the robot’s path is blocked would be a possible solution to this irritation. We discuss this further
in section 5.2 which discusses version 2.0.
As explained in section 4.4, we overlaid a cross-hair on the video screen to keep the user aware
of where the camera is pointed so they do not inadvertently drive with the camera off center.
Driving with the camera off center can put the robot, environment, or victims in danger. It has
been observed in another study that users have spent 10% of their runs driving unknowingly
with the camera off center [Yanco, Drury 2004]. We did not want to enforce automatic camera
re-centering in our design because we anticipated that some users would want to scan sideways
while driving. In this round of testing, with the cross-hair overlaid, none of our subjects
mistakenly drove with the camera off center. One subject, however, chose to drive with the
camera off center in order to scan along a wall while the robot drove forward.
Yanco and Drury [2004] also found that due to the lack of SA behind the robot, rear hits
accounted for 41% of the total collisions. We theorized that adding a rear camera would
improve SA behind the robot, leading to fewer collisions. In this study, only 1 hit (4%) was
recorded while backing up. There are at least two reasons for this. First, the rear camera allows
the user to see obstacles behind the robot. Second, with ADR, the need for backing up is
essentially eliminated because the user can just swap camera views and drive straight out.
Every user made use of the ADR mode. Sometimes, however, to get out of a tight area a user
would either turn the robot around or back out using the rear-view display instead. One subject
commented that the lights weren't very bright in the back. To support ADR mode, we have
tried to make the front and back of the robot act identically; however, at the time of this test,
there were lights only on the front of the robot. It is possible that users turned in place because
it was too dark to see while backing up and so the front lights were needed. Despite this, the
users still made frequent use of ADR mode.
This study was also a precursor to Michael Baker’s thesis work on the suggestion system [Baker
2006]. An explicit goal of the suggestion system is to teach the novice user about the interface
and remind the experienced user. We felt this goal was accomplished in this study because every
subject remarked that suggestions helped them in some way. One subject who did not explicitly
35
Kommentare zu diesen Handbüchern