jFuzzyLogic: A Robust and Flexible Fuzzy-Logic

Inference System Language Implementation

Pablo Cingolani Jesús Alcalá-Fdez, Member, IEEE
School of Computer Science Department of Computer Science and Artificial Intelligence
McGill University University of Granada, CITIC-UGR
Montreal, Quebec, H3A-1A4, Canada Granada, 18071, Spain
Email: [email protected] Email: [email protected]

Abstract—This work introduces jFuzzyLogic, an open source In this work, we introduce an open source Java library
library for fuzzy systems which allow us to design Fuzzy named jFuzzyLogic. This fuzzy systems library allows FLCs
Logic Controllers supporting the standard for Fuzzy Control design and implementation, following the standard for Fuzzy
Programming published by the International Electrotechnical
Commission. This library is written in Java and is available Control Language (FCL) published by the International Elec-
as open source from jfuzzylogic.sourceforge.net. The use of trotechnical Commission (IEC 61131-7) [15]. The IEC-61131
jFuzzyLogic is illustrated through the analysis of one case study. norm is well known for defining the Programmable Controller
Languages (PLC), commonly used in industrial applications.
In the part 7, this standard offers a well defined common
I. I NTRODUCTION understanding of the basic means to integrate fuzzy control
applications in control systems. It also defines a common
Fuzzy rule based systems (FRBSs) are one of the most language to exchange portable fuzzy control programs among
important areas for the application of the Fuzzy Set Theory different platforms.
[1]. Classical rule based systems deal with IF-THEN rules. The main goal of jFuzzyLogic is to bring the benefits of
FRBSs constitute an extension to classical systems, having open source software and standardization to the fuzzy systems
antecedents and consequents composed of fuzzy logic state- community. Our library offers several advantages:
ments. • Standardization, which reduces programming work and
A Fuzzy Logic Controller (FLC) [2]–[5] is a FRBS com- learning curve. This library contains the basic program-
posed of: i-) a Knowledge Base that comprises the information ming elements for the Standard IEC 61131-7, alleviating
used by the expert operator in the form of linguistic control developers from boiler plate programming tasks.
rules; ii-) a Fuzzification Interface, that transforms the crisp • Extensibility, the object model and API allows to create
values of the input variables into fuzzy sets; iii-) an Inference a wide range of applications. This is of special interest
System, that uses the fuzzy values from the Fuzzification for the research community.
Interface and the information from the Knowledge Base to • Platform independence, allows to develop and run on
perform the reasoning process and iv-) the Defuzzification any hardware and operating system configuration that
Interface, which takes the fuzzy action from the Inference supports Java.
System and translates it into crisp values for the control This work is arranged as follows. The next section presents
variables. a comparison on non-commercial fuzzy software and the main
FLCs are suitable for engineering applications in which benefits that the jFuzzyLogic offers with respect to other
classical control strategies do not achieve good results or when libraries. Section III describes jFuzzyLogic’s main features.
it is too difficult to obtain a mathematical model. FLCs usually Section IV, illustrates how jFuzzyLogic can be used in a
have two characteristics: the need for human operator experi- control application. Conclusions are presented in Section V.
ence, and a strong non linearity. Many real-world applications
use FLCs [6] such as mobile robot navigation [7], [8], air II. C OMPARISON OF FUZZY LOGIC SOFTWARE
conditioning controllers [9], [10], domotic control [11], [12], In this section we present a comparison on non-commercial
and industrial applications [13], [14]. fuzzy software (Table I). We center our interest on free
FLCs are powerful for solving a wide range of problems, software distributions because of its important role in the
but their implementation requires a certain programming ex- scientific research community [16]. Moreover, we do not want
pertise. In the last few years, many fuzzy logic software to establish a comparison among all software tools or to
tools have been developed to reduce this task. Some are emphasize the advantages of one over another. Our objective
commercially distributed, for example MATLAB Fuzzy logic is to detect the major differences in the software and then to
toolbox(www.mathworks.com), while a few are available as categorize jFuzzyLogic as an alternative to these suites when
open source software (see section II). other research requirements are needed.
 We analyze twenty five packages (including jFuzzyLogic), The specification defines six programming languages: In-
mostly from SourceForge or Google-Code, which are consid- struction list (IL), Structured text (ST), Ladder diagram (LD),
ered to be amongst the most respectable software repositories. Function block diagram (FBD), Sequential function chart
The packages are analyzed in the following categories: (SFC), and Fuzzy Control Language (FCL). While IL, ST,
• FCL support. Only four packages (∼ 17%) claim to sup- and FCL are text based languages, LD, FBD and SFC are
port IEC 61131-7 specification. Notably two of them are graphic based languages.
