In-Air Un-Instrumented Pointing Performance

I present an analysis of in-air un-instrumented pointing and selection. I look at the performance of these systems and how this performance can be improved, with the eventual goal that their throughput reaches that of the mouse. Many potential limiting factors were explored, such as latency, selection reliability, and elbow stabilization. I found that the un-instrumented in-air pointing as currently implemented performed significantly worse, at less than 75% of mouse throughput. Yet, my research shows that this value can potentially reach mouse-like levels with lower system latencies, user training, and potentially improved finger tracking. Even without these improvements, the large range of applications for un-instrumented 3D hand tracking makes this technology still an attractive option for user interfaces.