About wiring, etc.:

- Surely, it must be wireless: https://marvelmind.com/pics/architectures_comparison.pdf
- One of the greatest disadvantages of UWB, for example, that is not obvious, is that the anchors typically require PoE for power because they are power-hungry and Ethernet for the connectivity - it is complex and costly
- Wireless in a license-free 915/868 MHz perfectly works for clock synchronization, data exchange, and telemetry in our case - ultrasound. It wouldn't work UWB though

Other comments - about analog, ARM M4, etc., are very valid. We use DSP based on STM32. Analogue-based systems are too limited. There are a lot of things behind the curtain to make it work well. Doable. But not a one-week project :-)

I will post a few replies. Otherwise, my single reply would be toooooooo long :-)

- UWB is the next best thing after us in terms of accuracy: https://youtu.be/EaZY79O_UII?si=FMOR5UcocTdseLy-

Guys, "going through the wall" is the biggest misconception that kills the majority of UWB projects. UWB cannot "go through the walls" or, at least, without going one layer deeper and defining what the wall is.

If it is a radio-transparent wall—glass, paper, thin wood, thin plaster—it is not a problem. UWB can go through the wall. That is what you see in the lab. But then you go to the real factory, and oops...

Examples of obstacles that UWB cannot handle:
- Your own body - place the tags on your shoulder, on your cap - anywhere where it shouldn't go through the body, for example, to have distance measurement to an anchor behind the user
- Brick wall
- Concrete wall
- Metal walls

So, if you want a working, precise indoor positioning system, ALWAYS build with the line of sight between the tags and serving anchors.

For sure, ultrasound-based systems require even more in terms of line of sight, for example, a basic sheet of glass is radio-transparent for UWB but a complete block for ultrasound. But even tougher requirements are for optical systems. You hide an ultrasound-based mobile beacon under your breathable cloth or in the thin bushes and still be perfectly tracked. Not the case for optical systems.

More on the subject:
- https://www.youtube.com/watch?v=zg3oW_U_jdY&t=1608s
- https://www.youtube.com/watch?v=zg3oW_U_jdY&t=2114s

The non-line of sight is perfectly solvable by properly installing the mobile beacons (tags) and stationary beacons (anchors):
- https://marvelmind.com/pics/Marvelmind_Robotics_ENG_placement_manual.pdf
- https://youtu.be/MqfoZDU52fs?si=n2iqs_rlAsdXKu-l

Well, we were slow to discover the project, but since we focused on the same task at about the same time, I believe we have a few cents to contribute to the discussion :-)

A few short comments based on the topics raised above - not in the order of importance or time:

Range:
We recommend a range of up to 30 meters in a real-life industrial environment. In the lab - up to 50 meters. With a Horn - 1D tracking or precise linear distance measurements - up to 120 meters or so. Using submaps, you can cover as large an area as you wish - similar to cells in cellular networks:
- https://marvelmind.com/forklift-tracking-and-monitoring/ - 450x450-meter T-shaped warehouse with 120 forklifts
- https://marvelmind.com/solution/cranes/ - overhead cranes - up to 480 meters long indoors

With narrow beam - 120 meters - 1D tracking: https://youtu.be/OYrLOpGvbig?si=U67lt1CqTPqbpGXV.

There have been a few projects who already tried this. For example: http://cricket.csail.mit.edu/

I have tried this in the past (for a company). It was not really a success.

For starters the range was limited 10-15 meter.I am not an analog engineer so maybe it can be improved. But also direction of the beam, to get a good range a narrow beam was needed. 

Maybe with the new 32bit micro's and some clever processing a lot can be improved. 

 I think/hope the product from pozyx (http://www.pozyx.io/) works (currently a kickstarter). Because it would solve this problem. It is smaller than ultrasonic, does not need to be open to the air. Also communication is also solved with this product. Only thing is: it has to work first ;)

@Armin, thanks for the links & info! You are very right about the range being a very difficult challenge I think. We shall see what I can do. 

The links are great you sent. I really think the posyx is a great device. I'm very impressed with it. I'll have to learn more about what this wide-band radio stuff means.  The fact that it uses radio waves and can go through walls is a huge leg-up on any ultrasonic or sonic solution requiring traveling pressure waves and direct line-of-sight through the air.

