BRN Discussion Ongoing

[TECH]

Tesla's pot-hole detection is achieved through "hive mind technology", which uses over the air updates enabling cars to share information with each other to better manage changing weather conditions, traffic incidents and road conditions. However this requires connectivity and most of the roads that have the worst pot-holes are in areas with low to no connectivity.

Another disadvantage of this system, aside from concerns about updates being hacked) is that it requires that at least one vehicle to drop into a pot-hole in order for the system to recognise and communicate that pothole to other vehicles.

Peter Van Der Made has spoken about BrainChip's future networks (AKIDA 2000 and AKIDA 3000) being able to learn the difference between a plastic bag or a rock because they will learn from the sequences of events and the behaviour of objects in the physical world. I expect that if plastic bags can be detected, then potholes would be able to be detected too and no-one will need to worry about falling into a giant pothole in the middle of the Bungle-Bungles because AKIDA isn't reliant on connectivity.

That was an excellent explanation...and that's really the main difference, but it's a MAJOR difference, the internet as we all know it today just won't be able to cope with the increased traffic, the bandwidth bottleneck will be replaced with the hangmans noose, from which there will be no escape, it's either continue on the way we have been, or expect that the upcoming surge in IOT devices, IIOT devices, or from our vantage point, AIOT will ultimately mean the world will have to change, grow and at some point, honour Peter & Anil.

More Data Centres, More Server Farms, More ways of Accelerating the process, More Power, More Heat...you only had to listen to Jerome this morning, it's everything Peter has been saying for years, but the major difference now is that, it's no longer a dream...

IT'S REAL, IT'S HERE, IT'S GLOBAL....BUT IT DID ALL START HERE IN AUSTRALIA...I love that final part...Brainchip xx

 

[Slade]

That was an excellent explanation...and that's really to main difference, but it's a MAJOR difference, the internet as we all know it today just won't be able to cope with the increased traffic, the bandwidth bottleneck will be replaced with the hangmans noose, from which there will be no escape, it's either continue on the way we have been, or expect that the upcoming surge in IOT devices, IIOT devices, or from our vantage point, AIOT will ultimately mean the world will have to change, grow and at some point, honour Peter & Anil.

More Data Centres, More Server Farms, More ways of Accelerating the process, More Power, More Heat...you only had to listen to Jerome this morning, it's everything Peter has been saying for years, but the major difference now is that, it's no longer a dream...

IT'S REAL, IT'S HERE, IT'S GLOBAL....BUT IT DID ALL START HERE IN AUSTRALIA...I love that final part...Brainchip xx

To be honest I think Tesla has taken the right approach to the problem of autonomous driving and potholes. It would take some fancy LiDAR to detect potholes in time to warn the car or driver. Having a map of potholes already on the cars system from data fed from other vehicles seems to me to be the only viable option at the moment. Would love to be proven wrong.

 

[Deleted member 118]

https://www.eetimes.com/podcasts/embedded-ep13/

 

[Fullmoonfever]

Trying to figure out if Texas Instruments (as @Bravo has also been hunting from memory) has any connection anywhere.

Found a couple of things that were interesting but not sure if we may be in the background via the ARM NN relationship with TI and the use of ARM M4 co-processors for some edge, IoT, gesture etc. I don't think so but....

Anyway, maybe @Diogenese can have a squiz whenever has time.

I note the ARM NN runs on Cortex-A cores as highlighted however wondered if that is in connection with the M4 co-processor?

Links:

https://www.ti.com/microcontrollers-mcus-processors/processors/products.html#p2271=1%20Arm%20Cortex-M4F;2%20Dual%20Arm%20Cortex-M4;2%20Arm%20Cortex-M4;4%20Arm%20Cortex-M4

3.15.5. Arm NN and Arm Compute Library — Processor SDK Linux Documentation

software-dl.ti.com

AI and ML

Develop and optimize ML applications for Arm-based products and tools. Join the Arm AI ecosystem.

developer.arm.com

New processors make edge AI more accessible while cutting power consumption in half

Expanding on its history of delivering highly integrated processors, Texas Instruments (TI) (Nasdaq: TXN) today introduced new Sitara™ AM62 processors that help expand edge artificial...

news.ti.com

 

[toasty]

That was an excellent explanation...and that's really to main difference, but it's a MAJOR difference, the internet as we all know it today just won't be able to cope with the increased traffic, the bandwidth bottleneck will be replaced with the hangmans noose, from which there will be no escape, it's either continue on the way we have been, or expect that the upcoming surge in IOT devices, IIOT devices, or from our vantage point, AIOT will ultimately mean the world will have to change, grow and at some point, honour Peter & Anil.

More Data Centres, More Server Farms, More ways of Accelerating the process, More Power, More Heat...you only had to listen to Jerome this morning, it's everything Peter has been saying for years, but the major difference now is that, it's no longer a dream...

