AIR Institute

Smart cities and artificial intelligence (AI) are among the most popular discourses in urban policy circles. Most attempts at using AI to improve efficiencies in cities have nevertheless either struggled or failed to accomplish the smart city transformation. This is mainly due to short-sighted, technologically determined and reductionist AI approaches being applied to complex urbanization problems. Besides this, as smart cities are underpinned by our ability to engage with our environments, analyze them, and make efficient, sustainable and equitable decisions, the need for a green AI approach is intensified. This perspective paper, reflecting authors’ opinions and interpretations, concentrates on the “green AI” concept as an enabler of the smart city transformation, as it offers the opportunity to move away from purely technocentric efficiency solutions towards efficient, sustainable and equitable solutions capable of realizing the desired urban futures. The aim of this perspective paper is two-fold: first, to highlight the fundamental shortfalls in mainstream AI system conceptualization and practice, and second, to advocate the need for a consolidated AI approach—i.e., green AI—to further support smart city transformation. The methodological approach includes a thorough appraisal of the current AI and smart city literatures, practices, developments, trends and applications. The paper informs authorities and planners on the importance of the adoption and deployment of AI systems that address efficiency, sustainability and equity issues in cities.

The predominant means of communication is speech; however, there are persons whose speaking or hearing abilities are impaired. Communication presents a significant barrier for persons with such disabilities. The use of deep learning methods can help to reduce communication barriers. This paper proposes a deep learning-based model that detects and recognizes the words from a person’s gestures. Deep learning models, namely, LSTM and GRU (feedback-based learning models), are used to recognize signs from isolated Indian Sign Language (ISL) video frames. The four different sequential combinations of LSTM and GRU (as there are two layers of LSTM and two layers of GRU) were used with our own dataset, IISL2020. The proposed model, consisting of a single layer of LSTM followed by GRU, achieves around 97% accuracy over 11 different signs. This method may help persons who are unaware of sign language to communicate with persons whose speech or hearing is impaired.

The urbanization problems we face may be alleviated using innovative digital technology. However, employing these technologies entails the risk of creating new urban problems and/or intensifying the old ones instead of alleviating them. Hence, in a world with immense technological opportunities and at the same time enormous urbanization challenges, it is critical to adopt the principles of responsible urban innovation. These principles assure the delivery of the desired urban outcomes and futures. We contribute to the existing responsible urban innovation discourse by focusing on local government artificial intelligence (AI) systems, providing a literature and practice overview, and a conceptual framework. In this perspective paper, we advocate for the need for balancing the costs, benefits, risks and impacts of developing, adopting, deploying and managing local government AI systems in order to achieve responsible urban innovation. The statements made in this perspective paper are based on a thorough review of the literature, research, developments, trends and applications carefully selected and analyzed by an expert team of investigators. This study provides new insights, develops a conceptual framework and identifies prospective research questions by placing local government AI systems under the microscope through the lens of responsible urban innovation. The presented overview and framework, along with the identified issues and research agenda, offer scholars prospective lines of research and development; where the outcomes of these future studies will help urban policymakers, managers and planners to better understand the crucial role played by local government AI systems in ensuring the achievement of responsible outcomes.

Artificial intelligence (AI) is a powerful technology with an increasing popularity and applications in areas ranging from marketing to banking and finance, from agriculture to healthcare and security, from space exploration to robotics and transport, and from chatbots to artificial creativity and manufacturing. Although many of these areas closely relate to the urban context, there is limited understanding of the trending AI technologies and their application areas—or concepts—in the urban planning and development fields. Similarly, there is a knowledge gap in how the public perceives AI technologies, their application areas, and the AI-related policies and practices of our cities. This study aims to advance our understanding of the relationship between the key AI technologies (n = 15) and their key application areas (n = 16) in urban planning and development. To this end, this study examines public perceptions of how AI technologies and their application areas in urban planning and development are perceived and utilized in the testbed case study of Australian states and territories. The methodological approach of this study employs the social media analytics method, and conducts sentiment and content analyses of location-based Twitter messages (n = 11,236) from Australia. The results disclose that: (a) digital transformation, innovation, and sustainability are the most popular AI application areas in urban planning and development; (b) drones, automation, robotics, and big data are the most popular AI technologies utilized in urban planning and development, and; (c) achieving the digital transformation and sustainability of cities through the use of AI technologies—such as big data, automation and robotics—is the central community discussion topic.

