Spring RTS Engine

Hey, guys. I have a little question here.

I'm competing on a robotics challenge, and need some help. The problem is, how can I find out the exact direction and distance of a object on camera in relation to the robot itself? btw, the cameras have a offset, they are on the sides, 4pcs. Pic included.
I did some research in the wiki about... uh... homography but it's looking quite complicated.
http://en.wikipedia.org/wiki/Homography
This baby is going to play football.
http://www.robotex.ee/eng
P.S. It's running on .net

Columbus wrote:it's looking quite complicated.

You're doing friggin robotics, of course its complicated.

if you know the dimensions of the object already- ie, a football- it shouldn't be hard at all

I need accuracy, deviations caused by the lighting can cause the program to miscalculate the distance this way.
The idea is to create a simple 2D map of the surroundings of the robot, so we can decide, which ball is the best target. We are also featuring a dome mirror on top, to get a close-range overview of the surroundings.

Use a radar.

hur hur
Never mind, I got it working.

Columbus wrote:I need accuracy, deviations caused by the lighting can cause the program to miscalculate the distance this way.
The idea is to create a simple 2D map of the surroundings of the robot, so we can decide, which ball is the best target. We are also featuring a dome mirror on top, to get a close-range overview of the surroundings.

I find it remarkable that people consider this a fatal error when humans regularly make these sorts of mistake on an hourly basis.

The same is true of object recognition, we regularly mistake A for B

AF wrote:The same is true of object recognition, we regularly mistake A for B

Or at the very least 8 for B... I hate captchas.

The difference is that humans can immediately switch gears using logic and reasoning which immediately makes us better than machines at it.

If a machine makes a mistake, it will take the machine longer to realize and recover from that mistake.

Forboding Angel wrote:The difference is that humans can immediately switch gears using logic and reasoning which immediately makes us better than machines at it.

If a machine makes a mistake, it will take the machine longer to realize and recover from that mistake.

Hi there.

hit it! hit it with a stick!

Arm vs. Core. You decide which is which.

CarRepairer wrote:Arm vs. Core. You decide which is which.

However you choose, core wins.

Core:


arm:

Forboding Angel wrote:The difference is that humans can immediately switch gears using logic and reasoning which immediately makes us better than machines at it.

If a machine makes a mistake, it will take the machine longer to realize and recover from that mistake.

Not true, sometimes humans can perpetuate these mistakes for years at a time, there are even entire fields of research dedicated to exploiting these things, aka optical illusions, camouflage, magic tricks, make up

That's before we come across simple misunderstandings and mis-recognizing things due to incomplete or misleading information.

Eitherway your point is utterly destroyed by teh fact that your using a lack of feature A to justify perfection of feature B. The obvious solution here is not perfecting B to an unreasonable accuracy, but implementing basic feature A.

When building a tower, when ti gets too tall, do you research and build antigravity thrusters to hold it up? Or do you widen the base and use stronger materials?

AF wrote: Not true, sometimes humans can perpetuate these mistakes for years at a time, there are even entire fields of research dedicated to exploiting these things, aka optical illusions, camouflage, magic tricks, make up

All of these things you mention are things that humans can ignore what their eyes tell them and function regardless.

So wait, your arguments are that machines are better at reasoning and logic? That's a pretty weak position.