based on jFuzzyLogic. Only two packages that support Instruction list is similar to assembly language: one in-
FCL are not based on our software. Unfortunately neither struction per line, low level and low expression commands.
of them seem to be maintained by their developers any Structured text, as the name suggests, intends to be more
more. Furthermore, one of them has some code from structured and it is very easy to learn and understand for
jFuzzyLogic. anyone with a modest experience in programming. The focus
• Programming language. This is an indicator of code of this work is FCL, which is oriented to fuzzy logic based
portability. There languages of choice were mainly Java control systems and its syntax is similar to ST.
and C++/C (column Lang.). Java being platform inde-
pendent has the advantage of portability. C++ has an
A. IEC Language concepts
advantage in speed, but also allows easier integration in
industrial controllers. All IEC-61131 languages are modular. The basic module is
• Functionality. Seven packages (∼ 29%) were made for called Programmable Organization Unit (POU) and includes
specific purposes, marked as ‘specific’ (column Notes, Programs, Functions or Function Blocks. A system is usually
Table I). Specific code usually has limited functionality, composed of many POUs, and each of these POUs can
but it is simpler and has a faster learning curve for the be programmed in a different language. For instance, in a
user. system consisting of two functions and one function block
• Membership functions. This is an indicator of how com- (three POUs), one function may be programed in LD, another
prehensive and flexible the package is. Specific packages function in IL and the function block may be programmed in
include only one membership function (typically trape- ST. The norm defines all common data types (e.g. BOOL,
zoid) and/or one defuzzification method (data not shown). REAL, INT, ARRAY, STRUCT, etc.) as well as ways to
In some cases, arbitrary combinations of membership interconnect POUs, assign process execution priorities, process
functions are possible. These packages are marked with timers, CPU resource assignment, etc.
asterisk. For example, ‘M + N ∗ ’ means that the software The concepts of a Program and Functions are quite intuitive.
supports M membership functions plus another N which Programs are simple set of statements and variables. Functions
can be arbitrarily combined. are calculations that can return only one value and are not
• Latest release. In eight cases (∼ 33%) there were no supposed to have state variables.
released files for the last three years or more (see Rel.
A Function Block resembles a very primitive object. It
column in the Table I). This may indicate that the package
can have multiple input and multiple output variables, can be
is no longer maintained, and in some cases the web site
enabled by an external signal, and can have local variables.
explicitly mentions this.
Unlike an object, a function block only has one execution
• Code availability and usability. Five of the packages (∼
block (i.e. there are no methods). The underlying idea for these
21%) had no files available, either because the project
limitations is that you should be able to implement programs
was no longer maintained or because the project never
using either text-based or graphic-based languages. Having
released any files at all. Whenever the original sites were
only one execution block, allows to easily control execution
down, we tried to retrieve the projects from alternative
when using graphic-based language to interconnect POUs.
mirrors. In three cases (∼ 13%) the packages did not
compile. We performed minimal testing by just following At first glance FCL is similar to ST. However, there are
the instructions, if available, and make no effort to correct some very important differences. FCL uses exclusively a new
any compilation problems. POU type: Fuzzy Inference System (FIS) which is a special
case of a Function Block. All fuzzy language definitions
In summary, only eight of the software packages (∼ 33%) should be within a FIS. Since a fuzzy system is inherently
seemed to be maintained, compiled correctly, and had exten- parallel, there is no concept of execution order, therefore there
sive functionality. Only two of them are capable of parsing are no statements. For instance, there is no way to create
FCL (IEC-61131-7) files and both are based on jFuzzyLogic. the typical “Hello world” example since there is no print
statement. A simple example of a FIS using FCL is shown
III. JF UZZY L OGIC in Table II, this FCL code calculates the tip in a restaurant
Fuzzy Control Language is an industry standard specifica- (the equivalent of a “Hello world” program in fuzzy systems).
tion released by the International Electrotechnical Commission Fig. 1 shows the membership functions.
(IEC) as part of the Programmable Controller Languages Table III shows the corresponding Java code to run the FCL
(PLC) defined in the IEC-61131 specification. code shown in Table II.
 TABLE I
C OMPARISSON ON OPEN FUZZY LOGIC SOFTWARE PACKAGES . C OLUMNS DESCRIBE : P ROJECT NAME (NAME ), IEC 61131-7 LANGUAGE SUPPORT (IEC),
LATEST RELEASE YEAR (R EL .), MAIN PROGRAMMING LANGUAGE (L ANG .), SHORT DESCRIPTION FORM WEBSITE (D ESCRIPTION ), NUMBER OF
MEMBERSHIP FUNCTIONS SUPPORTED (MF) AND F UNCTIONALITY ( NOTES ). NAME ∗ : PACKAGE IS MAINTAINED , COMPILES CORRECTLY, AND HAS
EXTENSIVE FUNCTIONALITY.

