TESTING PROTOTYPES
Once you have a prototype, the next likely thing to do with it is to test it with potential users. It is particularly important with gestural interfaces that you find representative potential users on whom to test the prototype, because a gesture that works fine with an office of technically savvy, educated, and healthy people may not work well when exposed to the harsh light of the public or a specific user group. Many research firms will help you find research subjects if needed.
Because the use of sensors and the data from them can change based on environment, if you are creating a free-form gestural interface, it makes sense to test your product in the actual environment (or in a set of potential environments) to make sure that the sensors and any other hardware, such as projectors, are calibrated and set up correctly and that the data coming in from the sensors is triggering the system correctly.
You'll want to put together a test plan—a written path through your product to test—and a moderator script—what to say when guiding people through the plan—that make sure your gestures are discoverable ("What would you do if you wanted to do X?"), meaningful ("Would you ever do that?"), useful ("Are you able to do what you wanted to do?"), and usable ("Can you do that action?").
Several test subjects(typically less than 10) will usually quickly reveal the flaws of your product. If possible, you should leave yourself time between testing sessions to recalibrate sensors ...