A Smart System for Haptic Quality Control

Introducing an Ontological Representation of Sensory Perception Knowledge

Bruno Albert

1,2,3

, Cecilia Zanni-Merk

, Franc¸ois de Bertrand de Beuvron

, Jean-Luc Maire

Maurice Pillet

, Julien Charrier

and Christophe Knecht

ICube Laboratory SDC Team, INSA de Strasbourg, 300 bd S

ebastien Brant, 67400 Illkirch, France

SYMME Laboratory, Universit

e Savoie Mont-Blanc, 7 Chemin de Bellevue, 74940 Annecy-Le-Vieux, France

INEVA, 14 rue du Girlenhirsch, 67400 Illkirch, France

Keywords:

Sensory Perception, Haptics, Smart System, Semantic Analysis, Quality Control, Perceived Quality.

Abstract:

Perceived quality has become an important factor in the choice of products by customers. The human per-

ception process involves complex phenomena at a physical and psychological level that enable us to sense the

world and extract information about it. Because of the qualitative way humans represent and communicate

sensations, the ﬁeld of sensory perceptions makes extensive use of semantics. The use of knowledge-based

systems in the ﬁeld of perceived quality is hence natural. This project focuses on haptics in quality control

in industry. In particular, the aim is to develop a smart system which will enable to make decisions about

the haptic quality of a product. This paper introduces the framework used for the development of this smart

system, based on the KREM model. An ontological structure is proposed in order to represent knowledge re-

lated to the measure of sensory perceptions in general, and of haptic ones in particular. The proposed domain

ontologies about haptic control, that were elicited using semantic analysis, are aligned with the SSN ontology.

1 INTRODUCTION

Humans perceive the world and particularly objects in

the world through their different senses, which allow

them to not only understand, but also make their own

opinion and judgment about these objects. Consider-

ing this fact, perceived quality has naturally become a

major factor of the choice of products by customers.

In an industrial context, controlling perceived qual-

ity is often limited to controlling the visual aspect of

products, which in some cases is enough, but in most

cases does not fully correspond to the perception a

customer can have when interacting with a product.

In particular, the action of touching the product usu-

ally comes right after a ﬁrst visual observation.

Touch involves complex physical and psycholog-

ical phenomena which lead to very precise but also

very subjective haptic perceptions. Haptic percep-

tions are a combination of tactile and kinesthetic per-

ceptions (Lederman and Klatzky, 2009). Tactile sen-

sations are obtained thanks to sensory receptors local-

ized in the skin. Kinesthetic sensations are obtained

thanks to receptors localized in muscles, tendons and

joints. As a simpliﬁcation, considering contacts with

a product, tactile sensations can be understood as the

sensations obtained locally on the surface of the prod-

uct and kinesthetic sensations as the ones obtained

more globally over the product.

Therefore, on the one hand the control of hap-

tic sensations involves the comprehension of the pro-

cess of creation of haptic perceptions, at physical and

psychological levels. On the other hand, it involves

the formalisation of this knowledge in order to ex-

tract control protocols and make this knowledge us-

able by an automated system. In this context, knowl-

edge based systems are especially suitable.

The KREM framework is presented in section two

for the development of a Smart System for haptic

quality control. This framework highlights the use

of four main components (knowledge, rules, experi-

ence, meta-knowledge) to take into account the speci-

ﬁcities of the system to be developed. In particular,

this paper details the Knowledge component. An on-

tological structure is introduced in section three. It

presents the different domains of knowledge involved

and the corresponding ontologies, as well as the upper

level ontology used to structure the concepts. A do-

main ontology structuring haptic knowledge is then

presented in section four, with the integration of the

proposed formalization of the haptic sensations.

Albert, B., Zanni-Merk, C., Beuvron, F., Maire, J-L., Pillet, M., Charrier, J. and Knecht, C.

A Smart System for Haptic Quality Control - Introducing an Ontological Representation of Sensory Perception Knowledge.

DOI: 10.5220/0006048300210030

In Proceedings of the 8th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2016) - Volume 2: KEOD, pages 21-30

ISBN: 978-989-758-203-5

2 A FRAMEWORK FOR THE

DEVELOPMENT OF A SMART

SYSTEM FOR HAPTIC

QUALITY CONTROL

One framework used in order to develop a smart sys-

tem is the KREM model (Zanni-Merk, 2015), that has

been successfully used in several different domains

(Zanni-Merk et al., 2015; Gartiser et al., 2014).

