BRN Discussion Ongoing

[Fullmoonfever]

Mercedes taps Intel Liohi2 for neuromorphic AI

Mercedes is using the Intel Liohi 2 for neuromorphic computing to improve the performance of forward-facing automotive radar systems.

www.eenewseurope.com

Mercedes taps Intel Liohi2 for neuromorphic AI​

Technology News | October 14, 2024
By Nick Flaherty
Automotive AI

---

Mercedes is leading a project in Germany to use neuromorphic computing to improve the performance of forward-facing automotive radar systems.​

Mercedes is using the Intel Liohi 2 neuromorphic event-driven AI processor in the Naomi4radar project, says Markus Schäfer, Member of the Board of Management of Mercedes-Benz Group and Chief Technology Officer for Development & Procurement. Using neuromorphic, event driven algorithms and processors can increase the speed of response of the radar systems, he says.

The project is funded by the German Federal Ministry for Economic Affairs and Climate Action to assess how neuromorphic computing can be used to optimise the processing of radar data in automated driving systems.

Mercedes is also working with Karlsruhe University of Applied Sciences on neuromorphic cameras, also known as event-based cameras.

The main focus for neuromorphic computing research is an agreement with the University of Waterloo in Canada with on research led by Prof. Chris Eliasmith.

The focus is on the development of algorithms for advanced driving assistance systems. By mimicking the functionality of the human brain, neuromorphic computing could significantly improve AI computation, making it faster and more energy-efficient.

While preserving vehicle range, safety systems could, for example, detect traffic signs, lanes and objects much better, even in poor visibility, and react faster. Neuromorphic computing has the potential to reduce the energy required to process data for autonomous driving by 90 percent compared to current systems.

The work with the University of Waterloo complements a series of existing Mercedes-Benz research collaborations on neuromorphic computing that includes end-to-end neuromorphic AI learning for autonomous driving.

Startups in the the Ontario Vehicle Innovation Network (OVIN) will also join the growing international Mercedes-Benz STARTUP AUTOBAHN network. The initiative aims to drive transfer research and development to industrialisation.

The OVIN Incubators Program will focus on identifying and fostering innovation in future software & AI, future vehicle components and future electric drive. Working with startups, and in partnership with OVIN, Mercedes-Benz will help progress promising projects through the provision of its specialist expertise and use cases.

“Innovation is part of Mercedes-Benz DNA. In our global R&D strategy, open innovation gives us rapid and direct access to the latest ideas and developments around the world,” said Schäfer. “We are therefore delighted to further expand our activities in Canada as a founding partner of the OVIN Incubators. In a fast-paced environment, it is another important channel for developing exciting future products and elevating our customer experience through new technologies.

In its pilot phase, the OVIN Incubators Program will conduct startup scouting to identify opportunities in Ontario relevant to Mercedes-Benz fields of research. The aim is to empower startups to engage with industry and establish a robust pipeline of companies whose growth can be catalysed. Together, OVIN and Mercedes-Benz will narrow down an initial longlist through a process of evaluation, ultimately arriving at individual projects that will progress to proof-of-concept based on Mercedes-Benz use cases.

https://group.mercedes-benz.com/company/news/open-innovation-canada.html

The key take away for me is the outlined main focus... algorithms.

Whilst they may be using Loihi within their R&D environment there has been nothing to confirm that we are no longer in the background especially given they proved up our offering in the EQXX.

Now, imo we are a processor primarily, HW or SW and as such clients need specialised use case specific algos to run on it.

Is it possible that this MB / UoW agreement is as it states, specifically around the ADAS algo dev needs that I suspect can be transferred to something like Akida if desired and ready to implement commercially?


The main focus for neuromorphic computing research is an agreement with the University of Waterloo in Canada with on research led by Prof. Chris Eliasmith.

The focus is on the development of algorithms for advanced driving assistance systems. By mimicking the functionality of the human brain, neuromorphic computing could significantly improve AI computation, making it faster and more energy-efficient.

 

[MDhere]

Agree, however it won't be a research chip forever. We are yet to see widespread commercial uptake and time is running out. We are talking a good game but are we starting to loose to competitors?