Do you say that your attempt with the company was "not really a success" because A) you never even got it working at all, B) you never got the resolution or precision you wanted, or C) you never got the range you wanted? ...or some other reason?

Main problem was it was not economical feasible.

First time we looked at it we build a prototype to test the range. This was still connected with a wire to signal the start of the burst. Found some issue with the amplification, signal to noise issues. This range was to 10-15 meters when transmitter and receiver were pointed at each other.

Second time we did some simulations first, we needed a soccer field size indoors. We ended up needing a lot of beacons. Making it very expensive to install/maintain. All based on ranges of 10-20 meters.

So my advice: use some good low noise amplification (instrumentation amplifier, but I am not an analog guy), use some 32bit M4 (floating point) high speed adc (with DMA) microcontroller because you will need some advanced tricks to get signal out of the noise.

Anyone see a commercial use for this project?  Where there'd be enough customers to make a successful kickstarter? That is certainly one factor I consider in thinking about which projects to complete. I'm itching to contribute something to this world.

There would definitely be a market for indoor UAS applications if the accuracy is there. If you can make the receiver small enough and interface easily to autopilots, you could easily set bounds to remain in and paths to follow.

The RC guys would go nuts over a rock solid packaged solution for under a couple of hundred.

Put it this way, I'd buy one.

Yes :-)))

Surprisingly solid interest from autonomous indoor flying guys:
- https://marvelmind.com/solution/drones/
- https://marvelmind.com/drone-competition/
- https://marvelmind.com/cases/indoor_positioning_system_for_small_drones/
- https://marvelmind.com/cases/indoor_positioning_system_for_drone_swarms/
- https://youtu.be/2Rn0On9aEKg?si=Hc4zBwkiQSs3JXVn

"I plan for the application size to be at most the size of a soccer field."
Can you show your calculations for the actual range for the transducer you are planning to use?  I found the dimensions to be an order of magnitude too high if you are only using what you say for that.

Wiki:  the preferred size for many professional teams' stadiums is 105 by 68 metres

@K.C. Lee, I haven't done calcs on acoustic range yet. I was going to start small, and get it functional. My distance calcs were to see what audio sampling rates I would need for a given distance *resolution*, based on the speed of sound, not a range. I considered radio systems too but determined the speed of light is too fast to be practical unless you plan on having beacons perhaps kilometers apart....that's my thinking thus far. I plan on initially attempting with an ATmega328 mcu....max audio sampling rate = 50kHz.  The problem is processing the samples.  The details will I'm sure each pose significant challenges as they come, but I know with enough time and persistence I'll figure it out. In all honesty though, this project is like 10th on my list, so if I don't make progress for 3 years, I won't be surprised. My detailed notes I spent a few hrs writing the other day were for myself to come back to....eventually in my life, when this becomes a priority. They were simply to serve as a basis to see A) if this project seems theoretically feasible for me to accomplish given my skillset, and B) to remind me of some of the possible solutions I've thought out, so that I can get my mind back into it quickly when the time comes.

FYI:  My contest entry project log: https://hackaday.io/project/5903-sonar-for-the-visually-impaired/log/18329-ultrasonic-module-virtual-teardown

If you want to look at the analog waveform or making your own design  http://www.kerrywong.com/2011/01/22/a-sensitive-diy-ultrasonic-range-sensor/
Here is an app note to show how to calculate the range and sensivity: http://www.prowave.com.tw/pdf/an0508309.pdf

@K.C. Lee, thanks for the links; much appreciated

- 6-7 x Super-Beacons on one long side

- 6-7 x Super-Beacons on another long side

It takes a bit of stretching because the longest distance would be about 40 meters, but it is still doable.

Have you given much thought to the ultrasonic transducers themselves? What are your plans there?

Are you referring to generating or receiving the sound? At this moment, I haven't put a lot of thought into the ultrasound hardware. I focused my efforts thus far on timing, distance (ie: spatial resolution of distance measurements, NOT distance as in range of the system), and speed calcs to see if it was even feasible from that  perspective. It definitely is.  Do you have recommendations?