Conventionally, a smart system is composed of a

fact base and a rule base, on which various types of

reasoning can be made. But the observation of the

drawbacks of this classic architecture (the difﬁcul-

ties in eliciting expert knowledge, mainly because ex-

perts operate tacit knowledge, and basically, the non-

completeness of this elicitation (Milton, 2008)) led to

the proposal of this model, based on the use of seman-

tic technologies.

Semantic technologies use methods from auto-

matic language processing, machine learning and

knowledge representation to build the ontologies and

the rules that will enable its implementation. Se-

mantic technologies are also intended to create new

meaningful relationships, and therefore new knowl-

edge, based on information of different natures and

forms. Enriching documents with meta-data or cre-

ating speciﬁc linguistic or terminological standards

are examples of the possibilities offered by seman-

tic technologies to facilitate decision making through

effective knowledge management.

But decision-making, to be effective, must re-

sult from reasoning and analysis on this knowledge

and must also take into account the experience and

expertise of decision-makers. Naturally, the cap-

italization of experience appeared as a possibility

of improvement of the architecture, in the form of

speciﬁc knowledge structures and reasoning mecha-

nisms, such as SOEKS (set of experience knowledge

structure) (San

ın and Szczerbicki, 2009) and CBR

(case based reasoning)(Aamodt and Plaza, 1994).

Finally, the use of meta-knowledge to lead the ex-

ecution of our knowledge-based systems became a

need. Meta-knowledge is knowledge about the do-

main knowledge, the rules or the experience. It can be

in the form of context, culture or protocols that steer

the use of that knowledge. Context is any information

that characterizes a situation related to the interac-

tion between human beings, applications and the sur-

rounding environment (Dey et al., 2001) and is iden-

tiﬁed as belonging to four types: identity, location,

status, time. Context is typically the location, identity

and state of people, groups, and computational and

physical objects. Time is information that helps to

Figure 1: The KREM architecture with its four interrelated

components: Knowledge, Rules, Experience and Meta-

knowledge.

recognize a situation using historical data. The Cul-

ture aspect of meta-knowledge intends to reﬂect the

different ways decisions are made in different cultures

while Protocols include typically the ways the other

pieces of knowledge are used to accomplish the task

we are facing to (for example, diagnosis); or strategies

for problem solving or heuristics. Meta-knowledge

may be closely related to experience knowledge.

To take these ideas into account, the KREM model

has four interacting components that can be declined

by project or application domain. The re-use of com-

ponents is, of course, encouraged. The KREM com-

ponents are (Figure 1):

• The Knowledge component that contains the do-

main knowledge to operate, by means of different

domain ontologies.

• The Rules component that allows different types

of reasoning (monotone, spatial, temporal, fuzzy,

or other) depending on the application.

• The Experience component that allows the capi-

talization and re-use of prior knowledge.

• The Meta-knowledge component, including

knowledge about the other three bricks that

depends on the problem.

The way the domain knowledge is formalized will

shape the way the rules are expressed. Experience

will complete the available knowledge and rules. Fi-

nally, meta-knowledge will directly interact with the

rules and the experience to indicate which rules (com-

ing from experience or from the initial rule set) need

to be launched according to the context of the problem

to solve.

A modular architecture, such as KREM, is one

of the main architectural design pattern for large and

complex systems. In this pattern, each module or

component has a speciﬁc functionality providing sep-

aration of concerns that, in turn, support re-use or

replacement (i.e. changes in a single module would

not affect the others, permitting the continuous oper-

ation of the system). Moreover the communication

KEOD 2016 - 8th International Conference on Knowledge Engineering and Ontology Development

between modules needs to be based on well deﬁned

interfaces to provide low coupling.

Formalising and structuring Knowledge are the

ﬁrst steps of the development of the Smart System.

The following sections are hence focused on the

Knowledge component, which will eventually be in-

tegrated to a larger KREM architecture and therefore

coupled with the other modules previously presented.

