SpudFiles

From my experience.. it's always better to choose a larger calibre

POLAND_SPUD wrote:it's always better to choose a larger calibre

"better" in this case is subjective, 3mm is good enough for me and it's as big as I'm willing to go for the purposes of this project.

This is what I'm currently thinking of to promote reliable and consistent feeding.

Edit: prototype curing, testing tomorrow

Edit 2: prototype with untightened detent - ie relying on seal only - held a 3mm BB without "popping" at 500 psi, and sprang a leak. Leak repaired, testing at 850 psi due later today.

I know this is old thread but I know the OP is still active. I was curious about the targeting software. Does the software just track movement or the does it identify the human form? Also, if it does track forms, can it be changed to track other shapes?

I really should get down to finishing this thing

Does the software just track movement or the does it identify the human form?

The software analyses changes in the image the webcam is seeing. It basically creates a black/white image, where black is the unchanged background and white is the pixels which are changing, so basically a silhouette of the target is created. It is possible to store these shapes in such a manner that the turret will seek out and target them specifically, but any moving target is trackable.

You can also define minimum and maximum shape sizes, for example you can make it track a person walking by, but not a plane flying in the distance.

Finally I put together something that is reliable and light enough to be acceptable as the "gun" element for this project. The concept is one expressed earlier and looks something like this:



A previous prototype on this principle jammed up, on account of the tube I was using between the valve and the barrel being of 2mm I/D - exactly double that of my projectiles, which is asking for trouble. The one I'm using now is 1.5mm I/D which is perfect - small enough for the blow of BBs not to jam, but large enough to allow enough blowby to shoot past the projectiles in the tube.

It does jam up occasionally at the breech, because the odd BB doesn't fit in the barrel, but otherwise the cloud mechanism which feeds the tube performs flawlessly.

The beauty of this is that the heavy chamber, valve and associated trigger servo can be mounted on the stand, and the only thing the elevation and traverse servos have to strain with is the barrel (and maybe a laser sight )

Power is what can be expected of a 1mm BBMG, a single BB will not penetrate a CD, but will happily chew up cardboard smileys:

[youtube][/youtube]

Some detail:

Just thought I would post this in case it's worth anything to you guys: http://www.sparkfun.com/products/10335

Pretty cool pan/tilt bracket.

Edit: As pointed out by Tech, maybe not so cool afterall.

None of that fancy stuff for me, it's cable ties and Araldite all the way

[youtube][/youtube]

As you can see it's pretty ghetto, but works quite well the high velocity flow causes the POI to shift upwards but that can be compensated for. There's also a pretty wide spread of shot, almost the size of an A4 paper at 10 feet, a "shotgun" effect which should increase hit probability.

The laser sight is hooked up as dewey-1 recommended here, and it's all powered by a 1A 6V power supply.

All that's left now is to mount a camera and calibrate it

Seems like it might be worth making a more balanced and rigid frame for a newer version. (that ball valve mount )

I'd love to work on a similar project, probably not an autonomous tracking turret, but a system capable of estimating ranges and adjusting angle to hit a target designated on screen.

Looking good anyway.

al-xg wrote:Seems like it might be worth making a more balanced and rigid frame for a newer version. (that ball valve mount )

That ball valve certainly needed better securing - first auto targeting test, the thing came loose on the first shot and didn't close...

[youtube][/youtube]

Still, plenty of hits

I have one more question about the targeting software. Will it accept multiple video inputs? It would be nice to be able to create 360 degree capable auto turret.

Will it accept multiple video inputs? It would be nice to be able to create 360 degree capable auto turret.

Not that I know of.... but you could mount the camera on a separate pan/tilt mount separately controlled by another version of the program

...A4 paper at 10 feet, a "shotgun" effect which should increase hit probability.

At this point I have just one question, what exactly are you getting into here? Do you have a personal vendetta against squirrels? Is a normal home security system not enough? Or are you just going to bolt it to one of those robo-puppy toys and let it wander around town, blabberring GLaDOS quotes as it mows down curious children?

Fnord wrote:Do you have a personal vendetta against squirrels? Is a normal home security system not enough?

All legitimate uses but no specific use in mind for this one.

Fnord wrote:Or are you just going to bolt it to one of those robo-puppy toys and let it wander around town, blabberring GLaDOS quotes as it mows down curious children?

I thought about that but mount it on a moving platform and the software sees the whole image as movement and the turret goes insane. While the concept evokes a pleasant picture, sadly it's technically unfeasible with this technology :roll"

Fnord in his signature wrote:I wonder how much of this christian radio station's traffic originates from people making a typo.

I typed kxcd in the address bar on firefox and hit enter, it directed me to the comic

Maybe I should make this into a machine pistol type weapon and mount it on a heavy duty turret.

this software/hardware package available for a mere $150

hmm not to be a dick but now I realized that you should have gotten Arduino instead

I am not sure how much work would it require to use the software with arduino instead of the servo controller provided by the guy but at least you'd get something that isn't limited to just one purpose.

POLAND_SPUD wrote:hmm not to be a dick but now I realized that you should have gotten Arduino instead.

I see your point but If you compare the time it took me to earn that $150 to the time I would have spent trying to figure it out on my own, it's a bargain

I repaired the trigger and added (epoxied ) a permanent camera to the mount in the meantime, I hope to post some more testing soon. There's a lot of tweaking to do with calibration till you get it right, for example I realised it was set to a level so sensitive that it was detecting the laser dot as a separate target and trying to track it, talk about a cat chasing its own tail