Airsoft Sentry Turret: Powered by Puppy

Message

jyore · #1 Post by **jyore** » Tue 14 Dec 2010, 23:25

Hey guys,

I have been working on a design project for a class (Georgia Tech ECE 4180: Embedded Systems Design). I decided to power it with Puppy, and wanted to share the project with the Puppy Community.

The project was to design an airsoft sentry turret that would automatically fire at any object that moves in front of it. We used an eBox 3310 with Puppy Linux (v4.20) as its OS.

This project uses the OpenCV computer vision library for motion detection, it predicts where the object is moving, moves servos to a position, and then triggers the gun to fire at the object.

The servos are controlled using Phidgets Advanced Servo Boards, and a Phidgets Relay Board. If anyone is interested in a PET for the Phidgets drivers, I have one laying around here somewhere.

I am hosting the project site at:
http://jyore.com/4180

You can review the project, hardware, software (documentation and download available), and watch the video demonstration of the project (hosted on youtube).

Direct Link to YouTube video: Click Here

If anyone is interested in learning more about this project, or more about things that were done in this project, please feel free to message me, and I will answer as soon as I can.

As further promotion goes, my professor said he will suggest using Puppy to any student that's needing a Linux based system in the future!

Enjoy!

#2 Post by **Flash** » Sat 18 Dec 2010, 04:00

Very cool. Can you make it shoot down mosquitoes using the laser from a DVD burner?

If you wished for a lower latency Puppy, Puppy Studio was made for musicians. It has a real-time Linux kernel.

abushcrafter · #3 Post by **abushcrafter** » Sat 18 Dec 2010, 14:00

Flash wrote:Very cool. Can you make it shoot down mosquitoes using the laser from a DVD burner?

. Would that be possible?

This does look interesting though. Me thinks I know a few people/friends that might be very interested...

#4 Post by **Flash** » Sat 18 Dec 2010, 20:52

abushcrafter wrote:
Flash wrote:Very cool. Can you make it shoot down mosquitoes using the laser from a DVD burner?
. Would that be possible?

Indeed it is:
http://www.youtube.com/watch?v=eYXPqrXZ1eU
http://www.wired.com/epicenter/2010/02/ ... toes-dead/
Rocket Scientists Shoot Down Mosquitoes With Lasers

abushcrafter · #5 Post by **abushcrafter** » Sat 18 Dec 2010, 21:32

Flash wrote:
abushcrafter wrote:
Flash wrote:Very cool. Can you make it shoot down mosquitoes using the laser from a DVD burner?
. Would that be possible?
Indeed it is:
http://www.youtube.com/watch?v=eYXPqrXZ1eU
http://www.wired.com/epicenter/2010/02/ ... toes-dead/
Rocket Scientists Shoot Down Mosquitoes With Lasers

Great!

#6 Post by **ttuuxxx** » Sun 19 Dec 2010, 14:18

Blu-ray Laser Phaser is also pretty cool
http://www.youtube.com/watch?v=xfj1n8vP ... ature=fvwp
ttuuxxx

jyore · #7 Post by **jyore** » Sun 19 Dec 2010, 23:07

Hahaha, that would be great to convert it to a anti-air cannon for mosquitoes...I'll have to look into that. I think the hardest part would be determining the difference between a mosquito and any noise the camera sees.

That is good to know about the Real-Time Puppy version. Wish I knew about it a few months ago. I'll try to give it a shot when I get some free time. I will also let my professor know so that he can let students know a nice real-time linux kernel is available.

To deal with the latency and overall performance, we ended up threading the program's main functions (computer vision, servo movement, firing the gun), suppressing the video output (since we don't need to see the images, just the gun shooting things), killing X, and increasing the process priority. This cut our latency down pretty good and the program ran very smoothly.

#8 Post by **Flash** » Wed 22 Dec 2010, 16:30

Mosquitoes in flight make a pretty distinctive high-pitched whine. I'll bet it wouldn't take long to make a filter that could with good reliability distinguish the sound of a mosquito in flight from other common noises. Perhaps a hardware (active) filter would be easiest and best, since it would not add to the control program.

dogle · #9 Post by **dogle** » Sat 25 Dec 2010, 21:56

Flash is onto something - could be a great benefit to mankind, and me.

