You are on page 1of 40

Fuzzy and Logic based Route Choice

Fuzzy and Logic based Route Choice

Jorge Cristhian Chamby-Diaz

Institute of Informatics
Universidade Federal do Rio Grande do Sul

May 10, 2019

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 1/40
Fuzzy and Logic based Route Choice

Summary

1 Route Choice
Problem
Route Choice Modelling
Random utility models
Fuzzy route choice models

2 Fuzzy logic
Overview
Fuzzy Set Theory
Membership Function
3 Fuzzy and Logic Route Choice
Fuzzy Models
Related works
4 Conclusion
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 2/40
Fuzzy and Logic based Route Choice

Introduction

As mentioned early, route choice is similar to the Traffic Assignment


Problem (TAP) but focus on the drivers and their decisions. Algo-
rithms for the TAP are mostly centralized while route choice allows
for the use of decentralized algorithms.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 3/40
Fuzzy and Logic based Route Choice
Route Choice
Problem

Route Choice Problem


The route choice problem concerns how drivers behave when
choosing routes between their origins and destinations (OD pair,
henceforth).

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 4/40
Fuzzy and Logic based Route Choice
Route Choice
Problem

Route Choice Problem

Given:
a mono or multi-modal transportation network (nodes, links,
origin, destination)
an origin-destination par
link and path attributes
Identify the route that a traveler would select.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 5/40
Fuzzy and Logic based Route Choice
Route Choice
Problem

Theory of User Equilibrium

Travelers will select a route so as to minimize their personal travel


time between their origin and destination. User equilibrium (UE) is
said to exist when travelers at the individual level cannot unilaterally
improve their travel times by changing routes

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 6/40
Fuzzy and Logic based Route Choice
Route Choice
Problem

Route Choice Problem

When route cost are precise and certain, the route choice pro-
cess is very easy to apply: with costs of all possible paths
C1 < C2 · · · Cn then the driver chooses path 1 and all other
paths are rejected.
The assumption of perfect knowledge of travel costs has been
long considered inadequate for travel behaviour. Consequently,
probabilistic route choice models were developed in which drivers
were assumed to minimize their perceived costs given a set of
routes.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 7/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Route Choice Modelling

For explicit route choice modeling, starting from this theory and the
consideration that only a sub-set of all the possible routes (choice
set) is actually perceived by the users, a two-step pattern is consid-
ered in the literature:
modeling choice set perception
modeling route choice behavior

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 8/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Modelling choice set perception (I)

In the classical explicit approach for simulating the route choice


process, a route generation model with a selective criteria ap-
proach is specified, calibrated and validated in order to simulate
the perception of a route choice set of a generic user when a
large number of alternatives are available.
A selective approach identifies only some of the routes which
are topologically admissible on the basis of satisfying certain
rules (criterion).

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 9/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Modelling choice set perception (II)

The criteria have to be defined according to the factors that


affecting route choice behavior (e.g. minimum route length,
minimum route travel time, maximum motorway route, etc.).
In order to generate the minimum routes for the criterion it
is necessary to minimize trip disutility measured with a single
attribute (time, distance, number of traffic-light regulated in-
tersections, etc.) or a combination of them.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 10/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Modelling route choice behavior

Modeling route choice behavior consists in evaluating the prob-


ability of each alternative contained in the extracted sets.
The choice models most widely used in the literature belong to
the family of well-established (Quattrone and Vitetta, 2011):
Random Utility Models (RUMs)
Fuzzy Utility Models (FUMs)

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 11/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Random utility models (I)

Also named stochastic choice models, they represent cost as a


random variable.
Thus, different routes cannot be ordered classically, but proba-
bility that the corresponding cost might be the smallest can be
calculated.
Such a probability can be seen as a score given to each path
and representing its level of optimality: the higher the score,
the better the path.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 12/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Random utility models (II)

In the context of RUMs, a generic user n, traveling between each


(o, d) pair, assigns to each route k, belonging to choice set In , a
utility ULk which might be expressed as:

ULk = VLk + ϵk ∀k ∈ In
where:
VLk is the systematic utility of route k;
ϵk is the random residual, that generally includes user errors of
perception as well as the modeling approximation of the
analyst.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 13/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Random utility models (III)

The systematic utility VLk is calculated as a function of the at-


tributes Xi , obtained with the generic criterion c, through some βLi
coefficients (that must be calibrated) and parameter θ:

VLk = βLi∗ Xi with βLi∗ = βLi /θ
i=1

If the ϵk values are assumed to be (i.i.d.) independent and identical


Gumbel variates of zero mean and parameter θ, then the well-known
Multinomial Logit model (Dial, 1971) of route choice is obtained.