IT'S REAL, IT'S HERE, IT'S GLOBAL....BUT IT DID ALL START HERE IN AUSTRALIA...I love that final part...Brainchip xx

Just like wifi which was also an Oz development....

 

[Fact Finder]

To be honest I think Tesla has taken the right approach to the problem of autonomous driving and potholes. It would take some fancy LiDAR to detect potholes in time to warn the car or driver. Having a map of potholes already on the cars system from data fed from other vehicles seems to me to be the only viable option at the moment. Would love to be proven wrong.

If you are correct that Tesla is correct then how does Tesla justify the use of cameras only for AV.

The Tesla argument is that you only need vision because humans only use vision.

If this is correct and if I being human can see and avoid pot holes using vision why can’t Tesla???

The reason Tesla is using the system they have come up with is they are intending to monetise the data and sell it to the authorities responsible for maintaining roads.

It will also allow targeted advertising for tyre and rim repairs.

Suck every penny they can out of the system just like Google, Amazon and Facebook.

Every move you make, every step you take, they’ll be watching you and making money.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Chipper]

This is so wrong. How is it even allowed. Bloody s#$@rs.

 

[Boab]

If you are correct that Tesla is correct then how does Tesla justify the use of cameras only for AV.

The Tesla argument is that you only need vision because humans only use vision.

If this is correct and if I being human can see and avoid pot holes using vision why can’t Tesla???

The reason Tesla is using the system they have come up with is they are intending to monetise the data and sell it to the authorities responsible for maintaining roads.

It will also allow targeted advertising for tyre and rim repairs.

Suck every penny they can out of the system just like Google, Amazon and Facebook.

Every move you make, every step you take, they’ll be watching you and making money.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Diogenese]

To be honest I think Tesla has taken the right approach to the problem of autonomous driving and potholes. It would take some fancy LiDAR to detect potholes in time to warn the car or driver. Having a map of potholes already on the cars system from data fed from other vehicles seems to me to be the only viable option at the moment. Would love to be proven wrong.

Hi Slade,

What is the size of the google Maps database?

I've seen estimates of 20 petabytes.
https://www.quora.com/What-is-the-size-of-Google-Maps-in-gigabytes-How-big-is-it-actually

Not something you could load into your boot server.

Some LiDaRs can detect line markings, so it is not impossible that it could spot potholes.

You'd probly want it mounted at the top of the windscreen rather than low down on the bumper to improve the parallax angle.

 

[Fact Finder]

Something positive for all the haters of manipulators:

Stock Swami may face over $443k in damages​

Miklos Bolza - 4m ago

Mining investor Tolga Kumova is seeking more than $443,000 in damages for what he's called a relentless and defamatory campaign by the Stock Swami Twitter account.

© Dan Himbrechts/AAP PHOTOSInvestor Tolga Kumova is seeking more than $443,000 in defamation damages over tweets.

On Thursday, Mr Kumova's barrister Sue Chrysanthou SC said her client had his reputation slammed through a series of tweets claiming he engaged in insider trading and fraudulent "pump and dump" schemes.


"He's a person who comes here needing to be vindicated by the court by the award of damages. There was huge hurt," she told the Federal Court.
Mr Kumova is suing the owner of the Stock Swami Twitter account, Alan Davison, for defamation over six tweets which alleged, among other things, that he was part of a syndicate operating through Twitter that raised share prices through hype for their own personal profit.
As well as a cap of $443,000 for general damages in defamation cases, Mr Kumova is also seeking additional amounts in aggravated damages.
Ms Chrysanthou said the small caps mining investor had a good reputation before the tweets but this had been destroyed after word of the allegations had spread and others did not want to deal with him anymore.


"It's a very, very serious allegation indeed, and it's gotten around. Everybody knows about it and it has had a huge impact on his reputation," the barrister told Justice Michael Lee.
Mr Kumova found out the identity of the person behind the Stock Swami account through others who also had been defamed by him, the court heard.
New Zealand-based Mr Davison is also alleged to owe aggravated damages for his conduct during the proceedings, including criticisms that Mr Kumova, his legal team and witnesses had fabricated evidence, documents and legal fees.
On Friday, Mr Davison apologised to the court for withholding documents and misleading Mr Kumova and his lawyers during the proceedings.
Ms Chrysanthou argued that for all his tweets about improper conduct in the defamation case, the only one acting improperly was Mr Davison himself.
A press release by Mr Davison's lawyers at Xenophon Davis released two days before the hearing was also attacked for presenting the swami as a hero who uncovered pump and dump schemes because of his morals.
"To put himself forward in that way as some sort of hero of the masses… we say was not proper and aggravates damages," Ms Chrysanthou said.
Justice Lee questioned why a law firm would be releasing such a statement in the first place.
"What are solicitors doing writing press releases in respect of matters which are about to go to hearing in any event?" the judge asked.

 

[Shadow59]

If you are correct that Tesla is correct then how does Tesla justify the use of cameras only for AV.

The Tesla argument is that you only need vision because humans only use vision.

