- il y a 2 semaines
How Will Driverless Vehicles Hit the Road?
Catégorie
🤖
TechnologieTranscription
00:03So, is it working? It is working. Good morning, everybody.
00:08We really appreciate you turning up here, knowing that you're competing with Jason and everybody trying to get space in
00:15Jason's session.
00:17And now I have a bad echo.
00:20Today's topic, autonomous vehicles.
00:23So, many of you might remember, a little further west, 10 years ago at CES, there was a big hype
00:29on autonomous vehicles.
00:31Back then, I think the two gentlemen left to me would agree, everybody thought it is taking us 12, 18
00:36months and we will all be floating around like in the sci-fi movies.
00:41We all learned that the technology challenge proved way to be more complicated and challenging, but not only the technology.
00:49I think the reason why autonomous vehicles and autonomous driving has its renaissance lately is mainly around AI and additional
00:57compute power.
00:59And Christophe and Yunwei will kind of explore that as the technical experts in a minute.
01:04However, solving the technology dilemma is not the only challenge for autonomous vehicles.
01:11As of today, we have a very scattered regulatory environment and the regulatory environment is of such that in Europe,
01:19country by country, we have even federal state by federal state, different regulations.
01:23In China, we have an autonomous vehicle law, which is governing the entire country.
01:28So, those are the two extremes of the spectrum.
01:30Next to this, it is customer and societal change.
01:35What you can see is that there was a, and you need to bear a second with me, the American
01:41Automotive Association has been polling more than 10,000 people.
01:479% only in 2023 and 13% in 2025 would trust an autonomously driven vehicle to be a passenger
01:56or to be driven by it themselves.
02:00On the flip side, in China, 94% of all participants in road traffic would feel comfortable to share roads
02:09with autonomously operated vehicles.
02:12That just is a little bit of a backdrop where we are going to go.
02:15And we are not touching today on how the commercial business case side is looking.
02:20We are really just trying to, what does it take to get autonomous vehicles on the street?
02:24And Christophe, I kick it off with you.
02:26So, in light of what you have been hearing, and having spent 25 years as the VP for LIDAR and
02:32ADAS systems in the market, now with Valeo, what is it that you see and what is your read of,
02:38and do your statistics resonate?
02:41Okay, thank you very much, Konstantin.
02:43Hello, everybody.
02:45A few words on Valeo to start with.
02:46So, we are really a global leader in driver assistance system.
02:53So, what are driver assistance systems, we'll come back to that.
02:56But that's the beep-beep that sometimes are annoying you.
02:59That's also the cameras that are detecting the lane.
03:02That's also the radars that are allowing you to keep the safe distance to the car in front of you.
03:07So, very important feature.
03:09And we are offering today the most comprehensive portfolio of products, including LIDAR.
03:15And that's a topic that I'm sure we will come back to in a few minutes.
03:19But not only.
03:20Also, a very powerful central computer and software.
03:24And we have a product line called Answers that is really able to provide a full stack of software related
03:32to driving assistance, but also automated driving.
03:37Valeo is actually in a position, thanks to its portfolio, not only to do the driver assistance, but to power
03:45the step-up towards driving automation and automated driving.
03:50And that's where we think is also a lot of value.
03:56There will not be any driverless vehicle without LIDAR.
03:59And I think my colleague, John Wei, probably will agree with me on this.
04:05And then, the last thing I wanted to say is that we are focusing on safety.
04:11And the survey you've just mentioned is showing that safety is a very critical point.
04:17Safety and then affordability.
04:19But safety first.
04:22The other thing I would like to do, and that's probably reflected in the data you've just shared with us,
04:28is that we need to do a clear distinction between driving assistance and automated driving.
04:37Behind those two terms that I'm trying to simplify, you have the SAE, level 2, level 3, level 4, level
04:455.
04:46But for me, in order to talk to everybody, I'm not talking about those acronyms.
04:51I'm really focusing on assisting the driving, meaning the driver is still in control, or taking the autonomy, meaning the
05:01vehicle takes the control.
05:03And I think that's a very clear distinction that we need to make.
05:06And that's maybe something we've not explained to the public enough to get the acceptance of those systems.
05:14Because today, if you ask a driver, would you like your car to be safe, they will all say yes.
05:20And that's, I think, where we need to go, and where we need to explain how we are going to
05:27get there.
