OSRAM Licht AG/media/Files/O/Osram... · In this topic, LIDAR is not new for OSRAM, actually, we...
Transcript of OSRAM Licht AG/media/Files/O/Osram... · In this topic, LIDAR is not new for OSRAM, actually, we...
Page 1 of 25
OSRAM Licht AG
Tech Teach-In LiDAR
June 20th, 2018 | 14:00 CEST
Speakers:
Dr. Joerg Strauss
Andreas Spitzauer
Juliana Baron
Page 2 of 25
Andreas Spitzauer Good morning, good afternoon to all of you, and thank you
very much for joining our Teach-In call on LIDAR. My name is
Andreas Spitzauer, Head of Investor Relations of OSRAM, and
we’re happy to have you on board. We will start with a short
introductory video showing OSRAM’s activities in the field,
before I hand it over to our expert, Dr. Joerg Strauss. Who is
heading the business segment in Emitter Sensor Laser at
OSRAM Opto Semiconductors.
Dr. Joerg Strauss Hello, ladies and gentleman, this is Joerg Strauss speaking,
also, welcome from my side, I’m very happy to discuss the
exciting topic of LIDAR together with you today. And I’ll get right
to it, going briefly through the agenda, and the contents I’ll cover
today in the presentation. I’ll give a quick intro where we stand
today with LIDAR, cover autonomous driving, the different levels
of autonomous driving, and LIDAR technologies, as well as our
own emitter technologies which we do as OSRAM Opto, and
then I’ll give a quick outlook. Now, if you think about light for
automotive applications, you probably think mostly about the
visible applications, headlights, rear lights, interior lights. But in
fact, non-visible applications, infrared light applications have
been, and will be also very important in this space. In the past,
we talked about rain sensors, ambient light sensors. Now, in
future, we mostly think about technologies which can enable
autonomous driving. For instance, LIDAR plays a very important
role here, which can do pre-crash sensing, but also capture the
entire 360 environment of your car.
In this topic, LIDAR is not new for OSRAM, actually, we work on
this topic for many, many years, we launched the first product in
that space already in the early 2000s. In that time, the LIDAR
lasers, our 905 nanometre pulse laser, which was introduced,
was doing, or helping for auto emergency breaking, and it is
used by many OEMs. And in fact, we have delivered more than
ten million lasers into that marketspace. Those laser drove
already 200 billion kilometres without any chip failures, and for
the future, we have already more than 20 or 20 design wins in
that space. This is based on our strong technology position, we
have very good performance with our lasers, in terms of
efficiency, also thermally. And we were the first ones to provide
an automotive qualified laser to the market, which is crucial for
safety providing features.
So, it’s not a new topic for us, and we strong believe that it will
grow in importance, specifically when we talk about autonomous
driving. And I would like to touch on that a little bit more in detail,
jump in directly to slide six. What is it, autonomous driving?
Page 3 of 25
Actually, it’s defined through different levels of automation, L0
means no automation whatsoever, the driver has to do the entire
job of driving, L5, the highest level, means full automation, full
autonomous driving. The car takes over the job of steering the
car, driving the car, and the driver can shut eyes, hands, mind
off, can do whatever he or she wants in the car, aside driving.
And intermediate, there are Level 1 to Level 4, already today
Level 2 is widely established. The system does jobs like lane
assistance, speed control, and so on and so forth, so the driver
can take off the feet and hands from the pedals and the steering
wheels. Level 3, also starting to be adopted where you even can
take your eyes of the streets for a certain amount of time. And
Level 4, only in very specific decisions, situations, the driver has
to take over. Now, to enable that autonomous driving, these
growing levels of autonomous driving sensors are crucial to
ensure that you capture your environment. And the three
primary sensor technologies we talk about are camera, radar,
and LIDAR. On this sketch here to the left of slide seven, you
see different functions, so sensors take over and can take over,
because they have different strengths. For instance, camera is
very good in colour vision, can see colour, that means you can
see the colour of a sign, but doesn’t really work that well in
inclement weather. Radar can look very far, works in the dark,
and in bad weather, which is good. LIDAR is really the only
sensor technology which can give you a full 3D capture of the
360 degrees surrounding of your car. And those sensors, they
have to work at all driving conditions to really enable full
automation and driving.
Now let’s look into that a little bit more in-depth on the next slide
where we, on the left-hand side have put down all the key
performance indicators for LIDAR for autonomous driving.
Which is range, resolution, 3D capabilities, detection of objects
and speed, read signs and colours, all environmental conditions.
But also, other factors like interference effects, cost for sure, it’s
always important in size, I mean you want to have the
technology, you want to have the functions. But it shouldn’t
assume a lot of space, you shouldn’t really see the sensors, so
it doesn’t infringe your design options for the car.
Now let’s have a look how those different sensor technologies,
camera, radar, and LIDAR perform against those KPIs, because
they all have their strengths and weaknesses. Camera, as
mentioned before, great to see colour, but not so great in dark,
for sure, and also, inclement weather. Radar on the other hand,
Page 4 of 25
you see this by the colour code, very good in range, also great
at all weather conditions, but it’s not really able to have high
resolution in far distance. So, it can tell you there is an object,
but it cannot tell you which kind of object is it really, so that’s a
weakness. LIDAR, on the other hand, can do exactly that, it can
give you a detailed description of the object, but can see colour,
and pricewise, size wise, so far, it’s still developing.
So, at the end of the day, there is no one fits all solution, none of
those sensors has all the capabilities you need, only the
combination of all three sensor technologies provide you with he
capabilities needed for autonomous driving. And in addition,
they also provide the option for redundancy, which is very, very
important for safety related features. If one sensor fails, others
have to take over at least for a certain amount of time. So, we
believe strongly that we see a co-existence of those three
sensors, but we’ll definitely see LIDAR in the autonomous driving
space.
Now, on the next slide, slide nine, you see the amount of sensors
depending on the different levels of automation, in L2, you see
we count round about three radars and one camera for instance.
Level 4,5, full autonomous driving, up to ten or even more radar
systems, eight or more cameras. On the LIDAR side, right now,
L2, either none or one sensor for emergency breaking. For
instance, going up on the autonomous levels, there can be up to
six or even more LIDAR sensors. To do a far looking job and
also a job on the side of the car, to really do object recognition
in the entire 360 space, which is a great big strength of LIDAR.
