On Alan's computer (Dell Precision 690), in C:\KinectKeyframeStudy, there is what should be a working demo. It is also located on tanglefoot in the public folder. To run it, make sure AutoHotkey is installed (and the MS kinect driver at least), then run Demo.ahk. And pray.

Connecting the wii remote is also useful of course, and several online guides show how to do this, although here is a video to document the process also.

Youtube video for connecting the wii remote and starting the demo

Studying the .ahk scripts will give insight into how to use the command line arguments, including some that aren't documented very well (sorry).

To get the source, clone via git: ssh://tanglefoot.cs.byu.edu:/home/public/git/TAI ( or it might be http://tanglefoot… ) You must have a user account on tanglefoot to do this. You need to have Qt built for the Visual Studio 2010 compiler ( >= 4.8 ), Microsoft Kinect SDK ( >= 1.5), Windows DDK, and maybe something else we forgot installed. There are several environment variables you must set. You can set these either in the Windows system “Environment Variables” area, or on a per-project basis in Qt. I recommend just setting system environment variables for less headache, although Alan (painstakingly) set the paths in the Qt projects.

The current state of the project is somewhat of a mess. Tyler and Tim were in the process of modularizing everything so each piece was a nice encapsulated component, but they both graduated and left before it was fully finished. As such, some things are in component form, and others are not.

There is some information about what each project needs in each “.pro” file in each subfolder within the TAI root folder. The critical projects to get Alan's version working are:

• GLInterface-Alan
• Kinect
• qWiimote
• RobotPose
• TroyDriver
• qextserialport

Much of the code has portions to switch between the MS Kinect SDK and OpenNI, but we really moved to the MS SDK and have not updated the newer code to ensure it works with OpenNI. At this writing, the MS SDK is much more reliable, accurate, and less painful than OpenNI, so at least get it working with that first, and then you can switch to OpenNI if necessary.

Along the right are some screenshots that show the relevant environment variables and values. Keep in mind that several of these are user-dependent paths, so you will need to adjust them to reflect your own system.

Many of the rest of the projects in the folder are the componentized versions of all the pieces of the interface, but they will be missing lots of functionality compared to the monolithic GLInterface.

There is also an expression branch to use a new concept virtual robot, but it is missing some functionality and is very rough code.

The RobotOperationFlow project is the visual programming/sequencing tool for mapping different actions to buttons on the Wii remote. This is what clinicians would use for designing a therapy. I don't know the current state of the tool or what it takes to get it working, although it should be pretty similar to the rest of the tools.