05:28By 2030, 70% of the cars will be equipped with level 2 systems.
05:35But level 2 systems are only assistants, meaning the driver is still in charge.
05:42And that's an important distinction that I would like all of us to understand and to make.
05:47Thank you, Christophe.
05:49Yulwei, you're the CEO of SAiond, currently one of the rising stars in that hemisphere.
05:56And, is it working?
05:59Yeah.
05:59So, as you're currently traveling between the Silicon Valley and China, I think you're one of the very exposed people
06:07seeing how different the cultures, the way of thinking and adopting is.
06:12Also, kind of, if we look at the latest study, 86 players in the autonomous vehicle part have been surveyed.
06:20And they said, 60% of them, regulatory challenges, is the biggest impediment.
06:26So, what is that you are seeing?
06:28Yeah.
06:29So, just like Constantine said just now, we're in a unique position as a company starting in Silicon Valley almost
06:3610 years ago.
06:37So, as part of the hype, you know, for autonomous driving back then.
06:42And we have started, you know, deploying a lot of LiDARs on level 2 vehicles, you know, right now mostly
06:50in China market.
06:51Up until now, we've shipped over half a million LiDARs, you know, each on one, you know, IDAS vehicles, you
06:57know, accumulatively.
06:59So, all these vehicles are running, you know, you know, like as regular passenger vehicles, helping the driver to be
07:08safer when they drive.
07:10In addition, we are also seeing some of our customers are starting to use our LiDAR for level 4 vehicles,
07:17you know, and, you know, also you guys have seen in the news.
07:20You know, Waymo has been expanding very aggressively recently, first started in four cities in the U.S., like San
07:28Francisco, Los Angeles, Phoenix.
07:31Now, they are expanding to many more cities.
07:33We saw the news now like Washington, D.C., even in Japan, and some other countries as well.
07:39In China, Baidu last year started deploying, you know, a couple thousand vehicles in Wuhan and maybe to some other
07:48cities now as well.
07:49So, all this we are seeing, that's reality now.
07:52In addition, let's don't forget, there are other type of level 4 systems as well, like the goods delivery vehicles,
08:01you know, just for, you know, no driver, no passenger, just deliver goods.
08:05One of our customers, Zellos, which we announced last week, that they will be shipping, we believe, tens of thousands
08:13of vehicles, just running automatically, almost 24-7 to ship goods, you know, from one station to another station.
08:21All this we are seeing that's coming now.
08:23So, technology-wise, I think it's getting closer to be ready.
08:28Thank you, you're right.
08:29And, Gustav, you already mentioned, so, level 2, Konstantin is not supported to drive and text.
08:37Level 4, he can text.
08:39As level 3 is kind of being perceived as the tipping point between assisted driving and really autonomous driving,
08:45What, in your view, from a value perspective, is really required to make that, like, the table stakes for everybody?
08:54So, you're right.
08:55There is a clear distinction where the driver is still in charge and, therefore, his job in the car is
09:01to drive.
09:01And the driver is not in charge anymore and the vehicle is driving the driver on its own.
09:07And there is a significant gap between those two steps.
09:11And those gaps require sensors that are a lot more precise and a lot more accurate.
09:19But not only sensors, but also the right software stack in order to ensure that the functions that we are
09:27providing, first, is safe.
09:29We need to remember that today there is still 1.2 million deaths per year on car accidents.
09:38And the car industry has progressed a lot.
09:4020 years ago, it was 2 million.
09:42So, the progress we made and driver assistance system are part of it, are helped already.
09:49But the next level, level 3, is creating a situation where we need to be much better than a normal
09:57driver,
09:57which means better sensor, better detection, and better software.
10:02And you've mentioned Waymo some minutes ago.
10:06They are doing 250,000 rides per week, if I'm correct.
10:10And there is a beginning of data that is available, saying that with those vehicles,
10:15they are reaching 90% less corporate accidents and an even higher number at cross-intersection,
10:26reducing by 96% the number of accidents.
10:30So, it's driving the entire industry towards more safety, at first.
10:37And RoboTaxi are here to demonstrate it's visible.
10:40And second, affordability, so we can also expand to passenger cars.
10:44And perhaps building on it, and you mentioned it in the intro,
10:47we currently still have the dilemma of technology and sophistication of technology versus cost.