Why don't you hint to your prof that international FFFUNDING might be attracted by a Puppy mosquito-zapper project?

jyore · #10 Post by **jyore** » Fri 31 Dec 2010, 02:27

Haha, I'm sure he'd enjoy any extra funding he could get for the lab.

I wouldn't mind some funding for myself too tho

I'd like to try to figure out the tracking for the mosquito, but fortunately, I don't have them around the place...so unfortunately, it makes it hard to test, lol

To do it purely by computer vision, you'd have to figure out a way to pick up the mosquito and not any random noise...maybe solid white backgrounds? might need a zoom or something to make the lil guy bigger?

To do sound, you'd probably need a series of microphones so that you can filter out noises other than the mosquito, and then use some triangulation to locate the mosquito...kind of how the xbox kinect does with its microphones.

Would be so fun to put out on a porch and watch the mosquitoes fall tho

droope · #11 Post by **droope** » Sat 01 Jan 2011, 15:32

abushcrafter wrote:This does look interesting though. Me thinks I know a few people/friends that might be very interested...

The military?

Master_wrong · #12 Post by **Master_wrong** » Sun 02 Jan 2011, 03:12

To do sound, you'd probably need a series of microphones so that you can filter out noises other than the mosquito, and then use some triangulation to locate the mosquito...kind of how the xbox kinect does with its microphones.

you mean, we need 3-4 microphone so we can track horisontal and vertical location of the mosquito ?

#13 Post by **Flash** » Sun 02 Jan 2011, 04:02

I'm not sure sound would be the way to go for actually tracking the mosquito. It might be practical to use sound to notice that a mosquito is present, then the video tracking software would take over.
Here's my scheme: boresight a laser down the optical axis of a video camera, with a 45/ mirror the camera looks through, that is fully reflective at the laser wavelength and transparent at all other wavelengths, or at least in the range the camera can see. Then it should be relatively simple to use software to point the camera/boresighted laser so as to keep a mosquito in the center of the camera's field of view. When mosquito is in the center of the camera's field of view, give it a zap with the laser. Walla, one dead mosquito, pdq.

jyore · #14 Post by **jyore** » Tue 04 Jan 2011, 21:10

Might work...You can have issues when moving the camera if you don't have a solid background...The issue is, if the camera moves, anything with edges will look like something moving...so given a porch with say woods in the background and support columns, you may detect many objects. You can implement something called optical flow that can help solve that problem, but I am not too familiar with that, and it seems pretty complex from what I have seen of it...With a static camera, you can have any background and only get things that are moving with simple segmentation, which simplifies the problem in my opinion. This is why we used a static camera on the airsoft project.

To answer the question about multiple microphones...it is basically the same principle as all propagating waves, where having multiple sensors can allow you to determine position and distance (i.e. triangulation). Also note, from there, simple derivatives over time, naturally get you velocity and acceleration

.

I was thinking about using sound to just trigger the camera to find the mosquito as you suggested, but how do you know where to look? That is why I thought of using multiple mics to find a relative location to look in and find it with a camera with some zoom applied. But as I thought of that, I was wondering why would you even need the camera if you have a location from the microphones??

Really, what I should do is setup a simple application and wave something about the size of a mosquito and see how sensitive it needs to be in order to pick it up and eliminate noise. I'll work on this a bit throughout this week when I get some free time and post my findings.

abushcrafter · #15 Post by **abushcrafter** » Wed 05 Jan 2011, 23:12

droope wrote:
abushcrafter wrote:This does look interesting though. Me thinks I know a few people/friends that might be very interested...
The military?

. No some people from my sci lessons.

(old)Puppy Linux Discussion Forum

(old)Puppy Linux Discussion Forum

Airsoft Sentry Turret: Powered by Puppy

Airsoft Sentry Turret: Powered by Puppy