| description |
The Nexus framework anticipates the influence of the general
computational development towards the so-called state of Ubiquitous
Computing, where not only most electronic devices but even those
objects which today have no electric supply will be aware of their
position and context and be able to communicate among each other.
This will be a kind of a by-product of the ongoing miniaturization
of electronic components, growing performance of processors and
memory and at the same time dropping prices. By using this
communication capability and the collected spatial information, the
Nexus framework will provide a platform for multiple spatial aware
services for mobile people and self organizing applications. For
making mobile objects spatial aware, numerous sensors are needed,
which can be placed on the object itself as well as in the
stationary environment. In addition, simulation of physical mobility
helps to bridge the lack of spatial sensor data or in a first phase
generates virtual sensor data for test purposes for Nexus components
and applications. The simulation and modeling research in our
subproject focuses on two aspects of user mobility: the mobility of
pedestrians and the mobility of motor-driven vehicles. As for our
proposal to the SIIV Congress, we want to place emphasis on aspects
of microscopic simulation of pedestrian mobility. As it is one of
the most important objectives of the Nexus project, our main concern
in the field of pedestrian simulation is the scalability of our
modeling methods to large areas, e.g. to large (pedestrian)
networks. Therefore we mainly focus on two dimensional cellular
automata. By this way, we can reduce long range effects in a complex
simulation environment to a relatively small calculation rule set.
We are integrating several, partly already well known approaches to
pedestrian simulation based on cellular automata into a
comprehensive simulation platform. This allows us to reproduce
phenomena like lane formation, crossing behavior, oscillations at
doors and even panic situations. We are testing our simulation
platform on a rather complex geometric environment (the campus of
the University of Stuttgart), which comprises indoor, as well as
outdoor situations. Because of the need of more precise geometric
data, we are also developing an interface for the exchange between
CAD and GIS formats and the spatial world model format of Nexus. Up
to now, there are unfortunately no satisfying ways of evaluating the
quality of microscopic pedestrian simulations. The most frequently
used method is a heuristic comparison between the visualization of
the simulation and the reality. For that reason, a comprehensive
microscopic census of pedestrian mobility within the campus area,
comprising video analysis and detailed counts will be carried out in
the first half of 2005. As for the SIIV Congress, we will present
first results of the evaluation through real pedestrian mobility
data of our microscopic pedestrian simulations based on cellular
automata.
|