USB cRIO Module for Kinect

[soxfan269]

Our team is considering purchasing a NI module for our cRIO II that gives it a USB port. This would give us the capability to plug the Kinect directly into the cRIO without having to rewire or change anything. The module cost $320 and we we're wondering if it's worth paying that to have the Kinect on it. So I pose two questions. 1) Is having the Kinect as a sensor a big advantage. 2) Is it worth $320 to make it easy to connect the Kinect.:confused:

[Andrew Lawrence]

Our team is considering purchasing a NI module for our cRIO II that gives it a USB port. This would give us the capability to plug the Kinect directly into the cRIO without having to rewire or change anything. The module cost $320 and we we're wondering if it's worth paying that to have the Kinect on it. So I pose two questions. 1) Is having the Kinect as a sensor a big advantage. 2) Is it worth $320 to make it easy to connect the Kinect.:confused:

1) IMO, It's not. While it has its uses, it's not only heavy, but easily replaceable by lighter sensors that perform the same way it does.

2) I wouldn't pay that just to have a large collection of sensors I already have, but that's just me.

[AllenGregoryIV]

You would need to reprogram the FPGA for this to work and as far as I know this is not allowed. You would also need to write your own library for the Kinect. It would be far easier to use a separate computer on board feeding the Kinect information to the cRIO.

As far as the usefulness of the sensor, I don't think having it on the robot is worth the effort at all.

[kinganu123]

You still need to power it

[Alan Anderson]

So I pose two questions. 1) Is having the Kinect as a sensor a big advantage. 2) Is it worth $320 to make it easy to connect the Kinect.:confused:

I'll answer the question you should have asked first.

0) Is it even possible to do this?

The answer is no. The FPGA image required for FRC does not support the USB module.

[mwtidd]

1) IMO, It's not. While it has its uses, it's not only heavy, but easily replaceable by lighter sensors that perform the same way it does.

2) I wouldn't pay that just to have a large collection of sensors I already have, but that's just me.

I would disagree with these statements. There is one key difference between the kinect and any other sensor. The ability to reliably see the backboard with little to no calibration. By taking the intersection of the depth and color images you can easily see the red or blue rectangle that surrounds the backboard.

It is my belief that this will be the most consistent and least vulnerable way of auto aiming. Also ball locating and collecting will be more reliable if done based on depth & color rather than just color.

[shuhao]

First of all, just purchase a netbook with that price.

As to the comment of seeing the backboard with the kinect. Its Max range is only 6 m, and I've been playing with it and found it to be very inaccurate with far range...

[mwtidd]

First of all, just purchase a netbook with that price.

As to the comment of seeing the backboard with the kinect. Its Max range is only 6 m, and I've been playing with it and found it to be very inaccurate with far range...

You could reverse triangulate based on the 4" barrier and then bridges. If you could triangulate based on both of those positions and had a turret I believe you would be able to do an assisted aim from nearly everywhere on your side, except along the player station wall.

[Greg McKaskle]

The working range of the Kinect with the MS drivers is 4ft to 11ft. And Alan is correct, this module will allow a physical connection to keep the USB cable from dragging on the carpet, but it will not do much more than that.

Greg McKaskle

[shuhao]

You would need to image recognize the wall or the bridge +bump.

The bridges are also not a completely still object