In recent years, artificial intelligence (AI) has started to manifest itself at an unprecedented pace. With highly sophisticated capabilities, AI has the potential to dramatically change our cities and societies. Despite its growing importance, the urban and social implications of AI are still an understudied area. In order to contribute to the ongoing efforts to address this research gap, this paper introduces the notion of an artificially intelligent city as the potential successor of the popular smart city brand-where the smartness of a city has come to be strongly associated with the use of viable technological solutions, including AI. The study explores whether building artificially intelligent cities can safeguard humanity from natural disasters, pandemics, and other catastrophes. All of the statements in this viewpoint are based on a thorough review of the current status of AI literature, research, developments, trends, and applications. This paper generates insights and identifies prospective research questions by charting the evolution of AI and the potential impacts of the systematic adoption of AI in cities and societies. The generated insights inform urban policymakers, managers, and planners on how to ensure the correct uptake of AI in our cities, and the identified critical questions offer scholars directions for prospective research and development.

There is a consistent rise in chronic diseases worldwide. These diseases decrease immunity and the quality of daily life. The treatment of these disorders is a challenging task for medical professionals. Dimensionality reduction techniques make it possible to handle big data samples, providing decision support in relation to chronic diseases. These datasets contain a series of symptoms that are used in disease prediction. The presence of redundant and irrelevant symptoms in the datasets should be identified and removed using feature selection techniques to improve classification accuracy. Therefore, the main contribution of this paper is a comparative analysis of the impact of wrapper and filter selection methods on classification performance. The filter methods that have been considered include the Correlation Feature Selection (CFS) method, the Information Gain (IG) method and the Chi-Square (CS) method. The wrapper methods that have been considered include the Best First Search (BFS) method, the Linear Forward Selection (LFS) method and the Greedy Step Wise Search (GSS) method. A Decision Tree algorithm has been used as a classifier for this analysis and is implemented through the WEKA tool. An attribute significance analysis has been performed on the diabetes, breast cancer and heart disease datasets used in the study. It was observed that the CFS method outperformed other filter methods concerning the accuracy rate and execution time. The accuracy rate using the CFS method on the datasets for heart disease, diabetes, breast cancer was 93.8%, 89.5% and 96.8% respectively. Moreover, latency delays of 1.08 s, 1.02 s and 1.01 s were noted using the same method for the respective datasets. Among wrapper methods, BFS’ performance was impressive in comparison to other methods. Maximum accuracy of 94.7%, 95.8% and 96.8% were achieved on the datasets for heart disease, diabetes and breast cancer respectively. Latency delays of 1.42 s, 1.44 s and 132 s were recorded using the same method for the respective datasets. On the basis of the obtained result, a new hybrid Attribute Evaluator method has been proposed which effectively integrates enhanced K-Means clustering with the CFS filter method and the BFS wrapper method. Furthermore, the hybrid method was evaluated with an improved decision tree classifier. The improved decision tree classifier combined clustering with classification. It was validated on 14 different chronic disease datasets and its performance was recorded. A very optimal and consistent classification performance was observed. The mean values for accuracy, specificity, sensitivity and f-score metrics were 96.7%, 96.5%, 95.6% and 96.2% respectively.

