Recently, with the spread of smart houses, the smartness of housing equipment and home appliances has progressed, and the functionality and usability of interfaces between people and equipment and between people and home appliances have become important factors. Currently, the main interfaces are remote controls, smartphone applications, and even voice recognition. Furthermore, research is also being conducted on interfaces that can be operated without having the device at hand using cameras and radio waves. However, special equipment must be installed for operation, and compatibility with room design has become an issue. In this research, we proposed a system to transform existing furniture into an interface rather than providing a new interface. The proposed system focused on vibration sensors that are small, inexpensive, and can be attached to existing furniture or hidden from view. To evaluate the proposed system, an experiment was conducted to transform existing furniture into an interface for swiping by simply attaching the vibration sensor to the existing furniture. Specifically, the system attaches four vibration sensors with synchronized output signals to a table and uses a CNN to learn the vibration data obtained from the sensors to predict the direction of the swipe. As a result, when the table and person swiping were fixed, the system could predict the swipe with an accuracy of over 0.86.
Augmented reality (AR) devices have gained a lot of attention in recent years due to their ability to enhance people’s abilities through 2D/3D spatial sensing and recognition functions. RGB cameras are most often used as sensors in this type of spatial recognition, and a particularly important task is the detection and tracking of objects and people in physical space. However, the camera positions and orientations on AR devices such as smartphones and smart glasses, frequently change due to the user wearing them on their head, leading to non-linear and complex motion in the video frames and reducing the accuracy of tracking people. To address this issue, the proposed method combines person re-identification based on deep metric learning with trajectory prediction to estimate the person’s sequential positions in 3D space around the camera. The experimental result shows 95.45% accuracy with our dataset.
The management of road traffic incidents is a problem faced by governments in many countries. Normally, road operators have the infrastructure in place to monitor such incidents, albeit in a reactive manner. In Spain, there are traffic cameras on major roads to check for possible incidents, however, incident notification is slow and not automated. As an alternative, this paper proposes a system for automatic real-time traffic alerts. Thus, 1,500 camera images from the Dirección General de Tráfico (DGT) deployed on the main Spanish roads are analyzed in real time every 4 minutes. These images are not preprocessed, they have different qualities and are also affected by weather conditions such as fog, rain, sun reflections, etc. The system uses several Deep Learning classification models trained on a well-known dataset of traffic images including flowing traffic, dense traffic, accidents and fires. These models are used to classify the DGT images in real time, with satisfactory initial results, detecting both flowing traffic and dense traffic.
Proximity detection is the process of estimating the closeness between a target and a point of interest, and it can be estimated with different technologies and techniques. In this paper we focus on how detecting proximity between people with a TinyML-based approach. We analyze RSS values (Received Signal Strength) estimated by a micro-controller and propagated by Bluetooth’s tags. To this purpose, we collect a dataset of Bluetooth RSS signals by considering different postures of the involved people. The dataset is adopted to train and test two neural networks: a fully-connected and an LSTM model that we compress to be executed directly on-board of the micro-controller. Experimental results conducted over the dataset show an average precision and recall metrics of 0.8 with both of the models, and with an inference time less than 1 ms.
IoT technology has spread throughout the world and many applications have emerged that utilize large numbers of sensors. The concept of a sensor cloud, in which sensors are shared, has also emerged, and there is a need to manage and operate a large number of distributed sensors properly. However, in order for sensors to measure accurate values, periodic calibration is necessary, and performing this for all sensors is very costly and impractical. An automatic calibration method has been proposed as a method to calibrate a large number of sensors at once, taking advantage of the fact that the measurements of neighboring sensors take close values. However, these methods aim to find the optimal correction value, and it is impossible to know how accurate the correction value is. In this study, we propose a new automatic calibration method to keep the errors of all sensors sufficiently small. The proposed method uses a probability distribution to estimate the magnitude of error for each sensor, and periodically calibrates selected sensors based on this distribution to maintain the overall sensor accuracy at the required level. Through simulation evaluation, it is shown that the proposed method can maintain the accuracy of the sensors.