So, again, we are very convinced that LIDAR will play a key role
in this autonomous driving space.
Let me switch gears a little bit, looking at the markets of LIDAR
in future, and in the past as well, we see two parallel streams in
the LIDAR market. One is our traditional automotive space,
which right now implements Level 1 and two, maybe Level 3.
And the market functions like the regular automotive market
does, the driver is your customer, and you want to sell cars to
your driver. And you do this for instance, by additional features,
additional features like safety related features, which can be
LIDAR. But you also want to do it at attractive price points,
because you want that driver to click that box for the option, or
for that model you have the LIDAR in, so, that’s a primary driver
for that market.
The other big market, which is starting right about now, is
mobility as a service, the Robo-Cars, Robo-Taxis, here it’s
Page 5 of 25
different, here the task is to replace the driver. And to be able to
replace the driver, you need to have full level autonomous
driving, otherwise you’re not able to take that cost position out of
transportation, which is the driver. So, this market looks at
different KPIs, the biggest and most important one is safe
autonomous driving already now. So, cost is always important,
but not as important as for the traditional market. Eventually
those markets will converge at the Level 5, but again, the real
driver for a Level 5 technology is mobility as a service. And the
traditional market follow different rules. We tried to illustrate this
a little bit more in detail on the next slide 11, where you see the
history and the future of this LIDAR market from our point of
view. Illustrated in the past has mentioned all traditional
automotive markets through this emergency breaking systems.
Up until let’s say now, where mobility as a service really takes
off, which is this light greyish-blue line, which is growing pretty
quickly.
And the reason is again, because this technology, or this market
is pushing for L5, that means quite a number of LIDAR sensors
in the cars. And a lot of laser emitters, which is really the number
we are tracking here in this chart, in the car, or in those sensors.
On the other hand, in parallel, traditional market, pushing for L2,
L3 right about now, that means less LIDAR systems in the cars,
less lasers in the cars. And that’s a reason why we see it
growing a little bit slower, but also steady. And as soon as we
talk, also hear about L5, you know, it will be growing as well quite
intensely. So, and also those two markets, they cover different
technologies for LIDAR, which I would like to cover a little bit
more in detail over the next couple of slides, in the next couple
of minutes. Jumping to slide 13, LIDAR stands for Light
Detection and Ranging, and in simple words, it’s a time of light
measurement.
So, you have a light emitter in your car infrared light, which you
don’t see with the eye, but there is a laser inside which emits
infrared light onto the street, and as soon as it hits an object, no
matter what it is, it gets reflected back. And there is a detector
unit waiting in the module, in your LIDAR module, which detects,
absorbs that light, which is reflected back. And by the time lag
of your light emission, photon emission and photon detection,
you can calculate the distance to the object. And if you do this
over and over again, you see even the velocity of that object.
And this goes super fast, because we talk speed of light, 100
metre distance translates in .6 microseconds, which is very, very
fast.
Page 6 of 25
The requirements for LIDAR systems are shown on page 14,
and they also depend on the level of autonomous driving. In L1
and L2, which is our current space, it’s rather short-range
sensor, LIDAR systems, they look into the front of the car, round
about 20 metre for stop and go traffic. Emergency breaking, pre-
crash sensing, those things. In L3, we want to look already
farther, 20 to 100 metre, and in L4, L5, we will have a mix of
technologies which is depicted in this scheme here on the slide
by those orange bars to the front, and the back of the car. Which
reach more than 100 metre far, here we talk about object
detection of this infamous piece of a tyre, which is 200 metre
away, if you want to see and detect. This is the task of long-
range LIDAR technology. You also have short-range LIDAR
modules on the corners of your car, which is here shown by
those grey triangles. And those modules really have the task to
provide a 360 coverage of your closer surrounding, of your
closer environment.
Juliana Baron Could I just shortly interrupt, I’m sorry for that. I heard that some
have problems to access the presentation, if you just refresh
your browser cache, and try again to access the link, probably
that helps, thank you.
Dr. Joerg Strauss Thank you, Juliana. Okay, so I’m flipping to slide 15, and here
we talk a little bit more about the building blocks of a LIDAR
system, and what building blocks build a LIDAR system. First,
on the right upper corner of this matrix, it’s the emitter, so you
need a light source which emits infrared light onto the street,
which is a laser. And this illuminates the scene in front, usually
using an optical element, maybe a mirror, a lens, optical phased
array technology, and then it hits an object and get reflected
back. And here then you have a photo detector waiting,
absorbing this light, and creating an optical signal. Now, we
need to translate this optical signal into data, and this is part of
the job the ASIC does, it drives both the emitter and the detector
to make sense out of the information which was received. And
also translating it into data, and then this data is pushed towards
the processing and control unit, and here you do really the data
handling to create a point cloud or do object recognition.
In our, today’s presentation and also currently is OSRAM Opto
Semiconductors, our focus is mostly the laser emitter today, and
I would like to talk a little bit more about emitters and the different
tasks for the emitter. To do that, we need to distinguish between
two primary LIDAR technologies, which are shown here on page
17, we talk about flash LIDAR systems, and scanning LIDAR
Page 7 of 25
systems. Flash LIDAR is really easy, or pretty comparable to a
regular camera flash, right. So, if you have a visible camera
flash, whenever you hit that button of the flash, the entire scene
in front of you gets illuminated by the light. And the same is true
for a flash LIDAR as soon as it’s switched on, the entire scene is
illuminated by infrared light, without any moving parts involved.
Scanning LIDAR instead, you have a very narrow laser beam,
which hits a mirror, can be either a mechanical rotating mirror,
or it can be a micro-mirror. But at the end of the day, this narrow
laser beam hits a mirror, and wherever the mirror is pointing to,
the light gets pointed to it. If it looks left, the light looks left, if it
looks right, the light goes right, so, you illuminate the scene by
using the movement of the mirror.