Name IEC Rel. Lang. Description MF Notes

Akira No 2007 C++ Framework for complex AI agents. 4
AwiFuzz Yes 2008 C++ Fuzzy logic expert system 2 Does not compile
DotFuzzy No 2009 C# .NET library for fuzzy logic 1 Specific
FFLL Yes 2003 C++ Optimized for speed critical applications. 4 Does not compile
Fispro∗ No 2010 C++/Java Fuzzy inference design and optimization 6
FLUtE No 2004 C# A generic Fuzzy Logic Engine 1 Beta version
FOOL No 2002 C Fuzzy engine 5 Does not compile
FRBS No 2011 C++ Fuzzy Rule-Based Systems 1 Specific
funzy No 2007 Java Fuzzy Logic reasoning 2∗ Specific
Fuzzy Logic Tools∗ No 2011 C++ Framework fuzzy control systems, 12
FuzzyBlackBox No - - Implementing fuzzy logic - No files released
FuzzyClips No 2004 C/Lisp Fuzzy logic extension of CLIPS 3 + 2∗ No longer maintained
FuzzyJ ToolKit No 2006 Java Fuzzy logic extension of JESS 15 No longer maintained
FuzzyPLC∗ Yes 2011 Java Fuzzy controller for PLC Siemens s226 11 + 14∗ Uses jFuzzyLogic
GUAJE∗ No 2011 Java Development environment Uses FisPro
javafuzzylogicctrltool No - Java Framework for fuzzy rules - No files released
JFCM No 2011 Java Fuzzy Cognitive Maps (FCM) - Specific
JFuzzinator No 2010 Java Type-1 Fuzzy logic engine 2 Specific
jFuzzyLogic∗ Yes 2011 Java FCL and Fuzzy logic API 11 + 14∗ This paper
jFuzzyQt∗ Yes 2011 C++ jFuzzyLogic clone 8
libai No 2010 Java AI library, implements some fuzzy logic 3 Specific
libFuzzyEngine No 2010 C++ Fuzzy Engine for Java 1 Specific
nxtfuzzylogic No 2010 Java For Lego Mindstorms NXT 1 Specific
Octave FLT∗ No 2011 Octave Fuzzy logic for Toolkit 11
XFuzzy3∗ No 2003 Java Development environment 6 Implements XFL3 specification language

