Topics Topics Edit Profile Profile Help/Instructions Help Member List Member List  
Search Last 1|3|7 Days Search Search Tree View Tree View  

Ideas on how to track a fly

:: EPE Chat Zone ­:: ­Radio Bygones Message Board :: » EPE Forum Archives 2005-2006 » Archive through 03 July, 2006 » Ideas on how to track a fly « Previous Next »

  Thread Last Poster Posts Pages Last Post
  ClosedClosed: New threads not accepted on this page        

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 81
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 05:27 am:   Edit Post Delete Post Print Post

A man from our local university was talking to me the other day and said they were experimenting with flies.His problem is they need some way of tracking them while they are flying.I asked him how far away will it need to pick up, and he said no more than 25 feet.I told him that i thought that it was impossible, but what do i know, im almost new to electronics.I know theres tracking cameras out, but dont know how precise they are.I think that is one thing hes wanting to do later on. For a camera to be able to track the fly. But the important part would be the pattern on a computer.I thought about it and came up with 3 differant possibilities.Sound,( from the fly), Imaging from a camera, or lasers.Any ideas?
Top of pagePrevious messageNext messageBottom of page Link to this message

Petelobus
Frequent Contributor
Username: Petelobus

Post Number: 138
Registered: 06-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 09:17 am:   Edit Post Delete Post Print Post

Is it just one fly or several? Is it permissable to mark the fly, or use say a black fly in a totally white room? If so I'd have thought a tracking camera might be possible. Otherwise how about something similar to a Laser Doppler Anemometer? I know that they can measure particle velocities in gas flows, so maybe fly velocities would be possible too? Reckon it all sounds quite expensive (or time consuming) -has he got time and /or money?

Ian
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 372
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 12:36 pm:   Edit Post Delete Post Print Post

Bats do it, in the dark, using ultrasonics, so in principle it's feasible !!

J
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 82
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 03:41 pm:   Edit Post Delete Post Print Post

I wasnt interested in this until he told me what they are doing with the flies. They are trying to engineer fireant eatting flies.Since im allergic to fireants, im all for it.I will ask him how many flies they are and what kind of budget the university has on that project.
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 374
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 07:45 pm:   Edit Post Delete Post Print Post

What's a fireant Scott? Not a beast known to use in the UK. Interesting topic generally the topic you've raised :-)
Top of pagePrevious messageNext messageBottom of page Link to this message

Mike_b
Frequent Contributor
Username: Mike_b

Post Number: 193
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 10:10 pm:   Edit Post Delete Post Print Post

This is a bit 'long distance' for your application perhaps ...

The long distance movement of Butterflies were tracked by means of radar. A Gunn diode plus a few surface mount bits were glued onto the thorax - the wire lead was cut to be a 1/4 wave aerial - this gave traceable echos at up to 2km away.

A two dimensional solution to the problem is not too bad using camera systems ... black fly, white background, say 10 frames per second scan rate and plenty of compute power.

To get 3 dimensional data you will need two cameras to compute range to target as well ... tricky, but not impossible.

rgds

mb
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 83
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 08 June, 2006 - 10:56 pm:   Edit Post Delete Post Print Post

Thanks for the replies. John, a fireant here in the states is a ant that will sting you if it gets on you or you step near, or on there mounds.Like a bee or a wasp.It's red in color.I think they are doing experiments on them because they are getting more potent and spreading more everyday.There taking over here, i can tell you that.Used to be when the kids played outside they were scared of bees and wasps. Now its fireants.I talked to him today and he said they keep about 15 to 20 flies in the room at one time. But he said tracking just one at a time would be a big help.As for as a budget, hes got 6 fast computers , but the budget is low and there hoping to get more money as the end of chool for the year has come.He said that they will spread more money around while the students are out for summer break.
Top of pagePrevious messageNext messageBottom of page Link to this message

Eagre
Frequent Contributor
Username: Eagre

Post Number: 67
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 09 June, 2006 - 12:54 am:   Edit Post Delete Post Print Post

As John pointed out, it would seem like ultrasonics would be the way. Bats routinely discriminate and track multiple targets, often smaller than flies. The 25 ft range may be a problem as I don't think bats achieve that, but with far greater transmitting power than a bat can muster that should not be a problem.