For the emitter, it has implications because scanning LIDAR is
mostly used for long-range LIDAR systems. So here you need
high directionality, you need very high powers, good thermo
performance, and we believe strongly that our edge-emitting
lasers are perfectly suited for that task. In flash LIDAR, it
depends a little bit of the distance you want to cover, and how
much laser power you need, but also here high powers are
needed. And therefore, also here the edge-emitting lasers play
a very important role. If it’s really short-range, very short range,
very low powers, VCSEL technologies might be applicable,
which I’ll talk a little bit more on my next slide 19. Here, you see
a basic comparison of the fundamental technologies of emitters
you can use for LIDAR. On the left-hand side, on the upper part
of that slide, you see VCSEL, a scheme of a VCSEL chip, and
it’s a vertical emitting laser. That means the entire surface area
of the chip emits directional laser light to the top. The right-hand
side, you see an edge-emitting laser, here only a tiny fraction of
your side corner is the light emitting part.
Now, we also put down the KPIs for laser light sources, which
are OPA power efficiency, sum of stability, cost, eye safety and
emitting area. And as you know, we have both the edge-emitting
lasers, and the VCSEL technology in-house. So, we try to
assess what’s the best solution from a technical standpoint, and
from a market standpoint for this application and this market.
And we try to show this by the colour code here, so let’s first
compare VCSEL and edge-emitters at below 1,000 nanometres.
Which is the left and the middle part of this colour coded column.
And the biggest issue for VCSEL is the emission area for the
power levels you need. Typical reasonable VCSEL size is two
to three square millimetre, and then we talk about the laser
emission of a couple of watts, two, three watts. As you could
Page 8 of 25
see before our current edge-emitting lasers for LIDAR emit 75
watts, and the customers ask us for more and more power. So,
we talk a couple of hundred watts, peak power to create a couple
of hundred watts peak power by VCSEL, we talk a lot of area,
we talk a lot of VCSEL space.
And this increases your VCSEL array, it also increases your
optic, it increases the size of your entire system, so you have a
large LIDAR module, if you want illuminating type scene, that’s
a big disadvantage. Also, OPA power efficiencies, thermal
stability are not as good as for the edge-emitting laser, which is
really very suitable, as you can see by the colour code. The only
disadvantage which we have for both, and that’s related to the
wave lengths of the system, is eye safety, and here I would like
to do the comparison of a edge-emitting laser. Which is below
1,000 nanometres, and an edge-emitting laser which is above.
The only real benefit for more than the 1,000 nanometres is that
the eye is not as perceptive for this kind of wave length at 1,550
nanometres, for instance, just doesn’t see that light anymore. It
still does a little bit, this 905 nanometre, so you cannot put as
much laser power onto the street, to fulfil the regulations, which
are very, very strict, for sure. And that’s the only real benefit, but
also with the limited power allowed for below 1,000 nanometres,
very good LIDAR systems can be built. All the other factors, cost
especially, but also power efficiency, are much worse for the
above 1,000 nanometres. So, long story short, we think if we
compare all the technologies we have available, that edge-
emitting lasers at a wave length of below 1,000 nanometres. Or
round about 900 nanometre, are the perfect fit for this LIDAR
applications, and this is also shown by a roadmap, which we
develop very closely aligned with our customers. If you look at
today, we talk single-chip, one laser, either laser bar, if you talk
scanning LIDAR, or a packaged version, which is our smart laser
for flash.
Now the customers ask for two things, one is more channels,
more lasers in one package, which just increases the power, and
also, easier handling, because handling a laser dye is not that
easy. So, they ask us for packaged version, SMT stands for
Surface Mount Technology, so that you can easily solder those
devices. And this is what we’re pursuing in close collaboration
with our customers, and then beyond that, we will further
increase the amount of laser channels, and we will further
improve our packaging. For instance, with integrated versions,
maybe we’ll include a driver IC. So, we are creating right now a
Page 9 of 25
toolbox which allows certain customisation for our customers,
but also, provides good economies of scale for ourselves. And
again, this is very well aligned with this, all the customers, and
we strongly believe that we talk to everybody who’s involved in
the LIDAR module business right now.
This is bringing me to my very last slide, slide 22, which talks a
little bit about OSRAMs positioning in the LIDAR space, and
LIDAR markets. Before I show the key building blocks, which
you need for a good LIDAR system, and I hope I could show you
that we are very strong in the area of emitters. We’re also
providing detectors PIN photodiode to the market already right
now. And we are looking in the space of detectors, but the other
spaces are equally important. Signal processing, ASIC, as I
showed before, optics and in general detection. And we use our
ecosystem, our partners, as you know, we invested into
Leddartech, and their big strength is signal processing. With
their ASIC technology, which by the way, fits very well to our 905
nanometre laser technology. Also, we have a strategic stake in
a company called Blickfeld, close to Munich, which is strong in
optics and speed control. And last but not least, we have a
smaller investment in TetraVue, and they are strong in detectors
again, high resolution depth images.
So, if you look at it, if also, if you remember those key building
blocks, we have a strategic investment into all those important
building blocks, emitter, optics, detector, and signal processing.
And this enables us to evaluate an extension of our portfolio in
the LIDAR market in future. With that, I’m through my
presentation, I think you very much for listening, and I’m looking
forward to our Q&A session starting right now.
Sven Weier Yes, good afternoon, Dr. Strauss, thank you for the presentation.
A couple of questions from my side, maybe we can go through
them one by one. The first question I had is, you were talking
about the design wins of around 20, which to me, seems to be
more around the emergency breaking systems. I was more
curious if you look more forward, how many design wins you
have already for Level 3 upcoming cars for the next two years.
Dr. Joerg Strauss Yes, thanks for the question, Mr. Byer. Actually, those design
wins really are talking about forward looking, we are in contact
with all types of technologies, both markets, both traditional and
also mobility as a service. And so, this is not only emergency
Page 10 of 25
breaking systems, but it’s really also towards a Level 5, and
Level 3 functionality with different companies.
Sven Weier Thank you for that. And then obviously, talking about the various
technologies, obviously, there’s quite a range of companies
competing ranging from Velodyne to Leddartech, Luminar,
Innoviz, you name them. I mean is there any of the major players
that you wouldn’t be able to supply to because they use a
different type of laser technology, or is everybody just basically
using the one you use, the 905?
