Information Technology Reference
In-Depth Information
Adaptive Decision Making in Microsimulations
of Urban Traffic in Virtual Environments
Fabian Krueger
1
, Sven Seele
1
, Rainer Herpers
1,2,3
,
Peter Becker
1
, and Christian Bauckhage
4
1
Institute of Visual Computing, Bonn-Rhein-Sieg
University of Applied Sciences, 53757 Sankt Augustin, Germany
sven.seele@h-brs.de
2
University of New Brunswick, Fredericton, E3B 5A3, Canada
3
York University, Toronto, M3J 1P3, Canada
4
University of Bonn, 53115 Bonn, Germany
Abstract.
To improve the plausibility of driving and interaction as well as the
perceived realism of agents in interactive media, we extend cognitive traffic
agents based on personality profiles with emotions. As proof of concept a sce-
nario with a narrowing road was evaluated. To enable agents to handle these
scenarios, an existing lane change model was adapted to model the required de-
cision processes and incorporate the driving style defined by static and dynamic
aspects of the agents.
Keywords:
adaptive agents, serious games.
1
Introduction
The credibility of NPCs in current entertainment software, called agents in the follow-
ing, often suffers from unrealistic and incomprehensible behavior. Especially in
serious games, where individual agents are often closely observed by a player, dis-
played behavior must be plausible to enhance immersion. By adding personality pro-
files to agents, observed decisions become more consistent and may increase realism.
However, in deterministic systems, the profiles cause identical reactions in identical
situations. Such predictable behavior may become implausible if the agent does not
adapt to its surroundings or to the player. This contribution will outline how psycho-
logical profiles of agents are extended with a model of emotion to dynamically adapt
their behavior. The model is applied to a specific road traffic scenario demonstrating
the benefits of such an approach.
2
Modeling Adaptive Decisions
Our approach for adaptive decision making is based on two elements: (1) a static
foundation, based on the Five Factor Model (FFM) as presented in [1], to achieve
consistent behavior patterns for individual agents and (2) a dynamic model to allow
