M. Chang et al. (Eds.): Edutainment 2011, LNCS 6872, pp. 292–299, 2011.
© Springer-Verlag Berlin Heidelberg 2011
Learning from Internet of Things for Improving
Environmentally Responsible Behavior
Jun Hu, Bram van der Vlist, Gerrit Niezen, Willem Willemsen,
Don Willems, and Loe Feijs
Department of Industrial Design, Eindhoven University of Technology, NL
{j.hu,b.j.j.v.d.vlist,g.niezen}@tue.nl
{w.willemsen,d.a.j.willems.1}@student.tue.nl, l.m.g.feijs@tue.nl
Abstract. We present two designs in the area of Internet of Things, utilizing an
ontology-driven platform, namely Smart-M3, to connect domestic objects in the
physical world to the information world, for coaching the behavior or raising
the awareness in domestic energy consumption. The concept and architecture of
Smart-M3 are introduced, in which the domestic objects are knowledge
processors connected to the semantic information broker that contains the
ontologies, using a blackboard design pattern and semantic web technologies,
enabling the interoperability among both digital and physical entities. Two
designs based on Smart-M3 are presented, as examples for coaching with
and learning from the Internet of Things. Although both designs are in the area
of domestic energy consumption, they can be seen as good starting points
towards broader areas of ubiquitous learning with the Internet of Things.
Keywords: Internet of Things, ontology, semantic web, ubiquitous learning.
1 Introduction
The environments that people inhabit are occupied with a growing number of digital
and networked devices. We have not yet succeeded in seamlessly operating among
these devices. Especially when we consider the way user interaction was envisioned
in paradigms like Ambient Intelligence [1], Pervasive Computing, Ubiquitous
Computing [10] and the more recent notion of Internet of Things [8]. One of the key
goals of these paradigms is “serendipitous interoperability”, where devices which
were not necessarily designed to work together (e.g. built for different purposes by
different manufacturers at different times) should be able to discover each other’s
functionality and be able to make use of it [9].
One solution to solving the interoperability problem at the infra-structure level is a
software platform developed within the SOFIA project1. SOFIA (Smart Objects For
Intelligent Applications) is a European research project within the ARTEMIS
framework that attempts to make information in the physical world available for
smart services – connecting the physical world with the information world. Rather
than promoting the compatibility within one specific service-level solution in terms of

1
www.sofia-project.eu

Learning from Internet of Things for Improving Environmentally Responsible Behavior 293
protocols or software stacks, it addresses information-level compatibility and the
collaboration between different producers and consumers of information on a more
abstract level. It does not add, nor require an additional single service-level
infrastructure or middleware that all manufacturers must adopt, but builds on what is
already available in the industry.
The Internet of Things is often referred to as a network of RFID tagged everyday
objects. In the context of SOFIA, the Internet of Things is much more. It is a network
of smart objects which each have (from very limited to extensive) computational
power and connectivity. These smart objects form are from the “internet”, in an
environment or across environments, which makes it close to the concept of
ubiquitous computing, and creates new possibilities for innovative applications.
In this paper we present two designs that utilize these new possibilities, for
coaching the behavior or raising the awareness in domestic energy consumption. A
study by Burgess et al [2] indicates that of all the energy that people consume, 30% is
consumed in a domestic setting. Furthermore it states that, roughly 30% of domestic
energy consumption which can be attributed to behavioral choices. By stimulating
more Environmentally Responsible Behavior (ERB), the consumption can be reduced
by up to 10%. Instead of actively and explicitly teaching and coaching, the stimuli
and advices are woven into everyday objects that are connected to the energy
consumption information, for people to learn from the Internet of Things to stimulate
their ERB.
Next the basic structure and concept of SOFIA platform are explained, followed by
two design cases. Each case is presented with the design concept and prototype, as
well as the feedback collected from the user evaluations.
2 SOFIA Smart-M3
The SOFIA software platform utilizes the blackboard architectural pattern to share
information between smart devices, rather than have the devices explicitly send
messages to one another. When this information is also stored according to some
ontological representation, it becomes possible to share information between devices
that do not share the same representation model, using the semantics of that
information [5].
Ontologies are used to enable the exchange of information without requiring up-front
standardization. The first core component of the SOFIA platform is called Smart-M3
and an open source implementation is available online2. A notable feature of the SOFIA
platform is the capability to subscribe to changes of data (stored as triples) in the data
store, and be notified every time these triples are updated, added or removed.
Smart-M3 takes the blackboard and publish/subscribe concepts and implements
them in a lightweight manner suitable for small, mobile devices. These devices
(Knowledge Processors or KPs), can operate autonomously and anonymously by
sharing information through an information store. The Semantic Information Broker
(SIB) is the information store of the smart space, and contains the blackboard,