Ed Grens
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 376
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 09 June, 2006 - 01:40 am:   Edit Post Delete Post Print Post

One problem with the bat technique (assuming you can catch one!) is that of direction. The bats move their bodies and dont care the precise compass direction of the fly, if it's in front of them they catch it. But any mechanical/electronic method means that the system must orientate towards the target, or sense target direction in relation to where the sensor is. And as Mike B comments, 3D imaging of some sort requires at least 2 sensors and processing in order to triangulate on the target. Bats have binaural hearing from well-cupped ears which can move and locate the target's echo direction. Can we replicate such? Intriguing.

Yes Ed, the bats we watch in the evenings at home (Pipistrelles we think) seem to work well within a 25ft range of the target.

Scott - sounds real nasty!

J
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 84
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 09 June, 2006 - 02:25 am:   Edit Post Delete Post Print Post

John,how sesitive is the pic ultrasonic radar you built in last june's issue?
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 378
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 09 June, 2006 - 09:13 am:   Edit Post Delete Post Print Post

Not enough for this purpose Scott. It uses 40kHz transducers and you need something with far greater potential definition from higher freq transducers for the accuracy you're looking for.

J
Top of pagePrevious messageNext messageBottom of page Link to this message

Darren
Frequent Contributor
Username: Darren

Post Number: 112
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 09 June, 2006 - 09:48 am:   Edit Post Delete Post Print Post

Hi Scott,
Will the flies be tracked in a controlled environment? ie a lab room/test chamber? Perhaps infrared/thermal cameras may be a solution. I saw on TV last night a system using an infrared camera linked to a laptop with software that would only detect movement, and movement of a particular size. The system was designed to detect dingos (native Australian dogs) but I'm sure if the flies were in a controlled environment (temperature controlled - not too hard to do) it could be adapted to track much smaller objects, or perhaps the flies, and ants, could be tagged (perhaps painted or something) with a thermal/infrared reflective paint/dye.
Regards,
Darren
Top of pagePrevious messageNext messageBottom of page Link to this message

Obiwan
Frequent Contributor
Username: Obiwan

Post Number: 521
Registered: 12-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 10 June, 2006 - 01:57 am:   Edit Post Delete Post Print Post

Ultrasonics won't do it. Way too slow.

sad to say, but I doubt he'll be able to do it with a "low" budget.

you'd need something like fast doppler. The problem is they're very small, and fast. by the time you "pinged" a fly, and got the signal back, it's somewhere else.

Not to mention just the resolution of the wavelength.

And they'll be too fast for a camera too. It would be very hard to find a motor that could make that camera track and zoom in and out to keep an "eye" on one. And from 25' or so, I doubt the camera could even spot it.

Interesting idea, but I doubt you can do it on the cheap unless you can get some good military hardware second hand. Even then it'd be hard as that is long range, not set up for 25' or so.


I sure hope they have luck, I'll even be a test bed!! I HATE those things!! you can't kill them, they just move a few feet. And when I get bit, it's worse than any mosquiteo bite and then it's like the flesh around it just dies off and it takes weeks for it to heal.
Do Not Hit The Fly That Lands On The Tigers Head.
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 85
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 10 June, 2006 - 03:43 am:   Edit Post Delete Post Print Post

Yes, they are in a controlled room.Ill have to look around for more info.Thanks for the replies.
Top of pagePrevious messageNext messageBottom of page Link to this message

Eagre
Frequent Contributor
Username: Eagre

Post Number: 69
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Monday, 12 June, 2006 - 02:29 am:   Edit Post Delete Post Print Post

Since the velocity of sound in ambient air is of the order of 1000 ft/s, at 40 kHz the resolution (approximately wavelength) should be about 1/3 inch. Perhaps good enough for tracking - about a fly's size. And, flies are relatively quick for their size, but not fast. With sampling on the order of 1 MHz, a detailed track should be available. The problem remains the ultrasonic detection system range.

