Member Info for TheLongestShot

Yes, CP you have passed the first test of the apprenticeship. Now "wax on, wax off" is the next stage of your training

Thanks PM,
I can tell you I don't have money. But what I do have are a very particular set of skills, skills I have acquired over a very long career, skills that make me a nightmare for people like you.

Hi Sn,

There are a few things to note in that series of clips.

I can't date or identify the BMW but it has an old instrument cluster and no obvious DMS - so just there to remind us we won them.

The Hyundai is a fairly budget model and nothing to note apart from there appears to be a thick cable hanging down from the mirror area which is never seen.

Then we get to the car at 1:10
- The steering wheel is missing any branding
There is a sat nav running on a phone in an aftermarket holder which looks like it is attached to the air vent.
There doesn't seem to be any power to either the phone or to the phone holder, and no obvious connections to the phone's usb socket.
There is a USB memory stick with a rotating metal sleeve plugged into the car
The patents that we saw did provide external camera AND LED illumination and was connected electrically via the dock. Without a power source this would not be practical.

Just done a quick image search and it looks like that is just a phone holder
https://myretrak.com/products/car-phone-holder-car-air-vent-phone-holder

However, there is something above the mirror that I can't identify. Of course they all could be stock video clips

The video in the background of Yahoo Finance that we were squinting to see is:
https://www.qualcomm.com/videos/seeing-machines

Enjoy the Rolls Royce experience

JS eleventyone,
This is still early, plenty of growing room in the share price. DYOR of course, but it is moving fast now!

PEEL
GO!
CFEP
GO!
SCAP
GO!
WINS
GO!
CNKS
GO!
STFL
GO!
SING
GO!
JBER
GO!

...

In the old days we couldn't support many extra Tier1s never mind extra OEM projects. With the Gen 1 Fovio, could give chip to Tier1 and take the licence fees if the platform was OK. Now there are more ways to help with minimal input.

What is your platform?
"OK here is a demo system, off you go."
"If you want some customisation, here is the price list."

On one of the Smart Eye boards, they are convincing themselves that only the first 30 minutes of driving data is all they need for training DMS. One of the interns published his thesis and the data sets used for training were small.
There focus was still on eye gaze. Not the stages before a microsleep or detection of the driver's state
They just were not understanding the importance of catching and training on real edge case data, not simulations.

That is why they Smart Eye don't have the respect of the HF academics, that SEE do. I was trying to find a paper that was discussing the difference between mind wandering and out of the loop. The difference is crucial if you are supposed to be attentive to monitoring the car's driving, those details of driver state are the difference between life and death, they may not be watching TV or playing on the phone, but they may as well be. Only a really good DMS can be proven to tell the difference.

Then as Sandy says, there is our hardware partners and we have enabled all their options for efficient and powerful solutions, including updatable over the air. OTA, is a game changer as experience in the field can be fed back to the implementation.
If you consider our likely partnership with Cerence, they will need OTA to add functionality and additional languages

Wilson, imagine sitting in your executive car and saying "Hey Mercedes/BMW /*Audi, what is the building on the left? "

"I'm sorry Sarah, the light isn't right and I was distracted, please look again for longer. "
That is why Cerence work with SEE, you can specify cheaper, but you will soon be seen to be cheap!

Cerence's patent IS EXTREMELY RELEVANT!

Contextual utterance resolution in multimodal systems
https://patents.google.com/patent/WO2020146134A1/en

What is claimed is:

1. A method of disambiguating a vocal utterance in a multimodal input system, the method comprising:

a processor storing context data;

a processor receiving a vocal utterance; and

a processor applying context data to determine the meaning of the utterance.

2. The method of claim 1, wherein the method includes a processor in an electronic vehicle system storing context data, receiving a vocal utterance and applying context data to determine the meaning of the utterance.

3. The method of claim 1, wherein the system is responsive to gaze input to determine the meaning of the utterance.

4. The method of claim 1, wherein the system is responsive to at least one of a stylus input, a haptic input, and a text input to determine the meaning of the utterance.

5. The method of claim 1, wherein the system is responsive to text input.

6. The method of claim 1, wherein the system employs embedded and cloud processing.

7. The method of claim 1, wherein the processor employs natural language processing to determine a word of the vocal utterance.

8. The method of claim 1, wherein the processor employs a recent antecedent interaction with the system as a context factor in applying context to the determination of the meaning of the utterance.

9. The method of claim 1, wherein the processor employs gaze data as a context factor in applying context to the determination of the meaning of the utterance.

10. The method of claim 1, wherein the processor employs current media playing data as a context factor in applying context to the determination of the meaning of the utterance.

11. The method of claim 1, wherein the processor employs the status of an associated system as a context factor in applying context to the determination of the meaning of the utterance.

12. The method of claim 11, wherein the processor employs the status of a vehicle as a context factor in applying context to the determination of the meaning of the utterance.

13. The method of claim 12, wherein the processor employs a sensor reading as an indication of the status of the vehicle.

14. The method of claim 1, wherein the processor employs a speech analysis technique including at least one of: voice activity detection, automatic speech recognition and natural language understanding.

15. A multimodal input system for disambiguating a vocal utterance, comprising: a processor to store context data;

a processor to receive a vocal utterance; and

a processor to apply context data to determine the meaning of the utterance.

16. The system of claim 15, wherein the system includes a processor in an electronic vehicle system to store context data, to receive a vocal utterance, and to apply context data to determine the meaning of the utterance.
...

