2015
Report
Unknown
Human activity recognition using multisensor data fusion based on reservoir computing
Palumbo F., Gallicchio C., Pucci R., Micheli A.
Activity recognition plays a key role in providing activity assistance and care for users in smart homes. In this work, we present an activity recognition system that classifies in the near real-time a set of common daily activities exploiting both the data sampled by sensors embedded in a smartphone carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. In order to achieve an effective and responsive classification, a decision tree based on multisensor data-stream is applied fusing data coming from embedded sensors on the smartphone and environmental sensors before processing the RSS stream. To this end, we model the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks (RNNs) implemented as efficient Echo State Networks (ESNs), within the Reservoir Computing (RC) paradigm. We targeted the system for the EvAAL scenario, an international competition that aims at establishing benchmarks and evaluation metrics for comparing Ambient Assisted Living (AAL) solutions. In this paper, the performance of the proposed activity recognition system is assessed on a purposely collected real-world dataset, taking also into account a competitive neural network approach for performance comparison. Our results show that, with an appropriate configuration of the information fusion chain, the proposed system reaches a very good accuracy with a low deployment cost.Source: ISTI Technical reports, 2015
Project(s): DOREMI
Palumbo F., Gallicchio C., Pucci R., Micheli A.
Activity recognition plays a key role in providing activity assistance and care for users in smart homes. In this work, we present an activity recognition system that classifies in the near real-time a set of common daily activities exploiting both the data sampled by sensors embedded in a smartphone carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. In order to achieve an effective and responsive classification, a decision tree based on multisensor data-stream is applied fusing data coming from embedded sensors on the smartphone and environmental sensors before processing the RSS stream. To this end, we model the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks (RNNs) implemented as efficient Echo State Networks (ESNs), within the Reservoir Computing (RC) paradigm. We targeted the system for the EvAAL scenario, an international competition that aims at establishing benchmarks and evaluation metrics for comparing Ambient Assisted Living (AAL) solutions. In this paper, the performance of the proposed activity recognition system is assessed on a purposely collected real-world dataset, taking also into account a competitive neural network approach for performance comparison. Our results show that, with an appropriate configuration of the information fusion chain, the proposed system reaches a very good accuracy with a low deployment cost.Source: ISTI Technical reports, 2015
Project(s): DOREMI
See at: CNR ExploRA
2016
Journal article
Open Access
Human activity recognition using multisensor data fusion based on Reservoir Computing
Palumbo F., Gallicchio C., Pucci R., Micheli A.
Activity recognition plays a key role in providing activity assistance and care for users in smart homes. In this work, we present an activity recognition system that classifies in the near real-time a set of common daily activities exploiting both the data sampled by sensors embedded in a smartphone carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. In order to achieve an effective and responsive classification, a decision tree based on multisensor data-stream is applied fusing data coming from embedded sensors on the smartphone and environmental sensors before processing the RSS stream. To this end, we model the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks (RNNs) implemented as efficient Echo State Networks (ESNs), within the Reservoir Computing (RC) paradigm. We targeted the system for the EvAAL scenario, an international competition that aims at establishing benchmarks and evaluation metrics for comparing Ambient Assisted Living (AAL) solutions. In this paper, the performance of the proposed activity recognition system is assessed on a purposely collected real-world dataset, taking also into account a competitive neural network approach for performance comparison. Our results show that, with an appropriate configuration of the information fusion chain, the proposed system reaches a very good accuracy with a low deployment cost.Source: Journal of ambient intelligence and smart environments (Print) 8 (2016): 87–107. doi:10.3233/AIS-160372
DOI: 10.3233/ais-160372
Metrics:
Palumbo F., Gallicchio C., Pucci R., Micheli A.
Activity recognition plays a key role in providing activity assistance and care for users in smart homes. In this work, we present an activity recognition system that classifies in the near real-time a set of common daily activities exploiting both the data sampled by sensors embedded in a smartphone carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. In order to achieve an effective and responsive classification, a decision tree based on multisensor data-stream is applied fusing data coming from embedded sensors on the smartphone and environmental sensors before processing the RSS stream. To this end, we model the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks (RNNs) implemented as efficient Echo State Networks (ESNs), within the Reservoir Computing (RC) paradigm. We targeted the system for the EvAAL scenario, an international competition that aims at establishing benchmarks and evaluation metrics for comparing Ambient Assisted Living (AAL) solutions. In this paper, the performance of the proposed activity recognition system is assessed on a purposely collected real-world dataset, taking also into account a competitive neural network approach for performance comparison. Our results show that, with an appropriate configuration of the information fusion chain, the proposed system reaches a very good accuracy with a low deployment cost.Source: Journal of ambient intelligence and smart environments (Print) 8 (2016): 87–107. doi:10.3233/AIS-160372
DOI: 10.3233/ais-160372
Metrics:
See at:
Journal of Ambient Intelligence and Smart Environments 
| ISTI Repository | content.iospress.com 
| Journal of Ambient Intelligence and Smart Environments | CNR ExploRA
2016
Journal article
Open Access
Taking Arduino to the Internet of Things: The ASIP programming model
Barbon G., Margolis M., Palumbo F., Raimondi F., Weldin N.
Micro-controllers such as Arduino are widely used by all kinds of makers worldwide. Popularity has been driven by Arduino's simplicity of use and the large number of sensors and libraries available to extend the basic capabilities of these controllers. The last decade has witnessed a surge of software engineering solutions for "the Internet of Things", but in several cases these solutions require computational resources that are more advanced than simple, resource-limited micro-controllers.Surprisingly, in spite of being the basic ingredients of complex hardware-software systems, there does not seem to be a simple and flexible way to (1) extend the basic capabilities of micro-controllers, and (2) to coordinate inter-connected micro-controllers in "the Internet of Things". Indeed, new capabilities are added on a per-application basis and interactions are mainly limited to bespoke, point-to-point protocols that target the hardware I/O rather than the services provided by this hardware.In this paper we present the Arduino Service Interface Programming (ASIP) model, a new model that addresses the issues above by (1) providing a "Service" abstraction to easily add new capabilities to micro-controllers, and (2) providing support for networked boards using a range of strategies, including socket connections, bridging devices, MQTT-based publish-subscribe messaging, discovery services, etc. We provide an open-source implementation of the code running on Arduino boards and client libraries in Java, Python, Racket and Erlang. We show how ASIP enables the rapid development of non-trivial applications (coordination of input/output on distributed boards and implementation of a line-following algorithm for a remote robot) and we assess the performance of ASIP in several ways, both quantitative and qualitative.Source: Computer communications 89-90 (2016): 128–140. doi:10.1016/j.comcom.2016.03.016
DOI: 10.1016/j.comcom.2016.03.016
Metrics:
Barbon G., Margolis M., Palumbo F., Raimondi F., Weldin N.
