Ebook: Intelligent Environments 2018

Intelligent Environments (IEs) refer to physical spaces into which IT and other pervasive computing technology are woven and used to achieve specific goals for the user, the environment or both. IEs have the ultimate objective of enriching user experience, better manage, and increase user awareness of, that environment. The accelerating pace of today's technological developments urges the materialization of IEs with such innovative ideas and the whole community, from scientists and researchers to the general public, yearns for this.

The 14th International Conference on Intelligent Environments focuses on the development of advanced intelligent environments, as well as newly emerging and rapidly evolving topics. In the present edition, we are pleased to include in this volume the proceedings of the following workshops, which emphasize multi-disciplinary and transversal aspects of IEs, as well as cutting-edge topics:

• 2nd Workshop on Citizen Centric Smart Cities Solutions (CCSCS'18)

• 2nd International Workshop on Intelligent Systems for Agriculture Production and Environment Protection (ISAPEP'18)

• 3rd International Workshop on Smart Sensing Systems (IWSSS'18)

• 2nd International Workshop on Legal Issues in Intelligent Environments (LIIE'18)

• 1st International Workshop on Personalized Health & Intelligent Workplaces Transforming Ergonomics (pHIWTE'18)

• 4th Symposium on Future Intelligent Educational Environments and Learning (SOFIEEe'18)

• 6th International Workshop on Smart Offices and Other Workplaces (SOOW'18)

• 9th Workshop on Intelligent Environments Supporting Healthcare and Well-being (WISHWell'18)

• 7th International Workshop on the Reliability of Intelligent Environments (WoRIE'18)

In an attempt to support the technical foundations, design approaches and immerging methodologies for the above cutting-edge topics, a number of advanced tutorials were given by well-known experts in the field. The aims of these tutorials are to introduce these topics of interest to PhD students and other “early career researchers”, plus other people who are beginning to develop interests in those areas.

As it can be understood from this list, these workshops and tutorials, organized in conjunction with IE'18 main conference, provide a forum for researchers, scientists, engineers and developers to engage in many interesting, imaginative and active discussions that will engage further research in these key areas of Intelligent Environments:

• Building Interactive Environments by means of Capacitive Sensor Surfaces

• Semantic web techniques meet sensor data

• Choosing your ontologies for sensor data applications

• Intelligent Systems for Smart Building Management

• Computing in the Fog: Recent Technological Advances and Development Techniques

• Digital Object Memories in Instrumented Spaces

• Social Interaction with Cloud Network Robots

• Business Process Management approaches and techniques applied to smart environments

• From Awareness To Foundation: Human Behavior in Computing

The proceedings contain a series of contributions reflecting the latest research developments in IEs and related areas, focusing on stretching the borders of the current state of the art and contributing to an ever increasing establishment of IEs in the real world.

We would like to thank all the contributing authors, as well as the members of the Organizing Committees and Program Committees of the workshops for their highly valuable work, which contributed to the success of the Intelligent Environments 2018 event.

The Workshops and Tutorial Chairs would like to take the opportunity to thank Professor Massimo Mecella and Professor Juan Augusto, the general chairs of IE'18, and Professor Daniele Riboni and Professor Paul Lukowicz, the program chairs of IE'18, for their trust in our work, and all the other members of the IE'18 organization for the confidence they placed on us.

We are also grateful to the local staff that worked thoroughly for the success of this event. Thank you for your esteemed help, without which this event would not have been possible.

June 2018

Ioannis Chatzigiannakis, Sapienza University of Rome, Italy

Paulo Novais, Polytechnic Institute of Porto, Portugal

Workshop Chairs of IE'18

Yoshito Tobe, Aoyama Gakuin University, Japan

Oliver Amft, University Passau, Germany

Tutorial Chairs of IE'18

Any conductive surface can be turned into a sensor plane by measuring the increase in its capacitance brought about by conductive objects approaching the layer. Depending on the size, material and distance of the objects to detect, this capacitance measurement can become quite difficult as electrical perturbations in the sensor's vicinity may overshadow the usable signal. In our tutorial we present an efficient measurement hardware and appropriate filter algorithms to account for this challenge. The result is a self-calibrating sensor board, which can measure up to eight different sensor surfaces independently in real time. This offers a unique platform for creating interactive environments.

