Valve connected to the cloud make sense?

[h20]

Hi guys, i am new to this website.
After reading the kind of subjects i decided to write my own post.

As the title says i have had an idea of connecting a valve directly to the cloud, so you can command and read data from it using MQTT/API, no mistery there.

I will explain a bit about how i got here.

I am a professional embedded developer, in my own time we have developed an algorithm that detects the flow of water just by listening to the sound it makes on pipes using Machine Learning(this is the really cool part).
So far so good we actually can detect if there is flow and give a quantity of the instantanious flow with a max error of 10%, just by attaching a simple microphone to the pipe, sounds crazy but it works.
The thing is that if we could connect our device to a cheap valve we would convert a 5$ valve into a valve that can detect open/close, flow/noFlow and the amount of flow.

Given that i have no experience as a product manager i ask the following:

Does any of you see this as a useful product?
 
I have attached a picture of an initial proofs of concepts

pic1.pdf

[Bufofrog]

I saw this technology 40 years ago, not the machine learning but sonic flow detectors with feedback to a throttle valve.

[Sensei]

Google has Patent Search Tool.

https://patents.google.com/

You should start from it.

 

[h20]

On 8/7/2020 at 6:03 PM, Sensei said:

Google has Patent Search Tool.

https://patents.google.com/

You should start from it.

 

Cool nothing exists yet

[Bufofrog]

2 hours ago, h20 said:

Cool nothing exists yet

How about this?

https://en.m.wikipedia.org/wiki/Ultrasonic_flow_meter

 

[Strange]

2 hours ago, h20 said:

Cool nothing exists yet

https://patents.justia.com/patent/20150082911

https://patents.google.com/patent/US20050011278

There were hundreds more results, some probably duplicates, so the concept is not novel. You would need to think of something to make a product like this competitive (and avoid infringing any existing patents).

[h20]

52 minutes ago, Bufofrog said:

How about this?

https://en.m.wikipedia.org/wiki/Ultrasonic_flow_meter

 

Do you know the price and power consumtion of those devices? Not suitable for portable iot applications...

28 minutes ago, Strange said:

https://patents.justia.com/patent/20150082911

https://patents.google.com/patent/US20050011278

There were hundreds more results, some probably duplicates, so the concept is not novel. You would need to think of something to make a product like this competitive (and avoid infringing any existing patents).

None of them apply, we do not genrate a sound and then measure, we listen to the sound of the pipe and then infere ...

We are not focusing the point here.
There are two options when it comes to developing new tecnologies.
1 - New important and crucial use cases force development of new tecnologies.
2 - New available tecnologies encourage new use cases that nobody ever thought might exist or even make any sense.

So my point is, this is a new way of measuring flow through pipes, it might or might not have its place.
As always for me this technology has enhancement at the consumption of the device, a lot cheaper and non invasive, which might open new use cases or not.
Bear in mind we are in the Big-Data era, so you might get a glance of were things are going, we want data no matter what(more data = more value), so this might be a cheap way of pputting one kind of data on the cloud...
Now its up to clever people to find what to do with that data!

[Strange]

9 minutes ago, h20 said:

None of them apply, we do not genrate a sound and then measure, we listen to the sound of the pipe and then infere ...

OK. I hadn't looked in that much detail.

9 minutes ago, h20 said:

Do you know the price and power consumtion of those devices? Not suitable for portable iot applications...

I was going to suggest that your selling point could be very low cost, to open up a wider range of applications.

Portable, though? Where do you see that being used? (Not being negative, genuinely curious!)

[h20]

23 minutes ago, Strange said:

OK. I hadn't looked in that much detail.

I was going to suggest that your selling point could be very low cost, to open up a wider range of applications.

Portable, though? Where do you see that being used? (Not being negative, genuinely curious!)

We are not focusing the point here.
There are two options when it comes to developing new tecnologies.
1 - New important and crucial use cases force development of new tecnologies.
2 - New available tecnologies encourage new use cases that nobody ever thought might exist or even make any sense.

I will put an  example though, in the US and Europe, the pattern on how elderly people use their water taps in houses is a pre-indicator of demetia or Alzheimer, so if a person leaves its tap running all night or does not use it at all indicates mental health problems, you can imagine if we glue this data in the cloud many things are there to be explored.

The fact of it having to be portable is otherwise the cost of installation will not make the device so attractive.

