AAH #719 - Autonomous Cars on the Verge of a Breakout
Annotations will appear as you listen
0:00
61:39
LIVE
Speaker 1: Auto Line after Hours. It's brought to you by Bridgestone Tires,
Solutions for Your Journey and by Borg Warner. The automotive
industry continues to evolve, and so do the opportunities to define it. Borg Warner, one of the world's most admired companies,
gets partners where they need to go. Let's do something
big together.
Speaker 2: All right, kind for another show. All yah, I'm doing great.
I'm doing you too, breathing and breathing out. Okay, yeah, yes,
whole thing. So before you stump us, let's let the
audience know. We've got Henry Liu here. He runs this
operation called Mcity, which is like this mobility autonomous vehicle test track at the University Michigan. We're going to be
talking a lot about EV's I know you're very skeptical about him, and you know that I am like an Avy proponent, right, so this is going to be interesting.
And we also have Michael Robinett from SMP Global here that's going to be talking about all kinds of things about the industry.
Speaker 3: So I know you introduced both of those guys because you're gonna want help on this one.
Speaker 4: Yea.
Speaker 2: Gary always stumps us with the question about something that happened in the industry usually on this.
Speaker 3: Is always on this date.
Speaker 4: Always.
Speaker 3: It would be very easy were it not on the state This is okay, So November fourteenth, nineteen fourteen, a vehicle was introduced by a company that still exists, but a company that is arguably perhaps not long for this world.
So what was the company? And you'd get like, and
I know, you know all about these old cars, and if you if you were to identify this this car, it would be just flabbergastingly amazing.
Speaker 2: Well, nineteen fourteen, I mean, Europe is in the middle of World War One this point, so I'm guessing car companies are not introducing cars. But is it a car?
It is a car, Okay, it is a car, because I was going to say, nineteen fourteen, maybe it's the Renault tank came out then, but no, it's so I'm guessing it's got to be an American company. Garry's not
given out.
Speaker 4: Barely, hes not in the world of giving a clues.
Speaker 2: So it's a company that's still around but on teetering.
Holy moly, you're gonna guess it.
Speaker 4: May still actually may still very well be Europe. Maybe
maybe that was just before the word started, I think.
Speaker 2: No, August nineteen fourteen August. I know this very but
if you think about it.
Speaker 3: I mean, when World War One started, it wasn't like the entire western Europe.
Speaker 2: No, that's true.
Speaker 3: It was covered with that's true, but I would ignore.
I just said that, Okay, okay, okay.
Speaker 4: Hence it's European not long for this.
Speaker 3: That's one interpretation.
Speaker 2: Well let's see, I mean, yeah, yeah, who's teetering on the brink right now?
Speaker 3: Well, teetering might be an exaggeration.
Speaker 4: Wow, could be one of the Jaguar Jaguar land I'm not serious. They're not teetering.
Speaker 2: But no, but they're not selling cars right now, Jaguars. No,
but he's said it was okay, and it was.
Speaker 4: I'll the more clues.
Speaker 3: It was built in one of the most amazing facilities in all of automotive history.
Speaker 2: Well, then it's got to be Highland Park.
Speaker 4: Well, no, Rouge was after that.
Speaker 2: It was after that, so Highland Park nineteen. Actually Highland
Park had started this.
Speaker 3: Well, the factory started before the vehicle, all right, Okay, the audience is getting bored with this. It was the
Dodge Model thirty oh.
Speaker 4: Okay, that was there their first vehicle.
Speaker 3: And so the amazing thing is the Dodge main plant.
And so this was on sixty seven acres. It had
its own power plant, it had his own medical facility, it had his own fire department, it had its own power generating system. I mean, this had everything. I mean,
we think about the Rouge in that context, but Dodge Maine basically had the same thing. And because this is
online after ours we won, doubtedly talked about Tesla several times before the show is over. And so remember that
we were very surprised and amazed at how quickly Elon was able to get plants up and running. So this
was astonishing I think is that they broke ground for the plant in June nineteen ten and they were producing parts by November.
Speaker 2: That's extraordinary.
Speaker 3: Well, admittedly they were making parts the arts, not cars, right, and so you know, they ended up tearing down the plant in Factory zero, the former gim Detroit hand Trammick.
Speaker 4: Yep, that's on the footprint.
Speaker 3: It's is on the part of footprint. Actually, the gem
bought more Land even made it bigger, so it's it's still sort of with us today.
Speaker 2: Yeah, you know, that's uh was in a very Polish area of Detroit ham Traamick, and the old Poles used to pronounce it not Dodge but Dodge gut, and that's what they would say, I work a Dodge comin.
Speaker 3: And actually it was because of that plant that many Polish immigrants came due to that area, and so basically the Dodge brothers responsible for that. But it's interesting you
think about the Dodge brothers, you know, died in nineteen twenty and Chrysler ended up buying the plant in nineteen twenty eight from Dodge.
Speaker 4: But wanted to write a book about the ten most influential automotive plants and some most of them aren't around with us anymore. But you know, Dodge Main would be one,
Packard would be another. You know, I would say up
in Ashawa, Ontario, that whole auto plaques that started in the twenties, in the thirties, that would be another one.
I mean, there's there's a whole slew of them that just really helped build the industry, and even some over and over in Europe outside of Paris, where Renault used to build most of their vehicles right on the edge of Paris there's all kinds of really interesting facilities that unfortunately aren't with us anymore.
Speaker 3: So do you do you see facilities like that being built today or having been built within the last ten twenty years.
Speaker 4: Well, they used to build facilities, used to be like five six stories tall, and they don't do that anymore because it's not very efficient. But I would say most
facilities are built in modules. Is you got a hundred
or one hundred and fifty thousand unit module, will grow that a little bit, maybe a two hundred and fifty thousand module. That's the way that the most the automotive
planners think about automotive plants now. So the days of
building a fifty thousand unit module, you almost do. Don't
do that because your pink chop is not efficient. You
got it's all about the paint chop all the time.
And that's where most of the planners will tell you that's right.
Speaker 3: It was interesting that they actually did trycoding at the I mean, you could get a vehicle with three colors on it. I mean, just imagine how difficult that would
be to paint.