Pn (k, In ) = PLk = ∑ exp (VLk )


i=1...N exp (VLi )

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 14/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Random utility models (IV)

Other RUMs:
C-Logit model (Cascetta et al., 1996): It has the same sim-
ple mathematical structure but considers a modified systematic
utility, introducing a Commonality Factor CFk that takes into
account possible overlapping among routes.
Path-Size model (Ben-Akiva and Bierlaire, 1999): It is sim-
ilar to C-logit in that a correction term, the Path-Size (PS)
attribute, is added to the deterministic part of the utility.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 15/40
Fuzzy and Logic based Route Choice
Route Choice
Route Choice Modelling

Fuzzy route choice models

One other way of dealing with imperfect ordering of alternatives


given to drivers is to use fuzzy logic theory.
Fuzzy models use a fuzzy sets representation of costs and de-
termine a preference level for each route with a fuzzy logic
inference engine using fuzzy rules (Henn, 2000).
But, what is fuzzy logic?

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 16/40
Fuzzy and Logic based Route Choice
Fuzzy logic

Summary

1 Route Choice
Problem
Route Choice Modelling
Random utility models
Fuzzy route choice models

2 Fuzzy logic
Overview
Fuzzy Set Theory
Membership Function
3 Fuzzy and Logic Route Choice
Fuzzy Models
Related works
4 Conclusion
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 17/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Overview

Definitions

Definition of fuzzy:
Fuzzy - “not clear, distinct, or precise; blurred”
Definition of fuzzy logic:
A form of knowledge representation, suitable for notions that
cannot be defined precisely, but which depend upon their con-
texts.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 18/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Overview

Description

Fuzzy Logic resembles the human decision-making methodology. It


deals with vague and imprecise information (Dernoncourt, 2013).
This is gross oversimplification of the real-world problems and based
on degrees of truth rather than usual true/false or 1/0 like Boolean
logic.

True/Yes/1
Is RAM honest? Boolean Logic

False/No/0

Extremely honest (1.0)

Very honest (0.85)


Is RAM honest? Fuzzy Logic
Sometimes honest (0.35)

Sometimes dishonest (0.0)

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 19/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Overview

Fuzzy Logic vs Probabilistic Logic

Fuzzy logic and probabilistic logic are mathematically similar


− both have truth values ranging between 0 and 1 − but con-
ceptually distinct, due to different interpretations.
Probability is associated with events and not facts, and those
events will either occur or not occur. There is nothing fuzzy
about it.
Fuzzy logic basically try to capture the essential concept of
vagueness,all about degree of truth (membership value).

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 20/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Overview

Fuzzy Logic vs Probabilistic Logic

For example, the fuzziness of the


water is poisoned is 0.6 means that
the water is poisoned by containing
60% of poison. The probability of
the water is poisoned is 0.6 means
there is a 60% chance of the water
being poisoned.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 21/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Fuzzy Set Theory

Fuzzy Set

Fuzzy logic is based on the theory of fuzzy sets, which is a


generalization of the classical set theory (Zadeh, 1965).
Formally, a fuzzy subset is defined through a membership func-
tion µV defined in the interval [0, 1] such as µ : R → [0, 1] in
which R is a universal set.
Basically it allows partial membership which means that it con-
tain elements that have varying degrees of membership in the
set.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 22/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Fuzzy Set Theory

Fuzzy Set

Classical set contains elements that satisfy precise properties of


membership while fuzzy set contains elements that satisfy impre-
cise properties of membership.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 23/40
Fuzzy and Logic based Route Choice
Fuzzy logic
Fuzzy Set Theory

Operations on Fuzzy Sets

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 24/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice

Summary

1 Route Choice
Problem
Route Choice Modelling
Random utility models
Fuzzy route choice models

2 Fuzzy logic
Overview
Fuzzy Set Theory
Membership Function
3 Fuzzy and Logic Route Choice
Fuzzy Models
Related works
4 Conclusion
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 25/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Fuzzy Models

Fuzzy Models

In relation to route choice, two different classes of fuzzy models for


route choice can be distinguished (Henn, 2005):
Fuzzy rule-based models: It is based on linguistic rules with
fuzzy sets (e.g. if the travel time on route 1 is very short and on
route 2 it is intermediate, then the driver will certainly choose
route 1).
Fuzzy cost-based models: It is based on assessing some sys-
tematic utility attributes (mainly travel time, cost or link length)
by way of fuzzy numbers, making a comparison of these at-
tributes.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 26/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 1: Fuzzy route choice model for TAP (I)

Henn (2000) proposed a new route choice model taking account


of the imprecision and the uncertainties lying in the dynamic
choice process.
It is assumed that drivers choose a path all the more than it is
foreseen to have a lesser cost.
The predicted cost for each path is modelled by a fuzzy subset
which can represent imprecision on network knowledge as well
as uncertainty on traffic conditions.
The costs of all possible paths are compared and result in an
attractiveness degree for each path.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 27/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 1: Fuzzy route choice model for TAP (II)

Fuzzy predicted cost


The basic idea of this model is to represent accurately the im-
perfections lying in the route choice process.
There are two major types of imperfection that can arise in
prediction: imprecision (e.g. what the length of the route, or
the speed we will be able to drive) and uncertainty (e.g. which
events will occur during their trip).