These devices are very short-range, but worth looking into: http://www.ebay.com/sch/i.html?_sacat=0&LH_BIN=1&_nkw=arduino+ultrasonic+sensor&_sop=15

-I have a couple I've used only a little

Also, I plan on sampling with a microphone to get analog audio samples of the ultrasonic waveforms, and I need to look into using large audio speakers to generate ultrasonic waves.

Audio speaker is *Audio* frequency range.  By reducing frequency, you are reducing resolution.

@K.C. Lee I'm not sure how you're thinking about the problem, but frequency is irrelevant the way I see it. I could use any frequency and keep the same resolution, so long as I can accurately detect the edge of the first (or last) pressure wave in a sequence I send and receive. I'll probably begin in audible frequencies anyway for easier debugging (I can hear it working). Ultrasound is only useful for me so it isn't bothersome to people.

Also, unless you've verified that audio speakers absolutely cannot output ultrasonic frequencies, I will assume that they possibly can. I will do some tests. If i use a raw speaker and my own amplifier, it simply comes down to the time response of the speaker diaphragm...or however they work. I can send whatever frequency in, and see if it can come out. Piezo buzzers are also on the list of beacons to try. I suspect they can do high frequencies without a problem. 

Wavelength determines the resolution.  The size of transducer and the wavelength determines the beam pattern. 
Might want to see my other reply to see the amount of research I have done so far to pass your judgement on me.

http://www.sensorsmag.com/sensors/acoustic-ultrasound/choosing-ultrasonic-sensor-proximity-or-distance-measurement-838

>The resolution of a range measurement made with an ultrasonic sensor is 

influenced by many factors. Since the sensor is measuring the arrival
time of an acoustic pulse, the higher the ultrasonic frequency the
greater the resolution because both the wavelength and period of the
echo signal are smaller at higher frequencies.

I think it depends on how you're planning on designing the system. I've spent a decent bit of time working with industrial-style ultrasonic level transmitters and they can be fussy if they aren't set up correctly--interference, beam angle issues, etc. I'm assuming you're planning to sync timing on all devices then compute distances based on time delays? 

One thing to keep in mind--the speed of sound in air varies based on temperature (along with other factors like composition). You'll need to measure temperature and do something like this: http://hyperphysics.phy-astr.gsu.edu/hbase/sound/souspe.html

One thought--prove out your concept in a quiet, climate controlled, indoor, smallish room. If you can get it to work in your basement, then you can start scaling to football field dimensions.

@zakqwy Thanks, yes, speed of sound compensation is planned. a = sqrt(gamma*R*T), where a = speed of sound. gamma = 1.4 and R is a const, so speed of sound is only a factor of temperature. And yeah, a small room is the first plan.

And yes, that's correct about the time delays. PS do you have any documentation, sales pages, anything you can share, on your industrial systems you mention? What were they for? Which company?

I've worked extensively with Siemens level instruments; they bought Milltronics a number of years ago, and those folks pioneered industrial ultrasonic level measurement back in the 70s. Current state-of-the-art for this technology is the EchoMax series of transducers and the LUT400 controller platform.

We had to custom-produce the transducers.

Of course, all electronics, low-, middle-, and high-level SW had to be developed from scratch.

There are several factors affecting the frequency selection:

- External noise is the most contributing. However, if you increase the frequency too high, you have too high losses in the air. There is an optimum for your particular set of requirements: https://marvelmind.com/download/how_to_choose_ultrasound_frequency_for_beacons/

Also, there are some practical limitations of different sorts:
- If the frequency is too low, the transducers must be larger - difficult to produce, or you must use a different technology
- If the frequency is too high, they become very directive - difficult to provide coverage
And many other small things that cumulatively create many barriers.

(FYI for others, here's Blecky's "SubPos" system: https://hackaday.io/project/4872-subpos-positioning-system)

Well, I wish I could say this was a full-time project, but at any given time there are a dozen things I'm working on, many of which will make it to my website. Expect slow progress. :) Thanks for following.

I will definitely be following this project closely ;)

You could name this project UltraPos :P