Speaker 4: Well, you gotta run it through three times, yeah, I mean masket running through masket again. Running through and I
don't know how straight the masking was, but it got done.
Speaker 2: So well, you know, think about it. I don't want
to dwell on this all too long. We want to
get into the modern stuff. But you know they had
separate fenders front and rear, left and right, you know, different things, so you probably didn't have to run the car through the line three times. You just said, hey,
put red ones on that one, you know, put green ones on the next. And you had your different color matches.
Speaker 4: All right.
Speaker 3: So one more thing about that plant. So it was
interesting to see some of the name plates that came out of there. Now there was the Plymouth Valari was
built there, and the Plymouth Barracudu was built there. But
I'd completely forgotten that there was a Dodge Saint Regis.
Speaker 2: Yes, I don't remember that.
Speaker 4: It was a four door off of the basically off that diplomat platform, off that platform, and there was the.
Speaker 3: This goes back to nineteen forty nine to nineteen fifty four, so we may not remember this. The Dodge meadow Brook No, Nope,
Wayfarer Custom Deluxe. But it's it didn't last very long, No,
they didn't.
Speaker 2: So anyway, onto av's. Yeah, let's go on to autonom's vehicles,
and so, Henry, we're seeing slowly but surely more avs becoming more available in different cities around the country. Waymou
just announced a big expansion or.
Speaker 3: It was in La that they said, right, you used to have to be on a list.
Speaker 2: Now if it's right, Main mobility is talking about getting into more including ann Arbor. And but what you're saying is, hey,
wait a minute. Right now, every one of the fifty
states has its own regulations about autonomous vehicles or no regulations at all. You're saying, what the United States needs
right now is a national standard. Is that right?
Speaker 5: I wouldn't say this is national standard. I would say
this is a national guidance on national framework for autonomos vehicle deployment.
Speaker 6: I do think we.
Speaker 5: Are at the cups of large scale automa's vehicle deployment because the technology issues are being feedured out and so you can see there are you know, cities best cities in terms of deployment these autonomous vehicles, and so autonomous vehicle has been you know, on the roller coaster in the last like twenty years.
Speaker 6: Also, right, So but.
Speaker 5: If you if you look at in terms of the given the advancement of artificial intelligence, much learning, I think we are almost there in terms of figuring.
Speaker 6: Out exactly what the technology is.
Speaker 2: So hold on when you say what we need is guidance, what do you mean? What what are you looking for?
Speaker 6: Yeah?
Speaker 5: So what I mean by that is we I think if you look at from basically from I would say two tiers one is more and more companies are deploying.
There are townless vehicles on the street, and we see we definitely see benefit. But at the same time, we
also want to ensure these atomomous vehicle are be deployed safely, does not become public hazard.
Speaker 6: We want to ensure that.
Speaker 5: So there's needs some type of testing standard the basic behavior competency of these vehicles.
Speaker 2: So you're you're talking about essentially a driver's test exactly ad.
Speaker 5: System for AVAY systems because in the US we've been relying upon self certification on vehicle safety. Right, so essentially,
as long as you don't create problems, then you know you can deploy.
Speaker 6: However, we do licensed drivers.
Speaker 5: Because previously what we have been focusing on in the automotive industry is functional safety. As long as you can drive,
drive safely, and then we ensure you have active safety and protect passive safety type of functions. But now we're
rually adding drivers. But this is a computerise vers a
neural network based drivers into it.
Speaker 6: But we do not have any guidance of framework.
Speaker 5: To ensure the safety of these drivers, particularly the basic behavior competency.
Speaker 3: Okay, but we may not have a de facto rule, but we have adjure rule in that if I'm weim it isn't my best interest not to have my vehicles run people over right, And of course, and when we did have a situation where crews had the problem with running and dragging the person that they ended up getting in big trouble for that. So I don't understand why
we need to create an additional framework. When the companies
that are involved in developing this technology and putting it on the road, it is in their fiduciary interests.
Speaker 5: It is in their best interest to be safe. It
is in their best interests. However, we also see from
our experience internal testing of some of the commercial available vehicles they do have, I would say some of the scenarios they do run into problems and they are being deployed.
Speaker 6: So so that's why.
Speaker 2: This is why we have regulations.
Speaker 5: So well, that's why we we we advocate for some of the basic behavior competents. For example, I want to
make sure all the vehicles being deployed they can safely follow other cars, they can safely do the unprotected left turns, they can merge into the freeways or merge into you know, the road. And and also sometimes if you often see
those type of situations as well, and you know you're driving on one lane of the road, but opposing traffic is sort of in a round maybe, and you want to be able to enunce type of situation, you want to be able to react to it properly, and so you need some basic you know, behavior competence, you be, basic scenarios for you know, for a human right where we want to make sure that Ammo's vehicle also has the capability to handle that.
Speaker 2: Can you do this test virtually or do you think, i e. You know, test it on a computer, you know,
plug your your stack, your av stack into a digital twin and test it that way, or do you think that you need to have physical models being tested on a track.
Speaker 5: I would say you need both because majority of the tests should be done in the digital world, I would say ninety nine point ninety nine percent, but nothing can replace the physical test at end. You will need still
need to have some physical tests. And so that's why
the test track is exist because then we can design these type of scenarios to do the basic behavior competency test.
Speaker 3: So this would would would this be something that knitso would administer or who I mean, who do you see being the owner.
Speaker 5: The owners Yeah, yeah, yeah, yeah, Nizza, it certainly has the authority to do that. But they can also delicate
to serve party testing agencies of testing organizations to do that.
So there are serve party testing organizations exists and they can do some of these two. And so John just
want just to sort of response to your question. I
also advocate not only just for basic behavior competency tests in terms of ensuring these are not creating public you know hazard. As you know Garant mentioned, you know, these
companies do have they're in in their best interest not to create any hazardous situation.
Speaker 6: On the other side of the story.
Speaker 5: When way miles, you know, crews, these leading autonomous vehicles, leading autonomous vehicle companies deploy and you also see there are you know, at least from two groups. One is
the agencies, the traffic management agencies, for example, when they deployed in San Francisco or Los Angeles. I'm sure the
agency also have questions that the doot, the city doots.