3 TOWARDS AN ONTOLOGICAL

STRUCTURE FOR HAPTIC

QUALITY CONTROL

The use of formal models, such as ontologies, is es-

sential for the development of a smart system. How-

ever, very few studies have proposed ontologies di-

rectly related to the description of human perceptions,

or more speciﬁcally to the perceived quality. One rare

example of a similar domain ontology has been pro-

posed by (Myrgioti et al., 2013), but this study fo-

cuses on software development for haptic interfaces.

This section presents the proposed ontological struc-

ture for the measuring of sensory perceptions and its

particularization towards haptic quality control.

3.1 Upper-level Conceptual Model

The ﬁrst step in the construction of a conceptual do-

main representation is the identiﬁcation of the general

ontological structure in which the domain ontology

can be included. In particular, the use of a high-level

ontology is essential in order to form the ”skeleton” of

the structure. There are multiple upper-level ontolo-

gies, but we will focus here on the Semantic Sensor

Network (SSN) ontology (Compton et al., 2012), sup-

ported by the W3C

, that has been identiﬁed as par-

ticularly relevant considering the context of the study

and the opportunities of further development. Comp-

ton et al. (2012) introduced the SSN ontology in order

to describe sensors and observations. Besides the per-

spectives of future development of the present study

around the system of sensors, the SSN ontology pro-

poses a way to conceptualize the links between prop-

erties, sensors and observation. In addition, SSN is

a core ontology that is based on the well-known top-

level DUL ontology (DOLCE+DnS Ultralite, 2010).

Figure 2 is a reduced version of the SSN ontol-

ogy including the stimulus-sensor-observation pattern

proposed by (Compton et al., 2012), focused only on

some of the entities relevant to this study. This rep-

resentation involves the different concepts of inter-

World Wide Web Consortium (http://www.w3.org/)

Figure 2: Extract of the SSN ontology (Compton et al.,

2012).

est, regarding the aim of knowledge integration about

haptic perception, as well as future automation of the

process. An alignment between this extract of the

SSN ontology and the proposed domain ontology is

presented below.

3.2 Global Ontological Structure

Because of the diversity of domains involved in the

control of the haptic quality of products, an ontolog-

ical structure is proposed in Figure 3. It aims at or-

ganizing knowledge that composes each domain into

different domain ontologies, which can be aligned to

the upper-level ontology, i.e. the SSN ontology here.

This ontological structure is presented as a classical

ontology hierarchy (Roussey et al., 2011) with a top-

level ontology, a core ontology, a general ontology, a

task ontology and multiple domains ontologies. The

domains involved here are: the sensors, the objects

of study (products to be controlled), the sensory per-

ceptions (and haptic perceptions in particular) and the

context. In addition, the control process is represented

as a task ontology.

The haptic quality control will then make use

of the elements of each of the domain ontologies,

in a ﬂexible and adapted way following the context

of application and industrial constraints (formalised

throughout the Context ontology). The following sec-

tion focuses on the Haptic Perception ontology which

gathers knowledge about haptics and enables a di-

rect correspondence with human perceptions. The

other domain ontologies involved in this structure are

part of the development process for the formalization

of haptic quality control, but they will not be exten-

sively detailed in this paper. Here is a brief descrip-

tion of these ontologies : the Sensor ontology gath-

ers knowledge about industrial sensors which are rel-

evant for the measuring of sensations. The Object

of Study ontology contains information about the

A Smart System for Haptic Quality Control - Introducing an Ontological Representation of Sensory Perception Knowledge

products or the samples on which the quality control

is performed. The Control Process ontology aims at

gathering the tasks and protocols necessary in order

to perform the quality control. Moreover, the Context

and Control Process ontologies are mainly part of the

Meta-knowledge components.

In addition, the proposed ontological structure has

been made general enough in order to foresee possible

utilization with other kinds of senses, for instance vi-

sion, hearing or taste, as shown with dotted line boxes

in Figure 3.

4 HAPTIC DOMAIN ONTOLOGY

This section presents our ﬁrst developments of a do-

main ontology integrating haptic perception knowl-

edge and its alignment with Compton’s core ontology.

In order to develop this ontology it was necessary to

start with the gathering of knowledge about this do-

main, by understanding the main concepts involved

as well as structuring and formalizing the vocabulary

involved. The proposed haptic perception ontology is

then detailed and an example is presented.