Ed Grens
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 86
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Monday, 12 June, 2006 - 05:33 am:   Edit Post Delete Post Print Post

Anyone know where to find high powered transducers? It seems this is just a matter of more power needed.Like ed pointed out, the processing power needed can be found in any pentium class computer. I have a meade telescope. It will track an object you can barely see for 8 hours, ( iv tested it) before having to recenter it. If you can get that kind of resolution out of a hand held device, i dont see any problem in the software side of the problem.
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 387
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Monday, 12 June, 2006 - 11:30 am:   Edit Post Delete Post Print Post

Thomas Scarborough tells me that RS are doing hi freq (200kHz) UX transducers and others; might be worth a search of their site.

J
Top of pagePrevious messageNext messageBottom of page Link to this message

Thomas
Frequent Contributor
Username: Thomas

Post Number: 106
Registered: 04-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Wednesday, 14 June, 2006 - 09:14 pm:   Edit Post Delete Post Print Post

I gave some hard thought to this one. A simple solution seems impossible. The closest I could come was -- say a fly emits a “signature” frequency with its wings -- then one might (roughly) track that through two dimensions with three very sensitive microphones, suitable bandpass filtering, and charge pumps. Thomas. (I wish someone would invent a search-and-destroy circuit for mosquitos)!
Top of pagePrevious messageNext messageBottom of page Link to this message

Obiwan
Frequent Contributor
Username: Obiwan

Post Number: 539
Registered: 12-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 15 June, 2006 - 03:59 am:   Edit Post Delete Post Print Post

I agree with Thomas, it can be done, just not cheap.

There is one alternative though.

If they can attach something to the fly. Something similar to the resonant devices they use in store security.

You might be able to get a 3D tracking system that way.

Possibly some reflective pain might make the visual method easier. It would give a camera MUCH more to look at.

But, flies frown on that. Not to mention PETA, man, they may even file a cruelty suit against ya for painting flies.

Scott, tracking a dimly lit slow moving Pixel is far easier than tracking a buzzing fly at high speed. Flies can change direction very quickly. I think you would be hard pressed to have a Pentium track a single fly in full motion, much less 3 or more.

Maybe ya'll have much bigger flies over there? But here they're small and very quick, They change direction very fist. If you can't catch one with a fly swatter, a computer isn't going to have much more luck.

Yes, it could be done. But not cheaply.
Do Not Hit The Fly That Lands On The Tigers Head.
Top of pagePrevious messageNext messageBottom of page Link to this message

Mike_b
Frequent Contributor
Username: Mike_b

Post Number: 212
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 15 June, 2006 - 11:01 am:   Edit Post Delete Post Print Post

Obiwan ... Thats what my post was referring to... its similar to systems used to track modern commercial jets. When a radar pulse is detected in the aircraft, a transponder transmits back a signal.

To track Butterflies, a single Gunn diode was stuck to the thorax, with the connection lead cut to 1/4 wavelength .. plus a couple of surface mount tuning capacitors.

The energy from the radar pulse stimulated the Gunn to reradiate, hence the power supply payload was very cleverly eliminated - essential if the 5 gram butterfly is ever to take off again.

I thought that this method of 'remotely powering' a circuit might have application to the fly problem as now perhaps you could place a transponder and even a microprocessor onto the fly which might help.

rgds

mb
Top of pagePrevious messageNext messageBottom of page Link to this message

Vlf
Regular Contributor
Username: Vlf

Post Number: 25
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 15 June, 2006 - 12:13 pm:   Edit Post Delete Post Print Post

This fly thing... ultra-violet for UK house fly's are good to attract them, in a low light situation; have a one-way trap system so they go in, but can't get out; use known low tech simbionics to get your fly... just a thought !!

Regards, Brian. L.
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 88
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 16 June, 2006 - 03:25 am:   Edit Post Delete Post Print Post

Vlf, we are not trying to kill the fly.We are trying to track them. Theres a thought. Make a device to shoot down pest.The money that could be made off of that is unimaginable.
Top of pagePrevious messageNext messageBottom of page Link to this message

Zeitghost
Frequent Contributor
Username: Zeitghost

Post Number: 227
Registered: 01-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 16 June, 2006 - 08:59 am:   Edit Post Delete Post Print Post

I've got a design for an 8051 controlled minigun for use as a fly killer.