They also want to know when these vehicles are deploys on my street, how does that impact my traffic safety, how does that impact my you know, mobility, does that create congestion? Things like that. And you also oftentimes also
hear from regular consumers this is this is fully safe and how you know, how we can ensure this is safe?
Speaker 6: So you also see right now, you also.
Speaker 5: See some of the companies start to work with party testing organizations two provide more independent evaluation of the safety, trying to you know, provide these type of independent evaluation to give more confidence to consumers as well as the public agencies.
Speaker 6: So there's two sides of that good.
Speaker 2: Pr I guess this would be fairly easy to put together in the sense that I've been hearing about this complaint.
We need a standard that everybody can build too. Yeah,
and but am I right? I mean there's been a
lot of talk about it. I'm sure there's a lot
of ideas of what that kind of test or guidance should be. Is it just a matter of getting the
political will to do it or what are your thoughts?
Speaker 5: Well, unfortunately, it's not so easy. I have to say
that it's actually quite difficult. If it's easy to do,
I feel that my we have we probably already built some sort of consensus on how to do the test.
The problem is really it's not easy, and the main problem is really we rely upon these machine learning models and machine learning models itself regardless, how do you you know, design these things easier. You put it into you know,
perception function, planning functional, you put it all together, become an end to und type you know machine learning models.
These models does have capability to handle very complex, you know situations, but not uristly, these models oftentimes create problems in these rare, low probability safety critical situations.
Speaker 6: The so called edge so called educases exactly.
Speaker 5: And then so I actually have a paper this year earlier this year, I call it crucial raty.
Speaker 6: It's truly a curs.
Speaker 5: To these automos vehicles because once you up a bar magnitude in terms of your safety you know, performance, then the cases become rare. The data collection, the effort, the
costs is one magnitude higher. So it's easy. Now it's
very easy to get you know, on average in terms of these vehicles can navigate through that, you know, a townis lay. Some of the students can do that now.
But if you want to provide a commercial product to be able to rule and navigates through complex, you know, urban driving situations, it's just very difficult. And I heard
many of the different stories, you know, you know, one of the story I heard that when some of the vehicles, this is in San Francisco being deployed and they're following another truck. The truck has a tree, Christmas tree in it. Well,
the vehicle is being treed. If you have if you
see a tree, you need to stop. But that truck
carry that tree. And so when the truck move away,
the descence becomes longer than this.
Speaker 6: Vehicles start, you know, pulling forward.
Speaker 5: But then when the truck slow down and you know there's this is a stationary object, I have to stop.
Speaker 2: And so it's because it thinks it's seeing a tree, realizing it's a tree in a truck.
Speaker 6: Yeah, So that those are the things. It's not.
Speaker 5: You don't see that that often, and so that's why you don't get that a lot of the data. And
that's also why I think, you know, at the beginning, like twenty fifteen, if you road dial back ten years back, everybody want to invest in a way because we think the deep learning, ourgorism will solve everything. You know, we
have the argorism, we have the technology. Now let's just
throw it out engineering development and then make it into mess production. That's why ten years ago many companies you know,
announced that. Then they find out this age case or
corner case situations, this cursed word, the.
Speaker 6: Things kick in.
Speaker 5: They find, oh, I need so much data to handle some of the situations. Well, the even more difficult situation
is what we find is that they also have these machine learning models also have the so called seesaw effect. Okay,
I find my model have difficulty to handle one type of cases. Okay, now let's collect more. We got more
data on that type cases. No, hunt, the model is
trained and they can handle this type of situation. Now,
well another problem pops up, and sometimes sometimes even those are the problems they can handle.
Speaker 6: Before. After you're between this, it.
Speaker 5: Might become problematic, like whack them all exactly exactly. So
that's that's why the problem it's it's it's so difficult.
Speaker 6: It's it's not it's not.
Speaker 5: So easy, particularly when you up to you know, the safety performance become nineteen nine point ninety ninety nine nine.
Speaker 2: Then, but it's very difficult, and there's all these edge cases, and you never know when an edge case is going to show up that no one ever thought of before.
And yet you started out by saying we're not on the verge of seeing widespread TV deployment.
Speaker 1: So what your.
Speaker 6: Guys is that.
Speaker 5: That's because truly I am very optimistic in terms of I would say artificial intelligence developed advancement. You know, we
start rule leads become full of in the commercial world, autonomous vehicle attract commercial investment really because of deep learning algorisms.
That twenty thirteen, fourteen years and then there's this major development in terms of transformers and and that. But once
one not in terms of the motion learning algorism. Now
in the last two years, particular the generative AI, the foundational models, and they have the ability to not they have the ability to learn I would say, the driving skills rather than mimic how people drive, and so that's very different.
Speaker 6: That's very different. That's very different.
Speaker 5: So the deep learning algorithm, the transformers befall, they're trying to mimic how people drive. Okay, I see the situation.
This is how people drive. Okay, now I am, you know,
trying to essentially mimic how people drive. And you, guys,
you can imagine most of the cases we see is sort of like discreet cases. You see this, Okay, I learned,
you see this. I learn no, I know both. Sometimes
sometimes you know, things happen in the middle you have not really seen those. So the world itself is certainly
continuous and it's not discreet. So it's sometimes it's quite
difficult to learn to cover all of the situations. And so,
but human you think about our you know, as a human driver. My son he's nineteen now, but three years
ago when he learned how to drive thirty hours in the driving school, right, and then he started to drive on the public road with parent inside for the first fifty hours. But there's not a lot of training data.
I think about that for human drivers. We don't really
rely upon that much of training data.
Speaker 3: Human drivers would be terrible coming edge cases.
Speaker 5: Well, see, they learn and then they learn and learn and they become you know, experienced forever they can handle some of the cases. And so what I think is
that we have, first of all, we have the general contest.
It's not contextual information. We not really just rely upon
you know how look at how people right. We also
understand you know, the situations in terms of this is Halloween situations and there's pedestrians wearing different costumes, and we have this background knowledge.
