Trajectory Collection and Reconstruction - Mobility Data - page 30

use the coordinates, speed, and orientation of the current position in order to
calculate a safe area where the next position might be located. If the next
incoming position lies in the calculated safe area, it can be ignored. There are
two options for the definition of the safe area. It is either calculated by using
the last position, whether it has been previously ignored or not, or by using the
last chosen position. In order to achieve better results, a combination of the two
algorithms is also proposed. Both areas are calculated, but only their intersection
is defined as the safe area.
These trajectory compression approaches are primarily based on the exten-
sion of geometric methods such as the DP algorithm. However, they are not
suitable for network-constrained trajectories. Therefore, recent works proposed
another kind of trajectory compression model that makes use of the underlying
road network. Through map matching, trajectories can be reconstructed (or rep-
resented) by only the matched road segments, without the need for keeping the
original movement points.
2.4 Reconstructing Trajectories
Chapter 1 introduced the differentiation between raw and semantically enriched
trajectories. Here we present reconstruction techniques for both types. Trajectory
reconstruction refers to the task of transforming raw spatio-temporal positions
into meaningful trajectories. An interesting note here is that different applica-
tions may need different trajectories. For instance, there may be a considerable
difference between the semantic definitions of a trajectory given by a traffic ana-
lyst and, on the other hand, a logistics manager. Let us consider a fleet of trucks
moving in a city and delivering goods in various locations. The logistics manager
may consider, for each truck, a number of different trajectories (e.g., between
the different delivery points) while the traffic analyst may consider a single tra-
jectory for the whole day. Thus, in order to satisfy these two, quite different in
semantics, requirements we would have to retrieve raw spatio-temporal position
data from a common repository and then execute two different reconstruc-
tion tasks so as to produce trajectories that are semantically compliant to each
domain. For instance, Figure 2.4 a illustrates a raw data set of spatio-temporal
positions. Different needs may result in different set of reconstructed trajectories
(Figure 2.4 b-d, respectively). Recalling the previous example of the truck data
set, let us consider Figure 2.4 b and c, which illustrate the reconstructed trajecto-
ries for the logistics manager and for the traffic manager respectively. Another
example of trajectory reconstruction is presented in Figure 2.4 d, which considers
a compressed trajectory of the movement. The exact number of reconstructed
trajectories depends on the different semantic definitions that can be given to a
trajectory. In this section, we present reconstruction techniques that can be used
to produce either raw or semantically enriched trajectories.