Ammo costs are pretty high though... :o)
Top of pagePrevious messageNext messageBottom of page Link to this message

Zardof
Valued Member
Username: Zardof

Post Number: 18
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 16 June, 2006 - 12:01 pm:   Edit Post Delete Post Print Post

Hi,
How about a touch of this paint http://www.evidenttech.com/applications/quantum-dot-pigments.php on the fly and a suitable light source and an IR camera/recorder . This should give you the flys path.
A high tech solution is going to cost lots.

Gary
-------------------------------------------------
Daft Leprechauns are not big and they are not clever.
Top of pagePrevious messageNext messageBottom of page Link to this message

Wattyei9feb
Just joined
Username: Wattyei9feb

Post Number: 3
Registered: 06-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Sunday, 18 June, 2006 - 02:55 pm:   Edit Post Delete Post Print Post

Some of the minature 40KHz ultrasonics and some minature piezo will work 80kHz to 150Kz. Small electret microphones are usable up to 100kHz.

Ultrasound is very feasible up to 10m range.

A vibrating reflector on a tweeter can "scan" an ultrasonic beam at high speed. The "ears" don't need to move, three well spaced (corners of room?) will give additional information regarding position. The reflected pulse delay to each ear gives distance of Tx to fly to ear.
Top of pagePrevious messageNext messageBottom of page Link to this message

Thomas
Frequent Contributor
Username: Thomas

Post Number: 107
Registered: 04-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Sunday, 18 June, 2006 - 07:06 pm:   Edit Post Delete Post Print Post

Hello Wattyei9feb: This is what I worked out for a past Bat Detector project -- I did some thorough tests at the time. Standard 40kHz ultrasonic receivers were generally effective to about 80kHz, although sensitivity dropped off as one went higher or lower than 40kHz. Quality piezo sounders were effective to about 50kHz, again with sensitivity dropping off the higher one went -- and budget piezo sounders went up to about 30 kHz. That need not disagree with your figures -- this is simply the results of my own experiments. Thomas.
Top of pagePrevious messageNext messageBottom of page Link to this message

Muenchlaender
Valued Member
Username: Muenchlaender

Post Number: 11
Registered: 03-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Wednesday, 21 June, 2006 - 12:20 pm:   Edit Post Delete Post Print Post

Hi Scott,

If your man has plenty of PC's, I would recommend to go with a video camera based approach and to grab a couple of software students. A half decent digital camera doesn't cost much these days.

Depending on the room layout, and whether the fly can hide behind objects, will determine how many cameras you will need - possibly up to three to really aquire a 3D position.

As long as the cameras themselves are fixed in position, it is relatively easy to compare two frames of video, and determine where the 'black spot' was, and where it has moved to.

Of course, depending on the camera resolution, it can take a lot of CPU power to compare all those pixels, hence why you need a PC. However, you could also simply record an hour or so of video onto tapes, and then post-process the images later. It doesn't need to be real-time.

I think this would really be the easiest, cheapest way forward, and would provide a proof of concept to see if it is worth investing more money into the project.

There are quite a few companies who specialise in developing motion detection software for video cameras, although the look for people, cars and bags normally, but the above approach is the one they use. They may even be interested in working with your friend.

Hope this helps,

Stuart
Top of pagePrevious messageNext messageBottom of page Link to this message

Muenchlaender
Valued Member
Username: Muenchlaender

Post Number: 12
Registered: 03-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Wednesday, 21 June, 2006 - 12:39 pm:   Edit Post Delete Post Print Post

By the way - just found one of those companies developing motion detection systems:

http://www.objectvideo.com/

I've seen this working at a trade fair, and it is scarily accurate!
Top of pagePrevious messageNext messageBottom of page Link to this message

Joe
Frequent Contributor
Username: Joe

Post Number: 135
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Wednesday, 21 June, 2006 - 01:20 pm:   Edit Post Delete Post Print Post

I think that some flies can detect ultrasound. If this is the case, then you may disorientate them using this approach - who's interested in dinner when they have a screaming headache.