Micro-controllers such as Arduino are widely used by all kinds of makers worldwide. Popularity has been driven by Arduino's simplicity of use and the large number of sensors and libraries available to extend the basic capabilities of these controllers. The last decade has witnessed a surge of software engineering solutions for "the Internet of Things", but in several cases these solutions require computational resources that are more advanced than simple, resource-limited micro-controllers.Surprisingly, in spite of being the basic ingredients of complex hardware-software systems, there does not seem to be a simple and flexible way to (1) extend the basic capabilities of micro-controllers, and (2) to coordinate inter-connected micro-controllers in "the Internet of Things". Indeed, new capabilities are added on a per-application basis and interactions are mainly limited to bespoke, point-to-point protocols that target the hardware I/O rather than the services provided by this hardware.In this paper we present the Arduino Service Interface Programming (ASIP) model, a new model that addresses the issues above by (1) providing a "Service" abstraction to easily add new capabilities to micro-controllers, and (2) providing support for networked boards using a range of strategies, including socket connections, bridging devices, MQTT-based publish-subscribe messaging, discovery services, etc. We provide an open-source implementation of the code running on Arduino boards and client libraries in Java, Python, Racket and Erlang. We show how ASIP enables the rapid development of non-trivial applications (coordination of input/output on distributed boards and implementation of a line-following algorithm for a remote robot) and we assess the performance of ASIP in several ways, both quantitative and qualitative.Source: Computer communications 89-90 (2016): 128–140. doi:10.1016/j.comcom.2016.03.016
DOI: 10.1016/j.comcom.2016.03.016
Metrics:
See at:
Computer Communications | ISTI Repository | Computer Communications | INRIA a CCSD electronic archive server | www.sciencedirect.com | CNR ExploRA
2016
Doctoral thesis
Unknown
Ambient intelligence in assisted living environments
Palumbo F.
The paradigm of Ambient Intelligence (AmI) aims at supporting humans in achieving their everyday objectives by enriching physical environments with networks of distributed devices, such as sensors, actuators, and computational resources. AmI is not only the convergence of various technologies (i.e. sensor networks and industrial electronics) and related research fields (i.e. pervasive, distributed computing, and artificial intelligence), but it represents a major effort to integrate them and to make them really useful for everyday human life. In particular, the recognition of human activities, coupled with the knowledge of the user's position in the indoor environment, represent two of the main pillars of the so-called "context-awareness". A context-aware system is aware of what sensory data mean: it is able to associate meaning to observations, and to make the best use of sensory data once their meaning has been assessed. In this context, one of the most important research and development areas is represented by assisted living environments. The general goal of Ambient Assisted Living (AAL) solutions is to apply ambient intelligence technologies to enable people with specific demands, e.g. with disabilities or elderly, to live longer in their preferred environment. This thesis deals with two major problems that still prevent the spreading of AAL solutions in real environments: (i) the need for a common medium to transmit the sensory data and the information produced by algorithms and (ii) the unobtrusiveness of context-aware applications in terms of both placement of devices and period of observations (i.e. long-term care offering services or assistance on a daily basis over a long period of time for people who are not independent). In order to address these challenges, this thesis contributes to the Ambient Intelligence research field, applied to assisted living environments, by means of a holistic solution composed of a beyond the state-of-the-art middleware infrastructure, providing interoperability and service abstraction, and a suite of unobtrusive applications, built on top the proposed middleware, that allows the detection of the user's context and behavioral deviations of his routine in indoor activities. The proposed solution has been thoroughly evaluated in the laboratory and in real testbeds offered by European FP7 projects, namely GiraffPlus and DOREMI, that showed its effectiveness in dealing with the requirements coming from the application of the AAL paradigm in the real world.Project(s): GIRAFF+
Palumbo F.