10:52We all know that there have been billions and billions poured into autonomous driving
10:58with mixed successes, as we know.
11:01What do you read from a Valeo perspective on how can we balance to make it economically viable
11:06while not foregoing kind of the quality and safety,
11:10and therefore making it really mass-adoptable?
11:13Okay, so on the Valeo side, and that's where Valeo is really stepping into being a tech supplier,
11:21you know, we are looking at this as a high technology, and safety is our main focus.
11:28Safety is what we need to take care about at first,
11:31because you've also talked about adoption and some skepticism on the market.
11:35You've also just hinted right now on some delays.
11:39Delays are coming due to accidents, due to bad advertisements,
11:43and we need to make sure that those systems are safe.
11:47When we've done that, affordability is the next step,
11:51and Valeo is one of the best positions globally in order to achieve that.
11:56If you're looking at what we achieved for Level 2 already,
11:59Level 2+, sensors and functions that were initially only on high-end vehicles,
12:06you can now find it on a Dacia.
12:08On a Dacia today, you have a camera, you have a parking system,
12:13and you have an automatic emergency braking in case a pedestrian is stepping in front of the car,
12:18and that's really an entry-level vehicle.
12:21Level 3 will follow the same path.
12:23First, we achieve the safety, we get the confidence,
12:26then the volume will help, and we will reach the mass market with them.
12:29Thank you.
12:30And speaking of volume, Yunwei,
12:32you just mentioned that, especially in China,
12:35there is a ton of cars being deployed.
12:36Most of them are Level 2, Level 2+.
12:38So what is your read?
12:40How much will China be also kind of almost an incubator or catalyst
12:44to accelerating, let's say, the European adoption or in other areas?
12:48Yeah, so from the world's way to understand the China market,
12:52you know, everybody knows the EV competition in China market is very tough.
12:57And initially, people compete, you know,
13:00on the acceleration of the EV vehicles,
13:04on the battery, size, mileage, and also on the cabin.
13:08But now, another competitive pillar is the autonomous driving or ADAS features.
13:15Okay, and all the EV makers, they try to educate the customer
13:19that our or their solution is better than the other.
13:24And they got the user, the customer, to experience it as well.
13:27And the fact is, the reason we have so many LiDAR deployed in the China market
13:31is because with LiDAR, as Christoph mentioned just now,
13:35the experience definitely is better.
13:37Although the full stack for Level 3 or Level 4 is not ready yet for passenger vehicles,
13:42but with LiDAR, it definitely gives you better performance,
13:47better reliability, and give you a higher MPI, miles per interruption.
13:53Thank you.
13:54And just building on the safety argument of Christoph earlier,
13:57when we prepared for the panel,
13:58you mentioned that you're also starting to deploy LiDAR sensors in the infrastructure.
14:03Can you give a bit of color what that means
14:05and what it is kind of adding to the ecosystem safety?
14:08Sure.
14:09From what we see, LiDAR's infrastructure have probably two main purposes here.
14:15One of them is to replace the current sensors
14:19that's been used in the traffic systems.
14:22I think in many European countries, in the US as well,
14:26there are about 30-40% intersections have sensors,
14:30like including the old magnetic loop, inductive loops,
14:34like cameras recently, and the radar as well.
14:37They were used to detect the presence of the vehicle
14:40as a basic function.
14:41In addition, to detect vehicles that are approaching the intersection
14:45to avoid the dilemma zone,
14:47and also detect the pedestrians,
14:49and detect near misses, and so on.
14:51Those are advanced ITS functions,
14:54intelligent transport system functions,
14:56many innovative traffic engineers want to do.
14:59And with our LiDAR today,
15:02with the cost is very affordable,
15:04and the results are very reliable,
15:06it can help to replace the old existing systems.
15:10And right now, we are seeing the markets picking up now with our solution.
15:14As part of this, we can also see
15:16the LiDAR infrastructure is a unique feature
15:20that can act as a kind of god's eye
15:23for the complicated scenarios at intersection.
15:26And with V2X means vehicle-to-infrastructure,
15:31the data can be sent either in real-time
15:34or with a little bit of delay,
15:36that can help complement the sensors on the vehicle
15:39that can offer much, much reliable results.
15:43Very often, it's impossible to detect from the vehicle
15:46whether it's a person, whether it's a driver,
15:49human-being driver, or the vehicle sensor itself.