2
Available from http://sourceforge.net/projects/smart-m3/

294 J. Hu et al.

Fig. 1. SOFIA infrastructure model
ontologies and required service interfaces for
the KPs. Fig. 1 shows a simplified overview
of the Smart-M3 infrastructure.
For applications, a Description Logic (DL)
based ontology can be created in OWL, the
Web Ontology Language [9]. In the current
ontology all user interaction within the system
is described in terms of interaction events [4].
To enable our semantic connections
interaction model (introduced in more detail
after this section), the connections between the devices need to be modeled.
A connectedTo relationship can be added or removed between two existing devices in
the ontology. It should be noted that this relationship is both symmetric and
irreflexive. A symmetric property is its own inverse, which means that if we indicate
a connectedTo relationship from device A to device B, device B will also have
a connectedTo relationship to device A. Another way to think of symmetric properties
is that they are bidirectional relationships. An irreflexive property is a property that
never relates an individual to itself. This allows us to restrict our model by not
allowing a connectedTo relationship from a device to itself. In our application to
energy streams, other properties such as transitivity and additivity can be added: if
device A has access to energy from B and B can take energy from C then indirectly A
has access to energy from C (transitivity). For a node which is neither a source nor a
sink, the sum of the incoming energy flows equals the sum of the outgoing energy
flow (additivity).
In this structure, to determine which other smart objects a specific device with
a deviceID is connected to, a simple SPARQL query suffices:



To get the last event belonging to a specific device, for example the event triggered
by Near Field Communication (NFC) when the device comes close to the other, the
SPARQL query is a little bit more complex, but still surprisingly manageable
(see [4] for more details):



At a more conceptual level the term “semantic connections” is used in the SOFIA
project, referring to meaningful connections and relationships between entities of the
Internet of Things [6, 7]. Semantic connections exist in both the physical and the digital
world. They have informative properties, i.e. they are perceivable in the physical world

Learning from Internet of Things for Improving Environmentally Responsible Behavior 295
and have sensory qualities that inform users about their uses. However, these physical
qualities might be hidden at some times, or only accessed on-demand, by a special
purpose interaction device. The digital counterparts of semantic connections are
modeled in an ontology. There may be very direct mappings, e.g. a connection between
two real-world entities may be modeled by a connectedTo relationship between the
representations of these entities in an ontology.
3 Design Cases
Based on the concepts from SOFIA and Smart-M3, two products are designed in the
application area of stimulating people to improve their ERB in domestic
environments, i.e, their home environments. These environments are conceptually
smart spaces enabled by SOFIA smart objects. Energy consuming appliances in these
environments are KP-enabled objects that are connected to the SIB hence they are
interconnected, providing energy consumption status and history to the SIB and
accepting and reacting on queries, events and commands from the SIB.
3.1 Doormate
Concept. The Doormate is first of all a doormat for wiping your feet, but also a
coaching mate supporting lowering of electricity consumption [11]. It does the latter
by communicating information through an integrated low resolution LED display
(Fig. 2). The Doormate gathers data from the smart appliances in the home, such as
time of use, frequency of use, intensity of the appliance during use and duration of
use. By combining this data and evaluating changes, information on improving usage
behavior can follow. It allows people to easily turn of devices when leaving the house
as well as improving their energy consumption behavior by learning from tailored
coaching when returning home.