Speaker 6: That's that's number one. Number two is we learn skills.
Speaker 5: We don't really mimic how my coach teach me, but I learned skills. And the general to AI the foundational
models I think gave us the capability to lick at those and to train the model to start to learn the skills. In this case, I also see the possibility
to actually reduce the data we need. We may not
need that much of data for us to train the model anymore if we can learn the skills. And so
there's this technology development I think all other ones been.
I feel it gave us the capability. Now I feel
we rually at them.
Speaker 6: You know.
Speaker 2: The last very interesting. So I want to run this
back to make sure that I understand it. What you're
saying is instead of teaching av systems to mimic what other drivers have done, just feed them all kinds of cases and they know, okay, in this situation that driver did this. What you're saying is teaching them basic driving skills.
Speaker 6: Basic driving skills so.
Speaker 2: When they encounter something they never encountered before, they encounter an edge case, they know how to.
Speaker 6: Deal with it exactly exactly.
Speaker 3: So would it be they are driving in a different way than human drivers do.
Speaker 7: Not necessarily, so it could be that it would say, hmm, it'd be more efficient if I were to tailgate this vehicle, I will I will get an arrow advantage from doing that, and plus it reduces the amount of space, therefore there will be less backup.
Speaker 3: But I mean, when I talk to the guys at rg AI, they basically said, no, no, no, we want to train our AI to drive like human beings do exactly.
Speaker 5: They're going to be in the world with exactly so you can you can steal. So again, look at the
interms of human drivers.
Speaker 6: We we knew.
Speaker 5: We will learn when we learn, we're not really mimic in terms of you know, like you know my coach is teaching me. We're really looking at Okay, we look
at the vehicle in front of us. We don't actually
exactly know how far we are from that vehicle. Our
eyes measures the distance, but the distance is not it's father distance.
Speaker 6: We don't really know.
Speaker 5: It's it's that fifty meters of fifty five meters. We
don't really know that, right, But we can still drive with you all. When we get too closed, we need
to slow down, and then we will when when the front vehicle has you know, something you know stops and then we I want to stop.
Speaker 6: Quickly as well. So those are the skills we learn.
Speaker 5: And now this generated AI, I see, we have the capability to learn such things as well. We don't have
to Again, we don't have to with this generally, the AI.
We don't have to measure the distance that accurate because again, people don't act.
Speaker 2: So what you're saying is not only do you need less data to teach the system, maybe you don't need all the sensors and everything too.
Speaker 5: As many Potentially I can, I can, Yeah, potentially I do.
Speaker 6: See.
Speaker 2: So does TESSELA have it right? Do you think can
you do it all with video?
Speaker 6: Well? That's a question of being time.
Speaker 5: Yeah, I mean time people ask me, you know, after I take over I'm CD as director of People, many people ask the same question. I would say, from the
first principle point of view human being, we don't really rely upon light, our sensors and you know, very accurate.
Speaker 2: We use our video.
Speaker 6: Yeah, exactly, exactly. So from long term perspective, I do
see there's a pathway. M hm. Short term I'm a
s capital.
Speaker 3: But I mean, isn't the point for an automated system to outperform the human being? Okay, we're talking about manufacturing before, right,
they came up with Nu Miracle control because they could make parts more accurately than the guy could we see in manual jig boar. Right, So if I'm going to
have an autonomous vehicle, I want the autonomous vehicle to be a better driver than people, not better than job, better than both.
Speaker 5: Yeah, yeah not because if you look at the international accident rate fatality raid on the road, there's a lot of uses because of the driving on their influence, the drugs that are alcohol and are not attentive to driving, reading cell phone, something like that. With autonomous vehicle, you
don't have any of those right, So if we can drive muster the same driving scale, even just muster of the same driving scale as human drivers, they obviously have much better driving performance than human drivers already because there's no influence, there's no distraction.
Speaker 3: What we do know that AI hallucinates, so if be careful.
Speaker 6: There, cluc nation definitely is a problem.
Speaker 3: So they may not take drugs, but it's.
Speaker 5: Well, see, that's that's the thing Cluse Nation generated a lot of these is because the the general the so called a GI artificial general intelligence, and there's there's so much data.
Speaker 6: Need to be sucking to learn.
Speaker 5: Actually, when you use less data, use generative approach, not social weather, that's that will still exist. So we can
see their pathways to overcome that HM and so so so that's why I do feel, given the last two years of a town, I would say artificial intelligence advancement AI advancement, we are at the word, you know, breaking through.
So I'm assuming all the leading companies they I mean, they must know this already, and that's why they really start to roll out.
Speaker 4: I'm interested in this certification or I guess this, you know, setting up guide length. But more if if I've got
a new company, and I want to be able to put my vehicle on the road under certain circumstances. Maybe
you you pass the test that gets you to hear, and then you've got certification. Then unless you change the
vehicle dramatically, you've got license to put your vehicles on these roads under these circumstances. And maybe you take another
test that says, hey, I've been able to hit some of these other edge cases or other capabilities, and then that broadens the capability or or the areas that you could function or you operate your vehicle. I mean to
me that that would add some level of assuredness from the public that oh, that's a level A vehicle, Level B vehicle, or whatever however you score it. Would it
would give people some additional comfort that that vehicle is interesting.
I mean, when you go to Phoenix, you see these way moold vehicles running around all day. Yeah, and you
know some some level of confidence that, oh, that vehicle has been able to pass these particular tests, just like you know, you and I had to take a driving test, same thing. I feel pretty sure that it can handle
most of what's in front of it. If it's on
the road in front of me.
Speaker 6: Yeah.
Speaker 5: Yeah, So so Mike, I would separate these two things.
One is I feel is small regulatory type of frame, which is we want to guarantee basic behavior. So that's
pretty competitive. Everybody should have it. If you want to
deploy on public growth, then above that, then above that in terms of rule, are you better than ten times better one magnived better, two times two magnituldes better, or you know the situation. I can handle that situation. I
cannot handle these. We're getting into more competitive space. So
those I feel we should offer framework in terms of these could be conducted, but you don't have to regulate anything like that. So that's why I always advocate, you know,
we have some basic behavior competence they want to ensure, but the rest of it will provide some guidance.
