Rivian R2 with Lidar = Waymo?

[tpberding]

This question calls for speculation based on a few facts. Will an R2 with Lidar function like a Waymo? I ask because, as an older person, could we eventually rely on an R2 to get around when driving becomes more difficult? If so, would that potentially expand the R2 market?

Sponsored

 

[ElGuano]

Waymo has like 7-8 lidars, multiple radars, and a dozen cameras. It's a sensor rig with a vehicle attached to carry it all around. R2 is going to have a single lidar.

I think Rivian is aiming for "universal eyes-off" hands-free driving. In reality, it's possible they never get to "full self-driving" start to finish with that tech.

Waymo is already there, and operating in multiple cities reliably, 24/7.

Tesla is pretty damned close and looks like they'll get there. Their full-self driving is achieving some impressive milestones (like a one-off coast to coast drive) but we don't know how much they need to do to get it *reliable and repeatable*.

My realistic guess - Rivian will never get to true FSD with R2, just like Tesla never got to FSD with its anticipated hardware, or the hardware after that. If you want to sit in a robotaxi without touching the steering wheel soon, you'll have to look elsewhere.

 

[UnsungZero_OldTimeAdMan]

Eventually, point to point handsfree is possible. When Rivian will reach that point? No one knows. Autonomy day demo was done on a pre-planned route. It would be unrealistic to expect full capability right away and everyone should be cautious. Waymo has a lot more development time and still isn’t without flaws. It’s not like their tech is open source and Rivian could pick it up and be its equal immediately.

 

[mikehmb]

(warning - computer vision nerd checking in)

There is an ongoing debate as to the efficacy of cameras as the primary vision systems used in ADAS++ (aka Self-Driving) applications - specifically, in this example, L4 autonomy. The R2 G3 will have lidar + cameras + other sensors, and that's great, but the Waymo (and other) application has lidar at every corner plus a top-mount, radar, and cameras which essentially gives it FAR greater disparity information than is available with a single lidar. This is why it can operate at L4 and do so relatively seamlessly*.

While technically possible, there are inherent limitations of using a single directional lidar with cameras. They can be overcome, but the amount of compute - and data - required, not to mention the efficacy of the perception stack, is not a small technical challenge.

I think this can get them to L3 autonomy, but L4 is a different league.

* - Waymo is still finding corner cases. Those will continue to exist, which is why they have backup tele-op.

 

[UnsungZero_OldTimeAdMan]

The other reality is legal liability. Who is responsible? The company that made the autonomous car or the driver/owner? We know Rivian to be very conservative in their risk-management, understandably so. And the legal responsibility question has yet to be answered definitively in the legal system. While Rivian may be talking up potential capabilities now, don’t take it to be a promise on product. The two things are not the same. Changing legal landscape could easily change how much autonomy capability they are comfortable putting in customer hands.

 

[savethemanual]

What's wrong with using a good old fashioned Homo Sapien driving for a car ride service?!? I get it that these giant corporations need to always make more money so a human is a liability, but at what cost? I know there are full time drivers who love that type of work.

 

[iansriv]

What's wrong with using a good old fashioned Homo Sapien driving for a car ride service?!? I get it that these giant corporations need to always make more money so a human is a liability, but at what cost? I know there are full time drivers who love that type of work.

Because a human would most probably not do this:

NTSB investigation

 

[ndmiller]

No

 

[Nixapatfan]

Full self driving will not happen anytime soon especially on a vehicle with single lidar sensor. Even Waymo with all its computers and sensors is really limited to decent weather conditions only throw any curveballs at it and it goes haywire like during the power outages recently. For a fully autonomous car it would need to demonstrate confidence in all conditions which the technology is still a longways away.

 

[BigSkies]

No. Waymo is moving towards custom built vehicles for their own purposes.

Tesla's deceptive marketing has made people think there's overlap between the capabilities of a car made for being a taxi and a car you buy to drive yourself around. There will not be overlap between these for many years, if ever.

 

[skyguyscott]

I have often said that the current state of the art in hands-free driving is that a 15-year-old with a learners permit will do as good or better job than some of the autonomous vehicles that have been running into emergency vehicles, pedestrians, and doing other bizarre things that most 15-year-olds would likely avoid.

Now it's possible that some autonomous vehicles might be doing a better job than a particular inattentive 15-year-old might do, but the real question is how comfortable are you letting a 15-year-old with a learners permit drive your brand new very expensive electric vehicle?

I expect that this will eventually improve, but I do find it ironic with all of the hype about how miraculous AI is, accidents involving AI driven vehicles continue to happen, and some of these accidents are ones that a 15 year-old would likely have avoided

 

[Just Passing By]

This ...

Waymo is still finding corner cases. Those will continue to exist, which is why they have backup tele-op.

... and this ...

The other reality is legal liability.

People continually underestimate corner cases and that's in the US. Driving conditions in European cities as an example can be far more complicated and not just due to older city layouts. Think I'm exaggerating? Then look at the Magic Roundabouts in the UK with video of Swindon in action. The Arc de Triomphe roundabout in Paris is another example requiring complex decision making but there are many many more everyday examples.

The other challenge is that fully autonomous vehicles are held to a higher standard than human drivers, at least at the moment. One death caused by an autonomous vehicle is national headline news, while ~40k deaths per year in the US associated with human driven vehicles hardly register in the collective psyche and are apparently considered acceptable, since nothing much changes with this statistic. Legal liability, whether held to be personal or especially corporate, exacerbates this situation.

 

[therealcmj]

Because a human would most probably not do this:

NTSB investigation

Except people do it all the time. Which is why states are rolling out camera enforcement.

https://www.ncsl.org/transportation/state-school-bus-stop-arm-camera-laws

 

[NoMoreGas71]

Full self driving will not happen anytime soon especially on a vehicle with single lidar sensor. Even Waymo with all its computers and sensors is really limited to decent weather conditions only throw any curveballs at it and it goes haywire like during the power outages recently. For a fully autonomous car it would need to demonstrate confidence in all conditions which the technology is still a longways away.

So you are questioning Elon Musk? They getting ready to roll out Cybercab with no controls using vision only.

Sponsored