Mercedes are adopting Brainchip for Hey Mercedes, as evidenced in previous articles by Mercedes. The topic of intel research chip is just that - research. I looked back on articles, LinkedIn posts, etc and MB is hot for neuromorphic computing and its reasonable that they would like to do research, as they mentioned the leaders in neuromorphic computing IS BRAINCHIP and intel.
Hey akida, Hey Mercedes, music to my ears
Even tata took note
Of course my opinion based on my research.
Hey fellow brners happy Tuesday

 

[FJ-215]

Not concerned about Mercedes right now. Our tie up is with the 'furureistic' car not current or near term production models.
As a matter of course they would test other producers.
IMO our relationship with say Tata should produce results sooner. Sean and Charman Tony said at the AGM we have engagements in progress.
Most BRN competitors specialise in 'niche' areas such as vision, robotics only etc whereas BRN does the lot.
Basically only BRN, Intel and IBM are into neuromorphic 'with the lot'. I am not sure how advanced Intel and IBM are???
Being a player with the 'lot' means we are attractive to big business such as Tata who have fingers in many industry pies.

Hi manny,

Quick question, what do you have against Antonio?

I'm not a fan of his "The S/P will do what it does" comment but other than that I don't have any concerns with him.

Charman = Janitor

 

[Diogenese]

100% agree.
With the $$ costs we have been told in the past of the Loihi research chip, Im just not buying it being in any production run. The value of the car would skyrocket. I reakon its all just a continuation of laboratory research of the ability and prospects of the technology in general. When it comes to production runs, surely it will need to be at Akida prices??

Hi Wags,

I saw something the other day from Valeo which said the value of software in the software defined vehicle, ie, all new cars, will increase from 10% today to 30% by 2030. So our new algorithm product gives us a toe in that door.

I'm prety sure that, in addition to the up front costs, there will be ongoing subscription fees to keep the software updated - money for jam.

 

[Esq.111]

Yet another day in the life of a BrainChip holder... nothing to see here, it's all under control.



Regards,
Esq.

 

[7für7]

Yet another day in the life of a BrainChip holder... nothing to see here, it's all under control.



Regards,
Esq.

 

[manny100]

Hi manny,

Quick question, what do you have against Antonio?

I'm not a fan of his "The S/P will do what it does" comment but other than that I don't have any concerns with him.

Charman = Janitor

Hi FJ-215,
Not sure why you think i have something against the Chairman.
You said "Quick question, what do you have against Antonio? " My post in no way indicates that i have anything at all against Chairman Tony.
I call him Chairman Tony because i can never remember his full name. We also have another Tony CTO.
I am supportive of the Chairman.

 

[FJ-215]

Hi FJ-215,
Not sure why you think i have something against the Chairman.
You said "Quick question, what do you have against Antonio? " My post in no way indicates that i have anything at all against Chairman Tony.
I call him Chairman Tony because i can never remember his full name. We also have another Tony CTO.
I am supportive of the Chairman.

All good,
Maybe because I'm in Victoria and we lived under Dictator Dan, Chairman Tony comes across as a diss.

 

[Pom down under]

Megachips seems to be pushing the university’s just like BRN

Contribution to Society | MegaChips Corporation

www.megachips.co.jp

 

[MegaportX]

OCP Global Summit 2024, San Jose Oct 15-17.

 

[MDhere]

@buena suerte :-) why do I have your name in my diary for sat 26 Oct marked Champagne??

 

[Rach2512]

Knowing we are partners with Intel Foundries is there any chance Intel could be incorporating Akida IP into Liohi?

Good question, also why would they promote Brainchip, where is that photo where we appear on stage on the screen in the bottom left corner of one of Intel's presentations. Sorry struggling to put my sentences together, been staring at spreadsheets all day, my brain is fried.

 

[DingoBorat]

Mercedes taps Intel Liohi2 for neuromorphic AI

Mercedes is using the Intel Liohi 2 for neuromorphic computing to improve the performance of forward-facing automotive radar systems.

www.eenewseurope.com

Mercedes taps Intel Liohi2 for neuromorphic AI​

Technology News | October 14, 2024
By Nick Flaherty
Automotive AI

---

Mercedes is leading a project in Germany to use neuromorphic computing to improve the performance of forward-facing automotive radar systems.​

Mercedes is using the Intel Liohi 2 neuromorphic event-driven AI processor in the Naomi4radar project, says Markus Schäfer, Member of the Board of Management of Mercedes-Benz Group and Chief Technology Officer for Development & Procurement. Using neuromorphic, event driven algorithms and processors can increase the speed of response of the radar systems, he says.