Speaker 3: So the basic is regulatory, that's what I and and then thrust is frusting on the cake.
Speaker 6: It's more in the competitives.
Speaker 3: But I mean, isn't basically the SAE levels do what you're talking about. I mean, so when you see those
weimos rolling around, somebody has said that.
Speaker 4: That I think that's more descriptive of the capability of the possible capability vehicle, not so much. Can it do
it under all circumstances. Maybe I'm wrong, but I thought
that level format it's supposed to it should be able to operate under these criteria given the content that's on the vehicle. But I don't. But past that, I don't know.
If there's some sort of tests that said, by the way, it's got the content and it and it passes all these tests and it is, you know, certified or something like that, I don't. I don't know the standard goes
quite that far.
Speaker 5: So so Mike, you're right, we are we're talking about only level four vehicles. This is you know, this is
not because it's not because these vehicles, say, past the basic behavior competency, then they's become level two and then past another launch become level three.
Speaker 6: It's told No, it's not. We're talking about only level
four level five. These are all level four, level five vehicles.
Speaker 5: And then for level two vehicles, there are driver assistance systems, so you have a driver taking responsibility whatever you do, that person take responsibility. We're talking about a vehicle without
a driver, so there's no person taking responsibility. So level
three is sort of in the middle, right, so, but really the focus is the level level four.
Speaker 2: It's really interesting to hear you talk about how jen Ai has really accelerated this development and I love what you're saying too. Of that, we're going to soon see
quick deployment of it. Well, we're going to have to
wrap up this segment of the show, but Henry Lou thanks so much for coming on Autoline after ours very interesting discussion.
Speaker 6: Thank you for the emait hes. Yeah, very good, thank you.
Speaker 2: We're going to take a quick commercial break. We're going
to come back talk about more industry news. But first
a shout out to our sponsors.
Speaker 8: Performance that shines even in the rain. That's what really matters.
Rich don't pretends to tires, improved grip and wet conditions.
The automotive industry continues to evolve, and so do the opportunities to define it. Borg Warner, one of the world's
most admired companies, gets its partners where they need to go.
Let's do something big together.
Speaker 2: All right, we're back, but now we're getting into a whole bunch of other things. And Gary, why don't you
kick us off?
Speaker 3: So Michael SNP Global Mobility has been doing a lot of research into the implications of what the Trump presidency will do to transportation. Tell us about that research, So, I.
Speaker 4: Would say, back in August ish we started. We obviously
knew the election was coming, and we knew we would get a number of questions. The campaign had already been
underway well back then at least it was Trump and Biden.
But nonetheless we kind of knew what was coming, and we more or less built for scenarios of different political outcomes that could impact the industry from a trade perspective, emissions perspective, you know, how swift autonomous vehicles could be approved, whole number of What would happen to the TIV total industry volume? What would happen to the PSD. That's an
acronym we use propulstional system design. Is it a p HAVE,
is it an E REV, is it a BEV? Whatever
case is. So we looked at a number of different situations.
We looked at for I would say, our two highest probability, where was going to be a Democratic split or Republican split split being split the House, you know, basically Congress or split Congress. House goes one way, Senate goes another way.
That didn't work out so we actually looked at other scenarios, which would be our Republican sweep or our Democratics sweep.
Obviously the Republican sweep worked this way through, and so we're working under that scenario to kind of build some guidelines in terms of how we're going have to modify our forecast. You know, it really would have worked for
us to kind of guess out loud and say, well, we think it's going to be an X and our forecast we more or less forecasted the status quo because we that's a known at least a known of sorts, and then if there's going to be different, well, obviously we'll have to integrate that in our forecast. And that's
really what we're doing right now is integrating those changes in the forecast, so that that kind of talks about what we did in terms of discovering or trying to understand what all the different facets would be. Now it's
down to brass tacks, what does it really mean? And uh,
you know, we're literally having calls every day because you learn new things every day, but we're having calls every day about what do we think it's going to happen in emissions policy, what's going to happen to the IRA, what's going to happen to build back better? The trade
policy is a big deal. I do a lot of
work in the supply community. They are getting gray hair
thinking about what the trade situation could be and in the end, U S m C A. So there's all
these different levers that can get pulled and UH, and the order that they get pulled and how far they get pulled and who pulls them back is all up for discussion.
Speaker 3: So so you guys, I mean, so so you look at tariffs, you look at the California waiver, you look at general emissions regulations across across the board, the IRA incentives that the tax credits and UH and US MCA and what's going to happen there.
Speaker 4: So you want to tackle one by one.
Speaker 6: Well, let's take one.
Speaker 2: EV subsidies the seventy five hundred dollars are a little lesser depending on the battery and all that done killed.
What's that doe to EV sales?
Speaker 4: Yeah, well, I mean there's a couple of different facets.
If you think about it, anybody, any oem that really needed that incentive to be competitive. That's obviously going to
be a problem.
Speaker 2: Didn't they all not need the incentive?
Speaker 4: Maybe our maybe our friend e Lon didn't need it for all of his vehicles. Yeah right, And so there's
a strategic thinking there that maybe that that's kind of strategic in terms of how that might work out. But
so let's face it, if the thirty seven to fifty times two goes away, uh, that that's going to be problematic.
It's gonna be problematic for some of the major OEMs that we're trying to drive down the road, especially what we call bebbing in the middle. You know, for mom
Pos Smith in des Moines, I always use I don't know if there's all a mom Pos Smith. I'm sure
there is always point them out in des Moin and never been in to des Moines, but I'm sure they live there, that they own an Equinox right now, or are they owning an edge or something, and they at some point, how do we get them into ABEV and that thirty seven to fifty times too seventy five hundred that would have been instrumental in getting them into above. Now the
market's going to have to basically build scale enough to bring the cost down to get them into that vehicle, or other factors like maybe you know, fuel prices go sky high and it drives them into that vehicle because it's more cost effective. All kinds of possibilities.
Speaker 2: But yeah, let's talk sales though. So we're on track
to do one point two million evs in the US this year. It's a little over eight percent of the market.