Your tracking solution shouldn’t interfere with their normal behaviour as it wouldn’t be representative of what happens when there in the wild.

I would probably try to avoid any active type of search system. To get an accurate 3D map you would need at least 3 TX/RX pairs and possibly more to get a suitable "view angle". They would need to be on different frequencies so that they didn't interfere with each other or some other method employed. Either way, it all sounds real complex.

A passive system like cameras or perhaps even IR cameras if fly’s show up on them, would probably be simpler in the long run.
As a previous poster has said, record a couple of hours, say using a multi-input video card direct to hard disk, and then process the data off-line at your leisure. Removes any problems related to the processing hardware not being able to keep up as you can play back the video data at a fraction of the original speed if required.
Do one thing each day that scares you – work here !
Top of pagePrevious messageNext messageBottom of page Link to this message

Ant
Valued Member
Username: Ant

Post Number: 17
Registered: 05-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 22 June, 2006 - 11:12 pm:   Edit Post Delete Post Print Post

Hello all,

It's not directly - or easily! - applicable but it struck me just, how do moths etc manage to track potential mates? They can smell the pheremones at incredibly weak dilutions, we're told, and obviously they need to be essentially down-wind. But it's difficult to see how such a facility, however sensitive, can be stereonasal or whatever word one might coin for it. It must be very efficient though.

Another bit of lateral thinking - might it be possible to track the fly indirectly, perhaps by watching the eyes of a frog which is looking for a meal?

As regards IR sensing of a fly, insects are essentially cold-blooded, although some insects do generate heat by buzzing their wings - bees for instance.

UV might be a better bet, a lot of insects use UV to find food etc and it seems a fair bet that they may somehow emit that sort of wavelength too?

Regards Ant
Top of pagePrevious messageNext messageBottom of page Link to this message

Thomas
Frequent Contributor
Username: Thomas

Post Number: 108
Registered: 04-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 23 June, 2006 - 06:35 pm:   Edit Post Delete Post Print Post

Hello Ant,

I've observed fruit flies, if one has a glass of beer on the table. They have a fantastic ability to zoom in on the glass. They go into a zig-zag pattern, at first wide, then narrower and narrower. They must be able to compute from that pattern where the glass is located.

Thomas.
Top of pagePrevious messageNext messageBottom of page Link to this message

Mike_b
Frequent Contributor
Username: Mike_b

Post Number: 216
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 23 June, 2006 - 07:46 pm:   Edit Post Delete Post Print Post

Slightly off topic, but I never fail to be astounded that a fruit fly or mosquito that can have no more than 10,000 to 50,000 neural switches say can solve flight stability problems, have a basic visual system with recognition/orientation control and make a host of basic decisions ....... yet it takes all my brain power (I grant you not a great deal) to get my model helicopter to hover without inverting and turning into a 'one time' lawn mower.

... but I still feel a sense of satisfaction every time I squish one of the little buggers ....

rgds

mb
Top of pagePrevious messageNext messageBottom of page Link to this message

Ant
Valued Member
Username: Ant

Post Number: 18
Registered: 05-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 24 June, 2006 - 11:05 am:   Edit Post Delete Post Print Post

Hello all,

It's not helping to stay on topic, but thinking about eyesight - as I understand it humans see by dithering the vision very very slightly and it's presumably that action which enables us to distinguish edges. I think that in bees, for instance, the dithering is done by the bee's body balancing it's wing vibrations.

So presumably in its reducing sweep approach Thomas' fly is improving its eyesight acuity. Perhaps we should try waving our "EPE"s about rather than looking for our reading glasses!

To get back to the original post, could a fly be trained to follow a beer glass? You could then have a stepper-motor driven beer glass and you'd _know_ where the fly was!

