotected right turns onto high-speed roadways, and double-right and left-hand turn lanes.

Secondly, we’ve expanded the autonomous driving capabilities of our robotaxi in Las Vegas and Foster City to include driving at speeds of up to 45 mph, in light rain, and at night.

While our robotaxi can reach 75 mph, we started out driving routes with limits of 35 mph last year and are now driving routes with a higher limit of 45 mph.

They are working to improve their stack to handle more complex scenarios, nighttime and rain:

Our perception team had to overcome the challenge of reliable long-range detections when making unprotected turns onto high-speed roads. And to navigate the new geofence, we had to tackle lane changes, which require the AI stack to simultaneously plan the trajectory in space and time.

For nighttime driving, we needed to be able to detect unclear shapes in darkness. This requires training and refining our machine-learning models, based on data from both our test fleet and our robotaxis.

Rain can be tricky for AVs as it could create a distraction. Our Lidar system, for example, may pick up reflections from droplets and puddles. So we had to make sure our vehicle’s Perception system was robust to such lidar distractors.

The next steps for our robotaxi in Las Vegas, Foster City, and beyond

Ground-up autonomy. Zoox isn't an updated car—it's a whole new form of transportation. With every detail designed around riders, not drivers.

zoox.com

 

[EVNow]

Zoox update.

They are expanding their ODD:

Are they just testing or actually carry passengers like Waymo ?

 

[diplomat33]

Are they just testing or actually carry passengers like Waymo ?

Just testing right now. They hope to start carrying passengers later this year.

 

[EVNow]

Just testing right now. They hope to start carrying passengers later this year.

So, Waymo remains the only one carrying passengers for now .... since Cruise is grounded.

 

[diplomat33]

So, Waymo remains the only one carrying passengers for now .... since Cruise is grounded.

I think Motional might be carrying passengers in their robotaxis in Las Vegas but with a safety driver. In terms of driverless, yes, Waymo is the only one carrying passengers for now in the US. Outside of the US, China has driverless robotaxis carrying passengers.

 

[spacecoin]

I think Motional might be carrying passengers in their robotaxis in Las Vegas but with a safety driver. In terms of driverless, yes, Waymo is the only one carrying passengers for now in the US. Outside of the US, China has driverless robotaxis carrying passengers.

I don’t think Motional will make it, but we’ll see this year:

EXCLUSIVE: Hyundai-backed Motional buys time with a bridge loan

Motional, the autonomous vehicle startup, has secured bridge financing to keep the company operating as it seeks longer term funding.

techcrunch.com

 

[diplomat33]

I don’t think they will make it, but we’ll see this year:

EXCLUSIVE: Hyundai-backed Motional buys time with a bridge loan

Motional, the autonomous vehicle startup, has secured bridge financing to keep the company operating as it seeks longer term funding.

techcrunch.com

No, I don't think Motional will make it either. They are too small and not scaling.

 

[spacecoin]

No, I don't think Motional will make it either. They are too small and not scaling.

Yeah getting ArgoAI vibes with owners in the auto industry too.

 

[heltok]

Some cool pics from Xiaomi:s new system:

So an occupancy network. Then it doesn't seem like they do hdmaps, rather than that the NN is outputting the lane connectivity.

Basically it seems that they implemented Tesla autonomy day 2022, but slightly shinier version of it. Imo wp for a first attempt. Will convince a lot of uninformed buyers to buy their product. VW, Toyota etc should be ashamed that Xiaomi could do what they could not.

 

[JB47394]

VW, Toyota etc should be ashamed that Xiaomi could do what they could not.

The Chinese don't have Not Invented Here syndrome which accompanies the hubris of being an established brand.

 

[ThomasD]

Wait until the Criminals figure out that the cars will not run them over and take off when they bust the Windows out and rob people.

 

[mongo]

Crash investigation with possible Blue Cruise involvement:
US to investigate Texas fatal crash that may have involved Ford partially automated driving system

 

[Dewg]

Crash investigation with possible Blue Cruise involvement:
US to investigate Texas fatal crash that may have involved Ford partially automated driving system

Yet another in a long list of reasons the media absolutely sucks today. I read the article. Where are the facts?

A Ford crashed into a Honda, which was on the side of the road, and the Honda driver was killed. That's the end of the facts. No where does it say that the ADAS system was in use. So the entire article is just speculation and click-bait. I love (with EXTREME sarcasm on "love") articles that make liberal use of "may have" and "the facts are not clear". SO WHY ARE YOU REPORTING THIS AS NEWS?!?!?!? You don't have all the facts, so guessing now qualifies as news?

Gee, I guess I'll be a reporter! BREAKING NEWS: There may have been alcohol in the Ford driver's system at the time of the crash. And the driver may have been legally blind while behind the wheel. The facts are not clear at this time.

*sigh*

 

[mongo]

Yet another in a long list of reasons the media absolutely sucks today. I read the article. Where are the facts?

A Ford crashed into a Honda, which was on the side of the road, and the Honda driver was killed. That's the end of the facts. No where does it say that the ADAS system was in use. So the entire article is just speculation and click-bait. I love (with EXTREME sarcasm on "love") articles that make liberal use of "may have" and "the facts are not clear". SO WHY ARE YOU REPORTING THIS AS NEWS?!?!?!? You don't have all the facts, so guessing now qualifies as news?

Gee, I guess I'll be a reporter! BREAKING NEWS: There may have been alcohol in the Ford driver's system at the time of the crash. And the driver may have been legally blind while behind the wheel. The facts are not clear at this time.

*sigh*

It wasn't on the side of the road, it was in the middle lane.
One might assume that AEB and FCW may have both failed to alert the driver, so that part of driver assistance is also part of the reason for investigating.
"NTSB is investigating this fatal crash due to its continued interest in advanced driver assistance systems and how vehicle operators interact with these technologies,"