Eight point six percent or something like that. Where's your forecast?
Speaker 6: Go?
Speaker 1: Now?
Speaker 2: Do sales drop? Do they stay? Even what happened?
Speaker 3: No?
Speaker 4: I think it's a complicated factor because right now there is a regulatory framework and it's kind of like a bus.
You just can't stop it on a dime. That regulatory framework.
As we all know, the iiams are basically committed through twenty seven. You know. There they can mix their their
propultion system designs. They can do a little bit more
of this, a little bit less of this, but they really can't do an awful lot. I mean, they're kind
of the kind of stuck through twenty seven. Our thought,
and again, you know, reality might be different, but our thought is probably by by twenty seven, whatever the REGs are, they're more or less going to get quote unquote frozen and so you know, folks, you're on that trajectory, just keep going. If we take to seventy five hundred bucks away,
that's definitely going to be a problem.
Speaker 2: That's it, because the marketplace has a vote no.
Speaker 6: No.
Speaker 4: But to your point, then we may bring back the ability to sell credits. I mean that was going to
go away anyways. Well maybe that gets brought back. So
there's this is this is kind of whackam mole. You
talked about wackam mole earlier. There's there's a lot of
moving parts here.
Speaker 3: Okay, but you just can't look at the you know, evs and isolation. No, no, it's the other the other
point that you guys looked at in terms of overall environmental regulations. So I mean we sort of treat it
as though this is you know, chisold and stone.
Speaker 4: It's not right.
Speaker 3: I mean, these these can change like that.
Speaker 4: I would say through twenty six. I mean, let's face it,
if there's an investment and you're ready to launch something in the middle of twenty six, there's so much too much risk and not investing because you've got or not deploying because you've got the toolings done. Everybody's ready, suppliers
are ready, you kind of have to do it. But
that's why we say like sort of that twenty seven ish that at that point, you know, basically there's gonna be a lot of changes in the in the portfolio we call them, you know, future product programs FPPs. What's
going to happen to that twenty twenty nine BEV that was supposed to launch then, well under the other under the old legislation, there's probably a good chance that was going to happen. Under the new one, maybe not so much.
So then you either descope it, or you push it, or or you or you literally you know, take it out of the portfolio and do something different. I think
this is you know, I got quoted in the media saying that, you know, every olym is war rooming right now.
They have been for months talking about, okay, if this goes all Republican, what are we going to do? And
and some oms have four or five plants that have gone to BEV and as you well know, you can undo it real quickly, and there's a lot of money involved in doing that. So I think there's a lot
of moving parts here, and and you know, we're going to find out over the next couple of months what's going to happen. By January twentieth, we're really going to
understand what's going to happen. And I think we're getting
indications ahead of time. Anyways, Yeah, we aren't.
Speaker 2: Right, but it's uh, you know, I think a bit of a sticky wicket if Elon Musk is involved in setting these policies, because there's a definite conflict of interest that he could tell Trump, hey, yank all the subsidies, knowing well, it's going to cripple his competition.
Speaker 4: I'm not sure of all the political ramifications, but I would imagine that those subsidies were probably on death row.
If they are in death row, maybe before Musk got involved. Anyways,
because I I you know, again, I don't speak for the administration by any means, but my thought was is that they're probably gonna want let's let the market decide what it wants, quote unquote. But I think folks that
are listening or watching today, the IRA is a very complex piece of legislation. You've got the consumer part, which
is the seventy five hundred bucks. If you by above,
then there's people know know a lot about the forty five dollars per kilowatt. If you make a battery plan
in the United States and you put it into a module that you will get your forty five bucks per kilowatt.
And then there's a couple of other facets we think, and again, who knows what's going to happen when we think the forty five bucks will probably stick around.
Speaker 2: It'll stick around. And I've talked to Washington and people
on this, and what they've told me is, look, this is money going to build plans that are doing what creating jobs? What is Trump all about creating jobs? And
guess what where all those battery plants are? Almost all
of them are in red state. But I don't think
that matters.
Speaker 3: I see, I don't.
Speaker 4: I don't buy this this they're in red states.
Speaker 3: For one minute.
Speaker 2: He doesn't care.
Speaker 3: He's not going to be elected again.
Speaker 2: He's already won.
Speaker 3: Right, So what's what's the blowback? There is none, Right,
There will be some senators and congressmen who will bitch about it. But if he if he doesn't want it,
he doesn't want it, I mean, and that's what we voted for, right, And so the thing that I really wonder about, though, is in terms of, you know, you work a lot of suppliers, and how does a supplier afford to do this BEV project that may not happen and ice project that is more likely to happen.
Speaker 4: To your point, I could probably count on one hand the number of suppliers that have enough cash to be able to work on any program they want. So most
suppliers have to pick and choose, and they say, okay, well, this bad program is with one of my favorite customers or a customer I'm very embedded with. So you know,
if I don't work on it and they really want me to, that's going to be a problem. So there's
all kinds of things that go into it. But you
are exactly right. I think you know. What we've been
telling suppliers is you know, you have to look at the program and say where is it in the chain?
Is it on the edges. Is it in a luxury
space where maybe they can afford to do a BEV.
Is it in the performance space, but it's in the middle of the market. Boy, you had better make sure
that that vehicle is going to be cost effective so that it's going to be able to drive volume under all this different circumstances like the seventy five hundred dollars going away, And to your point, I don't disagree. Forty
five dollars to help out on the battery is wonderful, But you know, to really get that vehicle over the finish line and get people into it, it's another story.
I think the other factor is what about the charge infrastructure?
Build back better? I had money for the charge infrastructure.
If that gets pulled back, that's been a major factor why a lot of people might say I'm not's so sure I want to get into that vehicle. I can't
drive from here to Grand Haven or whatever, because I don't know if there's a charger at the end. If
if in fact, some of those dollars are pulled back, that's also going to change the mentality of customers in terms of well can it Can I actually use this vehicle for what I want to use it for?
Speaker 2: Michael, I'm hearing that suppliers are talking about in their terms and conditions their contracts with the OEMs, that there's a volume guarantee if they're bidding on an EV program.