performing the required computations. This means that the IST

can represent the grammar, like and AST, but it also capable
of performing calculations. The parsed FIS can be evaluated
by recursively transversing the IST.
A FIS inference system is usually composed of one or more
Function Blocks (FB). Each FB has variables (input, output
or instance variables) as well as one or more Rule Blocks
(RB). Each rule block is composed of a set of rules, as well
as Aggregation, Activation and Accumulation methods. All
methods defined in the norm are implemented in jFuzzyLogic.
It should be noted that we adhere to the definitions of
Fig. 1. Membership functions for tipper example. Aggregation, Activation and Accumulation as defined by IEC-
61131-7, which may differ from the naming conventions from
other references (e.g. “Aggregation” may sometimes be called
B. jFuzzyLogic Implementation “Combination”).
jFuzzyLogic is fully implemented in Java, thus the package Aggregation methods define the t-norms and t-conorms
is platform independent. ANTLR [17] was used to generate playing the role of AND, OR and NOT operators. These can be
Java code for a lexer and parser based on our FCL grammar Minimum, Product or Bounded difference operators. Needless
definition. This generated parser uses a left to right leftmost to say, each set of operators must satisfy De Morgans laws.
derivation recursive strategy, formally know as “LL(*)”. Activation method define how rule antecedents modify rule
Using the lexer and parser created by ANTLR we are able to consequents, i.e. once the IF part has been evaluated, how
parse FCL files by creating an Abstract Syntax Tree (AST), a this result is applied to the THEN part of the rule. The most
well known structure in compiler design. The AST is converted common activation operators are Minimum and Product (see
into an Interpreter Syntax Tree (IST), which is capable of Figure 2).
 TABLE II
E XAMPLE OF F UZZY CONTROL LANGUAGE (FCL) CODE .

FUNCTION_BLOCK tipper

VAR_INPUT
service, food : REAL;
END_VAR

VAR_OUTPUT
tip : REAL;
END_VAR

FUZZIFY service
TERM poor := (0, 1) (4, 0) ;
TERM good := (1, 0) (4,1) (6,1) (9,0);
TERM excellent := (6, 0) (9, 1);
END_FUZZIFY

FUZZIFY food
TERM rancid := (0, 1) (1, 1) (3,0);
TERM delicious := (7,0) (9,1);
END_FUZZIFY

DEFUZZIFY tip
TERM cheap := (0,0) (5,1) (10,0);
TERM average := (10,0) (15,1) (20,0);
TERM generous := (20,0) (25,1) (30,0);
METHOD : COG; // Center of Gravity
END_DEFUZZIFY

RULEBLOCK tipRules
Rule1: IF service IS poor OR food IS rancid THEN tip IS cheap;
Rule2: IF service IS good THEN tip IS average;
Rule3: IF service IS excellent AND food IS delicious THEN tip IS generous;
END_RULEBLOCK

END_FUNCTION_BLOCK

TABLE III
E XAMPLE OF JAVA API TO EXECUTE FCL CODE .

public class TestTipper {

public static void main(String[] args)
throws Exception {
FIS fis = FIS.load("fcl/tipper.fcl", true);
FunctionBlock fb = fis.getFunctionBlock(null);
// Set inputs
fb.setVariable("service", 3);
fb.setVariable("food", 7);
// Evaluate
fb.evaluate();
// Get output
double tip = fb.getVariable("tip").getValue());
}
}

Sum.
Only two membership functions are defined in the IEC
standard: singleton and piece-wise linear. jFuzzyLogic also im-
plements other commonly used membership functions: trape-
zoidal, sigmoidal, gaussian, generalized bell, difference of sig-
Fig. 2. Activation methods: Min (left) and Prod (right).
moidal, and cosine. Furthermore, jFuzzyLogic allows to build
arbitrary membership functions by combining mathematical
functions.
Finally, accumulation method defines how the consequents
from multiple rules are combined within a Rule Block (see Because of the flexibility in defining membership functions,
Fig. 3). Accumulation methods defined in the norm include: we discretize them at a number of points. The number of
Maximum, Bounded sum, Normed sum, Probabilistic OR, and points, by default one thousand, can be adjusted according
Fig. 3. Accumulation method: Combining consequents from multiple rules
using Max accumulation method.