Semantic Web technologies have been gaining traction in the last decade as an important tool to enable data interoperability. They allow to represent, interlink, publish, query, and reason with heterogeneous data. The data is described using ontologies, formal definitions of the types of the entities that exist in the domain, and of relations that link them. Ontologies give formal semantics to the data, which allows for data exchange with shared and unambiguous meaning, logical reasoning, and data discovery. In addition, the Linked Data principles portray guidelines to publish semantic data on the Web, based on semantic web technologies, to ease the discoverability and reuse of data. Semantic Web technologies are used in a variety of fields, including intelligent environments, healthcare, life sciences, linguistics, and cultural heritage, among other. Ontologies are also present in industry whenever interoperability or heterogeneous data integration is required. Examples include knowledge graphs in large corporations, such as Google, Facebook, IBM, Adobe, or Yahoo. The goal of this tutorial is to present the basics on Semantic Web technologies, and Linked Data principles and best practices. The tutorial assumes no prior knowledge on the topics, and can serve as an introduction for people interested in attending the tutorial “Choosing your ontologies for sensor data applications.”

Semantics is increasingly seen as key enabler for integration of sensor data and the broader Web ecosystem. Analytical and reasoning capabilities afforded by Semantic Web standards and technologies are considered important for developing advanced applications that go from capturing observations to recognition of events, deeper insights and actions. The goal of this tutorial is to cover the fundamentals and best practices of Semantic Web technologies in a concise way, and show how can they are to model ontologies for Intelligent Environments, including notions of ontology modelling and presenting two of the most relevant ontologies for Intelligent Environments:

Managing smart buildings is a challenging task, particularly in presence of contrasting goals, such as satisfying users' needs and reducing the energy consumption. Artificial Intelligence allows to design smart buildings really capable of proactively support the users to reach their goals. Intelligent systems should be capable of exploiting the information gathered by sensors pervading the building, understanding the context, selecting the best actions to perform, and actively modifying the environment. The design of such systems is a complex task, because of the possibly wide set of functional and non-functional requirements, and the dependences among intelligent functionalities and their embodiment in the building's cyber physical space.

The concept of combining the resource-bound last-mile sensors of any Internet-of-Things-related application with computational capabilities is receiving increasing attention from researchers and practitioners. Recent technological advances in embedded devices has led to the production of small-sized heterogeneous multi-core processors that incorporate pattern machine engines on-the-chip and support the execution of power-efficient algorithms. We are now capable of analyzing the data collected from the sensors on the spot, classify the data, detect abnormal events and produce advanced alerts. The capability to locally process the data allows to transmit to the cloud infrastructure and store only the segments that correspond to an abnormal behavior. In this way the embedded device would conserve battery power and minimize memory requirements. Therefore, the so-called Fog computing approach extends the cloud computing paradigm by migrating data processing closer to production site, accelerates system responsiveness to events along with its overall awareness, by eliminating the data round-trip to the cloud. Offloading large datasets to the core network is no longer a necessity, consequently leading to improved resource utilization, protection of private and confidential information and quality of experience (QoE). Fog Computing can address the ever-increasing amount of data that is inherent in an IoT world by effective communication among all elements of the architecture.

In instrumented environments physical objects augmented with functionality and digital services play an important role. In this tutorial I plan to give an overview on existing infrastructures, that support the design and operation of such environments. We will discuss how object memories can be represented and how events related to those objects can be routed and handled. In particular, we will discuss an infrastructure that has been used for more than 5 years in the Innovative Retail Laboratory a joined research lab of a large German retailer (Globus) and the German Research Center for Artificial Intelligence. Here every item in the supermarket is considered to be augmented with information in a semantic product memory. The tutorial aims at researchers that plan to experiment and build complex intelligent environments and discusses the tools needed for those.

“Making Robots More Acceptable” – the words of Professor Gordon Chen who leads ICS (Institute for Cognitive System) at TUM. What exactly is a robot that does not give discomfort to us and we can easily accept its existence as a part of everyday life? Now we obtain capabilities to access ubiquitous information spaces and our human ability and cognitive performance will be gradually enhanced. Robots will also be integrated well into the human life and helping us naturally. They will have rich sensory perception and expressive facial signals, and are going to be social partners for us. In this tutorial, we are discussing what kind of “sociality” robots should have in human robot interactions (HRI).

