iOS App Acoustically Measures Distances Up To 25 Meters

n01 writes "A recently published app for the iOS platform uses the propagation of sound waves to measure distances of up to 25 meters in a dual device mode. The technique works through repeatedly sending a chirp signal from the master device to which the other (reflector) device synchronizes itself and then replies in a similar fashion. A novel combination of techniques has been engineered to enhance the robustness in noisy environments, such as using an optimum-autocorrelation-signal and semi-automatic frequency calibration together with an averaging over multiple cycles."

Comments

XO (OLPC)

[itamblyn]

http://wiki.laptop.org/go/Distance

Not impressive

[girlintraining]

That's not very impressive. Anyone who has two devices that are syncronized to a common timing source (which most cell phones are) can accomplish this. You just say "I started transmitting at x and you received it at y. x-y/speed of sound at sea level = your result. Now if it could be done with one device, and use doppler effect,etc., to map out the room and roughly what's inside it (like in Batman) then we'd be getting somewhere.

Re:Not impressive

[grumbel]

Yep, neat, but not exactly ground breaking. The OLPC had such an application for the last few years.

Re:Not impressive

[Solandri]

Fun story. While I was at MIT/Sea Grant working on robot submarines, we'd lay an array of underwater beacons for navigation. To conserve power, they'd listen for a certain sequence of sounds from the sub, then reply back with their unique ping. The sub could measure the time it took to receive each unique ping, and thus determine its position by using the ping times and knowledge of where the beacons were. Kind of an underwater GPS. The beacons could last a year or more when used like this, which was a big deal because it was really annoying to locate and retrieve one just to load it with a fresh battery.

On one particular deployment, we left the beacons because we were planning to return a few months later. When we got back, the beacons weren't working. We retrieved them and all the batteries were dead. So we recharged the batteries and redeployed them. After our tests were over, we left the beacons again. When we returned a couple months later, they were all dead again.

Eventually we figured it out. The dolphins in the area had figured out the sound sequence used to make the beacons respond (probably just listened in on our sub). They thought it was pretty cool to get an acoustic response every time they used that code, so they'd been merrily chirping away during those months, draining our batteries.

Re:Not impressive

[hedwards]

GPS containing units had better be able to do that or they'd never get your location down to something reasonable.

Re:Not impressive

[Hartree]

Mark Tilden tells a similar story about a prototype floor cleaning robot.

It took some effort to find out why it worked when he watched it, but when he came home after being away, it was always sitting still in the middle of the room without having cleaned most of the room.

The culprit: His cat would repeatedly trigger its collision avoidance sensor to make it turn away. It was a fun cat toy.

It's hard to design against active maliciousness. :)

Re:Not impressive

[Anonymous Coward]

You just say "I started transmitting at x and you received it at y. x-y/speed of sound at sea level = your result.

And then "your result" has at minimum a wavelength or two of precision, which sucks mightily at audio frequencies. This is why they use a nonperiodic (in this case chirped) waveform and correlation instead of "I started transmitting". You could have read this, at least, before making an ass of yourself.

Not that it's so novel as they try to make it sound, either -- SONAR and RADAR guys did all that long ago, and you'd get the basics needed to implement it in your first semester of DSP in any EE program. In fact, if they're even doing "semiautomatic frequency calibration", they're obviously using linear chirps -- exponential chirps are relatively immune to Doppler or other frequency shifts, and since there's no analog design, are no harder to implement -- suggesting they haven't had (or slept through) any formal education in the field.

  It just bugs me when people who know even less run down every decent, if not outstanding, project like this with their own mix of even lamer approaches ("just as good!") and pie-in-the-sky fantasy ("then I'll get excited")

Re:Not impressive

[Anonymous Coward]

That sounds pretty awesome. Do you know if anyone from the biology department at MIT went back there to study that behavior? Since dolphins already use echolocation to navigate, I just wonder if they were doing more than amusing themselves, and actually managed to adapt to use the beacon system for their navigation. I'm not a biologist, and don't know much about dolphins, so I don't know if that's feasible or not, but it would be pretty amazing.

wait, dual-device mode?

[Tyrannosaur]

I thought this was going to be a cool sonar thing- you'll need 2 iphones? get a tape measure...