Regards, Ant
Top of pagePrevious messageNext messageBottom of page Link to this message

Thomas
Frequent Contributor
Username: Thomas

Post Number: 109
Registered: 04-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 24 June, 2006 - 08:46 pm:   Edit Post Delete Post Print Post

Hello Ant, I did a constructional project not that long ago, for THAT magazine :-) -- a robot that had a single eye, yet was able to zoom in on a target. It homed in through using a zig-zag pattern. I like your innovative thinking. OF COURSE, why didn't we think of THAT -- why track the fly when the fly could track the detector?! It takes a true genius to figure something like that out. Thomas.
Top of pagePrevious messageNext messageBottom of page Link to this message

Boris
Frequent Contributor
Username: Boris

Post Number: 148
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 29 June, 2006 - 12:30 am:   Edit Post Delete Post Print Post

A small camera flash or strobe will reflect more dimly from more distant objects. A fly, in flight, should light up more than the background. Also, two successive flashes could produce an image with two bright spots from which the 2D position, direction and speed of the fly can be determined. Software could generate a predicted track for the motion of the fly and direct an articulated camera to a position on this track.

I took some experimental pictures. Here the ‘fly’ is actually the neck of a blue toy balloon dangled from black thread. The background is about 4m away and the ‘fly’ is about 2m away. It’s moving from right to left.

1. You can’t quite make out the fly yet.



2. The fly a moment later.



3. This is the two previous images mixed on Paint Shop producing a single ‘difference’ image.



4. Same as 3, but with enhanced contrast.



5. The predicted track. The camera should now move so as to point to the yellow X.



Obviously less than a perfect demo, but it illustrates the idea. Once the camera has been re-positioned the cycle begins again.
www.pyxia.com - BASIC was never less basic.
Top of pagePrevious messageNext messageBottom of page Link to this message

John_becker
:::: Super User ::::
Username: John_becker

Post Number: 452
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 29 June, 2006 - 01:44 am:   Edit Post Delete Post Print Post

Fascinating Boris, well done! :-)

J
Top of pagePrevious messageNext messageBottom of page Link to this message

Boris
Frequent Contributor
Username: Boris

Post Number: 149
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 29 June, 2006 - 02:11 am:   Edit Post Delete Post Print Post

It occurs to me that if the flash is separated from the camera and were directed so as to not illuminate the background, perhaps on the ground pointing upwards, or on the ceiling pointing downwards, the fly will stand out much better.

Also:

Two separate flash units, one placed on each side of the camera, maybe a foot or two apart, would cast two shadows onto the background, something like o-x-o where o=shadow on background and x=fly in foreground. The distance between the shadows would be inversely proportional to the range of the fly and would be quite accurate. You would of course need to know the distance to the background and it would need to be of uniform shape and colour.
www.pyxia.com - BASIC was never less basic.
Top of pagePrevious messageNext messageBottom of page Link to this message

Scott2734
Frequent Contributor
Username: Scott2734

Post Number: 94
Registered: 10-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 29 June, 2006 - 05:16 am:   Edit Post Delete Post Print Post

I wished i could see that better.Looks like a good idea tho.Very good boris.I have 2 spare webcams here,but dont think they would have the resolution to do that. What kind of programs could do a x y track with 2 cameras? I doubt it could be done in basic, maybe c++.
Top of pagePrevious messageNext messageBottom of page Link to this message

Joe
Frequent Contributor
Username: Joe

Post Number: 141
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Thursday, 29 June, 2006 - 07:46 am:   Edit Post Delete Post Print Post

Scott,
I was thinking about your last question so I wrote a quick VB program to see how long it might take to process frames of data. I created 3 arrays 640x480 to a 16 bit depth. I filled 2 of them with random data but in reality you would place one frame of video data in each array.
Then timed how long it would take to compare each pixel of the two frames and generate a 3rd "difference" frame based on the colour value changing by 10% from one frame to the next.
The P4 3GHz PC's we use here managed to average the comparision in 14ms - which is better than I thought.
Of course, that dosn't allow for the time it would take to grab each frame of video and buffer it and do somthing meaningfull with the result.

Now my maths is terrible, but at 30 frames per second, I rekon you have around 33.3ms to process each frame so it's quite possible to do this real-time... Of course, it would depend on video resolution, how long it takes to "grab" a frame etc.
You may be better using a language like C/C++ and even running this in DOS - that way you have no Windows overhead which can be significant.
But again, this depends on what hardware you decide to use and video capture driver availability etc.

In theory tho', it seems possible.