Business Process Management is an established discipline that deals with the identification, discovery, analysis, (re-)design, implementation, monitoring, and controlling of business processes. In turn, the Internet of things (IoT) is the inter-networking of physical devices, vehicles, buildings, and other items embedded with electronics, software, sensors, actuators, and network connectivity enabling these objects to collect and exchange data. Up to know Business Process Management (BPM) and IoT have developed as two different fields with relatively few touch points. However, there is an increasing amount of work demonstrating the benefits of a closer link of business process management and IoT in the context of smart environments. In this tutorial I will give an overview of existing business process management approaches applied to smart environments and illustrate examples of successful combinations of BPM and IoT to enhance smart environments. Moreover, the tutorial will cover challenges that need to be addressed to better unveil this potential and outline research opportunities.

In the advancing ubiquitous computing, relationship between human users and computer systems has been rapidly evolving and getting more complicated. Through various types of users' devices, such as notebooks, smartphones, watches and wearables, and also those that embedded inside our surrounding environment, we let the computer systems “sense” information about ourselves and the physical space, and let the systems “inform” (and provide) various types of value-added and services. In this talk, I present some of our latest work on understanding human activities and even some internal status (such as mood) through various types of sensing technologies spanning from device sensors, mobile sensors, and participatory sensing. Furthermore, I will also focus on the problem of “interruption overload” that occurs when human users' are overwhelmed by an increasing amount of proactive information provisions from the computer systems, and present the latest research on “human attention management”, a research challenge actively researched by the Ubicomp community in the recent years.

Smart city capabilities are currently realized in a staggered fashion or they exist in silos. However continuous improvement and personalization is expected by citizen for sustained engagements. This is possible only if the service enabling platform is able to continuously learn about the citizen persona and her need. This enables a platform to transform the way the services are delivered to an individual citizen. The current study provides insights to develop Digital Citizen Engagement Framework (DCEF) to build a platform for citizen to personalize the engagements with smart city services. The framework is arrived at through Content Analysis, a qualitative research methodology identifying various categories and themes to manage technical components and features that need to be part of DCEF that can be deployed in a Smart City to make it future ready. Our initial experiments on an Ambient Assisted Living (AAL) use-case for geriatric care proves the effectiveness of our proposed framework.

Smart Homes are environments that automate action and adapt themselves to user behaviours. In this sense, it is necessary to employ learning strategies to allow Smart Homes to truly become intelligent, in a sense that they anticipate needs and actions. This requires constant monitoring of environments, users and their actions, as well as, non-supervised dynamic learning strategies.

The purpose of this work is to develop a system capable of taking the best action possible based on its environment. In this document, we present a reinforcement learning approach to automate lights and appliances in a Smart Home environment. An intelligent agent perceives the ambient and the past interactions of the user with the home in order to learn what is the best action to perform, which action has a certain reward associated in order to inform the agent his behavior. A reinforcement learning algorithm learns a policy for picking actions by adjusting its weights through gradient descent using feedback from the environment.

With a new industrial revolution, Industry 4.0, changing the organisational and manufacturing processes of an organisation, the increase of importance of consumers interests and preferences is more evident. Organisations face a need to understand their consumers opinion, while being able to address their index satisfaction variables. The explosion of data that we face, results in an increase of information that is present on external sources (such online social platforms), that contain valuable insights about consumers opinion regarding an organisation or, more specifically, products. This information is correlated with the smart manufacturing process, and can also affect the decision-making processes. This work presents a vision for an intelligent component that is able to construct knowledge about the feedback gathered from external sources, in order to aid or take control of decision-making processes, by determining which supplier should supply a specific material, based on consumer opinion.

Intelligent environments are complex systems that may require a diverse set of hardware devices, software libraries, networking and human computer interactions. New tools and techniques that can facilitate the engineering of such systems are thus critical. However, given the size and heterogeneity of the literature and in the light of, to our knowledge, there being only informal surveys restricted to specific issues have been conducted, we have seen the need to organise and synthesise the existent research corpus to obtain a clear idea on the main approaches that have been utilised for the development of IEs. To address this research gap, this systematic literature review was carried out. This paper presents the review's preliminary findings that are expected to provide avenues for further research in this area. We find that there are different approaches for developing IEs and the development cycle consists of several phases. However, not all phases have received equal consideration. An evaluation framework which could offer guidance on the choice of the most suitable techniques per phase should also be the target of research efforts.

