Abstract
Warfighter safety can be significantly increased by offloading critical reconnaissance and surveillance missions to robotic assets. The subtleties of these tasks require significant operator involvement--usually carried out locally to the robot's deployment. Human soldiers use gestures to communicate movements and commands when engaged in this type of task. While considerable work has been done with robots visually observing humans to interpret their gestures, we propose a simpler, more field-appropriate system that allows robot operators to use their natural movements and gestures (via inertial measurement units [IMUs]) to teleoperate a robot while reducing the physical, as well as the cognitive, load on the soldier.
This paper describes an operator control interface implemented on a smartphone, in contrast to the proprietary robot controllers typically used. The controller utilizes the device's IMUs, or attitude sensors, to bypass the touchscreen while accepting user input via gestures; this addresses a primary concern for gloved users in dirty environments where touchscreens lack reliability. We also propose that it provides a less visually-intense alternative for control, freeing up the soldier's cognitive load toward other functions.
We present details of the attitude-based control software, as well as the design heuristics resulting from its iterative build-test-rebuild development. Additionally, results from a set of user studies are presented, showing that as a controller, this technique performs as well, or better than, other screen-based control systems, even when ignoring its advantages to gloved users. Twenty-five users were recruited to assess usability of these attitude-aware controls, testing their suitability for both driving and camera manipulation tasks. Participants drove a small tracked robot on an indoor course using the attitude-aware controller and a virtual [touchscreen] joystick, while metrics regarding performance, mental workload, and user satisfaction were collected. Results indicate that the tilt controller is preferred by 64% of users and performs equally as well, if not better to the alternative, on most performance metrics. These results support the development of a smartphone-based control option for military robotics, with a focus on more physical, attitude-based input methods that overcome deficiencies of current touch-based systems, namely lack of physical feedback, high attention demands, and unreliability in field environments.