Fig. 2. Doormate integrated with a low resolution LED display
Fig. 3 illustrates the interface. To switch off the depicted device the user steps with
one foot on the lit up power icon (top left) and with the other foot applies pressure on
the display, as if putting out a cigarette. If more devices are available to be switched
off, the arrows will light up and can be used to scroll through the icons. When
entering the house the user can spend a moment to learn (or get a cue to remember)
how certain behavior can be changed to be more energy efficient. As the contact time
between product and user is longer, animations are used to explain the coaching tips.
If the user does not understand the animation, he or she can get more information later

296 J. Hu et al.

Fig. 4. Doormate scenarios
on his/her smart phone or laptop by both pressing the lit up coaching icon (top right)
as well as the display. In the case of the coaching state the user is able to ‘flip’
through tips if more are advised.

Fig. 3. Doormate Interface
Fig. 4 shows two use
case scenarios. The top
scenario shows how a user
forgot to switch of the
lights and is reminded by
the Doormate. He then
decides to switch them off
using the Doormate. The
bottom scenario shows how
a person who is coming
home is detected and, while
he is taking off his coat, is
shown an animation on
how to be more electricity
efficient. If he does not
understand it immediately,
he can press the Doormate
to receive more information
on his smart phone.

Evaluation. Preliminary
tests were done with seven
participants. The prototype
was used to test three
aspects: visibility and
general understanding,
animation and icon
interpretation, and
preference in initial
coaching display. The effect of the light coming from a doormat, which is generally a
very uninteresting and low value object, surprised them and gave the mat more value.

Learning from Internet of Things for Improving Environmentally Responsible Behavior 297

Fig. 5. Bonsai Garden
All participants were enthusiastic about the control functionality, as they all recognized
the situation where they forgot to turn off appliances. The results and reactions on the
Doormate are promising. People recognize the benefits and see themselves using the
Doormate over a longer period of time.
3.2 Bonsai Garden
Concept. The product consists of
local feedback devices and a central
feedback device. The local feedback
devices give direct feedback to the
user on their consumption and the
central feedback device gives overall
feedback on ERB of the different
people in a household. The overall
feedback is represented by a tree and
the trees are placed together in a
“bonsai garden” (Fig. 5). These trees
consist of building blocks and each individual user can construct their personal tree in
any way they want. There are three different kinds of building blocks (straight, angled
and split pieces) and they provide endless building possibilities. The amount of
building blocks and thus the size of the tree represents the user’s personal effort on
reducing resource consumption. The user can earn building blocks with good ERB
and direct feedback on ERB is given by the local feedback devices (triggers). These
triggers show when the user earns points for ERB by changing shape and standing
upright. These points are represented by lights in the building blocks for that person’s
tree. When all building blocks are lit up, the user can add a piece to it.

Fig. 6. Gaming elements in Bonsai Garden
The target group of this design is families with young children (8 - 12 years old).
The involvement of the entire family adds to the social aspect of motivation. The
design contains game elements, derived from on the work of Chatfield [3]. These
elements were implemented in this project to create motivation for ERB (Fig. 6): 1)
gaining levels: the size of the tree represents the level of good ERB from a person; 2)

298 J. Hu et al.
long and short term goals: The trigger is short-term, the tree represent a long-term
goal; 3) always reward effort: users get rewarded for trying to behave well; 4) rapid,
clear and frequent feedback: a trigger responds to each resource consuming event; 5)
an element of uncertainty: users do not know what kind of building block they will
get next; 6) Involving other people: users can compare their trees and compete for the
best building results.