Speaker 4: I've got I've got swamp land in Florida for you.
Speaker 2: That's just you know, crazy wish on the part of suppliers.
Speaker 4: I say, I'd be careful what I say. But but
but what I would say is that the only ms will push back pretty hard on that unless they really really want you in their vehicle in terms of technology or whatever.
Speaker 2: If I'm a supplier, I got froduciary duty to my you know, shareholders and everything like that. I'm not going
to sign up for a program if I think I'm going to lose my ass on.
Speaker 4: No, no, no, I agreed, and and and you know, suppliers know it's all about risk. How much risk am
I'm willing to bring in on a particular program. So
you're gonna have volume risk, you've got startup risk. So
you know, we used to in the ice world, we used to think about, you know, after about two or three months, you should be up to pretty well full speed. Yeah,
some ollams kind of proved us wrong and said, well, it took us six months, but usually two to three months.
The Beverolds has shown us that it takes forever to get up to full speed. And once you're there, someone
else is designing as brought out of BEB and now they're in your market and maybe the volume's not there, so all those dynamics are are are really quite interesting.
But to your point, I hate to say that, but the Williams are you know, there's they're wily bunch. They're
smart people and they're always looking around. Can I find
another supplier to do what I can get Joe to do?
Maybe I can get Alice to do what Joe does, so that if Joe knows that Alice can do it, then Joe's not going to push me so hard on some of these. And to your point, you know the
number of ice programs out there that are really what we call cream puffs, you really want to be on them.
They get stable volume, good margins, very you know, very very stable. There's not as many of those as there
used to be, and the Oiams know which ones those are.
Speaker 3: By the way, So so okay, is it the cream puffs are going away because there are now smarter people who are running the finances at OEMs or are they going away because the money has been diverted or the attention has been diverted to BEVs.
Speaker 4: It's kind of both. I would say that the the
the OE purchasing or procurement community, they're pretty smart bunch.
They most of a lot of them were former suppliers, so they kind of understand how the world works. So
you can't get you know, you're not going to teach an awful lot. They've kind of seen it all, I
would say that. But also as we as we add
more BEVs into the mix, and we're gonna be adding more, not as many as we used to, but we are adding more, the volume per platform or per vehicle is going down. And so, but it depends on where you're
in the vehicle, or you're in the bottom of the vehicle, where you're across more vehicles, then you're good. If you're
at the top of the vehicle and the top had maybe the volume's not there. There's some different dynamics that
need to be worked in. But you know, suffice to say,
as we've moved over to batter Electric, a lot of suppliers that I really want to make sure I've got placement on these high volume ice programs, and we know are going to be kind of one of the last ones to move over, you know, like the F one fifties and T one xx GM, which is a full size pick up, and those are kind of the ones that are going to be last to move over, and so suppliers work pretty hard to be on those programs, and and but once they're on them, you know, they usually work out pretty well.
Speaker 3: So, so, to what extent do you see the change in administration changing the focus on propulsion systems within OEMs, which then obviously have a big effect on the supply base.
Speaker 4: Sure, No, Well, let's just say that the REGs are are basically flattened as of Let's say at the end of twenty seven, someone says, okay, we're flattening them out, and we're you know, California, you know, and there's a lot of discussion what might happen with their waiver. A
lot of things have to happen. But let's just say,
in some way, shape or form, that's going to happen, because I think that is really the administrations has been.
The new administration is that made no bones about it, that that's one thing I think they certainly want to do.
Speaker 3: If that happened tried the last time.
Speaker 4: So and they tried the last time, but I think they'll they'll probably have a little bit more luck this time.
They know what to do and what not to do.
So if that happens, I do feel, you know, we were on a trajectory, you know, from a BAV perspective, from a compliance perspective, BEV was going to be a major part of the equation. It's still a part, but
I think that they'll probably look at e revs basically extended range vehicles, and say, how do we treat those a little differently, so we give them a little bit more credit. Maybe they're more than a p have a
plug in hybrid. So there's a lot of different But
we do think that ICE will stick around a little bit longer, Mild hybrid will stick it around a little bit longer, full hybrids will stick around a little bit longer because I think they might change the way that the credit system works and maybe continue what we had before what we currently have right now. There's a lot
of things up, and I think unfortunately the OEMs are all of the suddener are going, well what do I do?
And and these are major capital decisions where billions of dollars are and we've already spent billions of dollars. That's
the problem. It's not so much Okay, you told us
what to do, and now we're going to deploy our capital.
They've already deployed it, and now they're trying to figure out do I read, what do I do to shut it down? Do I continue it? Do I what do
I do? And think of all these ice engines that
we have out there, and that, for lack of a better word, we have more or less of stop development on them, on the majority of the ice engines. I mean,
that's to me, that's as big an issue as what do you do with of bebs? Do redevelop some of
these engines? Do I try some new displacements? So I
try some new technologies to continue these along. I think
they'll choose some of that, but that had been really ramped down.
Speaker 2: I think they'll do ice development for erevs, because, as you know, it's a different application. You can do a
lot of interesting things with an engine if it just runs in a limited rev range and charges a battery.
But what's your thought about erevs? Because there's so much
talk about them all over the world, but especially here in the US right now, and this is I guess just more your opinion than anything. Will people actually buy
them and plug them in? And I know all the
arguments in favor of an EREV. You know, you don't
have range anxiety, don't have charging time anxiety. You know
you can plug it in, you can run on gas whatever you want. But will people really buy them.
Speaker 4: I'll be honest with you, if they have the opportunity to not have the range anxiety, that would be number one number to The performance you can get out of these vehicles is really superior at many different capabilities, not all capabilities, but many capabilities better than what you can get out of a nice vehicle. I think they're going
to market that pretty hard. You know, we did a
project on this, and the things we've learned was just jaw dropping in terms of the dynamics you got to think about and what do you do at altitude? Because
you know, engine can the engine run enough that if we don't have any power in the battery, it'll be able to propel the vehicle maybe towing something. So these
are all fastest that the engineers are thinking about. But
I think it's a nice natural evolution to eventually get to a BEV once that that battery range gets better, the charging time gets faster, it gets lighter. You know,
we may not quite be there for all of what we call the different vocations. But EREV is is quite interesting.