The current energy strategy of the European Union puts the end-user as a key participant in electricity markets. The creation of energy communities has been encouraged by the European Union to increase the penetration of renewable energy and reduce the overall cost of the energy chain. Energy communities are mostly composed of prosumers, which may be households with small-size energy production equipment such as rooftop photovoltaic panels. The local electricity market is an emerging concept that enables the active participation of end-user in the electricity markets and is especially interesting when energy communities are in place. This paper proposes an optimization model to schedule peer-to-peer transactions via local electricity market, grid transactions in retail market, and battery management considering the photovoltaic production of households. Prosumers have the possibility of transacting energy with the retailer or with other consumers in their community. The problem is modeled using mixed-integer linear programming, containing binary and continuous variables. Four scenarios are studied, and the impact of battery storage systems and peer-to-peer transactions is analyzed. The proposed model execution time according to the number of prosumers involved (3, 5, 10, 15, or 20) in the optimization is analyzed. The results suggest that using a battery storage system in the energy community can lead to energy savings of 11-13%. Besides, combining the use of peer-to-peer transactions and energy storage systems can potentially provide energy savings of up to 25% in the overall costs of the community members.

Smart grids concept benefits and leverage distributed management systems while allowing its players to actively participate in the smart grid. This paper merges the concepts of microgrid and transactive energy. The proposed model is tested in an office building with multiple tenants. An agent-based platform, running in single-board computers, for microgrid intelligent management with a peer-to-peer energy transaction model is proposed in this paper. This paper describes the peer-to-peer transaction auction model and the deployment of the platform in an office building. The results regard a one-week period where the use of peer-to-peer transactions is compared with a scenario where no transactions among agents are performed. The results are promising, showing the energy price inside the microgrid dropping for the majority of players/agents. The presented work demonstrates how smart grid players can decrease their energy costs using simple approaches that do not require load shifting, consumption optimization nor the acquisition of new equipment.

This paper presents an efficient cyberphysical platform for the smart management of smart territories. It is efficient because it facilitates the implementation of data acquisition and data management methods, as well as data representation and dashboard configuration. The platform allows for the use of any type of data source, ranging from the measurements of a multi-functional IoT sensing devices to relational and non-relational databases. It is also smart because it incorporates a complete artificial intelligence suit for data analysis; it includes techniques for data classification, clustering, forecasting, optimization, visualization, etc. It is also compatible with the edge computing concept, allowing for the distribution of intelligence and the use of intelligent sensors. The concept of smart cities is evolving and adapting to new applications; the trend to create intelligent neighbourhoods, districts or territories is becoming increasingly popular, as opposed to the previous approach of managing an entire megacity. In this paper, the platform is presented, and its architecture and functionalities are described. Moreover, its operation has been validated in a case study where the bike renting service of Paris-Vélib' Métropole has been managed. This platform could enable smart territories to develop adapted knowledge management systems, adapt them to new requirements and to use multiple types of data, and execute efficient computational and artificial intelligence algorithms. The platform optimizes the decisions taken by human experts through explainable artificial intelligence models that obtain data from IoT sensors, databases, the Internet, etc. The global intelligence of the platform could potentially coordinate its decision-making processes with intelligent nodes installed in the edge, which would use the most advanced data processing techniques.

A smart city is an environment that uses innovative technologies to make networks and services more flexible, effective, and sustainable with the use of information, digital, and telecommunication technologies, improving the city’s operations for the benefit of its citizens. Most cities incorporate data acquisition elements from their own systems or those managed by subcontracted companies that can be used to optimise their resources: energy consumption, smart meters, lighting, irrigation water consumption, traffic data, camera images, waste collection, security systems, pollution meters, climate data, etc. The city-as-a-platform concept is becoming popular and it is increasingly evident that cities must have efficient management systems capable of deploying, for instance, IoT platforms, open data, etc., and of using artificial intelligence intensively. For many cities, data collection is not a problem, but managing and analysing data with the aim of optimising resources and improving the lives of citizens is. This article presents deepint.net, a platform for capturing, integrating, analysing, and creating dashboards, alert systems, optimisation models, etc. This article shows how deepint.net has been used to estimate pedestrian traffic on the streets of Melbourne (Australia) using the XGBoost algorithm. Given the current situation, it is advisable not to transit urban roads when overcrowded, thus, the model proposed in this paper (and implemented with deepint.net) facilitates the identification of areas with less pedestrian traffic. This use case is an example of an efficient crowd management system, implemented and operated via a platform that offers many possibilities for the management of the data collected in smart territories and cities.

