Page 32 - Innovations in Intelligent Machines

20 M.L. Cummings et al.
towards them. High workload areas, or “bottlenecks,” are highlighted through
a reverse shading technique while the rest of the colors are muted, but still
visible. In addition to identifying areas of high workload, the computer also
recommends a course of action to alleviate the high workload areas, such as
moving a particular Time on Target (TOT).
The super-active LOA (Fig. 4d) also builds upon the passive level visual
timeline, but instead of making recommendations to the operator as in the
active LOA, a management-by-exception approach is taken whereby the com-
puter automatically executes the arming and firing actions when the rules of
engagement for such actions are met, unless vetoed by the operator in less
than 30 seconds (LOA 6, Table 1).

Experiment Protocol. Training and testing of participants was conducted
on a four screen system called the multi-modal workstation (MMWS) [15],
originally designed by the Space and Naval Warfare (SPAWAR) Systems Cen-
ter. The workstation is powered by a Dell Optiplex GX280 with a Pentium
4 processor and an Appian Jeronimo Pro 4-Port graphics card. During test-
ing, all mouse clicks, both in time and location, were recorded by software.
In addition, screenshots of both simulation screens were taken approximately
every two minutes, all four UAV locations were recorded every 10 seconds,
and whenever a UAV’s status changed, the time and change made were noted
in the data file.
A total of 12 participants took part in this experiment, 10 men and 2
women, and they were recruited based on whether they had UAV, military
and/or pilot experience. The participant population consisted of a combina-
tion of students, both undergraduates and graduates, as well as those from the
local reserve officer training corps (ROTC) and active duty military person-
nel. All were paid $10/hour for their participation. In addition, a $50 incentive
prize was offered for the best performer in the experiment.
The age range of participants was 20–42 years with an average age of 26.3
years. Nine participants were members of the ROTC or active duty USAF
officers, including seven 2 nd Lieutenants, a Major and a Lieutenant Colonel.
While no participants had large-scale UAV experience, 9 participants had
piloting experience. The average number of flight hours among this group
was 120.
All participants received between 90 and 120 minutes of training until
they achieved a basic level of proficiency in monitoring the UAVs, redirecting
them as necessary, executing commands such as firing and arming of payload,
and responding to online instant messages. Following training, participants
tested on two consecutive 30 minute sessions, which represented low and high
workload scenarios. These were randomized and counter-balanced to prevent a
possible learning effect. The low replanning condition contained 7 replanning
events, while the high replanning condition contained 13. Each simulation was
run several times faster than real time so an entire strike could take place over
30 minutes (instead of several hours).