15:52It can be blocked by some other things,
15:55but the sensors on the infrastructure,
15:57on the interaction can definitely give a much better vision.
16:00We believe that can help a lot
16:03to boost the safety
16:04and boost the confidence of the community
16:06to adopt autonomous driving going forward.
16:10Thank you, Yulwe.
16:14And Gustav, on the question of
16:17if you take it all together,
16:19you have the technology piece,
16:24you have the technology piece,
16:26you have the regulatory piece,
16:29we all love to talk about ecosystems these days.
16:32So, what is your current view
16:34on how good are we as technology suppliers like you,
16:41car makers, regulatory municipalities,
16:43how is the ecosystem coming together
16:45and is it kind of a,
16:46everybody's trying to solve it for one common goal?
16:49Is it kind of competing?
16:51What is your view as Valeo being one of the real leaders
16:55in that space and having interactions
16:57with many of these stakeholders?
17:00So, I think you're right
17:02mentioning tech companies
17:04and the fact that Valeo there
17:08is leading the pack
17:11with up to 9,000 software engineers
17:14that are working on algorithm
17:17and on what we call SDV,
17:20software-defined vehicle.
17:21But SDV is such a wide topic
17:24that having one company
17:26trying to handle it all is not possible.
17:29and we have to set a partnership,
17:32we have to be open
17:33and that's probably something
17:35that is a bit new
17:36for the automotive industry
17:38but not new at all for a tech company
17:40that is used to those type of things.
17:42And I can give you a few examples
17:45with partner with Amazon,
17:49with Google and with Qualcomm
17:52in order to work on the Renault software-defined vehicle.
17:58We can also have other partnerships
18:00with Moilai, for example,
18:02or with NVIDIA for other customers.
18:05You might have heard also about the work
18:07we are doing with BMW on the Neuer Klasse.
18:09It's also a collaboration
18:11between several companies.
18:12and it's very important
18:15that in order to go faster
18:17we are open to those kind of partnerships
18:20because without those
18:21we are all trying to do the same thing
18:23and only competing.
18:25But competition is good as well
18:27and there will still be companies competing,
18:29I'm sure.
18:31And real quick,
18:32just building on that stuff,
18:33we've all seen certain similar structures
18:35like ADECO and so on
18:38where people were supposed to share IP and data and so on.
18:43So what's your view?
18:44What does it need to really make sure
18:46that data, also your point,
18:49the more we share the data on autonomous driving,
18:51the more everybody can benefit
18:53because you have more data points,
18:54the AI can train and so on.
18:56So what's your view?
18:57Is that something that governments
18:58should kind of almost like
19:00an independent cloud
19:02that they should own it
19:03and make it accessible to everybody?
19:04Should it be still proprietary?
19:07So I think you're a little bit
19:10stepping into the legislative side
19:12and I have to say
19:13even as a tech company
19:15or as an automotive supplier,
19:18we have to follow legislation.
19:20We can, of course, support,
19:22we can explain,
19:23we can make recommendations
19:25but we are not the ones
19:26that are going to do the legislation.
19:27and we are following them.
19:30If you're looking at Europe
19:32and we were saying earlier
19:34that China is going very fast
19:36on the implementation of Level 2, Level 2+,
19:38having LIDARs on nearly all of those vehicles
19:41and so on.
19:42But the first Level 3 system ever launched
19:45was launched in Europe
19:46and was launched by Mercedes.
19:49and that's thanks to the legislation
19:51that is in place.
19:53Now, will Mercedes share the data
19:55they are collecting?
19:57It's a question for them, basically.
19:59Okay.
20:00On our side,
20:01we are working with the data
20:03that is available
20:04and where legislators decide
20:08that more data should be available,
20:09we will, of course, use it.
20:12Thank you.
20:13Yelme,
20:14just shifting gears a little bit.
20:15So, I think we have a good understanding
20:17of what is required
20:18in terms of technology
20:19and what is currently in the making.
20:22Where do you see the,
20:24you mentioned Waymo and Beidou,
20:25but what are the real business cases
20:27where we can see mass adoption
20:28of Level 3, Level 4?
20:30I mean, truck platooning
20:32is something that's being mentioned,
20:33Robotaxi is being mentioned,
20:34but what is it where we really get the scale?