Microgrids bring advantages to end-users and to the smart grid environment. However, adequate management software, enabling bringing to the field new energy management concepts, is not available yet. Small, single-tasked, software is usually proposed and tested while a clear overall system architecture for microgrid management required to take full advantage of the microgrids’ potential. Previous publications usually focus on energy-related problems and do not provide an efficient and viable solution for players’ representation and microgrid operation. This paper proposes a complete architecture for a microgrid management system based on a multi-agent approach – µGIM – allowing the easy implementation of different energy strategies. The µGIM agents can independently manage local resources while able to collaborate and/or compete with other agents. Designed to run in single-board computers, µGIM agents are light-weighted and easily deployed in buildings. To demonstrate these capabilities, the paper details and presents a microgrid deployment using µGIM solution.

Several factors are motivating the development of preventive, personalized, connected, virtual, and ubiquitous healthcare services. These factors include declining public health, increase in chronic diseases, an ageing population, rising healthcare costs, the need to bring intelligence near the user for privacy, security, performance, and costs reasons, as well as COVID-19. Motivated by these drivers, this paper proposes, implements, and evaluates a reference architecture called Imtidad that provides Distributed Artificial Intelligence (AI) as a Service (DAIaaS) over cloud, fog, and edge using a service catalog case study containing 22 AI skin disease diagnosis services. These services belong to four service classes that are distinguished based on software platforms (containerized gRPC, gRPC, Android, and Android Nearby) and are executed on a range of hardware platforms (Google Cloud, HP Pavilion Laptop, NVIDIA Jetson nano, Raspberry Pi Model B, Samsung Galaxy S9, and Samsung Galaxy Note 4) and four network types (Fiber, Cellular, Wi-Fi, and Bluetooth). The AI models for the diagnosis include two standard Deep Neural Networks and two Tiny AI deep models to enable their execution at the edge, trained and tested using 10,015 real-life dermatoscopic images. The services are evaluated using several benchmarks including model service value, response time, energy consumption, and network transfer time. A DL service on a local smartphone provides the best service in terms of both energy and speed, followed by a Raspberry Pi edge device and a laptop in fog. The services are designed to enable different use cases, such as patient diagnosis at home or sending diagnosis requests to travelling medical professionals through a fog device or cloud. This is the pioneering work that provides a reference architecture and such a detailed implementation and treatment of DAIaaS services, and is also expected to have an extensive impact on developing smart distributed service infrastructures for healthcare and other sectors.

Globalization has led to a new paradigm where the traditional industries, such as agriculture, employ vanguard technologies to broaden its possibilities into what is known as smart farming and the agri-food industry 4.0. This industry needs to adapt to the current market through an efficient use of resources while being environmentally friendly. The most commonly used approaches for analyzing efficiency and sustainability on farms are production efficiency based analyses, such as Data Envelopment Analysis and Stochastic Frontier Analysis, since they allow to see how efficient the outputs are generated regardless of the units of measurement of the inputs. This work presents a real scenario for making farms more profitable and sustainable through the analysis of the Data Envelopment Analysis and the application of the Internet of Things and Edge Computing. What makes this model interesting is that it allows monitoring the ambient conditions with real-time data from the different sensors that have been installed on the farm, minimizing costs and gaining robustness in the transmission of the data to the cloud with Edge Computing, and then to have a complete overview in terms of monthly resource efficiency through the Data Envelopment Analysis. The results show that including the costs of edge and non-edge data transfer have an impact on the efficiency. This small-scale study set the basis for a future test with many farms simultaneously.