Human activities represent a major source of information for smart home automation. While performing their daily activities, humans trigger sensors producing measurements that flow into a sensor log. Vast majority of techniques to recognize and exploit the occurrences of human activities are supervised, requiring the log to be manually labeled in correspondence of the onset and the end of each activity repetition. This task requires a considerable effort by the final user, resulting in imprecise labeling tampering the performance of algorithms. In this paper, we propose an unsupervised technique allowing to automatically segment smart home logs containing position sensor measurements. The proposed technique exploits information about the position of the human to automatically extract basic actions, which are then segmented on a temporal basis and clustered. The approach is evaluated against a state-of-the-art dataset.
Work-related disorders are a growing issue for office workers and represent a significant burden to public health. Work aspects such as sitting for prolonged periods and occupational stress are modifiable risk factors highly associated with occupational disorders in office workers. The PrevOccu-pAI Project (Prevention of Occupational Disorders in Public Administrations based on Artificial Intelligence) objectively investigates relationships between a variety of occupational risk factors and physiological outcomes. For this purpose, a data acquisition protocol was carried out at the Portuguese Tax and Customs Authority. Physiological, movement, and environmental signals from office workers were acquired during five consecutive workdays using a smartphone, a smartwatch, and two electromyography sensors. Additionally, demographic, occupational, and pain information were collected through questionnaires. The present manuscript provides a detailed description of the PrevOccupAI acquisition protocol. The collected data is used to gather knowledge regarding modifiable factors at the individual and organisational levels.
We posit that predicting sensor event sequence (SES) in a smart home can proactively support resident activities or recognize activities that have not been completed as intended and alert the resident. To realize this application, we propose a framework to support accurate SES prediction by leveraging online activity recognition. Our framework includes a novel method of applying a GPT2-based model, which is a sentence generation model, for SES prediction by taking advantage of the property that the relationship between ongoing activity and SES patterns is similar to the relationship between topic and word sequence patterns in NLP. We evaluated our method empirically using two real-world datasets where residents perform their usual daily activities. Our experimental results show the use of the GPT2-based model significantly improves the F1 value of SES prediction from 0.461 to 0.708 compared to the state-of-the-art method, and that using ongoing activity can further improve performance to 0.837. We found that the performance of the online activity recognition model required to achieve these SES predictions was about 80%, which could be achieved using simple feature engineering and modeling.
When Business Processes (BP) work with real-time data from the physical world, i.e., from data obtained from Internet of Things (IoT) devices, these are capable of taking more informed decisions to achieve a specific goal. However, such Iot-enhanced BPs, do not usually operated with the low-level data that is generated by such devices. Within this context, Complex Event Processing (CEP) systems are commonly integrated with BP engines to properly execute such Iot-enhanced BPs. However, IoT devices are heterogeneous by nature and CEP systems need to manage a technology heterogeneity in the data that make its processing a difficult task. In the same way, the myriad of technology solutions that exist for BP engines usually led to the creation of integration solutions that couple the CEP system with the integration API provided by a specific engine. All these issues make it difficult to adapt the system if technology requirements change. To facilitate decoupled interoperability among BP engines, CEP systems, and IoT devices in this paper we propose an architecture based on microservices, ontologies, and event-based communication which allows us to design a common communication model that is independent of technology. Specifically we propose. As a proof of concept, we have developed a tool to test and support the proposal against specific changes on technological requirements.
Essential skills for worker success in smart manufacturing encompass not only technical skills, but also decision-making skills to be able to interact effectively with machines, software, and other humans. This multidisciplinary project brings together insights from experimental cognitive psychology research and qualitative field research to present a framework that identifies decision-making skills at the individual and organizational system levels for success in smart manufacturing. The cognitive psychology research indicated that human memory should be augmented during training with novel equipment to maximize human performance and retention of information. The field research revealed several essential decision-making factors for workers and organizations. A proposal is presented to apply the research findings for designing a training paradigm for smart manufacturing workers and facilities to be adaptable for a variety of decision-making situations.
