Interface Evaluation for Mobile Robot Teleoperation
Robert Olivares, Chen Zhou, Julie A. Adams, and Bobby Bodenheimer
Abstract
Controlling mobile robots through teleoperation is a challenging task
that demands a flexible and efficient user interface. Mobile robots
are often equipped with numerous sensors (proximity sensors, system
status sensors, positioning and heading devices, multiple cameras,
etc.) that provide a high volume of data to the user. Because the
amount of data is vastly larger than what can fit on the screen, and
because the needed subset of data can change rapidly and unpredictably
depending on events in the robot s environment, modern teleoperation
interfaces often display user-selected data with a windowing paradigm
that facilitates quick display modification. In this paper, we
examine the possibility that too much fine-grained control over window
positioning and sizing could hamper user performance by interfering
with display modification. To test our hypothesis, three
human-computer interfaces for a mobile robot were designed and then
evaluated through performance studies consisting of 12 expert and 24
novice participants. The first two interface designs followed the
standard Microsoft Windows graphical user interface design paradigm
and provided participants with fine-grained control over the position
and sizing of sensor displays. The third interface was designed using
principles from cockpit and human-factors research, and provided
participants with limited control over display position and
sizing. User performance for each interface was assessed on a display
reconfiguration task consisting of adding, removing, positioning, and
resizing sensor data windows, a common task in robot navigation and
situational assessment. In general, expert participants preferred the
interface with limited control over position and sizing, while novice
participants pre-ferred the more traditional windows, icons, menus,
pointing device interfaces. No single interface accurately captured
all user preferences, but the feature-specific results found provide
direction for future designs.
Bobby Bodenheimer 
Last modified: Wed Jun 25 14:19:25 CDT 2003