The project is funded by the German Federal Ministry for Economic Affairs and Climate Action to assess how neuromorphic computing can be used to optimise the processing of radar data in automated driving systems.

Mercedes is also working with Karlsruhe University of Applied Sciences on neuromorphic cameras, also known as event-based cameras.

The main focus for neuromorphic computing research is an agreement with the University of Waterloo in Canada with on research led by Prof. Chris Eliasmith.

The focus is on the development of algorithms for advanced driving assistance systems. By mimicking the functionality of the human brain, neuromorphic computing could significantly improve AI computation, making it faster and more energy-efficient.

While preserving vehicle range, safety systems could, for example, detect traffic signs, lanes and objects much better, even in poor visibility, and react faster. Neuromorphic computing has the potential to reduce the energy required to process data for autonomous driving by 90 percent compared to current systems.

The work with the University of Waterloo complements a series of existing Mercedes-Benz research collaborations on neuromorphic computing that includes end-to-end neuromorphic AI learning for autonomous driving.

Startups in the the Ontario Vehicle Innovation Network (OVIN) will also join the growing international Mercedes-Benz STARTUP AUTOBAHN network. The initiative aims to drive transfer research and development to industrialisation.

The OVIN Incubators Program will focus on identifying and fostering innovation in future software & AI, future vehicle components and future electric drive. Working with startups, and in partnership with OVIN, Mercedes-Benz will help progress promising projects through the provision of its specialist expertise and use cases.

“Innovation is part of Mercedes-Benz DNA. In our global R&D strategy, open innovation gives us rapid and direct access to the latest ideas and developments around the world,” said Schäfer. “We are therefore delighted to further expand our activities in Canada as a founding partner of the OVIN Incubators. In a fast-paced environment, it is another important channel for developing exciting future products and elevating our customer experience through new technologies.

In its pilot phase, the OVIN Incubators Program will conduct startup scouting to identify opportunities in Ontario relevant to Mercedes-Benz fields of research. The aim is to empower startups to engage with industry and establish a robust pipeline of companies whose growth can be catalysed. Together, OVIN and Mercedes-Benz will narrow down an initial longlist through a process of evaluation, ultimately arriving at individual projects that will progress to proof-of-concept based on Mercedes-Benz use cases.

https://group.mercedes-benz.com/company/news/open-innovation-canada.html

So this guy Nick Flaherty, is supposed to be a "Freelance Technology Journalist" and he spells Intel's neuromorphic chip wrong, in the misleading Headline and throughout the piece??..

Although his services obviously weren't "free" It's apparent, whoever funded the article, didn't pay "that" much..
(And I know we shouldn't be throwing stones, about spelling errors etc )


I say the Headline is misleading, because it almost suggests, that Mercedes has "chosen" Intel, for neuromorphic.

And I definitely don't discount Intel (as LDN once also said) but I'm more concerned about their history of anti-competitive business practices, than any technological threat.



This is 7 years old (but hey it "is history" right?)
And at over half an hour long, it won't be watched by the Pom (although he may relate to the accent).
The title of the clip, says a lot.
Intel - Anti-Competitive, Anti-Consumer, Anti-Technology


It's just heresay and I've said it before (he also has reason to be biased) but Peter once commented, that AKIDA (1st Gen) was better than Loihi and TrueNorth put together.

AKIDA 2.0 with the addition of TENNs, isn't even in the same room, as Loihi 2, or the antiquated TrueNorth (this is obviously in my layman, biased opinion, but I believe it can be supported by facts).

I don't doubt at all/it wouldn't surprise me, if Intel was paying Mercedes, to play with Loihi 2.

The same is certainly not true, of BrainChip.

 

[Wags]

Hi Wags,

I saw something the other day from Valeo which said the value of software in the software defined vehicle, ie, all new cars, will increase from 10% today to 30% by 2030. So our new algorithm product gives us a toe in that door.

I'm prety sure that, in addition to the up front costs, there will be ongoing subscription fees to keep the software updated - money for jam.

Dio, and all,
Wondering if Intel could sprinkle a little akida IP in there future ingredient list for Loihi2 HW, but in the meantime MB have TENNS talking lingo to Intel as a simulation. Given Im guessing it is in research, the goal posts can move around. Then sure, the headline can read MB using Loihi2 which is not a lie, and maybe the security of Intel with its future fab and Govt contracts gives security.
But then again, maybe its just this tempranillo clouding my brain.
Who is that company mentioned at the AGM, if they take a licence, Brainchip becomes profitable overnight?
Anyways, the lamb is nearly ready, have fun.