20:37Also, Christoph's point earlier,
20:39the faster we can get scale,
20:40the faster we get the cost down,
20:41the faster we get mass adoption,
20:42the faster we get the entire ecosystem
20:44quality of safety
20:45to a different level.
20:46So, what's your perspective?
20:49I think we are seeing
20:50two approaches here.
20:52Since we started,
20:53you know,
20:53since the autonomous driving boom
20:55started about 10, 15 years ago,
20:58you know,
20:58there's always two approaches.
20:59One is from top down,
21:01from, I mean,
21:02go to the Level 4 directly, right?
21:04And in the last few years,
21:06there are quite a few Level 4 companies
21:08eventually we saw Waymo
21:10started making it,
21:11really demonstrating
21:12that it's scalable now.
21:14Okay,
21:14and economy-wise,
21:16then probably
21:16when they do the math,
21:18when they are scaling up
21:19more and more operations,
21:21I believe
21:21their analysis shows
21:23they can make money
21:24out of it.
21:25So,
21:25but that involves
21:26quite a lot of,
21:27you know,
21:28scale of operation
21:29and very few companies
21:30can do it.
21:32Another approach
21:32is from,
21:33you know,
21:34Level 2,
21:34Level 3 up.
21:35and luckily
21:36we are seeing
21:37first in the China market,
21:39we believe
21:39very soon
21:41in a couple years
21:42in the European market,
21:43you know,
21:44you'll see a lot of vehicles,
21:45maybe first with,
21:46you know,
21:46cars from China
21:47than many European car makers
21:49who start adopting,
21:50you know,
21:50getting better
21:51Level 2 systems as well.
21:53And with more systems
21:54being deployed,
21:55the cost can only come down.
21:58Okay,
21:58the sensor cost.
21:59You guys might have known,
22:01you know,
22:01for,
22:01for like,
22:02you know,
22:03mid-end LiDAR,
22:05okay,
22:05which can detect
22:06about 200 meters.
22:07Currently,
22:08the cost,
22:08the pricing to the OEM
22:10is already less than $200 now.
22:12Okay,
22:13which is a very small amount
22:14compared to overall
22:15bond price of a vehicle.
22:17And the high-end LiDAR
22:18like ours
22:18is also
22:19way below $1,000.
22:22With this,
22:23you know,
22:23we can see
22:24more and more deployment,
22:25more and more adoption
22:26by the OEMs
22:27and more systems
22:28on the road
22:29and OEM
22:30getting more confident,
22:31getting more data,
22:31and we can see
22:32the system
22:33will getting more reliable,
22:34you know,
22:35getting more,
22:35you know,
22:36accepted
22:36by the community.
22:39And just building on that,
22:41you want to add,
22:41Christoph?
22:42Yeah,
22:43just wanted to add,
22:44and I have a similar view,
22:46but I think
22:48we all need to understand.
22:49Level 2
22:50is mass market today.
22:53Level 2
22:54is mass market today.
22:55By 2030,
22:5670%
22:57of the vehicle
22:58driving
22:59will be equipped
23:00with some type
23:01of Level 2 system.
23:03And it is actually
23:05what we are seeing
23:06in China.
23:07It is a carmaker choice
23:08to make it standard
23:09on their vehicle.
23:10And we've seen
23:11BYD,
23:12for example,
23:13going with Odise
23:15across the entire range
23:16of their vehicle
23:17without any
23:18on cost
23:18to their final customer.
23:21So that's
23:22a way to go,
23:23I would say.
23:24and Europe
23:25is somehow
23:27a little bit behind,
23:28but we've been
23:29in the front
23:29in the beginning
23:30for Level 2.
23:32And we are now
23:33being a bit
23:34catch up by China
23:35on Level 2+,
23:36what I would say.
23:37And you can see
23:38step by step
23:39how passenger cars
23:41are taking
23:42more sensors,
23:44more features,
23:45and are
23:47a little step away
23:48from Level 3,
23:49but it's coming
23:50very soon.
23:51From the other end,
23:52I fully agree
23:53with Yongwei
23:54on the Level 4
23:57and Robotaxi,
23:58it's happening
23:58as well.
23:59We talked about
24:01Waymo,
24:01we've talked
24:02about Baidu,
24:03even in France,
24:04in South of France,
24:05we are doing
24:05some evaluation
24:08right now.