Jester, why not follow those in the know at Seeing Machines. There is no leak, but many sprinklers that all land on a common list of targets ;-)

Thanks Fonzoo, Sandy,
Rather than posting again, click on my name and read the posts from the 15th January.

or Read Colin's untangling of me jumping to conclusions, then follow up by reading all the comments!
https://www.linkedin.com/posts/colin-barnden-1081376_dms-activity-6755784736863473664-ivn-/

In short, it isn't our NPU; but ours must be so much better if they licensed it when their's was ready for release!

https://mobileidworld.com/cerence-look-answers-driver-questions-about-outside-world-011304/ January 13, 2021 Cerence has launched a new product that provides drivers with more information about the world around them. The hands-free offering has been dubbed Cerence Look, and it will first appear in the new Travel Knowledge feature in the MBUX voice assistant from Mercedes-Benz. New Cerence AI Tech Can Tell Drivers What's Passing By Their Window With Cerence Look, drivers will be able to ask about something outside the vehicle, and the in-car assistant will respond with basic information about that location. For example, drivers can use it to find out the name of a restaurant, or to learn more about a nearby landmark. The system leverages Cerence’s conversational speech recognition technology, so drivers do not need to use a specific wake word. Instead, they can ask “What is the name of that restaurant?” or “What is that café on my left?” while driving past a coffee shop and get the same response. Cerence Look is designed to promote on-road safety, since drivers will be able to learn more about local points of interest without turning to a separate mobile device. **The solution uses environment reconstruction and sensor data to determine the vehicle’s location and figure out what the driver is looking at when they pose a question**.

Not Smart Eye then!!

Green blobs who follow British Bulls were told to stay in cash!

But they are 160% of where they were at the start of September.

Stay in SEE since then you would be 333%

Now we know why they are green.

#TuesdayTips: Plan your itinerary.
Mapping out the specifics of your road trip is the best way to eliminate stress and even avoid hazards when driving. #roadsafety #innovation #SeeingMachines #DMS https://t.co/N1ZqToXJYN.

That's an easy one, along with Cerence. Speeking and looking at places is a natural way to set up your Sat nav and to get other info :

Meet Cerence Look, a new product that enables drivers to interact with POIs outside the car. #Cerence Look powers @MercedesBenz's new Travel Knowledge in #MBUX in the #SClass and via the #Hyperscreen in the #EQS. https://t.co/fXFsdzQLt8 #ces2021 #multimodality Video c/o Mercedes. https://t.co/WRAgpc3ciT

“It is usually the roughest road that leads to the heights of greatness.” #MondayMotivation #seeingmachines #roadsafety #driversafety #innovation #DriverMonitoringSystem https://t.co/tyBtVV7pdS I'm going out on a limb and think this is to do with active suspension and driving policy - the car looks at how the driver is moving and feeling to determine when to slow down or adjust suspension firmness and ride height as road conditions change when using hands free cruise control in a big 4 wheel drive.

No doubt from Sophie, that was a thumbs up at the end

https://www.linkedin.com/posts/colin-barnden-1081376_dms-activity-6755784736863473664-ivn-

Fonzoo, now we have the answers :

Colin Barnden
Principal Analyst, Semicast Research
You can generally spot market leaders because they willingly and openly engage in serious questioning from analysts like me with simple and straightforward answers. The suppliers in the strongest position and making the most progress always have nothing to hide. Really great work from OVT and looking forward to writing about the future DMS products.

10 Likes

Robert Sanders
Director at Deutsche Bank
How sticky is the SM IP block?

LikeReply
1d

Colin Barnden
Colin Barnden
Principal Analyst, Semicast Research
Sticky isn't anywhere close to a word I would use, but I guess it is sticky in the way the foundations for the Empire State Building are sticky; or the flight computers in an Airbus A320; or your heart and lungs. I guess you could kinda swop them all out. The most important thing to remember is we are talking about functionally-safe, mission-critical system design here, not FMCG. That is a point most of London doesn't seem able to grasp

Paul McGlone
Chief Executive Officer
Thanks Colin. To answer the follow in question our IP stack is very sticky as a consequence of the systems approach we take. Optimised optics, AI, HF design and now acceleration at the outer limit of price performance.

LikeReply1 Like
12h

Mengxi Liu
Mengxi Liu
(She-her) Head of Marketing Communications @ OmniVision Technologies, Inc.
Great Read! Brian Pluckebaum Andrew Hanvey Sophie Nicoll

LikeReply2 Likes
22h

Sophie Nicoll
Sophie Nicoll
Corporate Comms | Investor Relations | Marketing Strategy
Yes, great clarification from Colin Barnden as always ??

Hyms,
No, I don't think I could ever add that much value ;-)

I heard a rumour there was a conference going on and lots of expectations and people may have started to hear what Colin was saying. Helped along with another story about Apple, so I can't claim much credit.

Colin did a quick interview to clear up (my mess). We didn't invent the "NPU". From rough comparisons, based on minimal information ( look where that got me - I pursuaded myself to invest in SEE ) with the OmniVision part for SmartEyeAB and the OAX8000, it looks like it run on 40% of the power (so creates 60% less heat - which is often more important)

The NPU will be responsible for much of that gain. However, that NPU is now confirmed to be wholly the work of OmniVision. Yet they have sought to licence the Occula NPU from Seeing Machines. So it must be a significant improvement on several fronts.

I would imagine that DMS providers could look at running their code on the OAX8000 (if their code will run) and their ML models will run on the NPU (if it is compatible).

Perhaps in a few/many months they could look to run on OmniVision's newer chip *families* and run with better performance, and with extra capacity for larger models in the OCCULA NPU with faster processing and therefore higher frame rates, or perhaps the same performance with lower heat and power. The costs will be different so that would be a business decision. The fact that some of that money would go to pay a Seeing Machines license would not dissuade them. This is business not football team rivalry. This is not a zero sum game.