Dr. Joerg Strauss Yes, very good question. I think that’s a big strength of ours that
pretty much everybody can use our lasers, I mean as you said,
there’s many, many technologies out there in the market. There
are potentially a couple that use certain detector systems where
longer wavelengths, or even shorter wavelengths can work a
little bit better. There’s also for instance, optical phase array
technology, which is not perfectly suited for our current
technology. But the major part is really working with our lasers
at 905, or it’s actually the material system we are pursuing which
is not fixed at 905, it goes below 900 nanometres. And above,
almost to 1,000 nanometres. So, we cover a very broad space
which is applicable for pretty much all the different systems. And
we also see this in our contact to the market I would say that
more and more companies pursuing different technologies, are
even contacting us right now for our technology. Than the other
way around that they’re moving away from us.
Sven Weier And you were speaking about long-range LIDAR as well, and is
that also something that 905 is then the best technology for?
Dr. Joerg Strauss Yes, absolutely, I mean again, you need different, the
combination of different building blocks, and it depends a lot
which kind of detector you’re using. But fundamentally, the 905
nanometre space or technology has the highest output powers
right now. It’s very efficient so you don’t lose a lot of efficiency
when you apply this, and this is crucial, if you look at long-range.
Because otherwise you create too much heat, you assume too
much space, so I think specifically for long-range, our 905
nanometre is perfectly suited. It’s the best technology which we
have available, and again, I would like to emphasise this is not,
because we only have that technology, we also would have
other technology options in-house. But that’s our choice we
made, or we’re making right now that we will pursue that
technology for that specific application. It might be different for
other applications, but for that one definitely 905 nanometre
Page 11 of 25
Sven Weier And the last question from my side is I remember like one and
half years ago, I think you made an announcement together with
Infineon about a joint product, which back then I think was just a
prototype. I was just wondering where you stand on that
cooperation with Infineon.
Dr. Joerg Strauss Yes, also very good question, I mean the ecosystem is very
large in the LIDAR space, and our key, from an Opto
Semiconductor perspective, at that time the cooperation with
Infineon was specifically with Opto Semiconductors. We really
wanted to show a design, a reference design to be able to show
what LIDAR with a MEMS mirror, scanning LIDAR can do. And
I think we achieved that target, now we can use, or customers
can use any kind of MEMS, and for sure, we are in contact with
the entire ecosystem in that regard.
Sven Weier Okay, thank you very much Dr. Strauss.
Dr. Joerg Strauss Thank you.
Peter Olofsen Yes, good afternoon, I had a question on one of the earlier slides
where you showed the likely increase in sensors as we move
from Level 1 up to Level 4, 5. Because Infineon recently showed
a similar chart, and both your charts looked quite similar when it
comes to radar and camera. But for LIDAR, your assumptions
seem more aggressive. You talked about at least six LIDAR
sensors at Level 4,5, while Infineon was mentioning at least one.
So, to what extent do you think that this, at least six that you are
mentioning, is it a kind of consensual few? Or have you taken
in a difference sense from what others in the industry had been
assuming so far? And if your view is more bullish, than what
others seem to believe, why would that be?
Dr. Joerg Strauss Yes, thanks for the question also, Mr. Olofsen. I cannot really
speak about Infineon’s assumptions, but I can surely speak
about ours, which I’m happy to do. And we take our
assumptions primarily through the information we get out of the
markets, also out of our customers. And there are a couple, as
I mentioned before, which are pursuing Level 5 full autonomous
driving already today. And our belief is that you will have
coverage for both applications, long-range front LIDAR. And
this is very closely related to Infineon’s technology where you
apply a MEMS mirror. And this might be, but just guessing the
explanation, but you also see that you need to cover your 360
environment by short-range LIDAR systems, this is a
requirement, or a request our customers have. So, that’s the
reason why we believe there will be a coexistence of long-range
Page 12 of 25
sensor LIDARs, and also short-range counter LIDARs. And that
led to the assumption that there will be up to six or even more.
I can tell you that there are customers who put even more than
six LIDAR systems on the car right about now.
Peter Olofsen Okay, that’s helpful, thank you.
Dr. Joerg Strauss Thank you.
James Moore Hi, Joerg, thank you, great detail. Maybe I could ask a few
simple questions. I’m going back to the question of design wins,
can you say how many design wins you’ve had, and how many
design awards there have been to the market? I’m really trying
to understand what sort of market share you think you have on
design wins we’ve already seen across the three technologies.
Maybe we should go questions one at a time, that’s the first
question.
Dr. Joerg Strauss So, thanks for the question, James, yes, quite frankly I don’t have
the exact number of all the design wins we had over the last 15,
almost 20 years now in my head. But we had many in the
emergency breaking space, but we also have this number of
round about 20 in the upcoming technologies, and customers.
And we believe that we have a very strong market share, well,
well above 50% at this point in time. I mean, and I think we are
very well positioned, because our technology is very strong, but
for sure, competition is always there, we value our competition,
but we don’t fear our competition right now.
James Moore Thanks, and maybe I could ask about the ship set value for the
different sensors that you think what’s the average dollar value
of these things to you as we roll forward, when it becomes more
of a scale market in five years or so?
Dr. Joerg Strauss Yes, that’s a difficult one, because there is no winning
technology out there, and at the end of the day as you know,
always price and volume go hand and hand. There are systems
which have many, many lasers inside above 50, if those win,
then we will sell more than if systems win which have only four
channels for instance. So, this is too early to really tell you how
much the amount of value in the module will be with laser dyes,
or laser packages.
Andreas Spitzauer Also seen, kind of from the end of the day, as we have shown
on page number ten, so do we talk about kind of the Robo-
Cars, Robo-Taxis, of course when you can replace the driver.
And of course, such systems without a problem could cost
more than 10,000, €20,000 whereas when we talk about the
Page 13 of 25
consumer cars, any cost which are above thousand would be
difficult. And of course, which route you go, the offer of cost
also gives you an indication about the number of sensors you
need, and what is the ASP for them?
James Moore Thank you, Andreas, maybe I can try on the total size of the
addressable market potential, I don’t know if you can put any
rough numbers on how big you think the overall autonomous
driving sensor market will be. Your addressable market in say
2020 and 2025, and how much it will change between that
period. I’m trying to understand what we’re talking about in
revenue potential, really.