The paradigm of Ambient Intelligence (AmI) aims at supporting humans in achieving their everyday objectives by enriching physical environments with networks of distributed devices, such as sensors, actuators, and computational resources. AmI is not only the convergence of various technologies (i.e. sensor networks and industrial electronics) and related research fields (i.e. pervasive, distributed computing, and artificial intelligence), but it represents a major effort to integrate them and to make them really useful for everyday human life. In particular, the recognition of human activities, coupled with the knowledge of the user's position in the indoor environment, represent two of the main pillars of the so-called "context-awareness". A context-aware system is aware of what sensory data mean: it is able to associate meaning to observations, and to make the best use of sensory data once their meaning has been assessed. In this context, one of the most important research and development areas is represented by assisted living environments. The general goal of Ambient Assisted Living (AAL) solutions is to apply ambient intelligence technologies to enable people with specific demands, e.g. with disabilities or elderly, to live longer in their preferred environment. This thesis deals with two major problems that still prevent the spreading of AAL solutions in real environments: (i) the need for a common medium to transmit the sensory data and the information produced by algorithms and (ii) the unobtrusiveness of context-aware applications in terms of both placement of devices and period of observations (i.e. long-term care offering services or assistance on a daily basis over a long period of time for people who are not independent). In order to address these challenges, this thesis contributes to the Ambient Intelligence research field, applied to assisted living environments, by means of a holistic solution composed of a beyond the state-of-the-art middleware infrastructure, providing interoperability and service abstraction, and a suite of unobtrusive applications, built on top the proposed middleware, that allows the detection of the user's context and behavioral deviations of his routine in indoor activities. The proposed solution has been thoroughly evaluated in the laboratory and in real testbeds offered by European FP7 projects, namely GiraffPlus and DOREMI, that showed its effectiveness in dealing with the requirements coming from the application of the AAL paradigm in the real world.Project(s): GIRAFF+
See at: CNR ExploRA
2016
Conference article
Open Access
Implementing virtual pheromones in BDI robots using MQTT and Jason (Short Paper)
Bottone M., Palumbo F., Primiero G., Raimondi F., Stocker R
Robotic coordination is a crucial issue in the development of many applications in swarm robotics, ranging from mapping unknown and potentially dangerous areas to the synthesis of plans to achieve complex tasks such as moving goods between locations under resource constraints. In this context, stigmergy is a widely employed approach to robotic coordination based on the idea of interacting with the environment by means of markers called pheromones. Pheromones do not need to be "physical marks", and a number of works have investigated the use of digital, virtual pheromones. In this paper, we show how the concept of virtual pheromones can be implemented in Jason, a Java-based interpreter for an extended version of AgentSpeak, providing a high-level modelling and execution environment for multi-agent systems. We also exploit MQTT, a messaging infrastructure for the Internet-of-Things. This allows the implementation of stigmergic algorithms in a high-level declarative language, building on top of low-level infrastructures typically used only for controlling sensors and actuators.Source: 5th IEEE International Conference on Cloud Networking, pp. 196–199, Pisa, Italy, 3-5 October 2016
DOI: 10.1109/cloudnet.2016.22
Metrics:
Bottone M., Palumbo F., Primiero G., Raimondi F., Stocker R
Robotic coordination is a crucial issue in the development of many applications in swarm robotics, ranging from mapping unknown and potentially dangerous areas to the synthesis of plans to achieve complex tasks such as moving goods between locations under resource constraints. In this context, stigmergy is a widely employed approach to robotic coordination based on the idea of interacting with the environment by means of markers called pheromones. Pheromones do not need to be "physical marks", and a number of works have investigated the use of digital, virtual pheromones. In this paper, we show how the concept of virtual pheromones can be implemented in Jason, a Java-based interpreter for an extended version of AgentSpeak, providing a high-level modelling and execution environment for multi-agent systems. We also exploit MQTT, a messaging infrastructure for the Internet-of-Things. This allows the implementation of stigmergic algorithms in a high-level declarative language, building on top of low-level infrastructures typically used only for controlling sensors and actuators.Source: 5th IEEE International Conference on Cloud Networking, pp. 196–199, Pisa, Italy, 3-5 October 2016
DOI: 10.1109/cloudnet.2016.22
Metrics:
See at:
eprints.mdx.ac.uk | doi.org | ieeexplore.ieee.org | CNR ExploRA
2018
Report
Open Access
NESTORE - Evaluation of universAAL solution
Candea G., Staicu M., Candea C., Palumbo F.
This document is the results of work in task T6.2 Evaluation of universAAL solution. Within this task the technological partners from NESTORE reviewed and analyses the official universAAL sources and documentation, realized a Proof-of-Concept. Based on learned lessons and intersecting them with NESTORE requirements are presented conclusions and NESTORE approach relating universAAL support.Source: Project report, NESTORE, Deliverable D6.5, 2018
Project(s): NESTORE
Candea G., Staicu M., Candea C., Palumbo F.
This document is the results of work in task T6.2 Evaluation of universAAL solution. Within this task the technological partners from NESTORE reviewed and analyses the official universAAL sources and documentation, realized a Proof-of-Concept. Based on learned lessons and intersecting them with NESTORE requirements are presented conclusions and NESTORE approach relating universAAL support.Source: Project report, NESTORE, Deliverable D6.5, 2018
Project(s): NESTORE
See at:
ISTI Repository | CNR ExploRA
2019
Conference article
Open Access
A Decision Support System to Propose Coaching Plans for Seniors
Subias-Beltran P., Orte S., Vargiu E., Palumbo F., Angelini L., Abou Khaled O., Mugellini E., Caon M.
This paper presents the decision support system that has been defined and developed under the umbrella of the NESTORE project. The main goal of the proposed system is to help users in selecting coaching plans by proposing personalised recommendations based on their behaviours and preferences. Recognising such behaviours and their evolution over time is therefore a crucial element for tailoring the interaction of the system with the user. A three-layer system composed of pathways, coaching activity plans, and coaching events, constitutes the so-called coaching timeline on which the analysis is grounded. Various techniques are used to model and personalise the recommendations and feedback. Firstly, the indicators are extracted from disparate data sources, then these are modelled through a profiling system and, finally, recommendations on the pathways and coaching plans are performed through a scoring and a tagging system.Source: 32nd IEEE International Symposium on Computer-Based Medical Systems, CBMS 2019, pp. 592–595, Cordoba, Spain, 5-7/6/2019
DOI: 10.1109/cbms.2019.00123
Project(s): NESTORE
Metrics:
Subias-Beltran P., Orte S., Vargiu E., Palumbo F., Angelini L., Abou Khaled O., Mugellini E., Caon M.
This paper presents the decision support system that has been defined and developed under the umbrella of the NESTORE project. The main goal of the proposed system is to help users in selecting coaching plans by proposing personalised recommendations based on their behaviours and preferences. Recognising such behaviours and their evolution over time is therefore a crucial element for tailoring the interaction of the system with the user. A three-layer system composed of pathways, coaching activity plans, and coaching events, constitutes the so-called coaching timeline on which the analysis is grounded. Various techniques are used to model and personalise the recommendations and feedback. Firstly, the indicators are extracted from disparate data sources, then these are modelled through a profiling system and, finally, recommendations on the pathways and coaching plans are performed through a scoring and a tagging system.Source: 32nd IEEE International Symposium on Computer-Based Medical Systems, CBMS 2019, pp. 592–595, Cordoba, Spain, 5-7/6/2019
DOI: 10.1109/cbms.2019.00123
Project(s): NESTORE
Metrics:
See at:
ieeexplore.ieee.org | ISTI Repository | doi.org | CNR ExploRA
2019
Bachelor thesis
Restricted
Sleep behavior assessment via heart rate tracker devices and stigmergic receptive fields
Venturini L.