Honeybees, which play an essential role as pollinators, have suffered a significant decline in recent years. Different types of sensors, including acoustic, chemical, vision, mass and temperature, can provide important information to assess their well-being. However, a multi-modal sensor system would need to be economical and affordable in order to be used on a large scale, including by less wealthy farmers or beekeepers. We present details of a low-cost sensor network system to allow the continuous monitoring of honeybee hives in a non-invasive manner, discussing its advantages relative to other existing systems for the same purpose, and initial results from the deployment of such a system in four hives.

This paper investigates the importance of various environmental factors that have a strong influence on strawberry yields grown in greenhouse using various pattern recognition methods. The environmental factors influencing the production of strawberries were six factors such as average inside temperature, average inside humidity, average CO2 level, average soil temperature, cumulative solar irradiance, and average illumination. The results of analyzing the observed data using Dynamic Time Warping (DTW) showed that the most significant factor influencing the strawberry production was average inside humidity, while it was found that average illumination was the lowest influential environmental variable. In addition, an increase in the level of co2 significantly affects the decrease in strawberry yield. Therefore, in order to increase the harvest of strawberries cultivated in the farms, it is necessary to manage the environmental factors such as thoroughly controlling the humidity and maintaining the concentration of CO2 constantly by ventilation of the greenhouse.

During times of disasters, users can act as powerful social sensors, because of the significant amount of data they generate on social media. Indeed, they contribute to creating situational awareness by informing what is happening in the affected community during the incident. In this context, this article focuses on the text-processing module in CASPER, a knowledge-based system that integrates event detection and sentiment tracking. The performance of the system was tested with the natural disaster of wildfires.

Precision agriculture has created new opportunities to solve problems in the field of agriculture by balancing investment with higher returns. This paper focuses on the problem of frost in stone fruit crops. These frosts occur due to the large temperature changes caused by climate change, which anticipate the blooming of stone fruit trees due to high temperatures at midday, but damage these flowers with temperatures below zero that occur at night in the last days of winter. The aim of this paper is to perform a preliminary study to predict, with the least possible error, the possible frosts that can occur in crops. Data for this initial study have been obtained from three meteorological stations belonging to the Murcia Institute of Agricultural and Food Research and Development. The purpose of this paper is addressed using two of the techniques offered by intelligent data analysis. Specifically, the M5P regression tree for temperature prediction and the C4.5 decision tree to classify, whether or not there is frost, have been used. Initial results are satisfactory with more than 89% accuracy in classification and an error less than 0.5 degrees Celsius in temperature prediction. In addition, the results identify the most relevant attributes to predict temperature, being some of them dew point, vapor pressure deficit and maximum relative humidity.

Piecewise segmentation approaches of plant volume estimation methods are presented. The primary 3-D reference shape of a target plant is obtained through the 3-D point cloud generation using multi-view stereo techniques. Then, the entire shape region is segmented into multiple pieces to calculate the plant volume. Two different segmentation method of i) slice-based and ii) cell-based are adopted. In the slice-based model, the entire point cloud is horizontally split into slices, whereas in the cell-based model, it is divided into small cubical cells. After that, volume estimation procedures for the two models are applied. Various experiments were performed to test validity of presented methods. Experiments to find the proper number of segments for the methods were also performed.

The increasing coverage and heterogeneity of sensor-based systems due to the low acquisition cost, ease of deployment and data collection makes the range of possible applications in the context of smart cities bigger. This makes the process of quality measurement of the generated data as well as anomaly detection some of the toughest challenges to overcome. The absence of a common mean to express general information about the sensor deployment, measurement capabilities, output data and the conditions, under which the sensor could produce anomalies, casts a big shadow over the trustworthiness of the data. The large number of low cost sensors makes the description of their context information even more difficult. We propose an architecture that enables the semi-automatic creation of context information about the sensors through the use of available information about the output data and the available information about the deployment. The definition of the sensor with its relevant context information is done by populating ontology instances for every sensor using the SSN and SWEET ontologies. We take the use case of the luftdaten.info platform with particulate matter sensors. This architecture with its implementation lay the groundwork for further steps such as including anomaly detection rules and quality measurement conditions into the sensor model. The ontology instance produced is the input used for automatic generation of data stream processing queries with data quality assessment.