Andreas Spitzauer I think that’s also kind of again, not an easy question, I would
say kind of it depends which studies you use, kind of. If you
have kind of some from the investment banks, if you have
something from the Uhlig kind of institute, so the range is kind
of quite big from a three billion in 2018, coming up to kind of 11
billion in 2025. At the end of the day, it’s a young market, and
it’s also questioned when will kick in what? What we can say
today with kind of the products which we have in the market for
emergency breaking and kind of a first design wins. We are
currently on an annual revenue base of kind of ten million to 15
million, and of course the strategy is to grow this faster. At the
same time, as you can see with investments we have taken so
far, still have to find out where’s this market moving to? So are
we at the end of the day kind of only on the product side, on the
component side from the edge-emitting lasers which we sell to
this market. And VCSEL which we have now in-house, or kind
of do we offer more to the market, up to the module? Anything
in between is right, but it’s too early really to say how much
content will come from OSRAM, nevertheless as you can see,
with kind this nice market share already, and having contact to
all of them. It’s an important market, we are kind of a very
good player in this, and the growth rates are kind of quite
interesting.
James Moore Thank you, and lastly, quickly, I mean you mentioned that you
run the emitter sensor laser business unit of Opto, is that also
housing the overall infrared business? Therefore, are you
running a business that’s sort of 15% of Opto sales, or have I
got that wrong, does it exclude infrared? And if it excludes,
could you give us a rough size of the business that you manage,
please?
Andreas Spitzauer You are right, I manage the business which also includes
infrared, pretty much anything which is either not visible, or
and including laser, that’s my business. And yes, I’m not
sure…
Page 14 of 25
Dr. Joerg Strauss Yes, we kind of have the 15% turnover coming from lasers from
kind of infrared, which is across all the customer segments from
automotive up to industry up to consumer products, and this
across, but yes, no further split unfortunately.
James Moore Okay, thank you very much.
Andreas Spitzauer Welcome.
Sebastian Growe Yes, good afternoon, gentleman and Juliana, thanks for taking
my questions, three of them, and hopefully it’s not repetitive,
because I joined the call a bit late. Can you just let us know
what the actual sales volume is on these LIDAR products, to
start with? Second question would be on the customer mix, in
the slide number 11 you have illustrated that there’s obviously
traditional LIDAR customers, and on the other hand, you also
have the mobility service providers, Uber and the likes. So, can
you just give us a rough indication of what’s the breakdown of
the revenue mix between customer groups is for now, and
where you see the trending going forward? And then last one,
can you also give us an indication where the current share of
what is trend, if we are moving from each level to each level, are
you going from Level 3 to Level 4, etcetera, etcetera? Thank
you.
Dr. Joerg Strauss Sure, thanks for your question, yes, the first one, right now with
the emergency breaking systems, we talk double digit figures,
ten to €15 million revenue, and this is certainly increasing now.
And our customer base, design wind base is both in the mobility
as a service market, as well as in the traditional market. As
mentioned during the presentation, the traditional market is
using less LIDAR systems, and usually less lasers per system,
compared to the traditional market. And so, the importance of
mobility as a service is increasing for us as Opto
Semiconductors on the emitter side right now. But from a longer
perspective, both markets will be equally important for us, and
then, I cannot give you a Share of Wallet number. Because it
strongly depends, really which kind of technology, which kind of
system will win at the end, in both markets, and therefore it’s
tough. But as I said before, we follow both markets very closely,
we take them both equally seriously, and the good thing for us
is that both markets can use our lasers. And so, we participate
in both.
Sebastian Growe Okay, thank you. And if I may just ask a very, very quick follow-
up on the price level that you have observed over the last say,
two years or so, whenever the right say, starting point and pace
Page 15 of 25
might have evolved. Can you just give us an idea how that has
developed over time, and where you see that trending, what the
customer expectations are over time?
Dr. Joerg Strauss Do you mean the price for emitters, or for entire modules?
Sebastian Growe If you ask me that then I would like to have the answer to both.
Dr. Joerg Strauss Okay, unfortunately I cannot give you the answer for the
emitters, I can’t comment on that, but I can tell you, and this is
just coming out of all the studies which are available that we
have systems which cost up to $10,000 or even more right now.
And the expectation of the market is for entire module,
specifically we talk about traditional automotive market that it
has to be below $1,000. My personal assumption is that even
this is pretty high, but this is the range we are talking about, and
this is again, citing the studies which are out there.
Andreas Spitzauer This is also one of the reasons why we invested in Leddartech,
because the cost must come down, and one of the strengths of
Leddartech is that they are working on smart software to replace
more costly hardware. And this is kind of especially when it
comes to the ASIC, and this will be kind of one of the key driver
that LIDAR takes off, which is important that also autonomous
driving takes off.
Sebastian Growe All right, very helpful, thank you.
Dr. Joerg Strauss Thank you.
Marie Rupp Hello, good afternoon. I have a question, and because it is a
technology teaching, I want to go back a bit to technology. You
are telling that your clients want to have more power, I suppose
it’s more light power, but normally they don’t have infinite budget
for energy power. So, can you tell something about what is the
roadmap of the trends in terms of increasing the light power,
versus the energy, and what is the limit that can be used in this
system in the car? Because I believe the power can be also
used for something else in the car.
Dr. Joerg Strauss Okay, thank you for the question, Ms. Rupp, and thank you for
bringing it back to technology, I appreciate that. The answer to
your question is yes, customers require more power, and this
will go twofold. On the one hand side, we need to increase the
efficiency of the lasers further on, we are on a very good level,
but we need to further increase. That means you waste less
power on the car side, so that’s a task for us. In addition,
customers request more channels, so as I showed you before,
Page 16 of 25
our current chips have 75 watts peak power. And the customers
right now, the requests range from one to let’s say eight
channels per package, maybe even 16. So, this is the
magnitude we talk about, for instance if we talk about traditional
cars. In addition, what is very important, it’s the cycle time, or
the switch on, switch off time, the pulse time, which you use for
those lasers.
Because you switch them on and off very, very, very quickly, and
always only when they are on, they consume power out of your
car, so that’s a third area where you can improve, to improve
pulse time. And just have the laser on as long as you really need
it, and all those different factors, really for consuming power, the
efficiency of the laser has to go up, which it does. The pulse
time needs to go down, which it does, and last but not least, we
need to find the optimum of laser channels in the car, which the
customers work on. And we talk about one to eight, maybe 16
channels.