to the precision-speed trade-off required for a particular ap-

plication. Inference is performed by evaluating membership Fig. 4. Membership functions for wall-following robot.
functions at these discretization points. In order to perform
a discretization, the “universe” for each variable, has to be
estimated. The universe is defined as the range where the execution orders can be implemented by the user, which allows
variable has non-neglectable value. For each variable, each us to easily define hierarchical controllers.
membership function and each term is taken into account when
calculating a universe. Once all rules have been analyzed, the IV. A CASE STUDY
accumulation for each variable is complete. We present an example of creating an FLC controller with
The last step when evaluating a FIS is defuzzification. jFuzzyLogic. This case study is focused on the development of
The value for each variable is calculated using the selected the wall following robot as explained in [18]. Wall following
defuzzification method, which can be ’Center of gravity’, behavior is well known in mobile robotics. It is frequently
’Rightmost Max’, ’Center of area’, ’Leftmost Max’, ’Mean used for the exploration of unknown indoor environments and
max’ (continuous membership functions), or ’Center of grav- for the navigation between two points in a map.
ity’ (discrete membership functions). The main requirement of a good wall-following controller
is to maintain a suitable distance from the wall that is being
C. API extensions followed. The robot should also move as fast as possible, while
avoiding sharp movements, making smooth and progressive
Some of the extensions and benefits provided by jFuzzy- turns and changes in velocity.
Logic are described in this section. In our fuzzy control system, the input variables are: i)
Modularity. Modular design allows to extend the language normalized distances from the robot to the right (RD) and
and the API easily. It is possible to add custom aggregation, left walls (DQ); ii) orientation with respect to the wall (O);
activation or accumulation methods, defuzzifiers, or member- and iii) linear velocity (V ). The output variables in this
ship functions by extending the provided object tree. controller are the normalized linear acceleration (LA) and the
Dynamic changes. Our API supports dynamic changes made angular velocity (AV ). The linguistic partitions are shown in
onto a fuzzy inference system: i) variables can be used as Fig. 4 which are comprised by linguistic terms with uniformly
membership function parameters; ii) rules can be added or distributed triangular membership functions giving meaning to
deleted from rule blocks, iii) rule weights can be modified; them.
iv) membership functions can use combinations of pre-defined In order to implement the controller, the first step is to
functions. declare the input and output variables and to define the fuzzy
Optimization API. An optimization API is available, allow- sets (Table IV). Variables are defined in VAR INPUT and
ing fine tuning membership function rules and rule weights. A VAR OUTPUT sections. Fuzzy sets are defined in FUZZIFY
few optimization algorithms are already implemented, such as blocks for input variables and DEFUZZIFY blocks for output
gradient descent, partial derivative, and delta algorithm. Other variables.
optimization algorithms can be implemented based on these One FUZZIFY block is used for each input variable. Each
templates. TERM line within a FUZZIFY block defines a linguistic term
Data Types. Due to the nature of fuzzy systems and in order and its corresponding membership function. In this example all
to reduce complexity, jFuzzyLogic considers each variable as membership functions are triangular, so they are defined using
REAL variable which is mapped to a double Java type. the ’trian’ keyword, followed by three parameters defining left,
Excecution order. By default it is assumed that a FIS is center and right points (e.g. ‘trian 1 2 3’).
composed of only one Function Block, so evaluating the FIS Output variables define their membership functions within
means evaluating the default FB. If a FIS has more than one DEFUZZIFY blocks. Linguistic terms and membership func-
FB, they are evaluated in alphabetical order by FB name. Other tions are defined using the TERM keyword as previously de-
scribed for input variables. In this case we also add parameters [3] H. H. D. Driankov and M. Reinfrank, An Introduction to Fuzzy Control.
to select the defuzzyfication method. The statement ’METHOD Springer-Verlag, 1993.
[4] R. Yager and D. Filev, Essentials of fuzzy modeling and control. Wiley,
: COG’ indicates that we are using ’Center of gravity’. New York, 1994.
These membership functions can be plotted by [5] P. Bonissone, “Fuzzy logic controllers: An industrial reality,” in Com-
running jFuzzyLogic with an FCL file, having the code putational Intelligence: Imitating Life. IEEE Press, 1994, pp. 316–327.
[6] R. Palm, D. Driankov, and H. Hellendoorn, Model based fuzzy control.
shown in Table IV, as argument (e.g. java -jar Springer, Berlin, 1997.
jFuzzyLogic.jar robot.fcl). The corresponding [7] M. Mucientes, J. Alcalá-Fdez, R. Alcalá, and J. Casillas, “A case study
