Sonic ranging experiment on small scale

[Mr RB]

Hi, I had a bag of matching cheapo electret microphone inserts, and did an experiment on small scale sonic ranging.

This uses 2 or more mic's to detect where a surface has been "tapped", ie; detects and locates a sonic event.

Maybe this could be used for special keypads or musical instruments or some other fun application.

Here's a photo and 'scope capture of the test setup;

**broken link removed**

**broken link removed**

And for anyone interested in small scale sonic ranging using cheapo hobby level technology the full experiment is here;
https://romanblack.com/SonicRanging/Sonic_Ranging.htm

 

[ronv]

Cool...

 

[BeerBelly]

Stereoscopic hearing?
You sure have a lot of fun.

 

[misterT]

That is called "Differential Time Of Arrival" (DTOA). And the localization method is called trilateration, not triangulation. Helps with googling if you get your terms right.

We did a robot localization system using three beacons around the area and one in the robot. The robot sent a pulse and the beacons responded to that. The robot could localize itself by listening the responses. You would need three microphones (or beacons) to locate the sound source on a plane. With two microphones you get a line (hyperbola) where the sound source might be. If you want I can post the math for intersection of two hyperbolas.. If I find my notes. The solution comes down to few (~10) multiplications and divisions (~2) and finally solving an intersection of two lines.

 

[Mr RB]

That is called "Differential Time Of Arrival" (DTOA). And the localization method is called trilateration, not triangulation. Helps with googling if you get your terms right.
...

Thanks. I had no problems googling for "sonic triangulation".

...We did a robot localization system using three beacons around the area and one in the robot. The robot sent a pulse and the beacons responded to that. The robot could localize itself by listening the responses. You would need three microphones (or beacons) to locate the sound source on a plane. With two microphones you get a line (hyperbola) where the sound source might be. If you want I can post the math for intersection of two hyperbolas.. If I find my notes. The solution comes down to few (~10) multiplications and divisions (~2) and finally solving an intersection of two lines.

Cool! Maybe your project was one of the uni projects I found in my google searches? There are quite a few projects out there for robot and gunshot event localising etc, but I didn't see much for small scale localising ie keyboard type stuff.

Re the math I already have a ton of stuff on the traditional way of doing it, I'm more interested in solutions that DON'T require the hyperbolic math, and a few present themselves if the mic's and sound sources are in fixed positions on a 2D plane.

 

[atferrari]

That is called "Differential Time Of Arrival" (DTOA). And the localization method is called trilateration, not triangulation. Helps with googling if you get your terms right.

We did a robot localization system using three beacons around the area and one in the robot. The robot sent a pulse and the beacons responded to that. The robot could localize itself by listening the responses. You would need three microphones (or beacons) to locate the sound source on a plane. With two microphones you get a line (hyperbola) where the sound source might be. If you want I can post the math for intersection of two hyperbolas.. If I find my notes. The solution comes down to few (~10) multiplications and divisions (~2) and finally solving an intersection of two lines.

The hyperbola solution becomes less and less accurate when the distance between the target and sensors is reduced. After all you are working on a curve that approaches a line.

I find the "autolocalization" idea very interesting. Could you post?

 

[misterT]

I find the "autolocalization" idea very interesting. Could you post?

This was for Eurobot competition. The play area is about 3 meter by 2 meters and there are two robots playing at a time. Both players can install 3 beacons on the side of the table, 1 beacon on the opponent robot, and one beacon to their own robot. We built a system that can locate the opponent robot and our own robot. Here is the "paper" I wrote. It does not explain the system very thoroughly (it is a Mathematica worksheet) and I won't re-write it just for this.. you can ask questions though. I try to answer.