Geoscience Reference

In-Depth Information

comparing its results.
Section 5
discusses multiple improvements to be made to the

original algorithm to be more compatible with the specificities of indoor environ-

ments. This chapter is completed with a conclusion on the discussed issues.

2 Least Risk Path Algorithm

The ultimate goal of cognitive algorithms is to lower the cognitive load during

the wayfinding experience. Various cognitive studies have indicated that humans

during navigation value the form and complexity of route instructions equally as

much as the total path length (Duckham and Kulik
2003
). This is the reason why

several algorithms have been developed for outdoor space with the purpose of

simplifying the navigation task for unfamiliar users. In this chapter, the least risk

algorithm forms our focal point as it is implemented in a three-dimensional indoor

environment (Grum
2005
). More specifically, we want to investigate whether the

results of the least risk path algorithm have the same connotation and importance

in indoor spaces as in outdoor space where it was originally developed. Also, the

least risk path algorithm is analysed for its applicability in providing route instruc-

tions that are more adhering to the natural wayfinding behaviour of unfamiliar

users in indoor space.

The least risk path as described by Grum (
2005
) calculates the path between

two points where a wayfinder has the least risk of getting lost along the path, by

selecting all edges and intersections with a minimal risk value. The risk of getting

lost is measured at every intersection with the cost of the risk calculated as the cost

for taking the wrong decision at the intersection. This algorithm assumes (1) that

the person taking the path is unfamiliar with its environment (and as such local

landmarks). Also, (2) when taking a wrong path segment, the wayfinder notices

this immediately and turns back at the next intersection (Grum
2005
). While the

algorithm assumes that an unfamiliar user immediately notices a wrong choice,

Grum (
2005
) also acknowledges that the algorithm needs to be tested for its repre-

sentativeness of the actual behaviour of users (Fig.
1
).

The formula for the calculation of the risk value at a certain intersection and the

total risk of an entire path p is as follows:

Total
_
Risk
(
p
) =
risk
_
values
(
i
) +
lengths

(1)

Risk
_
Value
(
i
) =
2
∗
length
_
wrong
_
choices

possible
_
choices

(2)

Equation
2
indicates that the risk value is dependent on the number of street seg-

ments converging on the intersection, combined with twice the length of each

individual segment (as it assumes the user will return through the same edge

when going in the wrong direction). The risk value of an intersection increases

with more extensive intersections and with many long edges that could be taken