Dayly monitoring and analisys of is becoming more and more a popular and the detection of deviations of behavioural patterns is a crucial element if we want to assess the quality of daily living without disorders. The aim of this thesis is to present a sleep monitoring method based on the collection of heart rate data during sleep and a stigmergic analysis of the latter. In this regard, we present an adaptive anomaly detection method based on multi-layered stigmergic fed by a contact-free sleep tracker. We exploit heart rate data, gathered via the tracker, in order to identify subject's behavioural pattern over human's habits. This tipe of sistem allows the monitoring at home of any type of patient, to study the changes in the behavioural habits during sleep and detect anomalies. The detection of this variation allows the family or a doctor to become aware of behavioural shifting from the sleep normality of the subject and enable them to verify if this is related to the onset of disorders or diseases that could compromise the stability of the patient's lifestyle. Exploiting a technique called ballistocardiography the tracker detects the movements of the body, imparted by the ballistic forces (recoil and impact) associated with cardiac contraction and ejection of blood and with the deceleration of blood ow through the large blood vessels, and produces useful signals, one of which is the heart rate. Sample data are then processed by via computational stigmergy. Each sample is associated with a deposit of digital pheromone, called mark, dened in a mono-dimensional space and characterized by evaporation over time. Close samples are aggregated into functional structures called trails. A similarity between trails is then computed and a clustering algorithm is applied. The outcome is a similarity between days of the same subject. At the end an anomaly index ranks the anomaly level of new day or series of days.
Venturini L.
Dayly monitoring and analisys of is becoming more and more a popular and the detection of deviations of behavioural patterns is a crucial element if we want to assess the quality of daily living without disorders. The aim of this thesis is to present a sleep monitoring method based on the collection of heart rate data during sleep and a stigmergic analysis of the latter. In this regard, we present an adaptive anomaly detection method based on multi-layered stigmergic fed by a contact-free sleep tracker. We exploit heart rate data, gathered via the tracker, in order to identify subject's behavioural pattern over human's habits. This tipe of sistem allows the monitoring at home of any type of patient, to study the changes in the behavioural habits during sleep and detect anomalies. The detection of this variation allows the family or a doctor to become aware of behavioural shifting from the sleep normality of the subject and enable them to verify if this is related to the onset of disorders or diseases that could compromise the stability of the patient's lifestyle. Exploiting a technique called ballistocardiography the tracker detects the movements of the body, imparted by the ballistic forces (recoil and impact) associated with cardiac contraction and ejection of blood and with the deceleration of blood ow through the large blood vessels, and produces useful signals, one of which is the heart rate. Sample data are then processed by via computational stigmergy. Each sample is associated with a deposit of digital pheromone, called mark, dened in a mono-dimensional space and characterized by evaporation over time. Close samples are aggregated into functional structures called trails. A similarity between trails is then computed and a clustering algorithm is applied. The outcome is a similarity between days of the same subject. At the end an anomaly index ranks the anomaly level of new day or series of days.
See at:
etd.adm.unipi.it | CNR ExploRA
2019
Bachelor thesis
Restricted
Deep Learning of Sleep Quality based on Ballistocardiographic sensors, Stigmergic Perceptrons and LSTM Networks
Culcasi F. P.
The negative effects due to inadequate sleep in human beings of any age are well known. Neuroscientists and sleep experts work every day to understand how and what makes sleep more effective in its physical and mental recovery action. Among the most adopted techniques, the standard for the measurement of vital parameters on sleeping subjects is certainly Polysomnography. It is a method that involves many sensors and a laboratory environment, factors that introduce inconveniences that could negatively influence the sleep of the individual himself. In this thesis we will base the experimentation on signals obtained through Ballistocardiography, a portable and less intrusive technique that deduces the heartbeat and respiratory acts based on the accelerations of the body lying on the bed due to forces imparted by the heart to the mass of blood that is pumped to the peripheral body systems. In order to solve the problem of the parametric complexity of explicit analytical models of sleep quality in terms of "sleep architecture", in this work we propose a Deep Learning architecture based on multiple levels of Stigmergic Perceptrons. This approach describes a soft classification technique on time series with respect to a collection of archetypes, each representing a different behavioral class. Finally, the Deep Learning architecture is integrated with recurrent Long Short-Term Memory networks, known in literature for the classification of time series, above all for their ability to solve the problem of long-term dependencies over time. This technology is appropriate in order to establish a classifier able to recognize the level of quality of sleep declared by the subjects of the measurements and confirmed by a result obtained through the application of quantitative heuristics.
Culcasi F. P.
The negative effects due to inadequate sleep in human beings of any age are well known. Neuroscientists and sleep experts work every day to understand how and what makes sleep more effective in its physical and mental recovery action. Among the most adopted techniques, the standard for the measurement of vital parameters on sleeping subjects is certainly Polysomnography. It is a method that involves many sensors and a laboratory environment, factors that introduce inconveniences that could negatively influence the sleep of the individual himself. In this thesis we will base the experimentation on signals obtained through Ballistocardiography, a portable and less intrusive technique that deduces the heartbeat and respiratory acts based on the accelerations of the body lying on the bed due to forces imparted by the heart to the mass of blood that is pumped to the peripheral body systems. In order to solve the problem of the parametric complexity of explicit analytical models of sleep quality in terms of "sleep architecture", in this work we propose a Deep Learning architecture based on multiple levels of Stigmergic Perceptrons. This approach describes a soft classification technique on time series with respect to a collection of archetypes, each representing a different behavioral class. Finally, the Deep Learning architecture is integrated with recurrent Long Short-Term Memory networks, known in literature for the classification of time series, above all for their ability to solve the problem of long-term dependencies over time. This technology is appropriate in order to establish a classifier able to recognize the level of quality of sleep declared by the subjects of the measurements and confirmed by a result obtained through the application of quantitative heuristics.
See at:
etd.adm.unipi.it | CNR ExploRA
2013
Conference article
Restricted
Multisensor data fusion for activity recognition based on reservoir computing
Palumbo F., Barsocchi P., Gallicchio C., Chessa S., Micheli A.
