Kinect or no kinect?

[Joe Ross]

Originally Posted by Daniel_LaFleur

Do we know how much processing power we'll have at the Kinnect station? or will it be the teams OC running the Kinnect? (In which case, the best processor/OC should be running the Kinnect station always)

Per the kickoff video, the kinect gets plugged into the team's computer.

[Greg McKaskle]

The Kinect Station is a display/feedback device. The Kinect is not plugged into it, and team code doesn't run on it. The Kinect Station is relayed the same Kinect Server data as the robot, and it draws the standard skeleton so that the driver knows if the Kinect sees their hands, feet, and other body parts.

The Kinect USB cable is plugged into the OC laptop and the Kinect Server on that computer does the work. Other OC computers have a Kinect Server running on them too, but without a Kinect plugged in, they will mostly snooze.

This configuration will also allow teams to know ahead of time how the Kinect will respond when you get to the field -- it is your code, your computer, and your robot, you are just borrowing the field Kinect and using the display for driver feedback.

Greg McKaskle

[Wetzel]

We've decided to push it to our would be cool but only if we get our more important stuff done first. With little experience with our programmers, we are leery of trying lots of new things during build.

Wetzel

[Martian]

just wondring does anyone have any working hybrid code that tests the kinect.

[RufflesRidge]

Originally Posted by Martian

just wondring does anyone have any working hybrid code that tests the kinect.

There are examples built into the updates for C++ and Java and in a ZIP file right next to the Kinect Server download for LabVIEW. I've been having the students on the teams I work with start with these to help understand the Kinect APIs provided by WPILib.

[Zuelu562]

Currently I'm actually developing a true Hybrid mode; it will do some tasks autonomously, but it will get to a point where it's waiting for the human to make the movement that will start the shooter.

Pros:

1. something cool to demonstrate in the postseason and even during build (to demonstrate to potential sponsors for example)
2. relatively easy to do (in C++, i've already figured out how everything works without a drivable robot)

Cons:

1. If you have experience and a lot of time and/or sensors, true auto is probably simpler to achieve.

You can setup a wait time to prevent multiple balls entering the same hoop, but it's easier to make sure there aren't 6 balls trying to get into the top hoop in the first 3 seconds with the kinect.