pic: 358 has four eyes! [Archive]

Matt Krass

27-01-2006, 00:32

Prototype gimbal.

Now we're waiting for Kevin to help us interface four cameras with only two serial ports.....any ideas?

Arkorobotics

27-01-2006, 00:43

Why in the world do you need 4?

Gdeaver

27-01-2006, 07:36

How can you justify the cost of 4 cameras and the additional electronics rule?

Tristan Lall

27-01-2006, 08:54

How can you justify the cost of 4 cameras and the additional electronics rule?4 × $199.95 (USD) doesn't violate the cost limits. They are COTS electronics, so they don't violate the material utilization rules either.

Erics626

27-01-2006, 09:12

Do I spot a joystick base being used for the pan and tilt? How did you manage to do that?

Peter Matteson

27-01-2006, 09:27

Do I spot a joystick base being used for the pan and tilt? How did you manage to do that?

Replace the potentiometers with servos. Just takes a couple of brackets and a basic shaft.

Sachiel7

27-01-2006, 10:30

Again, I ask (as others have) why the need for four cameras?
If you're having that much trouble centering in on a goal, I think you might need to adjust the focus...dunno....
If its for something else, such as the proposed "tracking balls" theories teams have been putting forth, I still think you could accomplish that with less....

Mark McLeod

27-01-2006, 11:25

Again, I ask (as others have) why the need for four cameras?

One for the goal and three to track opposing alliance robots of course.

Billfred

27-01-2006, 11:53

One for the goal and three to track opposing alliance robots of course.
If you can see Chuck Norris, Chuck Norris can see you. If you can't see Chuck Norris, you may be only seconds away from death.

Sounds like you're planning for protection against any robots that may channel the spirit of Chuck Norris. ;)

Tom Bottiglieri

27-01-2006, 11:58

Prototype gimbal.

Now we're waiting for Kevin to help us interface four cameras with only two serial ports.....any ideas?

It's really easy!!

Cam1 = new Camera();
Cam2= new Camera();
Cam3= new Camera();
Cam4= new Camera();

RC.MakeThread();

Thread.Run(Cam[x].TrackRobotsAndStuff()) ;

Shu Song

27-01-2006, 12:09

You're gonna have a very hard time tracking the other robots, because they don't have illuminated targets. You'll be trying to track them like last year's vision tetras.

Secondly, how are you actually using the joystick to control the camera? Are you attempting to move the turret and as the change in acceleration moves the camera on the joystick and reading the X and Y values on the joystick? The measurement you get is acceleration and if you want to position you gotta integrate that twice and introduce a lot of error in the progress. Having a potentiometer to measure absolute position is the best way to do it.

Cory

27-01-2006, 15:06

One for the goal and three to track opposing alliance robots of course.

...what happens when they aren't in front of you?

It's just a hunch, but Im going to go out on a limb and guess that human eyes will be much more effective at tracking robots than robots will :rolleyes:

Mark McLeod

27-01-2006, 15:22

It's just a hunch, but Im going to go out on a limb and guess that human eyes will be much more effective at tracking robots than robots will :rolleyes:
Have you seen our shooter video featuring our guide dog?
Not all driver's see equally well.:cool:

...what happens when they aren't in front of you?
"What is behind me is not important." Unless it's Chuck Norris in a bad mood of course.
(what movie is that line from?)

Maybe we're simple bird watching enthusiasts.;)

Greg Marra

27-01-2006, 15:40

The joystick -> camera mount is extraordinarly effective. We used it last year to great success.

I suppose you could mount an EDUBot controller and have it run two CMUCams and then pass the values back through analog in's on the big RC...

Still, 4 cameras seems a tad overkill.

Mike

27-01-2006, 16:13

Duh, you guys are idiots. The obvious reason for having multiple cameras is for Matt's proposed method of vision multiplexing by algorithmic implementations of backwards-compatible and proprietary standards. This method is extremely nimble, with its low risk and high yield. The cameras must work in such a way that they resemble a cluster team of multi-scalable processes. From my statistical analysis, I may ascertain that the 358 robot will cause a paradigm shift in the area of FIRST autonomy.

Duh.

geeknerd99

27-01-2006, 19:53

Uh oh. Looks like its mounted on a vison tetra panel. That means that it'll distract our cannons... Oh wait, it's a vision tetra panel. It's not like we can see those anyways...