Ambient Assisted Living facilities provide assistance and care for the elderly, where it is useful to infer their daily activity for ensuring their safety and successful ageing. In this work, we present an Activity Recognition system that classifies a set of common daily activities exploiting both the data sampled by accelerometer sensors carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. To this end, we model the accelerometer and the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks implemented as efficient Echo State Networks (ESNs), within the Reser- voir Computing paradigm. Our results show that, with an appropriate configuration of the ESN, the system reaches a good accuracy with a low deployment cost.Source: EvAAL 2013 - Evaluating AAL Systems Through Competitive Benchmarking. International Competitions and Final Workshop, pp. 24–35, Madrid-Valencia, Spain, July and September 2013
DOI: 10.1007/978-3-642-41043-7_3
Project(s): UNIVERSAAL
Metrics:
Palumbo F., Barsocchi P., Gallicchio C., Chessa S., Micheli A.
Ambient Assisted Living facilities provide assistance and care for the elderly, where it is useful to infer their daily activity for ensuring their safety and successful ageing. In this work, we present an Activity Recognition system that classifies a set of common daily activities exploiting both the data sampled by accelerometer sensors carried out by the user and the reciprocal Received Signal Strength (RSS) values coming from worn wireless sensor devices and from sensors deployed in the environment. To this end, we model the accelerometer and the RSS stream, obtained from a Wireless Sensor Network (WSN), using Recurrent Neural Networks implemented as efficient Echo State Networks (ESNs), within the Reser- voir Computing paradigm. Our results show that, with an appropriate configuration of the ESN, the system reaches a good accuracy with a low deployment cost.Source: EvAAL 2013 - Evaluating AAL Systems Through Competitive Benchmarking. International Competitions and Final Workshop, pp. 24–35, Madrid-Valencia, Spain, July and September 2013
DOI: 10.1007/978-3-642-41043-7_3
Project(s): UNIVERSAAL
Metrics:
See at:
doi.org | link.springer.com | CNR ExploRA
2014
Journal article
Open Access
Sensor network infrastructure for a home care monitoring system
Palumbo F., Ullberg J., ?timec A., Furfari F., Karlsson L., Coradeschi S.
This paper presents the sensor network infrastructure for a home care system that allows long-term monitoring of physiological data and everyday activities. The aim of the proposed system is to allow the elderly to live longer in their home without compromising safety and ensuring the detection of health problems. The system offers the possibility of a virtual visit via a teleoperated robot. During the visit, physiological data and activities occurring during a period of time can be discussed. These data are collected from physiological sensors (e.g., temperature, blood pressure, glucose) and environmental sensors (e.g., motion, bed/chair occupancy, electrical usage). The system can also give alarms if sudden problems occur, like a fall, and warnings based on more long-term trends, such as the deterioration of health being detected. It has been implemented and tested in a test environment and has been deployed in six real homes for a year-long evaluation. The key contribution of the paper is the presentation of an implemented system for ambient assisted living (AAL) tested in a real environment, combining the acquisition of sensor data, a flexible and adaptable middleware compliant with the OSGistandard and a context recognition application. The system has been developed in a European project called GiraffPlus.Source: Sensors (Basel) 14 (2014): 3833–3860. doi:10.3390/s140303833
DOI: 10.3390/s140303833
Project(s): GIRAFF+
Metrics:
Palumbo F., Ullberg J., ?timec A., Furfari F., Karlsson L., Coradeschi S.
This paper presents the sensor network infrastructure for a home care system that allows long-term monitoring of physiological data and everyday activities. The aim of the proposed system is to allow the elderly to live longer in their home without compromising safety and ensuring the detection of health problems. The system offers the possibility of a virtual visit via a teleoperated robot. During the visit, physiological data and activities occurring during a period of time can be discussed. These data are collected from physiological sensors (e.g., temperature, blood pressure, glucose) and environmental sensors (e.g., motion, bed/chair occupancy, electrical usage). The system can also give alarms if sudden problems occur, like a fall, and warnings based on more long-term trends, such as the deterioration of health being detected. It has been implemented and tested in a test environment and has been deployed in six real homes for a year-long evaluation. The key contribution of the paper is the presentation of an implemented system for ambient assisted living (AAL) tested in a real environment, combining the acquisition of sensor data, a flexible and adaptable middleware compliant with the OSGistandard and a context recognition application. The system has been developed in a European project called GiraffPlus.Source: Sensors (Basel) 14 (2014): 3833–3860. doi:10.3390/s140303833
DOI: 10.3390/s140303833
Project(s): GIRAFF+
Metrics:
See at:
Sensors | Sensors | Sensors | Sensors | urn.kb.se | CNR ExploRA
2014
Report
Unknown
CEO: a Context Event Only indoor localization technique for AAL.
Potortì F., Palumbo F.
Ambient Assisted Living applications are deployed in smart environments that provide some basic services, a typical example being user localization. AAL applications generally have low accuracy requirements for indoor localization; this opens the opportunity for parasitizing the existing smart environment infrastructure without adding dedicated positioning sensors. In this scenario, one can exploit simple binary sensors that are usually present in the smart environment, such as light and appliance switches or intrusion detection sensors, to obtain a rough estimate of the position of the user. This application is device-free, meaning that the user is not required to carry any device in order to be localized. In this paper we present CEO, a software-only system which we evaluate along the technical guidelines of the EvAAL competition. While the localization performance of CEO is lower with respect to most EvAAL competitors of past editions, it has the benefit of being non-intrusive, easy to install and perfectly compatible with other software systems: these characteristics would made it a potentially significant EvAAL competitor. While developing CEO, we only exploited the definition of the EvAAL competition environment as it was presented to competitors. The only inputs to CEO are the context events generated during the competition, which in 2012 and 2013 were limited to pressing light switches and using a stationary bicycle. We compare the performance of CEO against the results of those editions of EvAAL and show how it can be used to easily improve the performance of any EvAAL competitor.Source: ISTI Technical reports, 2014
Project(s): GIRAFF+
Potortì F., Palumbo F.