Right now, it's not graded very well in terms of how they the emissions is looked at. They might revise
that it when the new administration and that could give that a little bit more breadth of life. So we know,
let's put this way, we know that virtually every major OEM is working on erevs right now. That we know
that for a fact, that should be nothing new to anybody out there, not a secret. That is not a secret.
They're all it's just a matter when they're going to get there. And the dynamics are very interesting. We talk
to a number of people. So you got this engine
in the front, the thermal's already done. If you have
a V six, you're already having your current vehicle, the thermal program's already done. You already know how to cool
and warm that engine. Like you said, you don't get
up to five thousand rpm. You're probably cruising at seventy
seven hundred and fifty rpm. You might go up to
maybe two thousand on a good day. But you're working
in a range and therefore you can really optimize how that vehicle works in that range. You probably don't need
displacement on demand, so there's a lot of content you can take out of.
Speaker 2: Those they cost out and make the engine more effic I hate.
Speaker 4: To say, the turbo suppliers probably aren't that happy about erevs.
Probably don't need a turbo in an E REV. So
I mean there's a lot of content that can be taken out, and all the drive line and NVH gets controlled because no longer do you have a drive shaft running through the vehicle, or you've got half chafts going to the side that's coming off the emotor. You can
better control that. So there are some advantages. It's just
the problem is is not all these platforms were designed to put an engine in the front, and and and all the crash guys are going, well, now you put an engine in the front, Now I got to crash the vehicle differently. So it's unfortunately nothing's easy and automotive.
Speaker 2: So if you're stuck with body on frame, it was a whole lot easier.
Speaker 4: Oh, John, you and I have had this discussion forever.
Oh my gosh.
Speaker 3: So okay, so you're saying that there's there's going to be some of this and some of that and some of the other thing, and the amounts of some are being adjusted. Yes, so can you put some metrics on
that in terms of you know, you maybe thought that we're going to be at X in terms of BEVs and we're going to be at Y in terms of ice.
Speaker 4: Sure, then the number a lot of the industry looks at is twenty thirty. Where are we going to be
by twenty thirty? They look at that as sort of
a pendulum point that you know, once we get to twenty thirty, the REGs have been really pushing BEVs for a good four or five six years in earnest, and then where are we going to be b twenty thirty?
Europe looks at it that way. Now they look at
twenty five, at twenty thirty, five, twenty forty, but we still look at twenty thirty from a North American perspective.
You know, I'll be honest, the bever rate's going to come down and and I think this is I'll get to your point in a minute, but this is really really important. If you're and I said, on a couple
of supplier boards, so I know this firsthand, it's it's a different thing if you set up I'm on the F one fifty and I've got the five liter Coyote and I've got the four point five or excuse me, three point five Eco boots, and I can put in one of those engines in. It's a little bit easier
to say I'm going to put more Coyotes in versus three point five Eco boots, and that if you're supplying the seats, you don't care. I mean, it's a seat,
that's a seat, and you don't care. But in the
BEV world, and maybe the F one fifty is not the best example, but usually it's a different platform. So
if somebody says I'm gonna I'm going to basically forecast or drive more BEVs, then that's a different program for that supplier versus let's say, being on the F one fifty.
This is why this this BEV situation is so important, and so to your point, you know, we given the Trump administration and the fact that it was a sweep, there's no doubt virtually every organization is out there driving their their their BEV rate down. How far it's going
to go down is really up for question, but you know, certainly it's gonna it's probably going to go south of thirty percent by twenty thirty, and then at some point where does it get by twenty seven or twenty eight, And then what is sort of that natural rate of evolution as new and better product is brought in and oh okay, that's cost efficient. Maybe I'll get into that BEV. Now,
I think it's gonna be more of a natural evolution and less of using sort of a ball peen hammer to to get it into the public. It is what
it is, but but that's unfortunately, that's the environment that the supply communities find itself.
Speaker 2: So I mean, essentially what you're saying is it's up to the car companies to come out with really cool electric cars that people want.
Speaker 4: To Yes, yes, that's it. Very much like hate SIB,
but very much like Elon did. He brought out cars
that were different, you know, had some of they had some issues, but they were different enough people like them that they bought them an en mass. I think that
the oms are going to have to do that. They're
going to start with longer range pehabs. Then they're getting
it into e revs and it eventually they're going to say, you know, I've got this BEV. I can get rid
of the engine and this thing does what you wanted to do and this cost effective.
Speaker 2: Hey, look, we promised we'd get you out of here at the top of the hour.
Speaker 4: I didn't talk about trade yet.
Speaker 2: I know, I know we didn't. You're going to have
to come back. That's all there is to but Michael
Robinat thanks so much for coming one after our Thank you gentlemen. All right, good.
Speaker 3: And next week we'll be talking more about autonomous vehicles, uh, with the guy from Plus. Oh that's right, Yeah, that's right.
Speaker 2: We'll go well from automotive, so you'll want to tune in for that one too, folks. And by the way,
thanks for having tuned into the show.
Speaker 1: Auto Line After Hours is brought to you by Bridgestone Tires, Solutions for Your Journey and by Borg Warner. The automotive
industry continues to evolve, and so do the opportunities to define it. Borg Warner, one of the world's most admired companies,
gets partners where they need to go. Let's do something
big together.
About this episode
A deep dive into the current state and future of autonomous vehicles (AVs) features insights from Henry Liu of Mcity and Michael Robinett from S&P Global. The discussion highlights the need for a national framework for AV deployment, addressing safety and regulatory challenges. Liu emphasizes the importance of basic behavior competency tests for AVs, while Robinett explores the implications of political changes on electric vehicle (EV) regulations and market dynamics. The episode also touches on the evolution of AV technology, including the role of generative AI in improving driving skills.
Original notes
TOPIC: Autonomous Vehicles PANEL: - Henry Liu, Mcity; Michael Robinet, S and P Global; Gary Vasilash, shinymetalboxes.net; John McElroy, Autoline.tv
Autoline After Hours
Autoline
0:00
61:39