Marie Rupp And then I suppose that you are talking here exclusively about
the edge-emitting laser, but for the VCSEL technology, is there
also efficiency gain curve ahead? So that eventually you know,
they could come closer in this energy consumption question.
Dr. Joerg Strauss I mean fundamentally, the edge-emitting laser and the VCSEL
use the same material system, but they are still totally different
from a technology perspective. So yes, for sure we also will and
can improve VCSEL, but I just, or we are just not strong
believers that VCSEL will make it in this market. And this is
purely because of size of the system, and usually the progress
you get in the performance, on the edge-emitter will then be
translated also in VCSEL. So, you will have a certain gap
between the two, until it comes to the theoretical limit, and then
they will merge. But the area aspect stays and remains, and
that’s a bit minus for VCSEL, so if you ask me, to really
specifically answer your question, there will be a gap between
edge-emitting and VCSEL for a couple of years. Until you really
step into the theoretical limit, yes, but the big disadvantage of
space consumption by VCSEL stays in this specific application
of LIDAR.
Marie Rupp Okay, and then on the technology, when we talk about these
design wins, are these design wins already approved by
regulation, in terms of the ECE? Or is it still a step that has to
be made ahead before we can consider that this car will really
hit the road one day?
Page 17 of 25
Dr. Joerg Strauss That’s a difficult question because I cannot fully answer, I mean
at the end of the day, the company which builds the module, and
introduces it to the car, has to assure and ensure that it meets
regulations. But it has to meet the regulation, no matter when
you put it in, so if you put a LIDAR system in today, it has to
meet the regulations as well as in future. So, you can assume
that every system which we’ve introduced into the market will be
compliant to those regulations. But for sure, if you first optimise
your system, and this is now going more into the detector area,
if you further prove your detectors. You can hopefully work with
less and less power, but it’s still substantially the power, it will
not go to only a couple of watts.
Andreas Spitzauer If you consider right now, the Audi A8 is kind of the highest level
which is on the road to be available. So, when we kind of talk
about Level 3 or above, this above is obviously not approved,
otherwise it would be on the market. But of course, it would be
working with the customer that we come to this.
Marie Rupp Okay, I have a final one about this number of LIDAR around the
car. I have the impression that this is the wrong debate, because
I think what is more important is the number of channels per
LIDAR. And apparently LIDAR will have different numbers of
channels, so, your revenue normally, or the endearment would
depend from the total number of channels, and not the number
of LIDAR. And then, apparently LIDAR will do different things,
you know, have different applications in the car function. Is it the
right way to think about this?
Dr. Joerg Strauss Yes, this is absolutely correct, this also makes it hard for us to
provide you with numbers, forward looking numbers, because it
absolutely depends on which systems will be implemented, and
how many lasers they will utilise. You’re absolutely correct, and
there is quite big differences in the market, and we have to see
which are the systems which will make it.
Marie Rupp Thank you.
Dr. Joerg Strauss Thank you.
Alok Katre Thanks for taking my questions, a couple that I have, one, I just
wanted to understand obviously you now have VCSEL and EEL
sort of technology as well, especially for the amateurs, like you
said. But how does the, let’s say the production process, or the
product development process, how does that relay across
those, let’s say product lines? I.e. if you decide at some stage
that you know, there are applications let’s say like you said for
Page 18 of 25
short range, LIDAR within the VCSEL, let’s say using the
VCSEL technology. I mean do you have to undergo a
completely different, and long let’s say process to make sure it’s
you know, you can use VCSEL in the commercials, you know,
say field? Or you know, because the underlying materials,
technologies are roughly similar, you can sort of let’s say do that
with limited, let’s say technological difficulty? So, that’s number
one.
Dr. Joerg Strauss Thanks for your question, I’ll probably answer this first, and then
we’ll step to the next question, thanks, Alok, for that one. Yes,
it is similar material systems, but quite frankly, the edge-emitting
laser technology, and the VCSEL technology are quite a bit
different, also from processing standpoint. But the VCSEL
technology is very close to our infrared emitter technology,
which we also which is a big and strong business of ours, which
I have the pleasure to also head. So, we have strong synergies
with our existing technologies, but it is on the AP side, yes,
between the VCSEL and the edge-emitting laser. But it’s
fundamentally and much more pronounced on the infrared
emitter and the VCSEL, there it’s very similar and that’s the
strengths we build on for the VCSEL point of view.
Alok Katre Right, so if at some point you decide that VCSEL will work for
let’s say short-range LIDAR, and that sort of picks up, then you
don’t really… It’s not like rocket science for you to use what you
know to translate that into a, you know, let’s say a proper
product for let’s say the short-range LIDAR, that’s not the case.
I mean you can do it with relative ease.
Dr. Joerg Strauss That’s absolutely correct, yes, absolutely correct, we are very
open to do that, we just have our doubts that this will actually
happen, but if it does, yes, we are prepared.
Alok Katre Right, and from a production equipment and so on, you know,
the MOCVD and everything else that you need, again, is even
there VCSELs are quite close to infrared or IR LEDs as such, or
is that, you know, slightly different ballgame?
Dr. Joerg Strauss It’s very similar.
Alok Katre Okay, great, thank you, thanks. My second question, you sort
of mentioned within you know, the building blocks, you know,
okay, you have emitters, you have detectors, you have the
signal processing, and so on. Now obviously, you are pretty
strong in the emitters side, you have a stake in you know, the
signal processing with Leddartech and everything else. But you
Page 19 of 25
said you’re exploring the detection side of things, and you know,
how you want to go ahead. You know, how should we think
about you know, the kind of technology that you have in-house?
Versus what you would perhaps need to let’s say look outside
as far as the area is concerned as well. And if I were to break
down let’s say those building blocks from a value chain
perspective, I mean where is the most value within those three
building blocks?
Dr. Joerg Strauss Yes, okay, thanks for that question. I think we have a very good
starting point because without light, there’s no LIDAR. And that
means our fundamental base, which is the emitters, is a great
base to be. But then, LIDAR is a very complex space, many
players are there, and it’s a complex ecosystem. And with our
strong stake in the emitters, we reached out to the ecosystem,
not only by working with the companies. But also, taking
investments, to be ready to take the next step, if reasonable.
