Targetting

[Clyd3]

Hello people
i have time, money and a brain, so I was wondering, anyone got tips on making this:

project made in both electronics and pc software / interface stuff

i want a sentry
it has a camera, whether its a webcam or whatever,
it detects movement, a significant change betwn. two frames software stuff... yeah

it decides, considering the area in the picture frame where the movement is, and sends the co ords down to the sentry to fire a blast a water in that direction

couldnt really decide to put this in projects or here, but decided here, 'cause it is a frikkin robot, i think...

whether you like it or not, i'm making it, this summer (southern hemisphere human)

got any suggestions or anything, i'm willing to discuss...

 

[Someone Electro]

You certanly need an DSP do do it fast enugh.

Also a huge problem is interfacing the camera.The USB or analong video will be very hard to interface.There is an camera on the maket that has an microcontroler frendly interface.

You cod also use an PC wich cod simplyfy things.Then you need an webcam that can be intrafeced trough USB no problem and an small MCU driving the motors.Also you have procesing power to spare.

 

[Clyd3]

the trick with the camera is my plan is to not have that integrated with electronics
preferably just a simple webcam to pc, got some info, i'll post shortly...
o wait, that's what you said
long time since i spoke to you eh?
anyway, got suggestions for co ordinates for firing???
maybe i'll use 2 servos or steppers???

 

[Someone Electro]

Stereps wod be beter becose servos are not as precise.

You will need to find out the corordinates on the picture and then translate that in to how many steps has the sevo make.This will probobly requre fine tuning to get it acurate.Also you need an valve of some sort to relese the water.Washing machines have such valves. So if you meaby rewind its coil to make it work at 12V or so,insted of mains.

You cod probobly use the printer port to do it.You just have to conect the transistors to control the stepers and the valve.

Do you have any programing experiance?

 

[Clyd3]

i have done a bit of nigel's pic stuff,
i've also done basics in most of the common languages or can if i haven't already

now that i know a bit about how to use pics, i'd like to bring in pics somehow
i think i'm sorted for the valve, but what are stereps???
i would like a less tiresome method, so they sound interesting...
i'll google, but a short explanation from you targetted at the project would help

what path do you think i should take, what languages and interfaces
do you think i should go the pic route or just use transistors and the parallel port???

finding out the co ordinates from a picture will require some pretty interesting programming, guess i'll just make them up from now, and focus on the electronics side, it is the more rewarding side anyway - but dont think i dont want to do the software too
it'll be fun to make a co ordinate system, maybe i should have like 1 point per pixel, like 640 by 480 to match a common camera resolution
i'm interested in these stereps, more accurate??? interesting... tell me more

 

[bloody-orc]

may i ask, how big is the area you are waching? i think you might be able to do it without the camera at all. just by using 2 long range distance meters (like sonars). then in software, you create a triangle and you get your cordinates very accurately. with only camera, it is impossible cause camera does not have 3D vision, so cant see the distance. the sensors must be in the corners of the area so that a triangle can be made. the sensors are also on servos that move them back and forth all the time so they cover the hole area. you can also add lasers to add a bit of woww factor to it.

 

[Roboticinfo]

Stereps wod be beter becose servos are not as precise.

You will need to find out the corordinates on the picture and then translate that in to how many steps has the sevo make.This will probobly requre fine tuning to get it acurate.Also you need an valve of some sort to relese the water.Washing machines have such valves. So if you meaby rewind its coil to make it work at 12V or so,insted of mains.

You cod probobly use the printer port to do it.You just have to conect the transistors to control the stepers and the valve.

Do you have any programing experiance?

Wrong berni servos are more accurate the steepers are
you forgot that steepers slip there for you can never be accurate.

Go with servos

 

[Clyd3]

can i get fairly powerful servos, the only servos i've come across are in r/c cars and stuff
the range isnt a problem, so 3d vision isnt that important, because the targets are limited to where they can be...
i would like to eventually have a generating 3d engine that has more than 1 camera monitoring the different views of the field, but that's not important
i am not sure if sonics or long range distance meters are cheap enough for the job, how much could i get one for???

 

[bloody-orc]

sonars cost about 15£. you can get servos in most electronics shops. just search the nearest one you have.there are 3kg servos (it can lift 3 kg on a 1 cm rod) and also 6 kg and maybe even more powerful. i recon that those are more than enaugh

 

[Someone Electro]

But threy are also more complicated to control from an LPT port.