The question I have is "how" do you represent this data.
You could, for example draw a cube and then plot in 3D a snail trail within the cube showing the fly's movements as two cameras mounted at 90deg to each other would give you a full 3D image.

Using Boris's flash idea may help make this quite simple I think, by helping to eliminate "background" noise from the data.

Facinating topic this one...
Joe
Do one thing each day that scares you – work here !
Top of pagePrevious messageNext messageBottom of page Link to this message

Eagre
Frequent Contributor
Username: Eagre

Post Number: 71
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Friday, 30 June, 2006 - 01:46 am:   Edit Post Delete Post Print Post

Joe -

Your previous post on flies responding to ultrasound is interesting. I had previously considered bombarding flies with high energy ultrasound at frequencies resonant with (parts of) their bodies. The idea of being able to explode flies in mid air was very attractive. Unfortunately, the energy densities required would likely be harmful to higher organisms (e.g. people).

Ed Grens
Top of pagePrevious messageNext messageBottom of page Link to this message

Boris
Frequent Contributor
Username: Boris

Post Number: 150
Registered: 05-2005


Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 10:56 am:   Edit Post Delete Post Print Post

"the energy densities required would likely be harmful to higher organisms"

Not if you focus the sound on the right place in 3D space.

Maybe 2 or 3 ultrasonic transmitters directed at the fly so as to interfere with each other at the 'fly resonance' frequency.

Maybe…
www.pyxia.com - BASIC was never less basic.
Top of pagePrevious messageNext messageBottom of page Link to this message

Mike_b
Frequent Contributor
Username: Mike_b

Post Number: 224
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 02:19 pm:   Edit Post Delete Post Print Post

Nice bit of work Boris and Joe.

If Scott (or anybody) wants to take it further I think its worth getting the CMU 2 unit from Total Robots for about £100.

It takes care of the basic frame grabbing, contrast correction etc and allows you to add your target processing code - it forms a good basis to solve machine vision problems.

rgds

mb
Top of pagePrevious messageNext messageBottom of page Link to this message

Mike_b
Frequent Contributor
Username: Mike_b

Post Number: 225
Registered: 05-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 02:27 pm:   Edit Post Delete Post Print Post

Ed,

Exploding insects - what a great idea (oh how I would have loved to see those clouds of mosquitos explode in the air when I was working overseas).

Perhaps the moving wing of a mosquito might modulate the return pulse from a 40Khz ultra sonic beam - thereby providing some something to track.

The only trouble is when the insect decides to take a rest.

rgds

mb
Top of pagePrevious messageNext messageBottom of page Link to this message

Thomas
Frequent Contributor
Username: Thomas

Post Number: 110
Registered: 04-2005

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 07:32 pm:   Edit Post Delete Post Print Post

A little off-topic. I once experimented with two ultrasonic transmitters with a view to transmitting highly directional audio on two ultrasonic beams. However, it didn't work -- I achieved a range of two metres at best. Thomas.
Top of pagePrevious messageNext messageBottom of page Link to this message

Ant
Valued Member
Username: Ant

Post Number: 19
Registered: 05-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 08:34 pm:   Edit Post Delete Post Print Post

Hello Thomas et al,

Somewhere in the darkest recesses of my circuit collection are vague details of some very good loudspeakers which were focussed to quite a definate extent, as I recall. Phillips? They were based on line source speaker columns, which in turn are based on column speakers but the sound is projected under some control by virtue of having differently sized drive units.

It might be much easier to do nowadays with computer-aided wossname.

The frightening possibility occurs of youths with even louder car stereos, focussed outside so's not to deafen themselves further...

Regards Ant
Top of pagePrevious messageNext messageBottom of page Link to this message

Ant
Valued Member
Username: Ant

Post Number: 20
Registered: 05-2006

Rating: N/A
Votes: 0 (Vote!)

Posted on Saturday, 01 July, 2006 - 08:39 pm:   Edit Post Delete Post Print Post

Hello all,

Going back to a slightly earlier point, it seems to me that if Boris were to use an over-rated flashgun the effect would be to burn out the background but the flies would then show up more clearly as black shapes, silhouettes really.

Regards Ant

Administration Administration Log Out Log Out   Previous Page Previous Page Next Page Next Page