4.1 Formalization of Knowledge about

the Haptic Domain

The sense of touch has been widely studied at a bio-

logical level in order to understand how it works. A

summary of the perception process is presented here-

after. Considering the description of sensations, only

few studies have proposed to analyse and select de-

scriptors. Moreover, these studies usually focused on

speciﬁc types of application and material. A more

general and generic approach is proposed here. It

intends to formalize and structure knowledge about

haptics by gathering relevant vocabulary and relations

from multiple sources, such as sensory analysis stud-

ies as well as vocabulary databases, and extracting

sensation classes.

4.1.1 Perception Process

The process of creation of a tactile perception starts

with touch, which can be deﬁned as the stimulation

of the skin by thermal, mechanical, chemical or elec-

trical stimuli. Sensory systems, and sensory receptors

in particular, activate as soon as a stimulus is detected.

They transform the energy received through the stim-

uli into electrical energy by a change of neuronal elec-

trical potential (transduction). Encoded information

is then processed by the nervous system in order to

produce sensations. Sensory systems located in the

nervous system interpret these sensations by compar-

ison to memories and known sensations. Perceptions

are the results of this process. A schematic view of the

tactile perception process is provided in Figure 4 (De

Boissieu, 2010) and De Rossi and Scilingo (2006).

In principle - and considering similar external

conditions such as temperature, humidity, level of

tiredness, etc. - a sensation is identical, or near-

identical, for everybody. However, a perception dif-

fers from one person to another, and depends on ex-

perience and/or culture.

4.1.2 Usual Haptic Sensation Descriptors

Humans communicate about sensations using words.

The ﬁeld of sensory perceptions makes hence great

use of semantics. In particular regarding haptic sen-

sations, more than 200 descriptors could be listed, as

a result of the search performed across the literature.

Some examples are provided below, along with the

speciﬁcity of these descriptors.

Descriptors found in the literature are usually re-

lated to speciﬁc types of product and material. They

also depend on the language and culture of the con-

trollers. Sensotact (Crochemore et al., 2003) is a

reference method introduced by Renault in order to

describe the tactile perceptions of vehicle interiors.

It employs ten descriptors distributed following the

exploration mode (hardness, responsiveness, mem-

ory effect, sticky, ﬁbrous, relief, scratchy, blocking,

slippery and thermal). Considering textile products

only, Issa et al. (2005) proposed six invariant de-

scriptors common to French and English languages

(ﬂexible/rigid, falling, thin/thick, soft, creasable, re-

sponsive). In the same ﬁeld of application, Pi-

card et al. (2003) found ﬁve pairs of descriptors

(soft/rough, thin/thick, mellow/hard, smooth/rough,

pleasant/harsh), from a set of twenty-four and Sola

(2007) listed fourteen descriptors. Considering paper

sheets, Summers et al. (2007) reduced it to two de-

scriptors (rough and stiff). In the ﬁeld of packaging,

Dumenil-Lefebvre (2006) suggested two groups of

descriptors, respectively for the tactile description of

ground-glass (sticky, rough, granular, slippery, cool,

greasy) and a multi-material group, including plas-

tic, cardboard, etc. (adherent, sticky, supple, elas-

tic, markable, rough, granular, slippery, scratchable,

cool).

The descriptors listed in the different studies are

hence very different from one product to another, as

well as from one type of material to another. While

some descriptors are common across different mate-

rials (for instance, soft that is used similarly for wood,

fabric, leather, ceramic), some others are very speciﬁc

to one type of material. For example, descriptors like

KEOD 2016 - 8th International Conference on Knowledge Engineering and Ontology Development

Figure 3: General ontological structure and domain ontologies of haptic quality control.

Figure 4: The human tactile perception process. Inspired

from De Boissieu (2010) and De Rossi and Scilingo (2006).

furry, fuzzy, ﬂuffy, etc. are mainly used for the de-

scription of fabric.

