Geoscience Reference
In-Depth Information
The ground floor standard deviations are generally larger due to a network with
higher complexity and connectivity. This trend can also be detected in the classifi-
cation of the paths and their respective increase in length by choosing a less risky
road. 80 % of the longest paths (compared to the shortest path) with an increase
of 50 % or more are found on the ground floor, while half of the paths on the first
floor are equal to their respective shortest path.
4.1.2 Analysis of Selected Paths
In this section, the authors dilate upon an example shortest and least risk path, visu-
alized in Fig.
5
, to examine whether the least risk path calculations actually result
in the selection of less risky routes compared to the shortest path calculations. As
shown in the example in Fig.
5
, there is a significant visual difference in path choice
of the example route with both the starting and the end point located on the ground
floor of the building. In this example, the least risk path is 43 % longer than its short-
est path equivalent, which minimizes its total length. This example shows a 'worst-
case scenario' as it has one of the biggest differences in total path length of the entire
dataset. While the shortest path takes the direct route following main corridors, the
least risk path avoids certain areas to (theoretically) prevent wayfinders from getting
lost as easily. However, from this figure alone, it is not entirely visible why the least
risk path deviates from the shortest path in favour of using its calculated route.
In Vanclooster et al. (
2013
), several benchmark parameters were identified
which objectively quantify the risk of getting lost based on research of wayfind-
ing literature (both in indoor and outdoor space). These parameters can be used to
understand whether the theoretically calculated least risk paths are selecting edges
that actually reduce the navigational complexity and as such lower the risk of get-
ting lost. Table
3
enumerates on the parameters used in the algorithm itself (first
3 lines) and on the selected benchmark parameters. The values show a lower total
risk value for the least risk path with a considerable lower risk value at the indi-
vidual decision points, by choosing a longer route. This is in line with the original
definition of the algorithm. The other parameters, however, show a different side
of the coin, with better results for the shortest path algorithm in terms of reducing
the risk of getting lost. For example, the shortest path has 7 turns in its description,
while the least risk path requires 12 turns. Wayfinding experiments have exten-
sively shown that more turns on a certain path considerably increase the risk of
disorientation making users more inclined to take wrong decisions at decision
points. The chosen corridors in the least risk path algorithm are also generally less
integrated, with less visibility towards the next decision points (4.68 vs. 5.17) and
a higher route complexity (more decision nodes passed on the total route, more
curves and more spatial units passed).
Above results indicate a less comfortable (and much longer!) route travers-
ing for unfamiliar users compared to the shortest path. It can be concluded that
the least risk path algorithm performs worse in terms of choosing less risky
edges which completely undermines the initial intentions of the algorithm.