24:09We heard Uber
24:09and Waymo
24:10announcing
24:11today or yesterday
24:13a fleet deployment
24:15in England,
24:16in London.
24:17So those two words
24:19will converge
24:19with different
24:20business cases.
24:21but both
24:22will happen
24:23in parallel
24:23and it's coming
24:25quite fast
24:25because Level 2
24:27is here today.
24:29I agree.
24:31So,
24:34now,
24:35I told you,
24:36you've got to bring
24:37your crystal balls,
24:38what's going to happen
24:39in the next 12 to 18 months.
24:41And I know
24:42it will be recorded,
24:43so we will look at it
24:44in 12 or 18 months,
24:45what are you saying.
24:46But anyway,
24:46starting out with you,
24:47what do you think
24:49is the braking point
24:50when we will have
24:51Level 3
24:52and I'll start
24:53with Level 3
24:54really in a relevant
24:56mass adoption
24:57outside of
24:58RoboTaxi's
24:59and stuff like,
24:59really,
25:00to Christophe's point,
25:01not necessarily
25:02in a Dacia,
25:02but perhaps something
25:03like a mid-range car
25:05that is kind of
25:05accessible to most people.
25:07Yeah,
25:08so,
25:08first is,
25:09as critical science
25:10does now,
25:11you know,
25:11Level 3
25:12has already been there
25:13by some of the OEMs,
25:15but with very limited speed.
25:18Okay,
25:18let's ignore that
25:20a little bit.
25:20Here,
25:21I want to talk about
25:21the real Level 3
25:22means it can drive
25:24at a regular speed.
25:25To achieve that,
25:26if you really do the math
25:28to the logic analysis,
25:29you know,
25:30you figure out that
25:31the sensors
25:32need to be better
25:33than what's available today,
25:35including LiDAR.
25:35the low-end LiDAR
25:37is not good enough.
25:38We need a high-end LiDAR.
25:39That's why I believe
25:40Christopher and our company
25:42are both working hard
25:43to develop those LiDARs now,
25:45and we think
25:46it's very close
25:46to be available.
25:48You said,
25:4912 to 18 months,
25:50that's about the time.
25:51Yeah,
25:52those high-end LiDARs,
25:54high-end sensors,
25:55that can allow you
25:56to detect the objects
25:58as far as
25:59150 meters away
26:00at,
26:01you know,
26:01the size that really matters
26:03for driving,
26:03that will be available
26:04very soon.
26:05And with those sensors
26:07and with more data
26:09from the,
26:09more breakthrough
26:10from the software side,
26:11some of the solution providers,
26:13we believe
26:13that LiDARs will be ready.
26:15Yeah.
26:16And just one question
26:18to that,
26:18because we also discussed it.
26:20How much of the,
26:21just looking at the technology
26:23side of the equation,
26:24we come to the rest
26:24in a second,
26:25how much is the sensor
26:26versus how much
26:28is compute power
26:28to really process the data
26:30and learn the system
26:31and kind of having
26:31the full stack?
26:35So the first thing,
26:3712 to 18 months,
26:38I think that's too short.
26:39We are already working
26:39on those vehicles now.
26:41They are nearly frozen,
26:43you know.
26:43So what's happening
26:44in the next 12,
26:4518 months
26:46is an incremental
26:48improvement
26:49of the Level 2,
26:50Level 2 Plus system
26:51and the Level 3 system.
26:53So what we are working on,
26:55thanks to software-defined vehicle,
26:57and coming back
26:58to this point,
26:59because those things
27:01are not going one
27:02without the other.
27:03You have the sensor set
27:04that needs to progress.
27:06It's already there
27:07to a certain level
27:08of maturity
27:09that allow a certain level
27:10of Level 3 system,
27:12but also the compute power
27:14that needs to be in the car
27:15and we've mentioned
27:17a couple of companies
27:17that are working
27:18on this already
27:19and the software.
27:21And one example
27:22I can give is,
27:23because it's one
27:24of our customers
27:25and it's public,
27:26we've launched
27:27the first Level 3 system
27:28with Mercedes
27:29two years ago
27:30and the maximum speed
27:32was 60 km per hour.
27:35Through a software update,
27:37software in the air,
27:38we could,
27:39in the same car,
27:39with the same system,
27:41without any sensor upgrade,
27:44bring the system
27:44up to 95 km per hour.
