Completely Autonomous, what to use?

[yelk11]

I was just playing around with the idea of a completely autonomous robot, and I was wondering what methods of tracking where the robot is relative to the field, and to the other robots were out there, and what sensors would be necessary?

To track where it is relative to the field you could use sonar sensors and bounce off the sides of the arena, however robots could get in the way and give you a false reading.

You could use the camera to track the targets (assuming there are targets) and use the size and orientation to calculate your position on the field. But that will take a lot of computing power away from your c-RIO.

Any thoughts on how to accomplish full autonomous and what to use?

Thanks in advance,

[EricH]

I seem to remember some halfway serious (or fully serious) discussion a couple years back. http://www.chiefdelphi.com/forums/showthread.php?t=84797

The camera tracking targets could work, if you had 2 pointing in opposite directions, and had at least one onboard coprocessor. But that's more work than you want to do, particularly right about now. Trust me on this one.


IMO, as a mechanical guy: Cameras for tracking targets (if done right, can give range and angle to target), ultrasonics for collision avoidance, a few beam-break sensors for figuring out if you have a game object and can score it, limit switches, pots, and encoders, along with an IMU-type sensor to figure out if you're tracking straight, would ALL be needed. Don't forget something to do some of the processing before the cRIO.

And don't forget to try this in the offseason. This is not an easy task.

[Pault]

If I needed to know my position on the field then I would probably use a 3-axis accelerometer and gyro with vision tracking, sonar sensors, and running into field elements to occasionally account for any error. But you don't necessarily need to know your exact position at all times. For example, in 2012 you would probably just use an axis camera to track the target when it is time to shoot, and a kinect to find balls and pick them up.

But honestly, the human brain is probably way more complex than anything one of us could code in our entire lifetime, let alone 6 weeks.