If this is correct and if I being human can see and avoid pot holes using vision why can’t Tesla???

The reason Tesla is using the system they have come up with is they are intending to monetise the data and sell it to the authorities responsible for maintaining roads.

It will also allow targeted advertising for tyre and rim repairs.

Suck every penny they can out of the system just like Google, Amazon and Facebook.

Every move you make, every step you take, they’ll be watching you and making money.

My opinion only DYOR
FF

AKIDA BALLISTA

Cutting out Lidar seems quite rediculous. More information to make decisions is always better.
I have to wonder if there are 2 things at play with Tesla.
1. They are reliant on internet connection, so bandwidth and volume of information transferred is critical.
and
2. Cutting out Lidar is giving them a power saving to make their travel distances look better.
I think sometimes Elon relies on the fact that he has a huge following that believe he's always right, so he can get away with saying you only need vision.

 

[Filobeddo]

To be honest I think Tesla has taken the right approach to the problem of autonomous driving and potholes. It would take some fancy LiDAR to detect potholes in time to warn the car or driver. Having a map of potholes already on the cars system from data fed from other vehicles seems to me to be the only viable option at the moment. Would love to be proven wrong.

Detecting potholes? Yeah we can do that

Just need to streamline the next model a little though

 

[Lex555]

Hi Slade,

Akida 1000 SoC is hardware - the main part of which is an array of 80 NPUs arranged in nodes of 4 NPUs. Each NPU is an arrangement of transistors forming digital logic gates adapted to compare input "spikes" (activations) with stored digital bits (weights) representing images in a previously collected library of images. The activations and weights can be selected to be either 1, 2, or 4-bits, trading speed and power for accuracy.


View attachment 10827

MetaTF includes a computer program to simulate the performance of the Akida SoC, although this will be slower and more power hungry than the hardware.

The main part of the Akida IP is data defining the physical configuration of the transistors which make up Akida SoC logic gates. This is encrypted so only BrainChip and Socionext/TSC have access to the information.

A+ post, appreciate the detailed explanation. For anyone new here, it can be helpful when you find a post like this to bookmark it and keyword tag it for future reference

 

[Fact Finder]

Detecting potholes? Yeah we can do that

Just need to streamline the next model a little though


View attachment 10839

These guys must have designed Daniel's F1 cause they would not have gone very far at Mercedes. Just brilliant. FF

 

[Bravo]

If you haven't seen this video from 2018 with Michael Dell (Dell Technologies), then do yourself a favour and watch it from approx 9.50 min mark an d another snippet at about 13 min mark. Michael talks about how he has a habit of following opportunities that may not have seemed obvious to others at the time. Check out what he then says about edge computing!

(9.50 min)

(13 min)

 

[Baisyet]

From Embedded World 2022

 

[Baisyet]

From Embedded world 2022

 

[KMuzza]

Just like wifi which was also an Oz development....

Hi Toasty,
yes Wi Fi was a CSIRO combination /invention.- BUT started out in the commercial market and was literally stolen by other users- CSIRO/WiFi had to engage in long /protracted legal battles to claim ownership.

BRAINCHIP- has started to go commercial with most important - PATENTS . and the JAST systems in place for commercial protection of their product.
I watch to see how others ( SoC Neuromorphic) will get around the JAST /Brainchip system.- See the ASXE- ALL ROADS LEAD TO JAST- @uiux thread .

AKIDA BALLISTA UBQTS.

 

[MDhere]

At Brainchip they have the ADE or AKIDA Development Environment. It is a software emulation of the AKIDA hardware. If you go to the AERO paper published by Brainchip late 2019 early 2020 they use the ADE to identify the 20 gas data set with state of the art performance using only 200 simulated neurons. They then predict that once implemented in hardware the performance will be improved.

So you can implement the AKIDA in software.

The IP or intellectual property is the design of the actual neural fabric so that you can build an AKIDA in hardware.

So in situations where you are not concerned with latency or power you could implement AKIDA in software and run it on old school Von Neumann hardware.

Now as I have said I am a technophobe, old and a retired lawyer who knows nothing so don’t rely on a single word I say and DYOR
FF

AKIDA BALLISTA

I dont know ware i would be, if you lot didnt explain to me what ware is what ware and why it is, i would have no idea what and when its best suited for and ware its best suited. I would beware of all the wares and when ive completely lost my mind i wouldnt know what to ware. lol

 

[alwaysgreen]

@alwaysgreen that poster on that sewage forum is possibly talking about this. LoL

Chips - Now it's really getting started! Intel, BrainChip, ARM, NVIDIA

Chip stocks like Intel have recently lost double digits on the stock market. But what about demand? Market researchers at McKinsey predict that the chip industry will grow by 8% annually until 2030 - and even more in selected niches such as the automotive industry. For chip stocks, this means...

news.financial

If he gets the thought that article means a possible agreement with Intel.

Thanks mate. I still check the IMU thread over there. Not sure why I bother popping into the BRN forum. What a joke!