Translations from one language to another can

also make these lists change, because the correspon-

dence between the meanings of translated words is

not always complete, or can be expressed with sev-

eral different words, following the context. For ex-

ample, frais in French could be translated into fresh

or cool. Some words also do not have any transla-

tion, which is the case of bouchard

e in French which

comes from a tool: the boucharde, which is originally

used to print marks on concrete (Sola, 2007). Differ-

ences between cultures, often represented by the dif-

ference in language, can sometimes induce a differ-

ence in the meaning of the same words, for example

a sensation of cold might not be perceived the same

way by people living in cold or hot areas.

4.1.3 Formalizing Haptic Sensations

Considering the diversity of the vocabulary used in

order to describe haptic sensations, there is a strong

need for the formalisation of the way these sensations

are described. For instance, studies performed in the

context of visual quality control (Baudet, 2012; Maire

et al., 2013) demonstrated the feasibility of reducing

the list of descriptors used and therefore formalizing

visual aspect anomalies.

The proposed method makes use of the seman-

tic characteristics of the descriptors in order to ex-

tract generic categories of haptic sensations. In par-

ticular, a classiﬁcation of the usual descriptors was

performed. First, semantic relations between descrip-

tors were used in order to group them. Synonym and

antonym links were drawn from semantic databases

like Wordnet

and the Thesaurus

. Then, a graphical

WordNet: https://wordnet.princeton.edu/

Thesaurus: http://www.thesaurus.com/

A Smart System for Haptic Quality Control - Introducing an Ontological Representation of Sensory Perception Knowledge

tool

and the OpenOrd method (Martin et al., 2011)

were used in order to gather elements with strong re-

lations and spread the ones with weak relations. The

result is shown in ﬁgure 5. The OpenOrd method in-

volves the computation of the distance between nodes

(descriptors), by minimizing a formula containing at-

tractive and repulsive terms. This grouping enabled to

extract categories from the meaning of the descriptors

contained in these main groups.

Descriptors were also classiﬁed following three

semantic axes proposed by Sola (2007), which focus

on the origins and meaning of the descriptors. These

semantic axes (source, effect and physical property)

highlight the semantic foundations of the descriptors

and the links with the characteristics of a surface. The

source axis refers to a perception (sensation comple-

mented with knowledge and experience), e.g. oily

refers to oil. One needs to have the knowledge of

what oil is to understand what an oily sensation is.

The effect axis refers to sensations, which involve a

judgement from the evaluator. These sensations are

hence subjective, or even hedonic. Finally, the phys-

ical property axis refers to a non-hedonic sensation,

which can be directly measured. This classiﬁcation

enabled to select descriptors which were relevant to

each category, and useful as an objective description

of speciﬁc levels of each category of sensation. Hedo-

nic elements (which form a separate group in ﬁgure 5)

were not selected, because of the strongly subjective

judgement they involve.

4.1.4 Elementary Haptic Sensations

As a result of the categorization previously presented,

a set of nine elementary haptic sensations is proposed.

These elementary sensations correspond to the groups

identiﬁed in the classiﬁcation step - with one excep-

tion of a central group (including homogeneous and

heterogeneous) corresponding to general characteris-

tics of the other descriptors. These elementary sensa-

tions are hence designed to cover all haptic descrip-

tors listed and to be used in order to describe them

in a formalised manner. Moreover, considering that

these classes were constructed using descriptors from

all kinds of domains of application, they enable a

generic description of tactile sensations. In particu-

lar, the descriptors found in the literature, and pre-

sented above, can be described by at least one ele-

mentary sensation. Table 1 shows the list of nine el-

ementary sensations, organised following tactile sen-

sations and kinesthetic sensations. In this Figure, an

association of stimuli and exploratory movements is

Gephi: Open graph visualisation platform, url:

https://gephi.org/

Table 1: Proposed elementary sensations, and associated

stimuli and exploratory movements.

also proposed. It was constructed thanks by extract-

ing relations between stimuli and descriptors in the

following studies : (Lederman and Klatzky, 2009; De

Boissieu, 2010; Bensma

ıa and Hollins, 2005; Jones

and Lederman, 2006; Crochemore et al., 2003).