Re:wait, dual-device mode?

[n01]

You can also use the device in a single device mode (with headphones), as shown in the second video. I just thought that the dual device mode would be more interesting and therefore emphasized it in my submission.

Star Trek?

[Anonymous Coward]

Weapons Officer: "Captain I can't get a fix on the enemies position."
Science Officer: "We could try using an optimum-autocorrelation-signal and semi-automatic frequency calibration together with an averaging over multiple cycles."
Captain: "Good idea."

Me at home: "Who makes up this stuff."

How about an echolocation app :)

[TheLink]

Some humans can learn echo location[1], but just wondering if we could have an app that sends clicks and chirps and processes the echos and creates a picture or 3D model.

But it might need two or more "ears" to quickly build a 3D image of the environment.

[1] http://www.youtube.com/watch?v=qLziFMF4DHA
http://www.youtube.com/watch?v=GYWpxmcHTOc

This is news because it's on iOS, right?

[QuasiSteve]

Because such an app already existed for PocketPC (That'd be Windows Mobile):
http://nerdipedia.com/tiki-index.php?page=Sonar+CE

Oh, and desktop PC:
http://nerdipedia.com/tiki-index.php?page=Sonar&structure=index

I hear there's a flashlight app for iDevices, too - Slashdot should really look into that. It's magical.

2009 : Sonar Ruler

[sugarmotor]

This didn't seem to do that good of a job, but was 2 years ago.

  * Sonar Ruler, By Laan Labs: http://itunes.apple.com/app/sonar-ruler/id324621243?mt=8
  * http://thenextweb.com/2009/08/20/amazing-iphone-app-lets-measure-distance-echoes-works/

Happy measuring!

"Novel"? Really?

[pipedwho]

When I see the word 'novel' applied to techniques that have been used for decades, I smell 'patent lawyer'. And be extension: astroturf.

Signal processing techniques applied to sonar (whether active, passive or beaconed) including signal correlation, spectral adaptation/equalisation, and filtering are standard fare in this field.

Maybe by novel they mean "on an iPhone"?

That being said, there is nothing that says this won't work - as it worked extremely well 20 years ago on dedicated systems with far less processing power. (Those systems, however, used multiple arrayed transducers and tailored beam patterns to significantly reduce the effective noise floor.)

Re:"Novel"? Really?

[pipedwho]

Keep in mind that all that 'specialised equipment' evolved out of a need to improve the simpler predecessor systems.

Sonar and sonic range finding systems use all that 'extra equipment' to achieve ranges far in excess of 25m and in mediums much more variable than air. The impulse response of miniature consumer grade condenser microphones and speakers are more than adequate for air use within an octave of the audible spectrum. The speakers in the iPhone are primarily limited by their output power, and the fairly omnidirectional nature of the microphones may lack overall sensitivity, but both are simple parameters that really only end up reducing total available range and accuracy (as compared to specialised custom hardware using the same algorithmic solutions).

Applying the same design principles that would normally be applied to a specialised system design to an iPhone implementation, would be very unlikely to provide anything unknown to someone in the industry. This is very similar to early stage engineering "proofs of concept" that are used to test various parameters within a system design, without the interactive complexity of implementing the entire system.

There is nothing within this extremely simple setup that hasn't been done as part of a larger system design. A single (consumer grade) speaker + microphone used in transmissive, active echo, or for passive echolocation is not unusual. Considering the iPhone has excessive processing capability to implement all the standard approaches (correlation, convolution, deconvolution, filtering, impulse response measurement, etc), there is no real need to be 'clever' as such.

'Back in the day', when trying to do this with a 10MIPS DSP in real time with moving objects, it was much more important to come up with better algorithms and shortcuts. Of course, this could otherwise have easily been done with standard theoretical methods and a modern processor a hundred times more powerful.

I see patents pop up all the time that describe things that are far from novel. Most of those patents are usually 'invented' by people with no real experience in the given fields. ie. Ideas that seem like earth shattering discoveries to the uninitiated, but are really just standard techniques used by properly skilled engineers.

I'm not saying that this iPhone app is bad/good, just that it is VERY unlikely to contain any actual improvements to the current state of the art (or the state of the art 20 years ago for that matter). I say this, because there is no real need to do anything new to achieve the results that they are claiming.

