Member Info for TheLongestShot

Phew, that was a private inventor, using his home address, so when Skynet takes over, once we have the time machine figured out, we know where to find him.

I hope there is a special place in hell for the guys at Harman who came up with this patent, or were they just taking the p1ss. Seriously, a little robot in the car that reminds you to buy flowers on your anniversary, so that the OEM can get a cut of the revenue and winks and nods in the direction of the shop.

DR777, I don't think CFP managed to read to the end of each of the 'plus' points that Cipia were trying to sell is. They were spelling out weaknesses to me

Interesting differences between Cipia and Guardian

For Guardian
-We have optional forward facing camera to capture external events to protect the driver
- The cameras are always on but overwriting, but the driver can press a button to save the last X seconds as evidence
- we have ongoing revenue - that is serious reliable money for us in the long term, allows us to offset some hardware cost (not that we do!)
- the ongoing revenue pays for the data traffic (I presume) which means we can do 2nd level analysis at Canberra or Tuscon with a human (although this is getting more automated)
- ***we get to see all the captured events for ongoing training data **** and they pay us!!!

For Cipia
- possibly some of the processing is done on a mobile phone, but not confirmed. That would reduce the hardware cost - but will require ongoing development of android (and perhaps iPhone) clients
- no data is sent to Cipia - while this may sound good for libertarians, it misses the opportunity to add value and to obtain
very valuable training data
- one off up front payment - sounds great for the customer, but it makes the hardware disposable, with no revenue stream, why should Cipia provide the service in the longer term, when they are bought out for their IP, they can drop the service.

It is worth digging up one of Colin's tweets from Feb 2020. It came out just after SEYE announced they were starting an aftermarket fleet product and aviation. - Well they have just got 4 devices in trucks starting to collect video for later analysis, only 5 billion Km to go!

https://twitter.com/semicast_res/status/1232978336378445824
Colin Barnden / Semicast Research @semicast_res
Feb 27
Quick thoughts: a co. with limited engineering resource, minimal HW experience & ~US$15m cash seeks in next 3yrs to be lead auto DMS vendor (~40% share); develop aftermarket production & distribution ops (AIS); & build share in aviation. 3 disparate verticals. V. bold strategy.
Quote Tweet
Smart Eye @SmartEyeAB
· Feb 20
Our Q4 report is out and even though (lets be honest) some of the numbers could be better we want to emphasize the following: "The decision is approaching for #DMS in the middle class segment. Soon all the major players..." Read full report here: https://bit.ly/2HFuZOk

Whereas the other GM patent was nominally about eye tracking, it was completely impractical, with moving mirrors to scan a single sensor across the field of view, its resolution would be low and the refresh frequency would be far too slow for anything other than, 'are the eyes open? '

The second GM patent describes how the HUD works, essentially it is a periscope. But a pertussis only works if your eyes are in the correct place, so they have movable mirrors so that they can move the 'eye box' to ensure it coincides with your eyes that needs eye tracking. The meat of this patent is that old nemesis 'the sun', since the HUD lives under the windscreen, it is bathed in sunlight, this patent is a method of directing the unwanted light down a light pipe, so that the driver isn't dazzled by the stray light bouncing around inside the HUD

This confuses me a Galvanic mirror is a moving mirror to allow scanning so that a single sensor can see in one dimension if two moving mirrors are used you can see in two dimensions. This is like going back to the early days of TV. Global shutter cameras that take a full frame without moving parts have made this obsolete. This is patenting for patent sake to close a research project.

https://amp.freep.com/amp/6496598002?__twitter_impression=true

Note average selling price is $50000.
Trucks seeking a fast as they reach the dealers.

https://www.seeingmachines.com/wp-content/uploads/2020/12/Seeing_Machines_Annual_Report_2020_WEB_SPREADS.pdf

We won't be appearing at wide bandgap conference. It is for electronics. EE Times isn't just about DMS

Glandore search out Mike Levine @mrlevine on Twitter. Last few days he has RT as dealers show there new stock of F-150

OldFool,
Comma is not a DMS system, it is a retrofit cruise control with steering and braking, it interfaces into the CANBUS to use the vehicle's sensors.

It is only available for certain cars such as Honda Civic. It can't be used on every car and certainly not in fleet. At least George Hotz did put a rudimentary DMS ontop, but as it hasn't got IR camera or illumination it is only partially working

https://www.theverge.com/2018/7/13/17561484/george-hotz-comma-ai-self-driving-car-scam-diy-kit

comma Two is a DIY kit, no IR camera, not for the faint hearted. The previous version got a cease and desist from the US government.