FCL file for this case study is available for download as for learning behaviors in mobile robotics by evolutionary fuzzy systems,”
Expert Systems With Applications, vol. 37, no. 2, pp. 1471–1493, 2010.
one of the examples provided in jFuzzyLogic package [8] C.-F. Juang and Y.-C. Chang, “Evolutionary-group-based particle-
(jfuzzylogic.sourceforge.net). swarm-optimized fuzzy controller with application to mobile-robot navi-
The second step is to define the rules used for inference. gation in unknown environments,” IEEE Transactions on Fuzzy Systems,
vol. 19, no. 2, pp. 379–392, 2011.
They are defined in RULEBLOCK statements. For the wall- [9] M. Gacto, R. Alcalá, and F. Herrera, “A multi-objective evolutionary
following robot controller, we used ’minimum’ connection algorithm for an effective tuning of fuzzy logic controllers in heating,
method (AND : MIN), minimum activation method (ACT : ventilating and air conditioning systems,” Applied Intelligence, vol. 36,
no. 2, pp. 330–347, 2012.
MIN), and maximum accumulation method (ACCU : MAX). [10] E. Cho, M. Ha, S. Chang, and Y. Hwang, “Variable fuzzy control for
We implemented the rule base generated in [18] by the WCOR heat pump operation,” Journal of Mechanical Science and Technology,
method [19]. Each entry in the rule base was converted to a vol. 25, no. 1, pp. 201–208, 2011.
[11] F. Chávez, F. Fernández, R. Alcalá, J. Alcalá-Fdez, G. Olague, and
single FCL rule (Table V). Within each rule, the antecedent F. Herrera, “Hybrid laser pointer detection algorithm based on template
(i.e. the IF part) is composed of the input variables connected matching and fuzzy rule-based systems for domotic control in real home
by ‘AND’ operators. Since there are more than one output enviroments,” Applied Intelligence, vol. 36, no. 2, pp. 407–423, 2012.
[12] G. Acampora and V. Loia, “Fuzzy control interoperability and scalability
variable, we can specify multiple consequents (i.e. THEN for adaptive domotic framework,” IEEE Transactions on Industrial
part) separated by semicolons. Finally, we add the desired Informatics, vol. 1, no. 2, pp. 97 – 111, 2005.
weight using the ‘with’ keyword followed by the weight. [13] Y. Zhao and H. Gao, “Fuzzy-model-based control of an overhead crane
with input delay and actuator saturation,” IEEE Transactions on Fuzzy
This completes the implementation of a controller for a wall- Systems, vol. 20, no. 1, pp. 181 –186, 2012.
following robot using FCL and jFuzzyLogic. [14] O. Demir, I. Keskin, and S. Cetin, “Modeling and control of a non-
linear half-vehicle suspension system: A hybrid fuzzy logic approach,”
V. C ONCLUSIONS Nonlinear Dynamics, vol. 67, no. 3, pp. 2139–2151, 2012.
[15] International Electrotechnical Commission technical committee indus-
In this paper, we have described jFuzzyLogic, an open trial process measurement and control. IEC 61131 - Programmable
source Java library for fuzzy systems which allow us to design Controllers - Part 7: Fuzzy control programming. IEC, 2000.
[16] S. Sonnenburg, M. Braun, C. Ong, S. Bengio, L. Bottou, G. Holmes,
FLCs following the standard IEC 61131. It allows us to reduce Y. LeCun, K.-R. Muller, F. Pereira, C. Rasmussen, G. Ratsch,
programming work and extend the range of possible users B. Scholkopf, A. Smola, P. Vincent, J. Weston, and R. Williamson, “The
applying fuzzy systems and FLCs. need for open source software in machine learning,” Journal of Machine
Learning Research, vol. 8, pp. 2443–2466, 2007.
We have shown a case study to illustrate the use of jFuzzy- [17] T. Parr, The definitive ANTLR reference: building domain-specific lan-
Logic. In this case, we developed an FLC controller for wall- guages, 2007.
following behavior in a robot. The example shows how FCL [18] M. Mucientes, R. Alcalá, J. Alcalá-Fdez, and J. Casillas, “Learning
weighted linguistic rules to control an autonomous robot,” International
can be used to easily implement fuzzy logic systems. Journal of Intelligent Systems, vol. 24, no. 3, pp. 226–251, 2009.
The jFuzzyLogic software package is continuously being [19] R. Alcalá, J. Alcalá-Fdez, J. Casillas, O. Cordón, and F. Herrera, “Hybrid
updated and improved. At the moment, we are developing learning models to get the interpretability-accuracy trade-off in fuzzy
modelling,” Soft Computing, vol. 10, no. 9, pp. 717–734, 2006.
an implementation of a C/C++ compiler for fuzzy inference
systems. This will allow easy implementation with embedded
control systems using different processors.
ACKNOWLEDGMENT
jFuzzyLogic was designed and developed by P. Cingolani.
He is supported in part by McGill Uninversity, Genome
Quebec. J. Alcala-Fdez is supported by the Spanish Ministry
of Education and Science under Grant TIN2011-28488 and
the Andalusian Government under Grant P10-TIC-6858.
We would like to thank R. Sladek and M. Blanchette
for their comments. Special thanks to P.J. Leonard for his
contributions to the project.
R EFERENCES
[1] L. Zadeh, “Fuzzy sets,” Information Control, vol. 8, pp. 338–353, 1965.
[2] C. Lee, “Fuzzy logic in control systems: Fuzzy logic controller parts i
and ii,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 20,
pp. 404–435, 1990.
 TABLE IV
WALL FOLLOWING ROBOT: F UZZY CONTROLLER IN FCL LANGUAGE : VARIABLE DEFINITIONS .