 

[Diogenese]

The key take away for me is the outlined main focus... algorithms.

Whilst they may be using Loihi within their R&D environment there has been nothing to confirm that we are no longer in the background especially given they proved up our offering in the EQXX.

Now, imo we are a processor primarily, HW or SW and as such clients need specialised use case specific algos to run on it.

Is it possible that this MB / UoW agreement is as it states, specifically around the ADAS algo dev needs that I suspect can be transferred to something like Akida if desired and ready to implement commercially?


The main focus for neuromorphic computing research is an agreement with the University of Waterloo in Canada with on research led by Prof. Chris Eliasmith.

The focus is on the development of algorithms for advanced driving assistance systems. By mimicking the functionality of the human brain, neuromorphic computing could significantly improve AI computation, making it faster and more energy-efficient.

Hi FMF,

The funny thing is, that since the advent of TENNS, research on algorithms for ADAS is reinventing the wheel. Actually, Valeo has several patents for signal processing, including for radar.

This is a recent one:

WO2024088937A1 METHOD TO ANALYZE AT LEAST ONE OBJECT IN AN ENVIRONMENT OF A SENSOR DEVICE FOR A VEHICLE 20221027

a method to analyze at least one object (23) in an environment (5) of a sensor device (2) for a vehicle (1). The method comprises:
Providing (S2) a sensor information (12) describing the at least one object (23) by at least one measurement point (11) located at a first radial distance (r1) to the sensor device (2);
determining (S4) an artificial sensor information (13) describing at least one artificial point (14) in the environment (5), wherein the artificial point (14) is located at a second radial distance (r2) to the sensor device (2) that is greater than the first radial distance (r1) by a predetermined distance value (Δr); and
analyzing (S6) the at least one object (23) by applying an analysis algorithm (21) on both the provided sensor information (12) and the determined artificial sensor information (13).

The analysis algorithm can be CNN.

Thus the Akida 2/TENNS simulation software could perform this function.

https://brainchip.com/brainchip-introduces-second-generation-akida-platform/

BrainChip Introduces Second-Generation Akida Platform​

Introduces Vision Transformers and Spatial-Temporal Convolution for radically fast, hyper-efficient and secure Edge AIoT products, untethered from the cloud
Laguna Hills, Calif. – March 6, 2023

• Supports all types of Convolutional Neural Networks (CNN), Deep Learning Networks (DNN), Vision Transformer Networks (ViT) as well as Spiking Neural Networks (SNNs), future-proofing designs as the models get more advanced.

 

[7für7]

So this guy Nick Flaherty, is supposed to be a "Freelance Technology Journalist" and he spells Intel's neuromorphic chip wrong, in the misleading Headline and throughout the piece??..

Although his services obviously weren't "free" It's apparent, whoever funded the article, didn't pay "that" much..
(And I know we shouldn't be throwing stones, about spelling errors etc )


I say the Headline is misleading, because it almost suggests, that Mercedes has "chosen" Intel, for neuromorphic.

And I definitely don't discount Intel (as LDN once also said) but I'm more concerned about their history of anti-competitive business practices, than any technological threat.



This is 7 years old (but hey it "is history" right?)
And at over half an hour long, it won't be watched by the Pom (although he may relate to the accent).
The title of the clip, says a lot.
Intel - Anti-Competitive, Anti-Consumer, Anti-Technology


It's just heresay and I've said it before (he also has reason to be biased) but Peter once commented, that AKIDA (1st Gen) was better than Loihi and TrueNorth put together.

AKIDA 2.0 with the addition of TENNs, isn't even in the same room, as Loihi 2, or the antiquated TrueNorth (this is obviously in my layman, biased opinion, but I believe it can be supported by facts).

I don't doubt at all/it wouldn't surprise me, if Intel was paying Mercedes, to play with Loihi 2.

The same is certainly not true, of BrainChip.

You smart little …
you made my day…I give you a 10/10…good point… GOOD POINT

 

[Tothemoon24]

The Software Defined Vehicle – the SDV – is a real turning point in automotive history.

Until now, the car as we know it has remained the same throughout its entire life. A 15-year lifespan, with no changes, no upgrades. Thanks to the SDV, the car will be updated, upgraded and customized, throughout its entire lifecycle. You’ll be able to add or remove functions according to your needs. Just like your smartphone.

