Geoscience Reference
In-Depth Information
(a)
(b)
FIGURE 9.2
(a) The step from the real-world to PARKAGENT and (b) schematic representation of the
major PARKAGENT objects - streets, parking places, destinations (buildings) and parking lots (not included
in the minimal version of the model). (From Levy, N. et al.,
Transportmetrica
, 9(9), 773, 2012.)
automatically generated when the street network is constructed (Kresse et al., 2012). These relation-
ships are necessary to simulate driving towards the destination. The relationship between street
segments and parking places (Figure 9.3a and b) is necessary for establishing a driver agent's view
of the parking situation and simulating the driver's ability to see if a parking place on the left or right
side of a street segment is occupied or free. A standard GIS linear referencing procedure (Brinkhoff
and Kresse, 2012) is employed for constructing this relationship in Figure 9.3b. In a similar way, the
relationship between parking places and driver agents represents which parking places are occupied
and where each driver parks (Figure 9.3c). In the full version of the PARKAGENT model, this rela-
tionship is employed for estimating a driver's fine for illegal parking.
Two more relationships are exploited in the PARKAGENT model for representing location infor-
mation. To represent driving towards a destination, destination buildings are linearly referenced in
relation to the street network links (Figure 9.3d), while the closest parking places are related to the
destination (Figure 9.3e). Note that the destination can have several entrances (dark grey triangles in
Figure 9.3a) and should be related, in this case, to several parking places (Figure 9.3e).
Four entities - street segment, parking place, building (destinations) and car and the relationships
between them - are expressed by the entity-relationship diagram in Figure 9.4.
Note that the relationships between street segments and between parking places and destination
buildings, and the relationship between parking places and destination entrances, reflect the infra-
structure of the street network and of the parking space and are static. When driving to a destina-
tion, driver agents will retrieve parking places along the segments and estimate when they should
park or when they can still drive closer to the destination. In contrast, the relationship between
parking places and drivers is dynamic, and the modeller has to update the relationship table every
time a driver occupies or leaves any parking place.
9.2.3 r
uleS
of
a
gent
B
ehaViour
Formally, the rules of agent behaviour describe how the properties and relationships of agents
change over time, including changes to the locations of the agents. The behavioural rules are the
most important part of every AB model, which enable the emergence of properties, patterns and
structures to form at a level above the individual. At the same time, the behavioural rules are inher-
ently model specific. Modellers appear to agree easily on agents, objects and relationships but differ