Ambient Assisted Living applications are deployed in smart environments that provide some basic services, a typical example being user localization. AAL applications generally have low accuracy requirements for indoor localization; this opens the opportunity for parasitizing the existing smart environment infrastructure without adding dedicated positioning sensors. In this scenario, one can exploit simple binary sensors that are usually present in the smart environment, such as light and appliance switches or intrusion detection sensors, to obtain a rough estimate of the position of the user. This application is device-free, meaning that the user is not required to carry any device in order to be localized. In this paper we present CEO, a software-only system which we evaluate along the technical guidelines of the EvAAL competition. While the localization performance of CEO is lower with respect to most EvAAL competitors of past editions, it has the benefit of being non-intrusive, easy to install and perfectly compatible with other software systems: these characteristics would made it a potentially significant EvAAL competitor. While developing CEO, we only exploited the definition of the EvAAL competition environment as it was presented to competitors. The only inputs to CEO are the context events generated during the competition, which in 2012 and 2013 were limited to pressing light switches and using a stationary bicycle. We compare the performance of CEO against the results of those editions of EvAAL and show how it can be used to easily improve the performance of any EvAAL competitor.Source: ISTI Technical reports, 2014
Project(s): GIRAFF+
See at: CNR ExploRA
2014
Contribution to book
Restricted
SALT: Source-Agnostic Localization Technique based on context data from binary sensor networks
Palumbo F., Barsocchi P.
Localization is a key component for many AAL systems, since the user position can be used for detecting user's activities and activating devices. While for outdoor scenarios Global Positioning System (GPS) constitutes a reliable and easily available technology, in indoor scenarios, in particular in real homes, GPS is largely unavailable. For this reason, several systems have been proposed for indoor localization. Recently, several algorithms fuse information coming from different sources in order to improve the overall accuracy in monitoring user activities. In this paper we propose a Source-Agnostic Localization Technique, called SALT, that fuses the information (coordinates) provided by a localization system with the information coming from the binary sensor network deployed within the environment. In order to evaluate the proposed framework, we tested our solution by using a previous developed heterogeneous localization systems presented at the international competition EvAAL 2013.Source: Ambient Intelligence. European Conference AmI 2014, edited by Emile Aarts, Boris de Ruyter, Panos Markopoulos, Evert van Loenen, Reiner Wichert, Ben Schouten, Jacques Terken, Rob Van Kranenburg, Elke Den Ouden, Gregory O'Hare, pp. 17–32. Berlin Heidelberg: Springer, 2014
DOI: 10.1007/978-3-319-14112-1_2
Project(s): GIRAFF+
Metrics:
Palumbo F., Barsocchi P.
Localization is a key component for many AAL systems, since the user position can be used for detecting user's activities and activating devices. While for outdoor scenarios Global Positioning System (GPS) constitutes a reliable and easily available technology, in indoor scenarios, in particular in real homes, GPS is largely unavailable. For this reason, several systems have been proposed for indoor localization. Recently, several algorithms fuse information coming from different sources in order to improve the overall accuracy in monitoring user activities. In this paper we propose a Source-Agnostic Localization Technique, called SALT, that fuses the information (coordinates) provided by a localization system with the information coming from the binary sensor network deployed within the environment. In order to evaluate the proposed framework, we tested our solution by using a previous developed heterogeneous localization systems presented at the international competition EvAAL 2013.Source: Ambient Intelligence. European Conference AmI 2014, edited by Emile Aarts, Boris de Ruyter, Panos Markopoulos, Evert van Loenen, Reiner Wichert, Ben Schouten, Jacques Terken, Rob Van Kranenburg, Elke Den Ouden, Gregory O'Hare, pp. 17–32. Berlin Heidelberg: Springer, 2014
DOI: 10.1007/978-3-319-14112-1_2
Project(s): GIRAFF+
Metrics:
See at:
doi.org | link.springer.com | CNR ExploRA
2015
Journal article
Open Access
CEO: a Context Event Only indoor localization technique for AAL
Potortì F., Palumbo F.
Ambient Assisted Living applications are deployed in smart environments that provide some basic services, a typical example being user localization. AAL applications generally have low accuracy requirements for indoor localization; this opens the opportunity for parasitizing the existing smart environment infrastructure without adding dedicated positioning sensors. In this scenario, one can exploit simple binary sensors that are usually present in the smart environment, such as light and appliance switches or intrusion detection sensors, to obtain a rough estimate of the position of the user. This application is device-free, meaning that the user is not required to carry any device in order to be localized. In this paper we present CEO, a software-only system which we evaluate along the technical guidelines of the EvAAL competition. While the localization performance of CEO is lower with respect to most EvAAL competitors of past editions, it has the benefit of being non-intrusive, easy to install and perfectly compatible with other software systems: these characteristics would made it a potentially significant EvAAL competitor. While developing CEO, we only exploited the definition of the EvAAL competition environment as it was presented to competitors. The only inputs to CEO are the context events generated during the competition, which in 2012 and 2013 were limited to pressing light switches and using a stationary bicycle. We compare the performance of CEO against the results of those editions of EvAAL and show how it can be used to easily improve the performance of any EvAAL competitor.Source: Journal of ambient intelligence and smart environments (Print) 7 (2015): 745–760. doi:10.3233/AIS-150343
DOI: 10.3233/ais-150343
Project(s): GIRAFF+
Metrics:
Potortì F., Palumbo F.
Ambient Assisted Living applications are deployed in smart environments that provide some basic services, a typical example being user localization. AAL applications generally have low accuracy requirements for indoor localization; this opens the opportunity for parasitizing the existing smart environment infrastructure without adding dedicated positioning sensors. In this scenario, one can exploit simple binary sensors that are usually present in the smart environment, such as light and appliance switches or intrusion detection sensors, to obtain a rough estimate of the position of the user. This application is device-free, meaning that the user is not required to carry any device in order to be localized. In this paper we present CEO, a software-only system which we evaluate along the technical guidelines of the EvAAL competition. While the localization performance of CEO is lower with respect to most EvAAL competitors of past editions, it has the benefit of being non-intrusive, easy to install and perfectly compatible with other software systems: these characteristics would made it a potentially significant EvAAL competitor. While developing CEO, we only exploited the definition of the EvAAL competition environment as it was presented to competitors. The only inputs to CEO are the context events generated during the competition, which in 2012 and 2013 were limited to pressing light switches and using a stationary bicycle. We compare the performance of CEO against the results of those editions of EvAAL and show how it can be used to easily improve the performance of any EvAAL competitor.Source: Journal of ambient intelligence and smart environments (Print) 7 (2015): 745–760. doi:10.3233/AIS-150343
DOI: 10.3233/ais-150343
Project(s): GIRAFF+
Metrics:
See at:
ISTI Repository | content.iospress.com | Journal of Ambient Intelligence and Smart Environments | CNR ExploRA
2015
Report
Unknown
GiraffPlus - System integration, final system and live demonstration
La Rosa D., Palumbo F.
