Analysis of a Gesture-based Interface for UAV Flight Path Generation

Abstract

Traditionally, flight paths for unmanned aerial vehicles (UAVs) are generated offline by trained pilots and engineers using assumed environmental conditions, terrain and obstacles as constraints. As new applications for UAVs emerge, their user base shifts from one of operators with knowledge of low level systems to that of non-experts. These new operators require a more intuitive method for building desired UAV flight paths such that they can leverage the full capabilities of the vehicle without needing to understand its system complexities. We present a gesture-based natural language interface for defining trajectory segments using a library of twelve simple hand gestures. A user study is presented to analyze the effectiveness, ease-of-use and accuracy of the gesture-based interface as compared to a baseline mouse interface. We explore differences seen between subjects given their hand dominance, their prior UAV flight experience (or lack thereof), and whether they chose to sit or stand while using the gesture interface. Given limited training time, subjects were able to accurately define an average of 74.36% of trajectory segments. Overall the user study highlights the favorable potential for the use of the gesture-based interface as an alternative input modality, as well as, feedback for future interface improvements and training methods.