Edited August 10, 2020 by h20

[Bufofrog]

If you can make your version of a sonic flow meter cheaper, more portable, more relable or more accurate you will have something.  Good luck with your project!

[Strange]

1 hour ago, h20 said:

The fact of it having to be portable is otherwise the cost of installation will not make the device so attractive.

So, effectively, you are using "portable" to denote low cost, ease of use, simple installation, etc.

Well, good luck. It is not always easy developing and marketing a new product but it should be valuable experience.

[MigL]

34 minutes ago, Bufofrog said:

Boy, this is going south in a hurry!?!

That's too bad.
I had an idea that the system could be adapted to a sensor which attaches to water pipes, and connects through wi-fi to an app running on your computer or phone. The app would warn you of excessive water flow, indicating a leak or bad valve, on that particular pipe, or even the main.

Had a bad flapper valve on the toilet in the basement, by the laundry room. The only time I'm down there is to do laundry, so the water is always running, and I didn't hear the toilet running on for months.  Only noticed something was wrong when the $725 bill came.

[h20]

11 hours ago, MigL said:

That's too bad.
I had an idea that the system could be adapted to a sensor which attaches to water pipes, and connects through wi-fi to an app running on your computer or phone. The app would warn you of excessive water flow, indicating a leak or bad valve, on that particular pipe, or even the main.

Had a bad flapper valve on the toilet in the basement, by the laundry room. The only time I'm down there is to do laundry, so the water is always running, and I didn't hear the toilet running on for months.  Only noticed something was wrong when the $725 bill came.

Yes, thats one of the applications for it, this proof of concepts makes use of the man with the stick principle:

https://www.lep.co.uk/business/meet-leak-detectors-and-their-wooden-listening-stick-656032

and takes it next level with artificial intelligence and a wide training set for a neural network. 

 

[Strange]

1 minute ago, h20 said:

Yes, thats one of the applications for it, this proof of concepts makes use of the man with the stick principle:

https://www.lep.co.uk/business/meet-leak-detectors-and-their-wooden-listening-stick-656032

and takes it next level with artificial intelligence and a wide training set for a neural network. 

That is a pretty cool idea. It is shame you have now lost the chance patent it!

[h20]

5 minutes ago, Strange said:

That is a pretty cool idea. It is shame you have now lost the chance patent it!

It cannot be patented

5 minutes ago, Strange said:

That is a pretty cool idea. It is shame you have now lost the chance patent it!

Your incompetence is so extended that you dont see that the value in this idea is not the idea itself is the training of the neural network, hence a brain with no training is just meat. You should train a bit more...

Edited August 11, 2020 by h20

[Strange]

14 minutes ago, h20 said:

It cannot be patented

There might be objections that it is not inventive enough (one of the criteria) but a good patent agent should be able to get round that.

[h20]

10 minutes ago, Strange said:

There might be objections that it is not inventive enough (one of the criteria) but a good patent agent should be able to get round that.

First patent legal framework is not uniform across the world and its economic areas, second you dont know what you are talking about, "inventive enough" is a subjective argument ...

So just as a little hint, this is what is not patentable in sweden:

According to the Patents Act, an invention cannot only constitute: 

1. a discovery, scientific theory or mathematical method,
2. an aesthetic creation,
3. a scheme, rule or method for performing a mental act, playing a game or doing business, or a computer program,
4. a presentation of information,
5. a procedure for surgical or therapeutic treatment, or diagnosis, to be practised on humans or animals. 

5 minutes ago, h20 said:

First patent legal framework is not uniform across the world and its economic areas, second you dont know what you are talking about, "inventive enough" is a subjective argument ...

So just as a little hint, this is what is not patentable in sweden:

According to the Patents Act, an invention cannot only constitute: 

1. a discovery, scientific theory or mathematical method,
2. an aesthetic creation,
3. a scheme, rule or method for performing a mental act, playing a game or doing business, or a computer program,
4. a presentation of information,
5. a procedure for surgical or therapeutic treatment, or diagnosis, to be practised on humans or animals. 

But again someone with ignorance in the neural network field would not see any break through in this, and would only see a man with stick, fooling himself he is so proud he discredited an idea, but is its lack of understanding giving him a false thought of confidence when he talks. And you sound VERY confident to not even have looked at the inners of the idea, or by other means not even knowing wat a neural network is! 