BTW, in the past I've worked on sonar/radar systems for air, ocean and rock. The biggest problem in 'noisy' environments is a lack of output level. Multipath isn't a major problem for a point to point (ie. line of sight, shortest path) ranging device - unless you're talking about wave guide shapes/sizes over long distances.

Stand-alone

[soundguy]

About 20 years ago, I had a hand-held device roughly the size of a smart phone but twice as thick that did distance measuring all by itself. It was infrared and as I recall, it was something like $25.00 from Rat Shack or Home Depot or some place like that. A 30 foot tape measure is about $8.00 and works a lot better.

Re:Just use a damn tape measure!

[DrgnDancer]

One suspects that the primary use case for this application is not, "Hey, we need to measure this, let's go get two iPhones!" It's "Hey we need to measure this and happen to have two iPhones, but no tape measure." Most people carry their phones around with them all the time, but unless they're contractors don't carry tape measures. The point of near ubiquitous mobile computers is that you can use them for lots of things. This is a cute and clever thing that you can now use them for.

Re:Nifty

[bonch]

So Slashdot can't report on anything, ever, if it's for sale somewhere. Got it.

Frequency?

[wisebabo]

Is there any benefit to moving to ultrasonic frequencies? Other than making it inaudible (so you don't bother people but maybe dogs!), would this improve the resolution? Does the range decrease? Do consumer level devices cover such a broad spectrum?

By the way, has anyone made an iOS or Android App that can record in the ultrasound (or infrasonic) ranges and change it so that we can listen in audible ranges? Might be neat to see/hear what the bats are doing!

Also, how DO bats build up a good 3D map of their surroundings using just one "speaker" and two "microphones"? Do they send out beams or are their ears swiveling? And, with the limited amount of computing power on a smartphone, would it be able to duplicate it? A bat's brain doesn't seem particularly large and they are doing this FAST (on the fly, ha ha).

Re:Frequency?

[Anonymous Coward]

Is there any benefit to moving to ultrasonic frequencies? Other than making it inaudible (so you don't bother people but maybe dogs!), would this improve the resolution? Does the range decrease? Do consumer level devices cover such a broad spectrum?

As I mentioned in another comment, I've been experimenting with a similar application on iOS devices. Yes, consumer devices do cover ultrasonic frequencies, but barely. For average humans, ultrasound begins above 18 - 19 KHz, and devices with 48KHz range can produce up to 24KHz frequencies... in theory. The problem is that the commodity speakers/microphones in smart-phones are optimized for the human perceptual range, and since ultrasound is beyond that, the transducer dynamic range and/or the in-built signal processing conspire to significantly attenuate and distort ultrasonic signals. Using an iPad, in preliminary experiments, I could only get a range of ~5m using ultrasound, whereas these guys say they can go up to 25m.

Moving to ultrasound also can affect resolution negatively. Since you're effectively using a much smaller bandwidth signal, your positioning accuracy reduces, on top of which, multipath problems get much worse. (Smaller bandwidth because by limiting the signal to ultrasound, you only get a bandwidth between ~18KHz and 24KHz for a 48KHz sampling frequency, and the iPad microphone strongly attenuates signals after the 20KHz range.)

Also, how DO bats build up a good 3D map of their surroundings using just one "speaker" and two "microphones"? Do they send out beams or are their ears swiveling? And, with the limited amount of computing power on a smartphone, would it be able to duplicate it? A bat's brain doesn't seem particularly large and they are doing this FAST (on the fly, ha ha).

Bat ears are highly specialized. This link gives a brief overview of how bats do echo-location:
http://science.howstuffworks.com/environmental/life/zoology/mammals/bat2.htm

I believe smartphones have, or will soon have, enough processing power to do the necessary signal processing if we can design the right algorithms. The problem is it would also need highly specialized audio transducers to get any useful signals, which may not necessarily fit into a smartphone form factor.

Re:Just use a damn tape measure!

[BitZtream]

You could measure room wall lengths close enough for basic estimates on how much paint to buy or fence length estimates and such where you don't need cut to fit accuracy.

Re:Nifty

[zevans]

I strongly prefer empirical methods in this case.