I had the opportunity today to share Valeo's vision of the SDV, its stakes, challenges and opportunities, during a keynote at Paris Automotive Summit, organized by the PFA - Plateforme automobile, on the occasion of Mondial de l’Auto - Paris de l’Auto .

SDV creates a completely new business for OEMs, as they will be able to perceive recurring revenues well beyond the sale of the vehicle, through subscription models similar to Netflix or Spotify. After video-on-demand, we will now have in vehicles features-on-demand.

You may wonder : what additional features? What if I don’t need additional features ? Well, people were asking the same questions at the beginning of smartphones. It will be up to our industry to invent new, desirable functionalities that bring real added value to users.

The widespread adoption of new functions will depend on the number of cars in the fleet that are sufficiently equipped to receive them. The car will have to be designed and equipped from the outset to ensure that it can evolve, be upgraded and updated for 10 to 20 years. Being too conservative at design with hardware equipment and investment would be to the detriment of its ability to be upgraded over time.

The shift to SDV must be collective, implying a whole ecosystem, including players well beyond automotive.
It requires closer collaboration between players with complementary skills, from operating systems, to middleware, to applications, to cloud and cybersecurity to name a few.
It also requires moving from a very top-down OEM and supplier hierarchy to a much more networked and partnership-based working model. The key words are co-development and open-innovation.

Valeo will be a major player in this revolution. Our ambition is to be the benchmark partner for our automaker customers, building the SDV with them, alongside them, in dynamic ecosystems. OEMs trust us and we’re proud to already work with many of them on very large and complex SDV projects, across the world.

We are fortunate to be living an historic opportunity to redefine mobility, offering end-users an increasingly rich and personalized experience over time.

Now, it’s up to us, all players involved in the SDV ecosystem, to collectively rise to the challenge of this transformation.



Edge processing

 

[Frangipani]

Correct ! As I stated in my post

Hi Tothemoon24,

when your 6:10 pm post popped up on my screen, while I was writing my own post about the upcoming livestream of an MB engineer’s presentation during Hochschule Karlsruhe’s ongoing AI Week, it did not yet have your comment on Loihi in it.

My screenshot of your original post (that I subsequently used to modify the image for my reply to you and ILL, adding a blue arrow) proves that you edited in the second half of your post 13 minutes later (or at least that was the last edit, in case there were several):

That same minute you edited your post (which I didn’t even become aware of until you replied to me), IloveLamp posted another screenshot of the same Markus Schäfer NAOMI4Radar announcement via LinkedIn, while I was still busy writing my own post.

Fast forward a few minutes, to when I was finally done with my HKA MB presentation post. In my FOMO to be the first one to point out that the NAOMI4Radar consortium led by Mercedes-Benz will use Loihi 2 for their research project, I rushed to scroll up to the end of your post in order to hit “reply” without actually rereading it, which I should have done, so apologies for that.

 

[JDelekto]

The cynical part of me says that MB just want a press release with 'Intel' on it but then will quietly use Akida because it is the superior product. If they don't know that already and all we've been told is true then they will soon find it out. Either way, it's not a bad outcome for us. What we need though is another public endorsement of Akida by MB or another party asap.

I don't think using Intel in this case is an "either/or" situation. The marketing/research materials may have mentioned Intel because it is a more recognizable brand. I think as BrainChip's IP is integrated into higher-tier manufacturing, its name may fade into obscurity except for those in the business of supplying these chips.

It is estimated there are about sixty (60) to one hundred (100) sensors in today's average vehicle. Some of these are in use while in motion (where power can be generated), while others are used while is considered "turned off". Akida excels in low power use, such as a vehicle being turned off, where it can detect wake keywords, driver presence, passenger (child) presence, and perhaps security sensors.

I do not think Akida is excluded from the picture when designing processors for use in the automotive industry, including their listed partner, MB.

As a side note regarding the Intel Foundry/BrainChip relationship: Intel Foundries are agnostic to their competitors. They provide the tooling and services to manufacture chips from IP, they are not competitive in the IP itself, but more so in the manufacturing process. They are more competitors to TMSC if anything.

 

[cosors]

"Google to buy nuclear power for AI datacentres in ‘world first’ deal

Tech company orders six or seven small nuclear reactors from California’s Kairos Power"

Google to buy nuclear power for AI datacentres in ‘world first’ deal

Tech company orders six or seven small nuclear reactors from California’s Kairos Power

www.theguardian.com