4.2 Haptic Perception Ontology

The formalized knowledge about the haptic domain

was then used to design a domain ontology of haptic

perceptions. Figure 6 shows an extract of the pro-

posed haptic perception ontology. It was obtained us-

ing OWLGrEd (B

arzdin¸

s et al., 2010). A UML-like

notation is used, where boxes are OWL classes, thick

lines are hierarchical relations and thin lines are re-

strictions, labeled with object properties. Only a sub-

set of the concept and relation restrictions are dis-

played for the sake of clarity, and in order to show

a speciﬁc example of the characterisation of the de-

scriptor Slippery. The complete ontology includes all

the elements of Table 1 as well as the full list of de-

scriptors.

The main concepts of the more general sensory

perception ontology are: Exploration, Stimulus, Sen-

sation and Descriptor. Exploration describes the way

the stimulus is generated. Receptor refers to hu-

man sensory receptors which detects stimuli. Stim-

ulus is the physical phenomenon that induces sen-

sations. Sensation is a formal description of human

sensations. It gives insights on Descriptor which in-

tegrates the vocabulary usually used to communicate

about perceptions. In addition, individual stimuli are

characterized by individual elements of PhysicalPa-

rameter. These elements are a decomposition of the

physical properties that characterise stimuli. They are

intended to be as low level as possible in order to pro-

vide elements for the control process as well as to es-

KEOD 2016 - 8th International Conference on Knowledge Engineering and Ontology Development

Figure 5: Extraction of categories from usual haptic descriptors using semantic classiﬁcation methods.

tablish future relations with sensors.

Regarding haptic perceptions in particular, the

concept of Exploration integrates the elements nec-

essary for the stimuli to be generated. This includes

parameters of the movements and contacts to be per-

formed. The concept of Stimulus integrates the phys-

ical characteristics measured by human receptors,

i.e. for instance Pressure, Stretching, Persistence,

etc. The concept of Sensation integrates the pro-

posed formalisation of haptic sensations previously

presented. The elementary sensations are distributed

on the two classes TactileSensation and Kinesthetic-

Sensation. The usual descriptors are grouped under

the concept Descriptor and are related to correspond-

ing sensation categories.

Figure 6 serves also as an example of the relations

involved in the description and characterization of

the descriptor Slippery. First, the performed formal-

ization of haptic knowledge provides information on

relations between the usual descriptors and the pro-

posed elementary sensations. Considering this spe-

ciﬁc example case, it is possible to establish that the

elementary Sensation of Grip uniquely describes the

Descriptor Slippery. Grip is sensed through the Stim-

ulus Stretching. This speciﬁc type of Stimulus is char-

acterized by the physical parameters TangentialDe-

formation and TangentialForce.

indent Thus, this representation provides all the ele-

ments necessary to link haptic sensations - and even

more speciﬁcally usual descriptors of these sensations

- to the physical properties of the stimuli that can be

measured in order to provide intensity values of hap-

A Smart System for Haptic Quality Control - Introducing an Ontological Representation of Sensory Perception Knowledge

Figure 6: Extract of the proposed Haptic Perception ontology.

tic sensations. Moreover, this ontology also provides

the related exploration parameters, as well as the sen-

sory receptors involved in the perception process (not

fully displayed in Figure 6).

4.3 Alignment to the SSN Core

Ontology

Considering the aim of applying haptic perception

knowledge to quality control, the higher level SSN on-

tology brings a formal structure to the proposed con-

cepts and will enable to reason about it in a more gen-

eral manner. Indeed, SSN is designed to provide a

formal ontological framework to represent the inter-

actions between sensors and properties, or more gen-

erally between a sensing system and features of in-

terest.The alignment to a part of the SSN ontology,

and more speciﬁcally to a part of the sensor-stimulus-

observation pattern, is relatively natural considering

the concepts proposed in the Haptic Perception (HP)

ontology. The following alignment is proposed :

HP : Sensation v SSN : Property

HP : PhysicalParameter v SSN : Property

HP : Descriptor v SSN : FeatureO f Interest

HP : Stimulus v SSN : Stimulus

HP : Exploration v SSN : Sensing

HP : Receptors v SSN : Sensor

In particular, in the SSN ontology, SSN:Property

is deﬁned as an observable characteristic of real-

world entities (SSN:FeatureofInterest), which are not

directly observable. HP:Descriptor can be aligned to

SSN:FeatureofInterest, because it corresponds to the

way people usually communicate about sensations,

which is not directly observable. On the contrary,

HP:Sensation and HP:PhysicalParameter correspond

to observable characteristics of HP:Descriptor. They

can hence be aligned to SSN:Property.