There is evidence that the lack of physical activity and unhealthy nutrition are among the major contributing factors to health conditions such as type 2 diabetes, cardiovascular diseases and hypertension. BehaviourCoach is a mobile application developed using an in-house framework which proposes an adaptation of the world-famous Monopoly boardgame incorporating exergaming features. BehaviourCoach includes features to promote sustainable behaviour changes in terms of increased physical exercise and healthier nutrition among participants. Following the development of a first application based on the framework, this paper focuses on its evaluation in terms of usability and perceived impact on individuals. A randomized controlled trial intervention involving sixty (n = 60) participants was undertaken during which data was gathered through questionnaires, observation and interviews. Participants perceived the application as easy to engage with and more importantly, BehaviourCoach had a positive impact on their level of physical exercises, which was more pronounced among those who were less physically engaged prior to the intervention.
Indoor localization provides important context information to develop Intelligent Environments able to understand user situations, to react and adapt to changes in the surrounding environment. Bluetooth 5.1 Direction Finding (DF) is a recent specification based on angle of departure (AoD) and arrival (AoA) of radio signals and it is addressed to localize objects or people in indoor scenarios. In this work, we study the error propagation of an indoor localization system based on AoA technique and on multiple anchor receivers.
In this work, we have developed a novel intelligent system capable of detecting and managing dynamic hazards in intelligent buildings. Our calculation of escape route strategies, numerical analysis, and visualization of evacuations, makes it possible to realistically investigate and evaluate hazards. For this purpose, we translated a real building into a static 3D model based on a building plan. For the analysis of evacuation scenarios, dynamic hazards were developed, which can also propagate dynamically over time. The computation of the escape route strategies is performed by using the Deep Reinforcement Learning (DRL) method Proximal Policy optimization (PPO). This work demonstrates that dynamic hazards have a great impact on the evacuation strategy in the building and can be analyzed by using this approach. Compared to traditional AI frameworks, scenarios can be created and analyzed both numerically and visually. As a result, the behavior of agents during training and evacuation can be examined for natural behavior.
Smart Materials (SMat) promise to open new opportunities in the area of Intelligent Environments (IE), whether as part of dedicated smart devices or as the fabric constituting everyday appliances and building infrastructure. Through the use of ontologies both IE engineers and the IEs themselves can be aware of, and predict, how novel configurable and changing materials react under different conditions. In contrast to conventional Smart Objects, however, as computational software/hardware-systems, lending themselves to the object-oriented perspective of conventional ontology specification languages, SMat and IE in the wider sense require a perspective focussing on extended spaces and numerical domains. Both are known to be problematic in terms of usability and computational complexity for the traditional object-oriented languages, with even very basic notions already leading into undecidability. Context Logic (CL), in contrast, is a formalism specialized for these domains. This paper demonstrates how terminology from this area involving extended spaces and numerical domains can be modeled in CL.
Greenhouses are complex systems where many variables are involved in order to optimize crops in an intensive agriculture framework. Therefore, monitoring and visualization of all these variables in real-time is mandatory to meet the trade-off between natural resource consumption and production maximization. In this article, we introduce an intelligent warning system to efficiently control agricultural activity in an operational greenhouse to increase productivity by optimizing crop production and energy consumption. The system includes a web application that allows the graphical and statistical representation of data measured by several sensors located inside a greenhouse. These sensors are located in strategic points that allow the reading of real-time data in a more accurate manner, therefore allowing the generation of information with the minimum percentage of error. In addition, the web application offers different data representations to allow a more exhaustive analysis of the data obtained. As a result, this warning system may help greenhouse managers to anticipate abnormal situations affecting their crops.