I wish you a good Christmas, but also for my fellow investors to be aware of what is out there in the market, we do not operate in a vacuum. Not all technology posts and analysis are claimed for SEE, but my experience is more technical than most, so hopefully I am helping everyone to SEE better.

The Honda patent was quite incredible, absolutely massive with over 100 pages of dense text, but the 143 diagrams were quite accessible. Essentially, it is how the car should modify its behaviour in a multitude of situations based on the state of the driver. It was inspiring to see his much thought had gone into the design and how embedded Human Factors are in Honda's designs.

These systems were planned out, based on a rock solid signal of how alert the driver is. That can only come from years of testing with the very best partner's equipment tools and techniques.

Would all 2nd and 3rd rate suppliers please take a step backwards. Who does that leave?

Your right Jazz, I'm on page 73 of 255 and still wading through. Was Tolstoy one of the authors?

That is a fair crop you have found today, so only time for a brief scan.

Veoneer - definately a Tier 1 implementation feature. A steering column camera that will break away in an accident.

Toyota Japan, looks like a DIY solution, if part of the face is hidden then detect based on the rest of the face. This can only be a layer on top of a terrible AI model. No threat to us we have worked for years with a partial face.

LG use the head up display to project an image onto the windscreen and detect if the eyes follow it to s if the driver is tired. Well, this looks like a work around, to get around poor research and to give an easily coded solution.

Simon Stent (Toyota) this is more about Tier 1 than two, it is about the car monitoring the important parts of the road scene, where Toyota Guardian spots a hazard that the drivers eyes haven't moved to, so initiates a take over. This one is good,

Smart Eye, trying to make a system do object selection on infotainment, simplify the UI and work on the direction that the eyes move not where they are looking. Allows a cheeper, lower resolution cameras, less accuracy and trying to work around a poor implementation of calibration. This really is a demonstration of the other problems that SEYE are facing.

Finally is Geely using face teaching to decide which camera feed to put on the screen.

CFP knows nothing and is keen to show it off as usual

https://web.cvent.com/event/51a51ecc-ce11-443b-b3c4-e3da350fd183/summary

Have a look at the Agenda for details

Semicast: Who Says DMS and ICMS Technology is Obsolete?

7:30 AM-8:00 AM

?

Colin Barnden

Principal Analyst

Semicast Research

With self-driving supposedly on the home stretch, it is all too easy to write-off driver monitoring and in-cabin monitoring systems as obsolete technology. In this session, Colin Barnden will talk through developments in the market and why legislative changes strongly suggest that driver-facing cameras will play a central role in automotive safety for decades.
 
---

Driver Monitoring Systems are here, and Occupant Monitoring Systems are coming. Are you ready?

8:00 AM-8:30 AM

?

Kevin Tanaka

Sr. Director of Automotive Marketing

Seeing Machines

?

Mike Lenné PhD

Senior Vice President, Fleet & Human Factors

New safety standards requiring high-performance, low-cost embedded driver monitoring solutions will be published very soon for near-term implementation, and it is obvious that pure end-to-end deep-learned neural networks will simply not be enough. Safety-focused regulatory bodies are looking for robust solutions to monitor the state of drivers to ensure that they are engaged and capable to retake control of the vehicle, if necessary. To this end, a scientific understanding of human driver impairment and driver engagement behaviors is required – this is a field called Human Factors (HF).  Seeing Machines, an industry leader in AI driven safety technology, roots all of its DMS and OMS solutions development decisions in HF science and research, and has deep expertise in optical path, algorithm development and embedding optimization. The company is on a mission to improve driver safety by working directly with government and safety bodies, along with Tier-1 and OEM customers, to get everyone home safely.

Thanks Fonzoo, definately worth a watch. Yes I did learn some things

JC, I was going to poo poo this, then I thought maybe, following on from Kinnect, maybe Microsoft are dogfooding* it at Redmond meeting rooms. But fast forward to 2020,and we could already be unwittingly contributing when on Teams meetings.

We have a camera pointed at us and are we looking at the screen, are we looking at the speaker or elsewhere?
How engaged are we in what is going on. Who in the meeting is paying attention? Who gets the meeting to perk up when they speak? Lots of KPIs and metrics for every meeting!

Unfortunately, Microsoft can do this today in Teams without our chop, the load would either be on the edge (or PC) or in the azure data centres where the video passes through on the way to the other attendees. Would this be marketable? Ethical? Possibly not,but it is possible for MS to do this without us. This isn't a safety system, it doesn't need high availability. tracking from the camera to the fixed faces on a grid that they control is relatively easy and if they get it wrong no one crashes.

*eating your own dog food (using your own systems) is a big thing at Microsoft