This document reports on the final integrated system at M36, which has been deployed in the testsites. It includes a brief description of the fifteen testsites, covering the particularities and technological issues experienced. The final set of hardware and software components are described, remarking the main improvements with respect to previous versions, while in-deep details are given in the correspondent deliverables from WP2, WP3, and WP4. The final prototype is the last evolution along this three years project and has been iteratively adapted to the users' feedbacks. The last users' impressions gathered by an on-line bug tracker tool during the last year have raised the enhancements integrated in the final prototype and described in this document.Source: ISTI Technical reports, 2015
Project(s): GIRAFF+
La Rosa D., Palumbo F.
This document reports on the final integrated system at M36, which has been deployed in the testsites. It includes a brief description of the fifteen testsites, covering the particularities and technological issues experienced. The final set of hardware and software components are described, remarking the main improvements with respect to previous versions, while in-deep details are given in the correspondent deliverables from WP2, WP3, and WP4. The final prototype is the last evolution along this three years project and has been iteratively adapted to the users' feedbacks. The last users' impressions gathered by an on-line bug tracker tool during the last year have raised the enhancements integrated in the final prototype and described in this document.Source: ISTI Technical reports, 2015
Project(s): GIRAFF+
See at: CNR ExploRA
2019
Report
Open Access
Nestore - D4.2 - First prototype of the DSS
Orte S., Subias P., Dauwalder S., Roecke C., Guye S., Palumbo F., Rizzo G.
The D4.2 is the software that conforms the first prototype of the Decision Support System. As the DSS is merely designed and implemented in form of a Software as a Service platform, and it does not have any graphical user interface, this document is intended to report a description of the DSS first prototype main features with a particular focus on the architecture, functionalities and technical implementation. Therefore, the aim of this document is to provide a picture of the actual development of the DSS starting from the scientific background from which it is grounded and going through the different elements that form the DSS. The DSS main objective is to help users in selecting coaching plans by proposing personalised recommendations based on users' behaviours and preferences. Recognising such behaviours and their evolution over time is therefore a crucial element for tailoring the interaction of the system with the user. A three-layer system composed of pathways, coaching activity plans, and coaching events, constitutes the so-called coaching timeline on which the analysis is grounded. Various techniques are used to model and personalise the recommendations and feedback. Firstly, the indicators are extracted from disparate data sources, then these are modelled through a profiling system and, finally, recommendations on the pathways and coaching plans are performed through a tagging system. With the aim of developing and testing the models and workflow prior to the pilot starting date, two simulators are also being implemented and reported in this document.Source: Project report, Nestore, Deliverable D4.2, 2019
Project(s): NESTORE
Orte S., Subias P., Dauwalder S., Roecke C., Guye S., Palumbo F., Rizzo G.
The D4.2 is the software that conforms the first prototype of the Decision Support System. As the DSS is merely designed and implemented in form of a Software as a Service platform, and it does not have any graphical user interface, this document is intended to report a description of the DSS first prototype main features with a particular focus on the architecture, functionalities and technical implementation. Therefore, the aim of this document is to provide a picture of the actual development of the DSS starting from the scientific background from which it is grounded and going through the different elements that form the DSS. The DSS main objective is to help users in selecting coaching plans by proposing personalised recommendations based on users' behaviours and preferences. Recognising such behaviours and their evolution over time is therefore a crucial element for tailoring the interaction of the system with the user. A three-layer system composed of pathways, coaching activity plans, and coaching events, constitutes the so-called coaching timeline on which the analysis is grounded. Various techniques are used to model and personalise the recommendations and feedback. Firstly, the indicators are extracted from disparate data sources, then these are modelled through a profiling system and, finally, recommendations on the pathways and coaching plans are performed through a tagging system. With the aim of developing and testing the models and workflow prior to the pilot starting date, two simulators are also being implemented and reported in this document.Source: Project report, Nestore, Deliverable D4.2, 2019
Project(s): NESTORE
See at:
ISTI Repository | CNR ExploRA
2019
Report
Open Access
Nestore - D4.3 - Plans and recommendations based on profiles
Orte S., Subías P., Fernández L., Palumbo F., Girolami M., Sykora M.
This document represents the deliverable D4.3 (Plans and recommendations based on profiles) of WP4 (NESTORE Decision Support System) and contains a description of the research carried out in the context of task 4.3 "Algorithms for modelling and profiling individuals". In order to preserve the Intellectual Property generated in this task, we have split the D4.3. in two parts: a public and a private document. We encourage the reader to ask for the private version of D4.3. to get the complete understanding of the development carried out in D4.3.; this public version only outlines what has been done. The purpose of this report is to: - characterize NESTORE users through Personas to create end-users' models; - explain the scope of user profiling implementation in the DSS; - demonstrate its functionality with examples; - present the methodology followed; - describe the personalization through recommendations; - explain users' models; - present the objectives and implementation details of a profiler simulator. In chapter 1, we introduce the Decision Support System and its main components with the aim of presenting the user profiling as the main element of personalization. In chapter 2, the user profiling process that has been carried out is listed. In chapter 3, we describe why and how Personas are designed going through all domains of NESTORE interest, and we present the approach we followed to integrate this concept in the profiling system. Chapter 4 introduces and explains the tagging system, which is the core of the recommendation system. In chapter 5, we introduce the recommendations proposed by domain experts and depict their integration into the system. Finally, chapter 6 describes how users are modelled and simulated.Source: Project report, Nestore, Deliverable D4.3, 2019
Project(s): NESTORE
Orte S., Subías P., Fernández L., Palumbo F., Girolami M., Sykora M.
