PNNL-13501 - Pacific Northwest National Laboratory

and interactions. The flexibility of the environment
enables any user to add, delete, or modify objects in real
time, within the environment.
This dynamic architecture naturally supports the creation
of dynamic avatars. Each avatar consists of a number of
independent dynamic objects (limbs) that are connected at
user-defined points (joints). The objects communicate
their relative positions, configurations, etc., to each other,
so that, for instance, rotating the torso of the object
automatically rotates the legs and arms, but the legs can
be raised independently of each other. In this way, the
user can posture the avatar in real time to create unique
and distinctive mannerisms. As people are very good at
learning and identifying people by their behavioral
mannerisms, this will enable others to trust that the user is
who it appears to be.
Usability evaluations conducted midway through this
project indicated that the dynamic avatars had significant
potential as collaborative tools by facilitating nonverbal
communication. However, these studies also indicated
the serious deficits of the mouse-and-keyboard control
mechanism. The second year of this project, therefore,
was focused primarily on developing alternative
mechanisms for controlling position and motion in six
dimensions. Three approaches were studied: the
traditional mouse/keyboard combination, a six-degree-offreedom
controller referred to as the Spaceball, and the
PNNL-developed gesture-based controller known as the
Human-Information Workspace (HI-space). Pros and
cons of each of these interaction tools were identified, and
a potential solution described with a prototype developed.
Results and Accomplishments
Human-form avatars were created; each with 14
independently flexible joints, including the neck,
shoulders, elbows, wrists, waist, hips, knees, and ankles.
The joints can be constrained to meet natural human
limitations (the head cannot rotate down through the
chest, for instance) or left unconstrained, at the user’s
choice. Figure 1 illustrates the avatars’ posture and
motion flexibility.
The second phase of this project explored the utility of
various interaction mechanisms beyond the mouse and
keyboard, which were deemed inadequate in usability
testing. Initial exploration was done using a Spaceball.
The Spaceball consists of a sensor ball and an array of
buttons. Sensors inside the ball detect pressure: push,
pull, or twist, and use that information to move the
appropriate object in the virtual space. The buttons can
Figure 1. Multi-jointed avatars “dancing”
be mapped to task appropriate commands. For this effort,
the buttons were mapped to the various limbs of the
avatar: arms and legs. For example, to use the Spaceball
controller, the user pressed a button to select for the upper
arm and then pushed on the ball to contort the arm as
appropriate.
The Spaceball controller provides one valid mechanism
for controlling the avatars in the virtual world. However,
like the mouse, it is still inherently limited to moving a
single limb or object at a time. Further, the mapping of
limbs to buttons must be hard-coded and cannot be
dynamically adjusted. This lack of flexibility suggested
that continued research needed to be done to explore
alternative control mechanisms. Accordingly, the final
phase of this project explored incorporating into this
effort other ongoing research into next-generation
interfaces. This effort uses the HI-space to posture and
move the avatars. The prototype controller, shown in
Figure 2, uses a video camera mounted over a table-sized
surface to track the user’s gestures. Selection of a body
part to move is done by extending the forefinger and
Figure 2. The HI-space being used to control an avatar
Computer Science and Information Technology 165