Electric vehicles have emerged as one of the most promising technologies, and their mass introduction may pose threats to the electricity grid. Several solutions have been proposed in an attempt to overcome this challenge in order to ease the integration of electric vehicles. A promising concept that can contribute to the proliferation of electric vehicles is the local electricity market. In this way, consumers and prosumers may transact electricity between peers at the local community level, reducing congestion, energy costs and the necessity of intermediary players such as retailers. Thus, this paper proposes an optimization model that simulates an electric energy market between prosumers and electric vehicles. An energy community with different types of prosumers is considered (household, commercial and industrial), and each of them is equipped with a photovoltaic panel and a battery system. This market is considered local because it takes place within a distribution grid and a local energy community. A mixed-integer linear programming model is proposed to solve the local energy transaction problem. The results suggest that our approach can provide a reduction between 1.6% to 3.5% in community energy costs.

Smart wearable technologies such as fitness trackers are creating many new opportunities to improve the quality of life for everyone. It is usually impossible for visually impaired people to orientate themselves in large spaces and navigate an unfamiliar area without external assistance. The design space for assistive technologies for the visually impaired is complex, involving many design parameters including reliability, transparent object detection, handsfree operations, high-speed real-time operations, low battery usage, low computation and memory requirements, ensuring that it is lightweight, and price affordability. State-of-the-art visually impaired devices lack maturity, and they do not fully meet user satisfaction, thus more effort is required to bring innovation to this field. In this work, we develop a pair of smart glasses called LidSonic that uses machine learning, LiDAR, and ultrasonic sensors to identify obstacles. The LidSonic system comprises an Arduino Uno device located in the smart glasses and a smartphone app that communicates data using Bluetooth. Arduino collects data, manages the sensors on smart glasses, detects objects using simple data processing, and provides buzzer warnings to visually impaired users. The smartphone app receives data from Arduino, detects and identifies objects in the spatial environment, and provides verbal feedback about the object to the user. Compared to image processing-based glasses, LidSonic requires much less processing time and energy to classify objects using simple LiDAR data containing 45-integer readings. We provide a detailed description of the system hardware and software design, and its evaluation using nine machine learning algorithms. The data for the training and validation of machine learning models are collected from real spatial environments. We developed the complete LidSonic system using off-the-shelf inexpensive sensors and a microcontroller board costing less than USD 80. The intention is to provide a design of an inexpensive, miniature, green device that can be built into, or mounted on, any pair of glasses or even a wheelchair to help the visually impaired. This work is expected to open new directions for smart glasses design using open software tools and off-the-shelf hardware.

The Internet of Things (IoT) paradigm allows the interconnection of millions of sensor devices gathering information and forwarding to the Cloud, where data is stored and processed to infer knowledge and perform analysis and predictions. Cloud service providers charge users based on the computing and storage resources used in the Cloud. In this regard, Edge Computing can be used to reduce these costs. In Edge Computing scenarios, data is pre-processed and filtered in network edge before being sent to the Cloud, resulting in shorter response times and providing a certain service level even if the link between IoT devices and Cloud is interrupted. Moreover, there is a growing trend to share physical network resources and costs through Network Function Virtualization (NFV) architectures. In this sense, and related to NFV, Software-Defined Networks (SDNs) are used to reconfigure the network dynamically according to the necessities during time. For this purpose, Machine Learning mechanisms, such as Deep Reinforcement Learning techniques, can be employed to manage virtual data flows in networks. In this work, we propose the evolution of an existing Edge-IoT architecture to a new improved version in which SDN/NFV are used over the Edge-IoT capabilities. The proposed new architecture contemplates the use of Deep Reinforcement Learning techniques for the implementation of the SDN controller.

Communication systems and networks are evolving as an integral part of not only of our everyday life but also as a part of the industry, fundamental infrastructures, companies, etc. Current directions and concepts, such as the Internet of Things (IoT), promise enhanced quality of life, greater business opportunities, cost-effective manufacturing, and efficient operation management through ubiquitous connectivity and deployment of smart physical objects. IoT networks are able to collect, pre-process, and transmit a huge amount of data. A considerable portion of this data is security- and privacy-critical data, which makes IoT networks a tempting option for attackers. Given that these networks deal with the actual aspects of our lives and fundamental infrastructures (e.g. smart grids), cybersecurity in such networks is crucial. The large scale of these networks and their unique characteristics and complexity bring further vulnerabilities. In this study we give an introduction to IoT, focusing on the IoT application layer, security requirements, threats, and countermeasures in this layer, and some of the open issues and future research lines.