5 CONCLUSIONS AND

PERSPECTIVES

This work sets the basis for the development of a

Smart System for haptic quality control. The main ar-

chitecture of the System was presented with the use of

the KREM model which provides a separation of the

modules involved in the development of this system,

while still enabling interaction between them. This al-

lows for more ﬂexibility in the development, consid-

ering future applications of the system, in particular

regarding the integration of experience. The Knowl-

edge component of this architecture was explored and

detailed in this paper, with the proposition of a general

ontological structure adapted to the constraints of the

project.The high-level SSN ontology was used in or-

der to structure knowledge corresponding to the mul-

tiple domains involved in the development of a Smart

System for haptic quality control. The domain on-

tology integrating haptic knowledge was speciﬁcally

presented. The formalization of haptic knowledge

was ﬁrst detailed with the proposition of generic el-

ementary haptic sensations. Haptic knowledge was

then conceptualised into the proposed Haptic Percep-

tion ontology and aligned with the SSN ontology.

The next steps of the development of a Smart Sys-

tem for haptic quality control include the develop-

KEOD 2016 - 8th International Conference on Knowledge Engineering and Ontology Development

ment of the Meta-knowledge and Rules modules of

the KREM architecture. In particular, the extraction

of control rules will be realized through the applica-

tion of the proposed formalization of haptic knowl-

edge on case studies. While expert knowledge en-

abled a validation of the semantic analysis, the in-

dustrial testing being performed will provide more

speciﬁc evaluation material. Moreover, the establish-

ment of the inﬂuence of the application context will

enable to select adapted rules. Furthermore, the pro-

posed system being intended to automate haptic qual-

ity control, knowledge about sensors and objects of

study will also be explored, as well as the relations

between data from the sensors and haptic sensations

which correspond to the problem of symbol anchor-

ing.

ACKNOWLEDGEMENTS

This work has been done within a thesis project

funded by the French technological research associ-

ation (ANRT) as well as the company INEVA

. This

work is the result of a collaboration between three

parties, which are all acknowledged here: the com-

pany INEVA, the INSA de Strasbourg (with the ICube

laboratory) and the University of Savoie Mont Blanc

(with the SYMME laboratory).

REFERENCES

Aamodt, A. and Plaza, E. (1994). Case-based reasoning:

Foundational issues, methodological variations, and

system approaches. AI communications, 7(1):39–59.

arzdin¸

s, J., B

arzdin¸

s, G.,

Cer

ans, K., Liepin¸

s, R., and

Spro

gis, A. (2010). Uml style graphical notation and

editor for owl 2. In Perspectives in Business Informat-

ics Research, pages 102–114. Springer.

Baudet, N. (2012). Ma

ıtrise de la qualit

e visuelle des

produits - Formalisation du processus d’expertise et

proposition d’une approche robuste de contr

ole visuel

humain. PhD thesis, Universit

e de Grenoble.

Bensma

ıa, S. and Hollins, M. (2005). Pacinian represen-

tations of ﬁne surface texture. Perception & psy-

chophysics, 67(5):842–854.

Compton, M., Barnaghi, P., Bermudez, L., Garc

ıa-Castro,

R., Corcho, O., Cox, S., Graybeal, J., Hauswirth, M.,

Henson, C., Herzog, A., Huang, V., Janowicz, K.,

Kelsey, W. D., Le Phuoc, D., Lefort, L., Leggieri,

M., Neuhaus, H., Nikolov, A., Page, K., Passant, A.,

Sheth, A., and Taylor, K. (2012). The SSN ontology

of the W3C semantic sensor network incubator group.

Journal of Web Semantics, 17:25–32.

INEVA : http://www.ineva.fr/

Crochemore, S., Vergneault, C., and Nesa, D. (2003). A

new reference frame for tactile perceptions: Senso-

tact. 5th Rose Mary Pangborn, Boston MA, USA,

pages 20–24.

De Boissieu, F. (2010). Toucher artiﬁciel

a base d’un micro-

capteur d’effort : traitement du signal et des informa-

tions associ

ees. PhD thesis, Universit

e de Grenoble.