VAR_INPUT
rd : REAL; // Right distance
dq : REAL; // Distance quotient
o : REAL; // Orientation. Note: ’or’ is a reserved word
v : REAL; // Velocity
END_VAR

VAR_OUTPUT
la : REAL; // Linear acceleration
av : REAL; // Angular velocity
END_VAR

FUZZIFY rd
TERM L := trian 0 0 1;
TERM M := trian 0 1 2;
TERM H := trian 1 2 3;
TERM VH := trian 2 3 3;
END_FUZZIFY

FUZZIFY dq
TERM L := trian 0 0 2;
TERM H := trian 0 2 2;
END_FUZZIFY

FUZZIFY o
TERM HL := trian -450 -450 -225;
TERM LL := trian -450 -225 0;
TERM Z := trian -225 0 225;
TERM LR := trian 0 225 450;
TERM HR := trian 225 450 450;
END_FUZZIFY

FUZZIFY v
TERM L := trian 0 0 1;
TERM H := trian 0 1 1;
END_FUZZIFY

DEFUZZIFY la
TERM VHB := trian -1 -1 -0.75;
TERM HB := trian -1 -0.75 -0.5;
TERM MB := trian -0.75 -0.5 -0.25;
TERM SB := trian -0.5 -0.25 0;
TERM Z := trian -0.25 0 0.25;
TERM SA := trian 0 0.25 0.5;
TERM MA := trian 0.25 0.5 0.75;
TERM HA := trian 0.5 0.75 1;
TERM VHA := trian 0.75 1 1;
METHOD : COG; // Center of Gravity
DEFAULT := 0;
END_DEFUZZIFY

DEFUZZIFY av
TERM VHR := trian -1 -1 -0.75;
TERM HR := trian -1 -0.75 -0.5;
TERM MR := trian -0.75 -0.5 -0.25;
TERM SR := trian -0.5 -0.25 0;
TERM Z := trian -0.25 0 0.25;
TERM SL := trian 0 0.25 0.5;
TERM ML := trian 0.25 0.5 0.75;
TERM HL := trian 0.5 0.75 1;
TERM VHL := trian 0.75 1 1;
METHOD : COG;
DEFAULT := 0;
END_DEFUZZIFY
 TABLE V
WALL FOLLOWING ROBOT. F UZZY CONTROLLER IN FCL LANGUAGE : RULE BLOCK .

RULEBLOCK rules
AND : MIN; // Use ’min’ for ’and’ (also implicit use ’max’ for ’or’ to fulfill DeMorgan’s Law)
ACT : MIN; // Use ’min’ activation method
ACCU : MAX; // Use ’max’ accumulation method