This document represents the deliverable D4.3 (Plans and recommendations based on profiles) of WP4 (NESTORE Decision Support System) and contains a description of the research carried out in the context of task 4.3 "Algorithms for modelling and profiling individuals". In order to preserve the Intellectual Property generated in this task, we have split the D4.3. in two parts: a public and a private document. We encourage the reader to ask for the private version of D4.3. to get the complete understanding of the development carried out in D4.3.; this public version only outlines what has been done. The purpose of this report is to: - characterize NESTORE users through Personas to create end-users' models; - explain the scope of user profiling implementation in the DSS; - demonstrate its functionality with examples; - present the methodology followed; - describe the personalization through recommendations; - explain users' models; - present the objectives and implementation details of a profiler simulator. In chapter 1, we introduce the Decision Support System and its main components with the aim of presenting the user profiling as the main element of personalization. In chapter 2, the user profiling process that has been carried out is listed. In chapter 3, we describe why and how Personas are designed going through all domains of NESTORE interest, and we present the approach we followed to integrate this concept in the profiling system. Chapter 4 introduces and explains the tagging system, which is the core of the recommendation system. In chapter 5, we introduce the recommendations proposed by domain experts and depict their integration into the system. Finally, chapter 6 describes how users are modelled and simulated.Source: Project report, Nestore, Deliverable D4.3, 2019
Project(s): NESTORE
See at:
ISTI Repository | CNR ExploRA
2019
Contribution to conference
Open Access
Preface to the short paper proceedings of the tenth international conference on indoor positioning and indoor navigation - Work-in-Progress (WiP) Papers
Potortì F., Renaudin V., Òkeefe K., Palumbo F.
This volume contains the Work-in-Progress papers presented at IPIN 2019, the Tenth International Conference on Indoor Positioning and Indoor Navigation (http://ipin2019.isti.cnr.it/posters) held in Pisa, Italy, on September 30th-October 3rd, 2019.Source: Aachen: CEUR-WS.org, 2019
Potortì F., Renaudin V., Òkeefe K., Palumbo F.
This volume contains the Work-in-Progress papers presented at IPIN 2019, the Tenth International Conference on Indoor Positioning and Indoor Navigation (http://ipin2019.isti.cnr.it/posters) held in Pisa, Italy, on September 30th-October 3rd, 2019.Source: Aachen: CEUR-WS.org, 2019
See at:
ceur-ws.org | ISTI Repository | CNR ExploRA
2019
Bachelor thesis
Restricted
Using contactless bed sensors for ballistocardiographic identification of sleep stages
Worku F. F.
The aim of this thesis is to analyze and identify sleep stages using ballistocardiography(BCG) sleep monitoring method. Identifying sleep stages by studying the behaviour of heart rate respiration rate, and movement properties is one of the important indicator of sleep problems. There is no doubt the consequence of Sleep problem has a huge impact on our mental , physical health, weight gain, and productivity. To analysis sleep problems golden standard method is Polysomnography(PCG). However, PCG method is expensive, intrusive ,complicated and measured in controlled environment. Due to this, for future use it is important to analysis non-intrusive long-term physiological monitoring ballistocardiography(BCG) method. In this thesis, ballistocardiography monitoring method is used to analyze and identify sleep stages.In BCG-based monitoring system developed to measure heart rate, respiratory rate, heart rate variability and movements, and able to drive other important parameters like respiratory rate variability and respiratory depth which are important to analyze and identify sleep stages. Nowadays, there are a different kind of BCG sensors are available in the markets.For this thesis, BCG sensor are selected based on the parameters they provide and accuracy. In this BCG-based monitoring system record of Murata sensor parameters used as to identify sleep stages and as a ground truth inferred Emfit qs sensor record.By studying behavior of parameters each stage and by performing k-means clustering able to identify each stage and ,furthermore experiments are performed to get better accuracy using classification on data by inferring as a ground truth Emfit qs sensor record. At the end all those experiment result are discussed.
Worku F. F.
The aim of this thesis is to analyze and identify sleep stages using ballistocardiography(BCG) sleep monitoring method. Identifying sleep stages by studying the behaviour of heart rate respiration rate, and movement properties is one of the important indicator of sleep problems. There is no doubt the consequence of Sleep problem has a huge impact on our mental , physical health, weight gain, and productivity. To analysis sleep problems golden standard method is Polysomnography(PCG). However, PCG method is expensive, intrusive ,complicated and measured in controlled environment. Due to this, for future use it is important to analysis non-intrusive long-term physiological monitoring ballistocardiography(BCG) method. In this thesis, ballistocardiography monitoring method is used to analyze and identify sleep stages.In BCG-based monitoring system developed to measure heart rate, respiratory rate, heart rate variability and movements, and able to drive other important parameters like respiratory rate variability and respiratory depth which are important to analyze and identify sleep stages. Nowadays, there are a different kind of BCG sensors are available in the markets.For this thesis, BCG sensor are selected based on the parameters they provide and accuracy. In this BCG-based monitoring system record of Murata sensor parameters used as to identify sleep stages and as a ground truth inferred Emfit qs sensor record.By studying behavior of parameters each stage and by performing k-means clustering able to identify each stage and ,furthermore experiments are performed to get better accuracy using classification on data by inferring as a ground truth Emfit qs sensor record. At the end all those experiment result are discussed.
See at:
etd.adm.unipi.it | CNR ExploRA
2020
Master thesis
Restricted
A survey on the use of 802.11 Channel State Information in device-free applications: indoor localization and human activity and gesture recognition
Uccheddu M. C.
It presents a survey on device-free applications using 802.11n Channel State Information (CSI). The survey analyzed device-free indoor localization works, human activity recognition works and gesture recognition works. For each work are described the system setting, the experimental environments and finally the evaluation. There is also the description of my personal implementation of a device-free indoor signal-based system setting, that was deployed at Consiglio Nazionale delle Ricerche (CNR) in Pisa.
Uccheddu M. C.
It presents a survey on device-free applications using 802.11n Channel State Information (CSI). The survey analyzed device-free indoor localization works, human activity recognition works and gesture recognition works. For each work are described the system setting, the experimental environments and finally the evaluation. There is also the description of my personal implementation of a device-free indoor signal-based system setting, that was deployed at Consiglio Nazionale delle Ricerche (CNR) in Pisa.
See at:
etd.adm.unipi.it | CNR ExploRA