True autonomous is better than a Kinect-driven "Hybrid" mode, but it is definitely not something to always dismiss, because it can (read: has a small-ish chance of us considering it to) be useful.
Take our current plan for using it, for instance:
The SmartDashboard accesses the Kinect. It takes a photo of them, and then uses the Kinect bone coordinate data to turn said photo into a paper doll of the person using the Kinect.
Then, it keeps updating the drawn parts of the paper doll to match the driver. The SmartDashboard then draws a sword in the driver's paper doll's hand, and starts shooting bombs at the driver's paper doll. The driver must chop the bombs, or he/she will be hit and blow up. The goal will be to last through all of auton.
This works well, as now we can do something while we wait for field issues to be resolved