And this is what we are exploring right now, and I think we are
very well prepared with the partners we have and established.
The Share of Wallet, it’s again, I cannot answer this right now,
because it strongly, strongly depends on the different building
blocks, which ones are being used? I mean for instance, if you
use an optical phase array against a MEMS mirror or a flash, it’s
just so different. Or if you use a PIN photodiode versus an
avalanche photodiode as clinic and photomultipliers, you know,
there is so, so many underlying technologies which all have
different fundamentals, all of different cost points. So, it’s just
you know, it’s just not… You cannot answer this question right
now.
Alok Katre I mean in terms of what I was trying to sort of get at was you
know, you have the emitters, you have the detectors, and they
have the signal processing as such. Normally if I were to sort of
think about you know, let’s say some related technologies,
typically it’s the electronics and the software that tend to, if I
were to equate it to that. Then that’s where most of the, you
know, value within the chain lies, versus the hardware. So, I
was just trying to think about it from that perspective, that you
know, the emitters and the detectors are the hardware. And
then you probably have the signal processing, which is the
software. And you know, is it logical to assume that the signal
processing is where most of the intelligence will lie. And
therefore, that’s where most of the value will be within these
three portions?
Page 20 of 25
Dr. Joerg Strauss Yes, I’m not so sure about the value per se, but you’re definitely
right that signal processing is very important, that’s the reason
why I invested into Leddartech, to be able to you know, have a
foothold in that space as well. And that you will see, but you
know, also what is an issue, and I believe that many start-ups
might struggle, is industrialisation. Quality, which is required in
the automotive space is very, very strict, and to be able to
manufacture and supply for the automotive market is a big
hurdle. We know how to handle that hurdle because it’s our
home turf, and that’s also an area which you shouldn’t neglect.
But yes, signal processing is definitely very important for LIDAR.
Alok Katre Fair enough, and my last sort of question on this I think was
following up you know, question around the energy let’s say
consumption, versus the energy ability within the car. Is you
know, obviously you know, given where we are moving with let’s
say electric vehicles as well and so on and so forth. Is there let’s
say a bit more pressure from your customers that okay, we you
know, suddenly the number of channels, etcetera, is fine? But
we also need a lot more efficiency, like you said, you know, do
you really need to go back to the drawing board for some of the
technologies for specifically for the use in electric vehicles?
Versus you know, let’s say the conventional powered vehicles,
which obviously might also have you know, this LIDAR. Or do
you think that actually LIDAR, by the time it comes on the market,
it will more or less be used only for electric vehicles, and
therefore that’s how development should go?
Dr. Joerg Strauss Yes, very good question. Quite frankly, so far, the power
consumption is not a big issue, is we discuss about LIDAR. My
reading is that it’s because it’s pulse systems, so you do not use
the peak powers for a very long time. It is more of an issue if
you talk automotive lighting, you always switch it, have the
device switched on, you know. This is where this discussion is
starting now, it’s not that much of a key issue in the LIDAR
discussions. The fundamental LIDAR discussion is really, if you
say in simple words, to make it work, you know, we are still in
the early, early phase. And there is no LIDAR system yet which
really fully is capable of doing what is required to do in the long
term, and this is the primary focus. Power consumption is not
the big focus right now.
Alok Katre Okay, fair enough. And sorry, just the last bit really, if I
remember I think you know, when the Leddartech investment
was made. I think one of the points that was mentioned really
was that you know, the aim was to get let’s say the price of the
Page 21 of 25
LIDAR unit down to less than 100. Now leaving some of the
commercial sort of aspects aside, you know, given what we
focused on, from technology perspective, where do you think
you need to go? What do you think you need to develop to get
this let’s say price… Am I right on the price point, first of all?
And where do we, you know, what needs to happen between
now and you know, by the time the price point sort of works? Is
it more efficiencies, you know, is it different system itself that
needs to be developed, or just want to understand from that,
and where do you stand, you know, from that perspective?
Dr. Joerg Strauss I would say when it comes to LIDAR, it’s really at early stage,
and as you kind of have seen, there are many kind of different
idea how to make LIDAR working, which then enables
autonomous driving. It’s a little bit too early, we have kind of
done our homework, and kind of have good own product, and
have done investments. And at the end of the day, the whole
system cost must come down, it’s a little bit too early to go very
specific into this, because it’s too early to see what is the
standard at the end of the day.
Alok Katre Okay, great, thank you.
Dr. Joerg Strauss Thank you.
Uwe Schupp Yes, thank you, Dr. Strauss. Two questions, please, hopefully
rather easy ones, basically just on the dye size of the edge-
emitting and the VCSEL lasers in a LIDAR system. I think earlier
on your presentation, you indicated that it may be, at least on the
VCSEL side, only two to three millimetre squared. Which
sounded a bit low to me, I was just wondering whether I got you
correctly there, and why that may be the case? Because I heard
different numbers, actually significantly higher than that.
Dr. Joerg Strauss Yes, thanks for that question, and sorry, I was not clear on that
point. No, you’re right, if you talk about a reasonable area which
you would like to consume, you talk two, three square
millimetres. Which is actually the VCSEL space you use in all
the mobile applications. But you only have, I don’t know, up to
three watts, and actually the emitting area for a edge-emitting
laser is much less than two to three micromillimeters, it’s just this
very tiny facet which is emitting the light. That was exactly my
point, which I wanted to stress, so for two to three millimetre,
square millimetre VCSEL size, we talk very few watts. But we
need maybe hundreds of watts, so by that magnitude, your
VCSEL area would need to grow so that you’re even able to put
up that amount of power. And this is then just a very tough to
Page 22 of 25
digest size in a car, I mean you don’t want to carry a trolley
behind your car to make the LIDAR functionality happen, right?
Sorry for not being clear.
Uwe Schupp No, that’s very helpful, thank you. And for the edge-emitter, how
big would the dye size be typically for a wallet, you know, 100
or whatever, system?
Dr. Joerg Strauss It depends, I mean what you do to increase the power, you
increase the length of the chip in the direction, but also the facet,
so we talk, I don’t know, 70 micrometres up to a couple of
hundred micrometres, so they’re fairly small.
Uwe Schupp And then just lastly, on the MOCVD side, is it typically made on
a gallium arsenide system, or can you also use these cheaper
gallium nitride systems using the phosphor trick, basically using
general lighting? Thank you.
