Environment Modeling in SVG and X3D
We have developed a systematic approach to create the usabilitybased
environment model. We started from a layer-based CAD
model, and converted the model to an intermediate SVG model.
This SVG model allowed us to scan it and convert it into a
discrete model consisting of spatial cells. The cells are then
assigned usability properties according to their design element
types. The discrete space model is also in SVG format, which has
proven to be an effective data structure for environment models
and behavior simulation.
Environment Geometry Modeling
To make the case study simple without losing generality, we used
the popular layer-based DXF format to build the CAD model ofUtilizing CAD models makes our simulation practical and
eventually applicable to everyday design practice, which is the
ultimate objective of this research. Our approach, however, is not
tied to any particular CAD software. Rather, it is designed as a
reusable tool for working with traditional graphic models and
Building Information Models. To apply the simulation to design
practice, a designer needs minimal extra effort to follow a
naming convention for the components of CAD models, e.g.
layers in layer-based models, or objects in object-oriented
models, in which doors, walls, windows, etc. are presented as
layers or objects, respectively. By naming the design components
in an easy-to-follow convention and marking the components’ zbuffer
Environment Usability Modeling
We first convert the CAD model into an intermediate model in
SVG format, and then built a discrete spatial model, which
became our usability model.
To model usability information, we applied SVG, which is a
language for describing two-dimensional graphics in XML (W3C
Recommendation). Using a graphics formatter, a SVG document
can be presented in a graphical way, e.g. a design drawing
(Figure 4). The following advantages of SVG make it very useful
for creating the usability-based model for our simulation:
SVG is semantically rich: the graphical information is always
associated with meaningful textual information. It allows
representation of both graphical and non-graphical data in an
open, published XML schema. Thus a rectangle represents not
only a shape, but also the meaning, which can be a door, a
User Modeling in SVG and VRML (H-Anim)
The Virtual User model includes three major components: (1)
geometric modeling and motion control, (2) cognitive modeling,
and (3) behavioral modeling.
Geometric Modeling and Motion Control of
Virtual Users
Geometric modeling includes 2D and 3D modeling. 2D modeling
is used for behavioral simulation and 3D modeling is used for
behavioral visualization.
The Virtual User’s 2D model is a fairly abstract symbol used for
user model design and checking purposes in the simulation phase.
The Virtual User utilizes SVG format for the same purpose of
presenting both geometrical and non-geometrical information.
As shown in Figure 8, its graphical view is made of a filled circle
and a short line indicating the facing direction of a Virtual User.
Cognitive Modeling of Virtual Users
Cognitive modeling defines the users’ ability to access and
interpret the environment model. It is the combination of four
components: (1) “seeing,” i.e. accessing the relevant parts of the
environment model within a circular area in front of a user in
real-time, and translating them into terms that correspond to the
virtual user’s cognitive model, for such purposes as avoiding
collisions and recognizing an acquaintance or an object; (2)
“knowing” the entire environment in advance to help make basic
decision of what to do and how to behave, much like a frequent
visitor has knowledge about the location and orientation of
benches, so they can seek them out in order to sit on one; (3)
“finding” paths using A* algorithm, which is widely used for
searching the shortest path in games. We optimized A* in our
simulation to reduce the search space from the total number of
cells to a subset of cells within a rectangle with a user’s starting
and target points as corners; and (4) “counting” the duration of a
specific behavior, such as sitting, to make a decision about what
to do next: continue sitting or walk away.
Behavioral Modeling of Virtual Users
Behavioral modeling is the most critical issue underlying the
simulation because it must mimic closely how humans behave in
phase. Public distance is not affecting users’ movement because
other persons’ present can be seen only peripherally in this
distance
Sample scenarios
Combining all the four components: Artificial Life algorithms,
social spaces, environmental effects, and randomization, we built
a user model to simulate individual and group behaviors. The
implementation details will be discussed in Section 4.
The following two scenarios are intended to test how the
behavior simulation works. They reveal that many behavior
patterns can be simulated. For testing purposes, we used only two
Virtual Users: Bob and Nancy. The implementation is achieved
with VRML and Java through External Authoring Interface
(EAI).
Simulation Results
We integrated the environment model and the user model into our
behavior simulation in the run time through a simulation engine.
The simulation engine first loads the environment model and
parses the model’s graphical and usability properties, then creates
a Virtual User group—a list that allows an unlimited number of
user models to be added into, and upon completion of a journey
removed from the list. The engine runs the simulation step by
step, and at each time step (one second) it adds users from the
entrances and moves all the users by one step. The Virtual Users
acquire environmental knowledge through the cognitive
processes (knowing, seeing, finding, and counting) so that the
users know, for example, where they can walk and where they
can sit. Then the engine lets the Virtual Users move following
behavior rules, e.g. shortest path, group movement rules, and
social spaces. The simulation engine uses Batik SVG toolkit with
Java2D rendering engine, and Document Object Model (DOM)
to traverse the design element tree of the environment model [see
Conclusions and Future Work
Our simulation has been considered by A/E/C experts a
significant topic dealing with evaluation of human spatial
behavior that can be made visible before construction. It raises
hopes that future modeling of users in designed environments
may be less crude than it is today. We expect, with our
environmental behavior simulation, human behavior analysis as
one of the most important aspects in environmental design can be
integrated into designers’ daily design practices seamlessly. The
evaluation of human spatial behavior can be made easier and
visible before the environment is built. This will encourage
designers to pay more attention to users and therefore innovative
environments concerning more about the needs of people can be
designed and built. We also expect in A/E/C industry, XML and
Web-based graphics technologies, because of their flexibility and
extensibility, will greatly facilitate the representation of design
information such as geometries, usability, materials or activities
such as design and construction.
Langgan:
Catat Ulasan (Atom)
Tiada ulasan:
Catat Ulasan