Using Kinect to Design Your Robot

[JesseK]

Forget driving a robot with the Kinect: go why not design your robot using multi-touch style Kinect libraries [arstechnica] and a projector? There is some interesting discussion in this youtube video's comments.

What would it take to use Kinect to manipulate any CAD software? What are the advantages? What missing features of a typical mouse/keyboard setup would need to be addressed?

Obviously there isn't a keyboard for shortcuts, but perhaps we could turn off the music for a minute to input voice commands for those things.

I wouldn't recommend modifying this season's build process to design a robot with Kinect (at least not yet), but perhaps it'd make a cool demonstration piece in the pits.

[Andrew Schreiber]

Originally Posted by JesseK

Forget driving a robot with the Kinect: go why not design your robot using multi-touch style Kinect libraries [arstechnica] and a projector? There is some interesting discussion in this youtube video's comments.

What would it take to use Kinect to manipulate any CAD software? What are the advantages? What missing features of a typical mouse/keyboard setup would need to be addressed?

Obviously there isn't a keyboard for shortcuts, but perhaps we could turn off the music for a minute to input voice commands for those things.

I wouldn't recommend modifying this season's build process to design a robot with Kinect (at least not yet), but perhaps it'd make a cool demonstration piece in the pits.

I've done some CAD work using the multitouch features of my computer. The primary thing missing from that is the fact that center clicking (on the scroll wheel) is impossible with the current drivers. I can right click fine though. The biggest obstacle I've found is the lack of a good way of entering numbers. Currently I use the pen to go to the left side of the screen and pull down the input panel in Windows then have it recognize my handwriting. It is effective but a little tedious (also 1, I, L, and i get mixed up pretty often).

The hardest problem to solve, for me at least, would be training. A keyboard and mouse is ubiquitous enough that nearly everyone knows how to use it. Any different interface system will have to be significantly more intuitive to overcome the initial hesitation of the users.

[davidthefat]

What is up with the flood of Kinect "hacks"? I mean its not new technology, there has been similar things in robotics for the longest time. I say KISS. Stick to the mouse and keyboard.

[Andrew Schreiber]

Originally Posted by davidthefat

What is up with the flood of Kinect "hacks"? I mean its not new technology, there has been similar things in robotics for the longest time. I say KISS. Stick to the mouse and keyboard.

You are correct of course, the technology behind the Kinect is nothing new. This is, however, one of the first times that technology like it has been available at low cost to tinkerers. Think of it as what happened when the Apple ][ hit. It wasn't the first one but it was the first one to be widely available.

[drakesword]

You got the thread wrong ... Should be: Design a Robot to use a Kinect.

hmm wonder what the g-force rating on these things are ...

[Andrew Schreiber]

I also want to share LadyAda's writeup on how to get started tinkering with USB devices. http://ladyada.net/learn/diykinect/ I found it quite an interesting read.

[JesseK]

Originally Posted by drakesword

You got the thread wrong ... Should be: Design a Robot to use a Kinect.

Yet then we'd have to begin all of the existential debates about whether or not the robot was beside itself...