Would it be legal to mount the Kinect instead of an Axis Camera on the robots? The Kinect has a much higher quality camera and my team’s Axis Camera (Its the older of the Models) has a tendency to fail and not work with certain C-Rios.
I’m pretty sure with a bit of work It could be done. Just a matter of “Is it Legal”.
Is it “legal”? Yes, 100% legal.
Is it going to be better than the axis camera? No.
I highly suggest using the axis camera instead of the kinect on your robot. If your current one doesn’t work, then get a new one.
Y is the axis camera so much better?
Because it is a proven concept in FIRST. The axis camera easily integrates with the FRC software. To connect the Kinect there are many custom things you have to do (i.e. power supply, transferring image data,etc)
PS: There are plenty of forums on it already
The Kinect may seem better than the axis camera and in fact may be a good replacement if you are short on cash. You will lack the stability and reliance with a Kinect, on a good Axis Camera you will not only have a sturdy base, but you very well may have a ‘fool proof’ way of interfacing with it should you decide to use it with the operation of the robot. If you look at some of the other Kinect posts, there are ways to get decent frame rates off the cameras, but using it directly with the CRIO is usually going to be harder than with an IP or USB camera.
It sounds like the kinect is better then the axis cAmera skill wise. Axis camera is just easier to set up.
I don’t know how proven the use of a camera is in FIRST. The one year I saw a camera work for a game was 2006 and that was because there was a single illuminated target. The issue with cameras, even when you use it with your own light source is you take certain elements for granted. Most practice fields are built out of wood, however the FIRST field is plexy and alluminum, meaning that when you get onto the field you will see reflections you never saw in testing. Further more if you are shooting from the key, the effect of ambient lighting on your calibration will be a further issue.
Designing a vision system that solely works with the cRio and the WPI ported libraries that shoots from the key, is a high risk endevour. However I am just relaying my experiences from last year so here as some tips if you move forward. Get your camera and light source as high as you can, you should avoid getting any reflected light off of the aluminum on the player station wall. WPI created an LED ring light that went around the camera lens and was green, I think this is a good idea.
Regarding the kinect, I think you should probably plan on 4, 5, or maybe 6 weeks of trying to get it set up (Hook it up to the robot, hook a laptop to the robot, get kinect library working (openkinect or ms library), get library exporting depth image to open cv, open cv processing the image, transferring useful commands from the laptop to the crio. And once that is all done you will have to figure out how to use the data you sent and test it with the robot
I hope this helps, and good luck!
So we are allowed to have a laptop on the robot?
of course not. there’s a crio which you can hook the kinect up to or something
WELCOME TO CHIEF DELPHI!
There’s a discussion of this topic going on at http://www.chiefdelphi.com/forums/showthread.php?t=99275