Physical aggression is a serious and widespread problem in society, affecting people worldwide. It impacts nearly every aspect of life. While some studies explore the root causes of violent behavior, others focus on urban planning in high-crime areas. Real-time violence detection, powered by artificial intelligence, offers a direct and efficient solution, reducing the need for extensive human supervision and saving lives. This paper is a continuation of a systematic mapping study and its objective is to provide a comprehensive and up-to-date review of AI-based video violence detection, specifically in physical assaults. Regarding violence detection, the following have been grouped and categorized from the review of the selected papers: 21 challenges that remain to be solved, 28 datasets that have been created in recent years, 21 keyframe extraction methods, 16 types of algorithm inputs, as well as a wide variety of algorithm combinations and their corresponding accuracy results. Given the lack of recent reviews dealing with the detection of violence in video, this study is considered necessary and relevant.

Over a billion people around the world are disabled, among whom 253 million are visually impaired or blind, and this number is greatly increasing due to ageing, chronic diseases, and poor environments and health. Despite many proposals, the current devices and systems lack maturity and do not completely fulfill user requirements and satisfaction. Increased research activity in this field is required in order to encourage the development, commercialization, and widespread acceptance of low-cost and affordable assistive technologies for visual impairment and other disabilities. This paper proposes a novel approach using a LiDAR with a servo motor and an ultrasonic sensor to collect data and predict objects using deep learning for environment perception and navigation. We adopted this approach using a pair of smart glasses, called LidSonic V2.0, to enable the identification of obstacles for the visually impaired. The LidSonic system consists of an Arduino Uno edge computing device integrated into the smart glasses and a smartphone app that transmits data via Bluetooth. Arduino gathers data, operates the sensors on the smart glasses, detects obstacles using simple data processing, and provides buzzer feedback to visually impaired users. The smartphone application collects data from Arduino, detects and classifies items in the spatial environment, and gives spoken feedback to the user on the detected objects. In comparison to image-processing-based glasses, LidSonic uses far less processing time and energy to classify obstacles using simple LiDAR data, according to several integer measurements. We comprehensively describe the proposed system's hardware and software design, having constructed their prototype implementations and tested them in real-world environments. Using the open platforms, WEKA and TensorFlow, the entire LidSonic system is built with affordable off-the-shelf sensors and a microcontroller board costing less than USD 80. Essentially, we provide designs of an inexpensive, miniature green device that can be built into, or mounted on, any pair of glasses or even a wheelchair to help the visually impaired. Our approach enables faster inference and decision-making using relatively low energy with smaller data sizes, as well as faster communications for edge, fog, and cloud computing.

The Internet of Things and the millions of devices that generate and collect data through sensors to send it to the Cloud are part of the life of users in many contexts, including smart farming and precision agriculture scenarios. This volume of data is stored and processed in the Cloud, with the purpose of obtaining knowledge and valuable information for organizations. Edge Computing has emerged to reduce the costs associated with transferring, processing and storing data from IoT environments in the Cloud. This paradigm allows data to be pre-processed at the edge of the network before they are sent to the Cloud, obtaining shorter response times and maintaining service even during communication breakdowns between the IoT and Cloud layers. Furthermore, there is a increasing trend to shared physical network resources among diverse user entities through Software-Defined Networks and Network Function Virtualization with the aim to reduce costs. In this sense, smart mechanisms are required to optimize virtual dataflows in the networks, as Deep Reinforcement Learning techniques. This paper proposes a Double Deep-Q Learning approach to manage virtual dataflows in SDN/NFV using an Edge-IoT architecture, formerly applied in smart farming and Industry 4.0 scenarios.