View Full Version : Weapon Accuracy?
scitobor 617
11-02-2006, 19:42
It's getting close to ship date and I'm sure that some very lucky programmers out there have already had some time to test out automated firing systems. I won't get to test until at least Wednesday and at the moment I have not even written a line of code for shooting(its a long story), yes I am getting a little nervous. I've been wondering what level of accuracy teams have been able to achieve? Also, what major problems have my fellow programmers been running into when approaching this interesting physics exercise?
Andrew Blair
11-02-2006, 20:04
Well, the old revision of the shooter was quite accurate, but we've made a serious revision, and I have no idea now...
Ours the camera sits seperate form the cannon so I have to triangulate!!
This is a programming nightmare!!Nope I don't even have nightmares this bad.
On paper everything works well but on the real thing all he$$ breaks loose.
Edit: And its not even done yet!!!!!!!!!!!!!!!I have no clue if the crap I am going to write now is going to work!
X-Istence
11-02-2006, 21:58
I am in the same boat as you fella's. No robot is done yet, or not far enough along for me to test the code. In theory it should all work, but theory is not good enough.
Our shooter is very accurate though, we just need to find a way to get it up to speed faster so we can shoot ten balls in 7 seconds during autonomous (accounting 3 for speeding up the shooter).
Jared Russell
11-02-2006, 22:02
Flywheel + flywheel speed feedback loop + backspin + camera = near perfection
CronosPrime1
11-02-2006, 22:36
0/10, seeing how my team doesn't have a shooter.
Code is very close to done, but the robot to test it on is not done. Logically everything works... how well it works depends on the time I'll actually get to test it...
Alan Anderson
11-02-2006, 23:29
...what major problems have my fellow programmers been running into when approaching this interesting physics exercise?
The programming team beat our collective heads against the compiler most of last week before finding the spot where the PID code was stubbornly using 8-bit arithmetic and overflowing. Today, while finally tuning the parameters, I found myself fighting the Victor deadband and the asymmetric response of the window motor.
The biggest hurdle to overcome is the simple fact that the robot isn't finished yet, so programming it is largely an exercise in faith that it's being wired the way we expect it to be.
Right now I have some code written and a few ideas. But its hard to know how well it will work without anything to test it on.
Oumonkey
12-02-2006, 10:09
I'm in the same boats as most of you, our robot isn't finished either. I hoping by mid this week. But I wanted to mount the camera on the shooter so in auto mode it actually moves the shooter instead of that little pan/tilt thingy it came with. But I know that servos and motors act differently so can anyone point me to something that could help, I have looked a little but I haven't had time to concentrate on that to much as of yet, and the little I have looked at didn't help me to much. I'm still doing some research on it though.
Thanks.
you will need a sensor(preferably a pot) to monitor the position of the cannon and then you have to set it up to be basicly controlled like a servo but with smoother motion because you don't want to damage the mechanisms. Basicly lower the power to the motor as the error gets smaller
error = where you want to be - where you are
SoftwareBug2.0
12-02-2006, 17:54
you will need a sensor(preferably a pot) to monitor the position of the cannon and then you have to set it up to be basicly controlled like a servo but with smoother motion because you don't want to damage the mechanisms. Basicly lower the power to the motor as the error gets smaller
error = where you want to be - where you areStraight proportional control is by no means the only possibility.
Straight proportional control is by no means the only possibility.
Well the power i give to motor is error/3 (based on a 0-254 error scale) + the minimum power required to get the thing moving. It slows it down as the error becomes less so it can settle. Its straight but I think it works. Its one part of the PID control? Still don't understand the other 2.
And ofcourse you can do more complex things...
SoftwareBug2.0
12-02-2006, 20:05
Well the power i give to motor is error/3 (based on a 0-254 error scale) + the minimum power required to get the thing moving. It slows it down as the error becomes less so it can settle. Its straight but I think it works. Its one part of the PID control? Still don't understand the other 2.
And ofcourse you can do more complex things...That sounds like a good way to do it. I sorta skipped over where you were qualifying with "basically".
Now I'm go to respond to your last message in a way that isn't quite literally there, but seems to be implied: it seems like you imply that other methods must be more complex.
There are actually simpler ways that you could potentially control the aim. For example, my team's manual aim mode works with only left, right, and off for the turning, and up, down, and off for the tilting. You could have your camera do the same thing.
And yes, proportional is the P in PID. The other are integral and differential, see a number of other threads for more information about them.
That sounds like a good way to do it. I sorta skipped over where you were qualifying with "basically".
Now I'm go to respond to your last message in a way that isn't quite literally there, but seems to be implied: it seems like you imply that other methods must be more complex.
There are actually simpler ways that you could potentially control the aim. For example, my team's manual aim mode works with only left, right, and off for the turning, and up, down, and off for the tilting. You could have your camera do the same thing.
And yes, proportional is the P in PID. The other are integral and differential, see a number of other threads for more information about them.
I have tried to understand the other ones a little bit but need to spend more time on it. I also don't have the time to impliment them.
Well your setup should work fine, but once you start having great error that constantly changes direction and your turret has to go back and fourth, it could create very jerky movements. Depending on how robust your system is, this could damage the motors and gears.
You might be smooth on the controls but the camera won't be(depends on the programming).
You are right they don't have to be complex.
nuggetsyl
12-02-2006, 22:16
i think you will see the elite teams hit 10 out of ten in auto and almost every shot after that. I strongly feel to win nationals you need to robots in auto to hit all ten balls if not 3 robots hit all ten balls to win it all. IF you win auto I think you pretty much have the game wraped up.
shaun
vBulletin® v3.6.4, Copyright ©2000-2017, Jelsoft Enterprises Ltd.