De Rossi, D. and Scilingo, E. P. (2006). Skin-like sensor

arrays. Encyclopedia of sensors, 10:1–22.

Dey, A. K., Abowd, G. D., and Salber, D. (2001). A

conceptual framework and a toolkit for supporting

the rapid prototyping of context-aware applications.

Hum.-Comput. Interact., 16(2):97–166.

DOLCE+DnS Ultralite (2010). ontologydesignpat-

terns.org.

Dumenil-Lefebvre, A. (2006). Integration Des Aspects Sen-

soriels Dans La Conception Des Emballages En Verre

: Mise Au Point D’Un Instrument Methodologique

Partir Des Techniques D’Evaluation Sensorielle. PhD

thesis, Ecole nationale sup

erieure d’arts et m

etiers.

Gartiser, N., Zanni-Merk, C., Boullosa, L., and Casali, A.

(2014). A semantic layered architecture for analy-

sis and diagnosis of sme. In International, K., edi-

tor, Knowledge-Based and Intelligent Information &

Engineering Systems 18th Annual Conference, KES-

2014, volume 35, page 1165-1174. Procedia Com-

puter Science.

Issa, M., Schacher, L., and Adolphe, D. C. (2005). Invariant

attributes in the tactile characterization of fabrics. In

Proceeding, Fiber Society Spring Conference.

Jones, L. A. and Lederman, S. J. (2006). Human hand func-

tion. Oxford University Press.

Lederman, S. and Klatzky, R. L. (2009). Haptic perception:

A tutorial. Attention, perception & psychophysics,

71(3):1439–1459.

Maire, J. L., Pillet, M., and Baudet, N. (2013). Measure-

ment of the perceived quality of a product - Character-

ization of aesthetic anomalies. International Journal

of Metrology and Quality Engineering, 4(2):63–69.

Martin, S., Brown, W. M., Klavans, R., and Boyack,

K. W. (2011). OpenOrd: an open-source toolbox

for large graph layout. Proc. SPIE, 7868(JANUARY

2011):786806–786811.

Milton, N. (2008). Knowledge Technologies. Polimetrica,

Milano, Italy.

Myrgioti, E., Bassiliades, N., and Miliou, A. (2013). Bridg-

ing the HASM: An OWL ontology for modeling the

information pathways in haptic interfaces software.

Expert Systems with Applications, 40(4):1358–1371.

Picard, D., Dacremont, C., Valentin, D., and Giboreau, A.

(2003). Perceptual dimensions of tactile textures. Acta

Psychologica, 114(2):165–184.

Roussey, C., Pinet, F., Kang, M. A., and Corcho, O.

(2011). An Introduction to Ontologies and Ontol-

ogy Engineering. In Ontologies in Urban Develop-

ment Projects, chapter 2, pages 9–38. Springer Lon-

don, London.

San

ın, C. and Szczerbicki, E. (2009). Experience-based

knowledge representation: Soeks. Cybernetics and

Systems, 40(2):99–122.

A Smart System for Haptic Quality Control - Introducing an Ontological Representation of Sensory Perception Knowledge

Sola, C. (2007). Y a pas de mots pour le dire, il faut sentir

: D

ecrire et d

enommer les happerceptions profession-

nelles. Terrain, (49).

Summers, I. R., Irwin, R. J., and Brady, A. C. (2007). Hap-

tic discrimination of paper. In Human Haptic Percep-

tion: Basics and Applications, number DECEMBER

2007, pages 525–535. Birkh

auser Basel.

Zanni-Merk, C. (2015). Krem: A generic knowledge-based

framework for problem solving in engineering - pro-

posal and case studies. In INSTICC, editor, 7th In-

ternational Joint Conference on Knowledge Discov-

ery, Knowledge Engineering and Knowledge Manage-

ment, pages 381–388. Science and Technology Publi-

cations, Lda.

Zanni-Merk, C., Marc-Zwecker, S., Wemmert, C., and

de Bertrand de Beuvron, F. (2015). A layered archi-

tecture for a fuzzy semantic approach for satellite im-

age analysis. International Journal of Knowledge and

Systems Science, 6(2):31–56.

KEOD 2016 - 8th International Conference on Knowledge Engineering and Ontology Development