Dr. Joerg Strauss Yes, no, it’s gallium arsenide based, it’s direct emission, you
cannot use phosphorous for this kind of application out of various
reasons. If you use a phosphor, you don’t keep the directionality,
you change from a directional light source to a Lambertian light
source, which direct light in all areas. And also, there is no really
good phosphorous available, which emit in the infrared space.
So, we talk direct infrared emission, typically gallium arsenide
space. If you talk below 1,000 nanometres, Indium Phosphide,
if you go above 1,000 nanometres.
Uwe Schupp That’s very helpful, thank you very much.
Dr. Joerg Strauss Thank you.
Peter Reilly Well if I’ve got one question, my question is, you talked about
20 wins in the backlog, can you talk about whether they are
single source or a multi-source? I’m just interested to
understand the competitor’s dynamics. Whether these new
technologies, people are typically partnering with a single
supplier for a single programme or project. Or whether they’re
specifying multiple suppliers for obvious reasons of price and
security of supply, and so forth.
Dr. Joerg Strauss Yes, thanks for the question, Peter. We do not know all of our
customers, whether they have a single or a multi-source
strategy. The amount of competitors in that space is limited, for
instance, if you compare it to LED. So, we know that for some,
we are single-source, for some, we just don’t know, and we
assume there’s multi-source, so it’s a mixed ballgame.
Page 23 of 25
Peter Reilly And if I could have a quick follow-up, the competitors you talk
about, are they big names I’d recognise, or are you competing
with small start-ups as well who don’t have a direct record in the
automotive space?
Dr. Joerg Strauss You would recognise most of the companies, definitely, and so
we also went through and defined key customers for ourselves,
and for sure you want to work with the big guys. But we do not
necessarily restrict ourselves, so it might be small companies,
but definitely the big players are among them.
Peter Reilly Thank you.
Dr. Joerg Strauss Welcome.
Guenther Holfelder It’s one question, hello, can you talk about the involvement of
you know, of the OEM and of tier ones in the design process?
Dr. Joerg Strauss It depends very, very strongly on the very specific customer, but
I think to make it simple, we talked to both tier ones and OEMs.
But you cannot, it really depends on the very specific project
we’re talking about. Sometimes we only talk to the tier one
supplier to the market, sometimes we talk only to the OEM,
sometimes we talk to both.
Guenther Holfelder One follow-up, I understand Audi is using emitters from a
competitor in the A8, so based on your design wins, when will
we see a car with your emitters on the road?
Dr. Joerg Strauss You can see already many cars on the road with our emitters
since 2002, so if you see a car with a working LIDAR system,
probably except one, there’s a high likelihood that it’s OSRAM.
Guenther Holfelder But I’m also referring to a real LIDAR system in a serous car, you
know, like saying the Audi A8 is the first car that will be able to
reach Level 3 autonomy supported by a LIDAR system.
Dr. Joerg Strauss Yes, I mean that’s hard to judge because on a specific level,
what I would think that the entire industry is shooting strongly for
the timeframe 2020, 21, to have LIDAR on the street. And as we
are a strong player, I’m very positive that this will also involve our
lasers. But I mean we have also full LIDAR systems already cars
on the street, it might not be your typical traditional player, but
so not only in the emergency breaking, but also in the full LIDAR
systems, there are cars.
Guenther Holfelder Referring to the Velodyne slayer more, okay, many thanks, yes.
Dr. Joerg Strauss Thank you.
Page 24 of 25
Sven Weier Yes, thank you for taking my follow-up question, it refers to the
front sector of LIDAR, because as we know, even the later
systems are still pretty bulky. So, I think in the interest of design,
what do you see long term as a possibility in more integration
with the lighting units in the headlight? I mean do you see some
synergies with your lighting colleagues there? Do you think that
could also then be another competitive advantage for you,
because you have all that know-how?
Dr. Joerg Strauss Yes, that’s a very good question. I mean certainly, we should
look into deliberation opportunities, and as I mentioned before,
we have already on our roadmap that we will involve an ICE,
driver ICE for the emitter. And then you have to figure out what
makes sense, how much functionality you put on the tube, or on
your own device. And certainly, in the market, there is thought,
I mean you need to put those devices somewhere, and one area
which could make sense is the headlight. And there are
companies out there which pursuing that route, and yes, I think
there can be synergies. Because we are strong in lighting
modules, as you know, we developed the first laser module for
visible light. So, we have knowledge and we definitely can
utilise this knowledge also for future LIDAR. So, I think yes,
there’s certain synergies in general.
Sven Weier Because I mean that goes also to the various entry, right,
because I wonder in terms of making the edge-emitting lasers,
what the real barriers to entry are for competition to come in. Is
it patents, or what is it really that protects you, because so far,
it’s been probably a bit of a niche market, it hasn’t really grown
that much. But all the sudden you have you have these new
applications opening up quite some nice growth.
Dr. Joerg Strauss Yes, I mean there’s a couple of things which are important if you
want to win in the automotive space. First one, as you
mentioned, yes, very important IP, and we have strong IP
portfolio both on LED as well as on lasers, that’s definitely a plus.
Second, our strong technology position, it is not that easy to build
an edge-emitting laser, and it’s just not as established as other
semiconductor technologies, so there’s just much less players.
This is a certain protection, at least it gives you a time benefit.
Third, the reliability which is required for automotive is very, very
important, as I mentioned before, we were the first ones to have
an automotive grade laser introduced to the market. So also that
expectation by the automotive customers on reliability is very
important. And last but not least, also, our system know-how,
when it comes to light systems and it actually doesn’t matter too
Page 25 of 25
much if it’s visible light or non-visible light. I think those are the
four key protection areas, IP, quality, technology, and know-how
about automotive systems.
Sven Weier Thank you.
Dr. Joerg Strauss Thank you.
Juliana Baron Thank you, Joerg, for all your insights into LIDAR technology,
and thank you all for joining our first Tech Teach-In. In order to
optimise our future Tech Teach-In, your feedback would be
highly appreciated to us, and we prepared a very short survey
taking about three minutes to respond. So, please follow the
link next to the presentation window, and we would highly
appreciate your feedback. Thank you very much.