ESP32 Forum

Hi everyone,

I've been developing for some weeks a system to manage capacity in sportive & educational places. I'm based in Spain and we are little by little decreasing the lockdown, which should go with some order and rules. We aim to control the entrance of people to public swimming pools and libraries with sensors.

I've built some prototypes with an ESP32 and 2 sensors, which push events to a cloud app that performs the calculations and shows a semaphore. I have tested it with a range of sensors, and it would be great to get some community feedback around it.

I started out with Ultrasonic HC-SR04, but it gets a lot of fake echo signals when you put it on an open environment. After that, I moved after it to IR where I tested 2 main options: the Sharp GP2Y0A21YK0F and the e18-d80nk.



We are starting next week the outdoor testing and I'm quite afraid of the IR performance on a swimming pool under the sun. I have considered trying out the VL53L0X or something similar but reading some thesis on the topic it seems that the results should be kind of similar, as all work under the time of flight principle and the photosensors tend to get saturated in open environments.



Do you have any tip that could help on the matter? If the post should be moved to a specific Hw room I will move it right away.



Thanks in advance!

Sorry to be brutal but this problem has already been solved.
It is a cornerstone of Phone based economies i.e. they want to know what you do so they can sell you s*&t, oops, stuff.
If you're primary purpose is dealing with risks/lives saved then....
I would guess that a $0.10 purchase & refund on entry/exit would give you the means. Ok, so $0.10 is not enougth to ensure ppl refund so location tracking seems the thing. Maybe entry on activation on tracking app.
Google tells me when a bar/venue is typically full/empty etc so this tech is out there. Just need to tune in to how that works.
Does leave the problem of how to deal with Gradma & her brick phone (if she even has one) but guess you keep a few burners.

Don't worry. No issue about the tone of the response.

The problem that I'm trying to solve is to have a real-time system that will stop the next person to enter on a library/swimming pool. I don't think the mobile approach would work for it.

The tech you mention apart of being Google's property and pretty shady on my opinion, it gives general statistics based on what you commented rather than counting in/out from an area.

mazaira wrote: ↑

Sat Jun 27, 2020 8:27 am

The problem that I'm trying to solve is to have a real-time system that will stop the next person to enter on a library/swimming pool. I don't think the mobile approach would work for it.

Turnstile? Problem solved.

Turnstile works, damn it! I hate it when the mechanical engineers have the best solution
Phone swipe/deposit and then refund works as well as a modern 'ticket' system.

The problem that I'm trying to solve is to have a real-time system that will stop the next person to enter on a library/swimming pool. I don't think the mobile approach would work for it.

Why would it not work? I flashed you mobile as I thought you were proposing a new set of hardware. Have you checked you're qualification costs? Gonna cost you Euro 2K per day at your local EMC test house & (just saying) unless you have sacrificed a kitten under a blood moon you might need a few days. Plus the other standards etc.

The tech you mention apart of being Google's property and pretty shady on my opinion, it gives general statistics based on what you commented rather than counting in/out from an area.

It was a general point meant to illustrate potential. If you check the api you will find that you can use this information to check entry.
If not precise enought for you then there is plenty of tech built into a phone. Lever that. RFID, Blue, Wifi etc.
Google & shady - well I think the Jury came in on that a long time back (& I would agree) but we cannot keep to our spears when others have guns.

Anyway I am going to push my agenda - that it must always be possible to beat a mechanical solution (soz. nvannote) with software
I have a picture of the evolution of man; crouched to upright. The left most 'man' (most bent over) is the mechanical engineer whilst on the right hand side (most upright) we have software engineers. Software must be able to provide a better answer!
PS
Think project managers were left out as they would be too far behind on the left for A4 or a computer screen to show.

PeterR wrote: ↑

Sat Jun 27, 2020 8:14 pm

Anyway I am going to push my agenda - that it must always be possible to beat a mechanical solution (soz. nvannote) with software
I have a picture of the evolution of man; crouched to upright. The left most 'man' (most bent over) is the mechanical engineer whilst on the right hand side (most upright) we have software engineers. Software must be able to provide a better answer!

LOL, Funny stuff. No offense taken. I am very much the tech/modern/software type.

Said turnstile could be a super cool "smart" turnstile, complete with the typical "click-whirl" sound when the gate is actuated (like those in modern airports and whatnot). Of course, you can even stick an ESP32 into it to control the show. Not like the super "dumb" turnstiles found in the NYC subway system, groan...

In any event, It's not very clear how the sensors in the original post are physically configured/shielded. Passive or not, exposing them to an even remotely open environment, false positives are going to happen.

Best Regards

Gonna have to dis 'Click-wirl'. Instead we need a GPP (“Genuine People Personality”) for the turnstyle.
The turnstile could indeed then have a satisfied sigh-like quality to it. “Hummmmmmmyummmmmmm ah!”
(HHG)

PS
You need to think the use cases through.

Someone walks up to the entrance. The IR sensor 'senses' them. They hover, they go back for the kids who (as always) are messing around. Then back through. How many entry/transactions is that? Let alone the dad carrtying junior in (who then runs back to ..).
The newly weds go for a swim. You know newly weds, only ever count as one IR signal. They have a fight and leave as two.
So not all kids have phones & so I would say both our tech solutions have issues.
Sometimes mechanical & human is the way, damn it!

Turnstile was rejected as it tends to create queues and where people aggregates together(supermarket cashier, waiting for entry to a shop, etc) is one of the most dangerous contagious points.

Some of the cases that were posted, as going near and only activating one sensor, a mother staying in front of the sensor which shouldn't create double reads, etc. have been taken into account when programming the sketch for it.

The core question was in regards of sensor type and technology. I started with ultrasound and the amount of fake positives from the echos received was too big, IR is going better but won't work so well on a sunny environment such the pool. From a reddit discussion some people was saying that it could improve with a photoresistor + laser, but the photoresistors also saturate....

A phototransistor plus a laser or IR diode work. Put the phototransistor in a small tube 'looking' at the laser/LED only, in the case of a laser put some red foil in front of it to filter out any non-red light, in the case of an IR led invest in an IR filter. For extra accuracy, let the sketch turn the laser off and on at a few KHz, and measure the *difference* in received light between the two. No difference means the laser is blocked by something, some difference means it's not.