27:47So software-defined vehicle,
27:49compute power,
27:50over-the-aid update
27:51and the sensor set
27:53that you're putting
27:53in the car
27:54are actually putting us
27:55in a situation
27:56where we can improve
27:58the existing system
27:59in order to have
28:00those incremental
28:02increase
28:03of what we call
28:04the ODD,
28:05so the availability
28:06of the system
28:07for the user.
28:08And I think
28:09short-term,
28:09that's what's coming.
28:12And timeline?
28:16So 12 to 18 months,
28:18we've already talked about it.
28:20I would say,
28:21so I don't see
28:22a tipping point
28:23or a big breakthrough.
28:24I think the automotive industry,
28:27even though we are facing
28:28a strong acceleration
28:29with electric vehicles
28:30and so on,
28:33is an industry
28:34of incremental improvement.
28:36And that's the best way
28:38to guarantee the safety
28:39and that's why
28:39we don't want
28:40to compromise with this.
28:41We cannot allow
28:42that the driver
28:43is not safe
28:46or taking risk
28:47to test the system.
28:49We have to do that.
28:51but we are already
28:52working on the systems
28:53that will come on the road
28:54in 10 years
28:56and they are level 3
28:59and they are level 4
29:00for the robotaxi.
29:01It's here.
29:02It's going to be
29:03an incremental increase
29:05of the usage
29:06and the time
29:08where you can use
29:08the system.
29:10So,
29:11as I have a lot
29:12of trust
29:12in the two of you
29:13and your companies
29:13in terms of solving
29:14the technology equation,
29:16if you would have
29:17kind of one wish,
29:20what would it be
29:21in order to expedite
29:23the journey
29:24on autonomous vehicle
29:25and really making
29:25a mass phenomenon
29:26on society,
29:28regulatory,
29:29anything,
29:30outside of technology?
29:31You want to go first,
29:32Gustav?
29:34I would say
29:35in the next years
29:37and we've mentioned it,
29:38the car will be electric
29:40and I think
29:41this will continue.
29:43The car will be safer
29:45and I think
29:46that's what we will need
29:47to get the adoption
29:49and the car
29:50will be also upgradable
29:52so we will be able
29:53to make some upgrade
29:54and make it better.
29:55So,
29:56my wish is
29:58we need to
29:59get
30:01final user acceptance
30:02as fast as we can
30:04in order to get volume,
30:05make the system affordable
30:06and have them
30:07on every car.
30:09Your way,
30:10your wish?
30:11My wish would be
30:13probably the more
30:14clear regulation
30:15by all kind of governments.
30:18You know,
30:19versus level two.
30:20Level two is
30:20too big a granularity.
30:22We'd like to be more clearly
30:23how scenario-defined
30:25regulation
30:26and this has to be done
30:27by the experts
30:29in the OEM,
30:30in the car industry community
30:32and by the government
30:33together as well.
30:34Clearly define
30:35what do you mean
30:35by level two?
30:36What do you mean
30:37by level two plus?
30:38What do you mean
30:39by level 2.8?
30:40Whatever you call it.
30:41Okay.
30:42Level three,
30:43that's even more important.
30:44And,
30:45you know,
30:45today is more defined
30:47from responsibility
30:48point of view.
30:48I hope
30:49there could be
30:50more clear definition
30:51from scenario-based case.
30:52and with that
30:54probably
30:55you will
30:56remove a lot
30:57of ambiguity
30:57by the car users,
30:59by the OEMs
31:00who develop the cars
31:01and also
31:02by the regular
31:03regulatory bodies.
31:06So,
31:07what we can take home
31:08is the technology
31:09is not a challenge.
31:10It is already there.
31:11It's in the making.
31:12Level three
31:13is most likely
31:14more or less
31:15around the corner
31:15for many of us.
31:17Those who want
31:18to enjoy level four,
31:19you've heard
31:20Baidu
31:20and Waymo
31:21are deploying
31:21more and more
31:22companies,
31:23fleets.
31:24With that,
31:25I think
31:25it is a bright future.
31:27Christophe,
31:28thank you for your time.
31:29Thank you for joining us.
31:31Thank you very much.
31:32Thank you all
31:32for listening
31:33and looking forward
31:34to more discussion.
31:37Thank you.
Commentaires