RULE 01: IF rd is L and dq is L and o is LL and v is L THEN la is VHB , av is VHR with 0.4610;
RULE 02: IF rd is L and dq is L and o is LL and v is H THEN la is VHB , av is VHR with 0.4896;
RULE 03: IF rd is L and dq is L and o is Z and v is L THEN la is Z , av is MR with 0.6664;
RULE 04: IF rd is L and dq is L and o is Z and v is H THEN la is HB , av is SR with 0.5435;
RULE 05: IF rd is L and dq is H and o is LL and v is L THEN la is MA , av is HR with 0.7276;
RULE 06: IF rd is L and dq is H and o is Z and v is L THEN la is MA , av is HL with 0.4845;
RULE 07: IF rd is L and dq is H and o is Z and v is H THEN la is HB , av is ML with 0.5023;
RULE 08: IF rd is L and dq is H and o is LR and v is H THEN la is VHB , av is VHL with 0.7363;
RULE 09: IF rd is L and dq is H and o is HR and v is L THEN la is VHB , av is VHL with 0.9441;
RULE 10: IF rd is M and dq is L and o is Z and v is H THEN la is SA , av is HR with 0.3402;
RULE 11: IF rd is M and dq is L and o is LR and v is H THEN la is Z , av is VHL with 0.4244;
RULE 12: IF rd is M and dq is L and o is HR and v is L THEN la is SA , av is HL with 0.5472;
RULE 13: IF rd is M and dq is L and o is HR and v is H THEN la is MB , av is VHL with 0.4369;
RULE 14: IF rd is M and dq is H and o is HL and v is L THEN la is Z , av is VHR with 0.1770;
RULE 15: IF rd is M and dq is H and o is HL and v is H THEN la is VHB , av is VHR with 0.4526;
RULE 16: IF rd is M and dq is H and o is LL and v is H THEN la is SA , av is VHR with 0.2548;
RULE 17: IF rd is M and dq is H and o is Z and v is L THEN la is HA , av is Z with 0.2084;
RULE 18: IF rd is M and dq is H and o is LR and v is L THEN la is HA , av is VHL with 0.6242;
RULE 19: IF rd is M and dq is H and o is LR and v is H THEN la is SA , av is VHL with 0.3779;
RULE 20: IF rd is M and dq is H and o is HR and v is L THEN la is Z , av is VHL with 0.6931;
RULE 21: IF rd is M and dq is H and o is HR and v is H THEN la is VHB , av is VHL with 0.7580;
RULE 22: IF rd is H and dq is L and o is Z and v is L THEN la is HA , av is VHR with 0.5758;
RULE 23: IF rd is H and dq is L and o is LR and v is H THEN la is SA , av is MR with 0.2513;
RULE 24: IF rd is H and dq is L and o is HR and v is L THEN la is HA , av is VHL with 0.5471;
RULE 25: IF rd is H and dq is L and o is HR and v is H THEN la is SA , av is HL with 0.5595;
RULE 26: IF rd is H and dq is H and o is HL and v is L THEN la is VHB , av is VHR with 0.9999;
RULE 27: IF rd is H and dq is H and o is HL and v is H THEN la is VHB , av is VHR with 0.9563;
RULE 28: IF rd is H and dq is H and o is LL and v is L THEN la is HA , av is VHR with 0.9506;
RULE 29: IF rd is H and dq is H and o is Z and v is L THEN la is HA , av is VHR with 0.4529;
RULE 30: IF rd is H and dq is H and o is Z and v is H THEN la is SA , av is VHR with 0.2210;
RULE 31: IF rd is H and dq is H and o is LR and v is L THEN la is HA , av is MR with 0.3612;
RULE 32: IF rd is H and dq is H and o is LR and v is H THEN la is SA , av is MR with 0.2122;
RULE 33: IF rd is H and dq is H and o is HR and v is L THEN la is HA , av is HL with 0.7878;
RULE 34: IF rd is H and dq is H and o is HR and v is H THEN la is SA , av is VHL with 0.3859;
RULE 35: IF rd is VH and dq is L and o is LR and v is L THEN la is HA , av is VHR with 0.5530;
RULE 36: IF rd is VH and dq is L and o is HR and v is L THEN la is HA , av is HR with 0.4223;
RULE 37: IF rd is VH and dq is L and o is HR and v is H THEN la is SA , av is HR with 0.3854;
RULE 38: IF rd is VH and dq is H and o is LL and v is L THEN la is HA , av is VHR with 0.0936;
RULE 39: IF rd is VH and dq is H and o is LR and v is L THEN la is HA , av is VHR with 0.7325;
RULE 40: IF rd is VH and dq is H and o is LR and v is H THEN la is SA , av is VHR with 0.5631;
RULE 41: IF rd is VH and dq is H and o is HR and v is L THEN la is HA , av is HR with 0.5146;
END_RULEBLOCK