Twisting an tube wont requre much power and there is not much force to make them slide.

You can control an steper just whith an feq. transistor on the printer port.This makes it more simple an therefore easyer to build and also less stuff can go wrong.

 

[ivancho]

If you want to use a PIC then I propose you do the following.

Obtain the coordinates with the PC + WebCAM. Send RS232 (serial) commands and coordinates to the PIC from the PC. Use the PIC to control a servo motor (This will simplofy your electronics quite a bit since the servo motor already has the controlling circuitry, something that you would have to build if you go the stepper route). Use a sprinker system valve and fountain pumps that you can find easily at stores.

The hardest part would be to code the WebCAM motion sensor program. There are some good articles on how to create a program that recognizes motion on the past 3-4 issues of the magaizne Nuts and Volts. But if you to test your electronics, you can use visual basic to control the serial port and send RS232 commands to your PIC circuitry.

 

[Someone Electro]

Its not that complicated to detect motion.You just store the last frame and then comare evry 10th pixel.If the pixel has changed above the limit then you can start aiming at that pixel.

This cod be done in visual basic.Its a bit slow but for this its enugh.

 

[hjl4]

roboticinfo,

Steppers will slip when they are overloaded or will miss steps if too much speed is asked of them.
Servos, are great but at the added expense.

If your system does'nt require much speed and your not moving great mass, I would go with steppers, as they are cheap and easy to setup.

It seems everybody has given up on the "STEPPER", but yet it is the most accurate system around in terms of positioning, cost wise.
Good Luck.

 

[Nigel Goodwin]

It seems everybody has given up on the "STEPPER", but yet it is the most accurate system around in terms of positioning, cost wise.
Good Luck.

But steppers have a number of 'problems', relatively low power and their high cost are two, but the biggest problem (for this sort of application) is that they only give relative positioning. A servo gives absolute positioning, so may be the better solution here?.

With a stepper you have no idea where it happens to be at any particular moment, you can tell it to move x steps clockwise, but there's no way of knowing if it actually happened (or even where it was in the first place). Adding a sensing system to do this adds greatly to the complexity of the system - anyone remember the old Commodore 64 disk drives?, they used a stepper to position the head - in order to know where it was they slammed the head repeatedly against a stop, then counted the steps from there.

With a servo you can tell it to go to a specific location, so there's no need to know where it was, or to check where it ends up - it's all done internally in the servo. Obviously if something stops it moving it can't get to where it wants, but it won't just 'give up' and get out of sync, it will continually try to move where you told it.

 

[hjl4]

Point well taken Nigel,

I don't know if this is a hobby project, or commercial, but if it were, a hobby, you could make your own encoder. In which you now have a closed loop system.

The low torque of steppers is inherent, therefore my suggestion to oversize the motor.

In a servo system, if the absolute is not met ie: false position,or failed position, the whole system fails, and all you are left with is a warning signal, showing the failure.

Maybe I'm stupid, but is this now possible to rectify via software.

Oh well , it's been a while.

 

[ChriX]

Have you seen this?

**broken link removed**

Pretty much exactly what you are trying to do, except with a BB gun instead of water. May offer a starting point.

 

[Nigel Goodwin]

Pretty cool! 8)

 

[Clyd3]

thanks for that

DAMMIT!!!!!!
plagiarizing copying little punks

i want those on every corner of my property

yeah, i'm just about to start work on this project again

so far i've almost finished the motion detection part in python
just finalizing how i'm sending the co ordinates feed
thats really the only connection of the robot the computer other than the webcam

 

[Oznog]

The stepper need not be forced against a stop, but against a limit switch where the controller can stop driving the rotation before it forces it. Actually a positioning pot sensor is still possible, though they can be hard to fit and not very long life (but then this project is unlikely to run for months or years on its own).

Actually the only problem that comes to mind is the elevation issue. A water stream is fairly slow thus drops more with distance, but the video sensor doesn't give you distance unless you implement stereo vision. One could look for motion low in the camera or a smaller distance between the highest and lowest moving fields and assume the distance is further. Even without elevation compensation, I'm sure it will be impressive whether it hits your face or feet.

 

[Someone Electro]

Just try to launch the water out at the highest velocety posible.The faster it goes the les elevation it will lose on the same distance.

Meaby an air compresor preshurising an water tank and an vale to let the water in the nozle.

Just make shurse evrything is tight becose if a hose comes off becose of the preshure you are gona have water evryware.