Evaluation. The prototype was used to test two aspects: motivation through
competition, and motivation through personalization. The evaluation was done with
five children from the target user group (Fig. 7). These children were all Dutch and
either in the final years of primary or the first years of secondary school. The
evaluation was performed in a home situation and the results were recorded with a
camera and by taking notes of events and comments. The evaluation was performed
with a prototype of the tree that allowed the participants to build a tree out of building
blocks. This prototype consisted of a base unit and 30 building blocks, which allowed
for complete freedom to build a unique tree.

Fig. 7. Children build their bonsai trees
The test started with an introduction to the design and how the participants could
build their own tree later in the test. The competition element was evaluated by
having each participant build their own tree and compare them in the final discussion.
The next step was a questionnaire about ways to improve ERB. Each right answer
would result in a point, and for each point a participant would get two building blocks
to build a tree with. In the discussion the nicest tree would be chosen by voting.
The results show that building the trees was a fun experience for all the children. It
was a social process, where they advised and commented on each other’s trees. Every
participant tried to make their tree unique and as different from the others as possible.
The prospect of earning building blocks and building their own tree was a big
motivation for the children and they were very concentrated on thinking of ways to
improve ERB.

Learning from Internet of Things for Improving Environmentally Responsible Behavior 299
4 Concluding Remarks
The SOFIA Smart-M3 platform enables the possibility to embed intelligence into
everyday objects and allows these objects to connect to each other and to information
entities and services, bridging different products and services from different
manufactures and providers. Two products are designed based on the concepts from
this platform for improving people’s ERB in energy consumption in domestic home
environments, implementing different learning strategies. The Doormate provides the
convenience of controlling the house appliances at the same time provide behavior
coaching, while the Bonsai Garden tries to raise the awareness by employing gaming
elements in the design. Although the Internet of Things is limited in one environment,
the idea of providing ubiquitous learning with smart daily objects seems to be
promising. In addition to smart home environments, in the SOFIA project we are also
experimenting with different scenarios such as personal spaces and smart city. The
technology can be applied for ubiquitous learning to a broader extent.
References
1. Aarts, E., Marzano, S.: The New Everyday: Views on Ambient Intelligence. 010
Publishers, Rotterdam, The Netherlands (2003)
2. Burgess, J., Nye, M.: Re-materialising energy use through transparent monitoring systems.
Energy Policy 36(12), 4454–4459 (2008)
3. Chatfield, T.: Fun Inc.: Why games are the 21st Century’s most serious business. Virgin
Books (2010)
4. Niezen, G., van der Vlist, B., Hu, J., Feijs, L.: From events to goals: Supporting semantic
interaction in smart environments. In: 2010 IEEE Symposium on Computers and
Communications (ISCC), pp. 1029–1034 (22-25, 2010)
5. Oliver, I., Honkola, J.: Personal semantic web through a space based computing
environment. In: Second IEEE International Conference on Semantic Computing,
Middleware for the Semantic Web, Santa Clara, CA, USA (August 2008),
http://arxiv.org/pdf/0808.1455
6. van der Vlist, B., Niezen, G., Hu, J., Feijs, L.: Design semantics of connections in a smart
home environment. In: Design and Semantics of Form and Movement (DeSForM 2010),
Lucerne, Switzerland, pp. 48–56 (2010)
7. van der Vlist, B., Niezen, G., Hu, J., Feijs, L.: Semantic connections: Exploring and
manipulating connections in smart spaces. In: 2010 IEEE Symposium on Computers and
Communications (ISCC), pp. 1–4. IEEE, Riccione (2010)
8. Van Kranenburg, R., Dodson, S.: The Internet of Things: A critique of ambient technology
and the all-seeing network of RFID. Institute of Network Cultures (2008)
9. W3C: OWL Web Ontology Language Use Cases and Requirements (2004),
http://www.w3.org/TR/webont-req/
10. Weiser, M.: The computer for the 21st century. Scientific American (September 1991)
11. Willems, D.: The doormate: Supporting people in decreasing their electricity consumption.
M11 report, Theme Realities, Department of Industrial Design. Eindhoven University of
Technology, Edinhoven (2011)