Perhaps it's for triangulation of their location by calculating the angles to the target?

spamified88

27-01-2006, 21:27

Wow, you have 4?! We don't even have one yet!!

Andrew Blair

27-01-2006, 21:36

Obviously, the four cameras will all return their own view on the subject of the illuminated target, where, in the brain, the information is fed into a highly complex social algorithm that allows all four streams to fight simultaneously. So, by the end of the match, hopefully, one will have established dominance of the others, just in time for the buzzer. Sounds great!

devicenull

27-01-2006, 22:55

I would think you are going to need some sort of co-processor.. there's no way you can have the RC monitoring 4 cameras and still doing other stuff. Co-processor for the camera is the way to go.

Matt Krass

28-01-2006, 12:47

We're going to keep color profiles of the primary colors of robots at the events and preselect them to track opposing alliance robots, it's also good for tracking multiple people and gunning them down with Poof balls.

We're working on pattern recognition for black and white stripes.

coldabert

29-01-2006, 19:31

if we're ever in the same pit, please stop by and explain to me how you managed to use all of that data in 10 seconds

Chriszuma

31-01-2006, 15:28

Replace the potentiometers with servos. Just takes a couple of brackets and a basic shaft.
Dude! You totally stole my idea from last year! http://www.chiefdelphi.com/forums/showthread.php?t=33418

Peter Matteson

31-01-2006, 15:32

Dude! You totally stole my idea from last year! http://www.chiefdelphi.com/forums/showthread.php?t=33418

Yeah we saw that and did the same thing because it was a nice simple solution. :D The difference was we got rid of the plastic we didn't need to lighten it up.

Chriszuma

31-01-2006, 15:34

And it seems like the 4 cameras might just be an extremely inefficient method of attaining hardware fault protection. In this way 3 cameras can be annihilated with poof balls before the robot is blind. Of course a simple protective cage with a window would work equally well, but if you're under budget why not go for the gusto?

Chriszuma

31-01-2006, 15:35

Yeah we saw that and did the same thing because it was a nice simple solution. :D The difference was we got rid of the plastic we didn't need to lighten it up.
Yeah, I was going to just rip out the actual movement mechanism in the joystick, but we ended up not even using the camera (not enough time to code + impossibility of using our robot to cap without human intervention)

Revolverx7

01-02-2006, 15:03

Auto turret targeting. I keep telling them i dont want to have to aim this thing i just wanna drive around pullin 360's so the camera's get to target and do all the work for me.

i wish.

Chriszuma

01-02-2006, 15:24

Auto turret targeting. I keep telling them i dont want to have to aim this thing i just wanna drive around pullin 360's so the camera's get to target and do all the work for me.

i wish.
That doesn't explain why you need 4. Turret tracking can be effectively accomplished with a single camera.

Rick TYler

01-02-2006, 15:39

"What is behind me is not important."

Well, Mark, I guess we are both pretty old and remember obscure movies. It's from Gumball Rally.

Revolverx7

02-02-2006, 15:31

That doesn't explain why you need 4. Turret tracking can be effectively accomplished with a single camera.

One facing each side of the robot.

Dillon Compton

02-02-2006, 18:34

Or, this COULD be a picture of 4 cameras mounted to their 2004 robot designed to keep you all busy talking about it instead of working on your own vision systems...

just sayin...

Chriszuma

02-02-2006, 21:50

One facing each side of the robot.
That's still overkill because you could cover 360 degrees with 2 cameras, which is absurd anyway.

Or, this COULD be a picture of 4 cameras mounted to their 2004 robot designed to keep you all busy talking about it instead of working on your own vision systems...

just sayin...
Yeah, but I am argumentative.

Mark McLeod

03-02-2006, 11:25

Or, this COULD be a picture of 4 cameras mounted to their 2004 robot designed to keep you all busy talking about it instead of working on your own vision systems...
Good eyes on the 2004 part. Kudos.

Well, Mark, I guess we are both pretty old and remember obscure movies. It's from Gumball Rally.
Give that man a camera!
Obsure but classic.:cool: Sort of like us old guys I suppose...

Revolverx7

03-02-2006, 11:25

Or, this COULD be a picture of 4 cameras mounted to their 2004 robot designed to keep you all busy talking about it instead of working on your own vision systems...

just sayin...

That is very true, but why test 4 at a time if we're not gonna use 'em all...

As to overkill, we could use one camera to track each opposing robot and one for the center goa.