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 28/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 1: Fuzzy route choice model for TAP (III)

Fuzzy predicted cost


Owing to the introduced representation, the authors define the
cost predicted by a driver for a given path. Let k be a route char-
acterized by ϕk possible states E1 · · · Eϕk (for example: “fluid”,
“congested”, “incident”, etc.). Every state Ei is given a confi-
dence level µi and an imprecise cost Vi which drivers predict to
pay if they choose route k and if event Ei occurs.
The global predicted cost for path p is:

UCk = µi · Vi (1)
i

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 29/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 1: Fuzzy route choice model for TAP (IV)

Fuzzy predicted cost: Example


Let us consider a path that might be either in normal condition (with
a confidence level of µ1 = 0.9 and a cost of approximately 30 min)
or in congested situation (with a confidence level of µ2 = 0.2 and a
very imprecise cost between 30 min − barely impossible − and 2h).
The predicted cost for this path is:

UC = 0.9 · V1 ∪ 0.2 · V2 (2)

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 30/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 1: Fuzzy route choice model for TAP (IV)

Figure: Example of predicted cost for a given path.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 31/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 2: Random and fuzzy utility models for RC (I)

Quattrone and Vitetta (2011) proposed a model simulate the


generation/choice process starting from the observation of a
sample of users in terms of route choice on the Italian national
road network.
The choice process is simulated in two phases:
Modeling choice set perception, in order to generate the possible
alternatives.
Modeling route choice behavior.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 32/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 2: Random and fuzzy utility models for RC (II)

In the fuzzy model, route systematic utility was considered the core
of a fuzzy number. Utility is calculated as with the RUM, but in this
case for each alternative the possibility (not the RUM probability)
of its utility being the greatest is calculated. These possibilities were
transformed into choice percentages.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 33/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 2: Random and fuzzy utility models for RC (II)

Similar to the RUM, the systematic utility VCk is calculated by:



VCk = βFi∗ Xi with βFi∗ = βFi /ρ
i=1

The utility VCk also represents the core value Vc of the associated
fuzzy number V = (Vs , Vc , Vd ), very often assuming it has a
triangular distribution.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 34/40
Fuzzy and Logic based Route Choice
Fuzzy and Logic Route Choice
Related works

Work 2: Random and fuzzy utility models for RC (III)

Figure: Graphical representation of the fuzzy number V = (Vs , Vc , Vd ).

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 35/40
Fuzzy and Logic based Route Choice
Conclusion

Summary

1 Route Choice
Problem
Route Choice Modelling
Random utility models
Fuzzy route choice models

2 Fuzzy logic
Overview
Fuzzy Set Theory
Membership Function
3 Fuzzy and Logic Route Choice
Fuzzy Models
Related works
4 Conclusion
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 36/40
Fuzzy and Logic based Route Choice
Conclusion

Conclusion

Similar to algorithms for the TAP, route choice allows for the
use of decentralized algorithms.
Choice set generation is not so easy due to the large number
of alternatives existing between each (o, d) pair.
A route choice model distinguishing the two main phases: mod-
eling choice set perception and modeling route choice behavior.
In particular, adoption of fuzzy theory and FUMs for route
choice allowed us to take into account the fuzziness related
to the perception of choice alternatives.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 37/40
Fuzzy and Logic based Route Choice
Conclusion

Questions & Answers

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 38/40
Fuzzy and Logic based Route Choice
Conclusion

References I

Ben-Akiva, M. and Bierlaire, M. (1999). Discrete Choice Methods


and their Applications to Short Term Travel Decisions, pages
5–33. Springer US.
Cascetta, E., Nuzzolo, A., Russo, F., and Vitetta, A. (1996). A
new route choice logit model overcoming iia problems:
specification and some calibration results for interurban
networks. In Proceedings of the 13th International Symposium
on Transportation Traffic Theory.
Dernoncourt, F. (2013). Introduction to fuzzy logic.
Dial, R. B. (1971). A probabilistic multipath traffic assignment
model which obviates path enumeration. Transportation
Research, 5(2):83–111.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 39/40
Fuzzy and Logic based Route Choice
Conclusion

References II

Henn, V. (2000). Fuzzy route choice model for traffic assignment.


Fuzzy Sets and Systems, 116(1):77 – 101.
Henn, V. (2005). What is the meaning of fuzzy costs in fuzzy
traffic assignment models? Transportation Research Part C:
Emerging Technologies, 13(2):107–119.
Quattrone, A. and Vitetta, A. (2011). Random and fuzzy utility
models for road route choice. Transportation Research Part E:
Logistics and Transportation Review, 47(6):1126–1139.
Zadeh, L. (1965). Fuzzy sets. Information and Control,
8(3):338–353.

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 40/40

You might also like