R2 has Lidar!!!! First look at windshield notch location on R2 (with R2D2 wrap)

[Dark-Fx]

I dunno. Rivian’s demo in the presentation yesterday showed the importance of lidar data, even in what seems low speed driving.

Philbin says lidar is part of the “trinity” of data that provides greater perception and depth to the environment around the car.

I can see front lidar only being plenty sufficient for Level 3 driving. But Level 4 seems like a higher standard, which Rivian seems to claim they can do with just front lidar. Whereas Waymo says it needs 360 lidar.

It seems like either Rivian is overpromising or Waymo is being too cautious

Two different level 4 systems could still have vastly different situations that they are able to handle. The important part of L4 is that it will pull over or stop safely in a situation it knows it cannot handle.

Sponsored

 

[DuoRivians]

Two different level 4 systems could still have vastly different situations that they are able to handle. The important part of L4 is that it will pull over or stop safely in a situation it knows it cannot handle.

I suppose. But both systems are meant for taking people on the same surface roads and use cases. It doesn’t seem like the Level 4 driving outcomes should be so vastly different.

 

[Dark-Fx]

I suppose. But both systems are meant for taking people on the same surface roads and use cases. It doesn’t seem like the Level 4 driving outcomes should be so vastly different.

Let's take RJ's example of dropping your kids off at school.

This is something you as a driver are going to do regularly. You might need to train your own vehicle on exactly how to handle this situation, to enable it to work on its own in L4. It would still be L4, even if it's on rails.

 

[DuoRivians]

Let's take RJ's example of dropping your kids off at school.

This is something you as a driver are going to do regularly. You might need to train your own vehicle on exactly how to handle this situation, to enable it to work on its own in L4. It would still be L4, even if it's on rails.

I’m not sure how this is a good example. Even if the route is the same, there could be a million of unknowns that happen along that route that require real-time responses by the vehicle.

If you’re saying that all the vehicle needs to do is to stop and pull over when faced any unknown, that seems very non-robust. I would imagine there would a standard for which vehicles can properly respond and recover to complete their journeys.

 

[portdirect]

It seems like either Rivian is overpromising or Waymo is being too cautious

I’ll try to be kind to Rivian: they are outlining their intended roadmap, none of this stuff works in a production suitable form today. You could say they are over-promising, but I think it’s more educated optimism at this point. How gen2 rolled out probably provides a good indicator of where they/we are:

1. They have announced product; following enough dev/lab work that they are now confident they can make it work in a form they can sell. (Late 2025)
2. They will ship hardware in product as soon as it’s working to parity with the prior generation. (Late 2026)
3. They will then finish the software in the hope (with a reasonable level of confidence they can) at some point after the hardware starts shipping. (Late 2027)

 

[DuoRivians]

I’ll try to be kind to Rivian: they are outlining their intended roadmap, none of this stuff works in a production suitable form today. You could say they are over-promising, but I think it’s more educated optimism at this point. How gen2 rolled out probably provides a good indicator of where they/we are:

1. They have announced product; following enough dev/lab work that they are now confident they can make it work in a form they can sell. (Late 2025)
2. They will ship hardware in product as soon as it’s working to parity with the prior generation. (Late 2026)
3. They will then finish the software in the hope (with a reasonable level of confidence they can) at some point after the hardware starts shipping. (Late 2027)

Sure, they can always develop more software. But only committing to front lidar AND saying this will be sufficient for Level 4 driving seems like a hard line in the sand.

 

[Dark-Fx]

I’m not sure how this is a good example. Even if the route is the same, there could be a million of unknowns that happen along that route that require real-time responses by the vehicle.

If you’re saying that all the vehicle needs to do is to stop and pull over when faced any unknown, that seems very non-robust. I would imagine there would a standard for which vehicles can properly respond and recover to complete their journeys.

Drop-off lines can be kind of complicated, so that's where my expectations went. It's something that could be inferred from the behavior of other vehicles, but what if your Rivian is the first one there?

And yeah, L4 is definitely not considered hardy to every problem. If it was, it would be L5.

 

[captainjp]

Sure, they can always develop more software. But only committing to front lidar AND saying this will be sufficient for Level 4 driving seems like a hard line in the sand.

Well, there’s this other company that thinks that L4 is achievable with a camera-only system

 

[macb00kemdanno]

Kyle has a very grounded take on what he saw. Bottom line, it’s not even close to Tesla FSD 14.x. His point being, if you’re going to come to market or showcase self driving tech, it needs to at least be on par with what’s already available.

 

[AlphaSnowbordergirl]

Kyle has a very grounded take on what he saw. Bottom line, it’s not even close to Tesla FSD 14.x. His point being, if you’re going to come to market or showcase self driving tech, it needs to at least be on par with what’s already available.

Did not like the review honestly. Kyle does show his Tesla fanboy side a bit, but he is comparing something that's only been turned on for 3 months and not set to lunch for another year (a lifetime in software years) to something already finished and out for consumer use. While I agree you cannot bank on promises, its hard to judge what is essentially a prototype with something already out for consumer production. While certain things do deserve critique, constantly saying, it's not up to Tesla standards and literally commenting about tesla products driving by and zooming in on them during your video is showing bias. It comes off as "if you can't be right at tesla's level, then why you even here" rather than looking at it as what it is, another car company entering the space, trying its own way of approaching the problem, and offering an alternative to people who don't want a tesla.

 

[captainjp]

Kyle has a very grounded take on what he saw. Bottom line, it’s not even close to Tesla FSD 14.x. His point being, if you’re going to come to market or showcase self driving tech, it needs to at least be on par with what’s already available.

I didn’t get the impression that this was intended as coming to market. This was a demo proving that it is indeed achievable with current gen2 hardware to have PTP. In addition to this milestone, the more exciting announcement is the introduction of their own AI chips and LiDar as part of the Gen 3 hardware. I anticipate that when this comes to market, it will make FSD look like a granny behind the wheel. FSD has pretty much plateaued at this point, or is close to it with current hardware arrangement.

 

[elektrode]

but he is comparing something that's only been turned on for 3 months and not set to lunch for another year

Part of his point is “so is everyone else”

 

[R2D2TOO]

They need to offer this as a color option.

Of course I love that wrap as well, but I won't be doing that on mine. I'm hoping for a new red.

Sponsored