Here you go:

https://link.springer.com/article/10.1007/BF01413870

 

Edited August 11, 2020 by h20

[Strange]

4 minutes ago, h20 said:

First patent legal framework is not uniform across the world and its economic areas, second you dont know what you are talking about, "inventive enough" is a subjective argument ...

OK. We will ignore the fact that I worked on patents for a couple of decades.

Yes, things like "inventive step" and "obvious" are subjective and it can be quite hard to persuade the examiner that a patent really is inventive.

You are correct, I am not that familiar with Swedish patent law. But I assume that if you were to develop a product based on this idea (as you stated in your first post) you would want to patent it in, potentially, the largest markets (e.g. the USA).

You are, obviously, correct that it is not possible to patent an idea or (in most countries) a piece of software. However, it is possible in principle to patent the sort of thing you talked about in your first post: a cloud-connected valve and acoustic-flow measurement system. You even posted a photo of a prototype device. That falls squarely in the domain of things that are patentable.

 

[Phi for All]

!

Moderator Note

I'm opening this again in case there are some further aspects to (politely) discuss that don't involve fundraising.

Posts generated from an earlier downward-spiraling misunderstanding have been split to the Trash. Further responses to modnotes will be split to Trash. If you have a problem with a moderator's note, USE THE REPORT POST FUNCTION so you don't take a thread off-topic.

 

[studiot]

On 8/10/2020 at 12:58 PM, h20 said:

None of them apply, we do not genrate a sound and then measure, we listen to the sound of the pipe and then infere ...

I have done quite a bit of both acoustic emission testing (your technique) and (ultra)sonic propagation testing over the years.

It is still an up and coming technology.

We still have much to learn, there have been good successes as well as spectacular failure with these techniques.

https://www.astm.org/DIGITAL_LIBRARY/STP/PAGES/STP15782S.htm

A technique allied to this is called acoustic signature analysis.

 

In my day processing the huge amounts of data required was a difficulty, as was calibrating a measuring device using these principles.

Neural networks are a modern method of attacking the data processing aspect, the second one still remains.

 

I suggest you look around the existing subject more widely as the applications are already legion, but there will be a ready market for genuinely better equipment in this field.

As a liquid velocity meter, how is your gadget affected by water quality eg suspensed material in the water ?

Ability to distinguish this could be a useful variant if it can provide an alert to contaminants in the piping.

 

 

 

Edited August 11, 2020 by studiot

[h20]

2 hours ago, studiot said:

I have done quite a bit of both acoustic emission testing (your technique) and (ultra)sonic propagation testing over the years.

It is still an up and coming technology.

We still have much to learn, there have been good successes as well as spectacular failure with these techniques.

https://www.astm.org/DIGITAL_LIBRARY/STP/PAGES/STP15782S.htm

A technique allied to this is called acoustic signature analysis.

There are many more...

In my day processing the huge amounts of data required was a difficulty, as was calibrating a measuring device using these principles.

Neural networks are a modern method of attacking the data processing aspect, the second one still remains.

 

I suggest you look around the existing subject more widely as the applications are already legion, but there will be a ready market for genuinely better equipment in this field.

As a liquid velocity meter, how is your gadget affected by water quality eg suspensed material in the water ?

Ability to distinguish this could be a useful variant if it can provide an alert to contaminants in the piping.

 

 

 

Hi studiot, glad we can talk specifics.

Just a note on the current state-of-the-Art regarding microcontrollers and edge computing.

You should take a look at TensorFlow + uTensor its what we have used and it allows a well trained neural network to run on a 250 K of RAM processor:

https://os.mbed.com/blog/entry/uTensor-and-Tensor-Flow-Announcement/

No calibration, no fft measurements(at least not for training), just plain 1 Dimensional data.

The regular approach on teaching a neural network to hear sound is actually to make it a spectrogram "picture" and then applying image recognition over that "picture".

Science Paper

We dont follow this approach, we just pass a liner vector of raw sound data to a trained neural network. Normally you were not able to just inyect raw data to a neural network, at least not into a microcontroller.

See sometimes things are simpler than they seem, we have designed an algorithm named: Wait-a-Minute.

The key is not how to evaluate that data but when, thats the key.

Imagine you tell an old man to listen to the pipes behind a wall and keep asking him: 'Hear anything?' he will be able to tell you yes or no most of the time, but if a truck passes by he will tell you : Wait-a-Minute...

 

Edited August 11, 2020 by h20