1 Introduction

Intelligent Environments are very complex artefacts, so much so we can even consider them as ’system of systems’. They represent the state of the art of technology today linking the real physical world with the virtual and digital one. These systems connect some of the latest sensing and actuation devices and interfaces through a network and contain software that analyzes the current and historical streams of data, making context-aware decisions to provide services to humans. Humans have been co-existing now with a large number of computing devices and systems and are growing increasingly demanding on what they expect to be provided by technology. These systems which aim at delivering services to the increasingly demanding users are made of many sophisticated components (both software and hardware), so there are many opportunities for systems containing errors or delivering suboptimally. Over ten years, the Journal of Reliable Intelligent Environments has attracted a number of contributions that have grown rapidly year after year. In this period, we have witnessed a more or less constant interest on the part of the authors in the field of health and well-being, so much so that eHealth is definitively the field of application with the most significant number of proposals, followed by Industry and then by Agriculture. What changed, of course, was the focus of the research and its results. Indeed, in the early years of publication, the journal focused on the foundations of intelligent environments, emphasizing aspects like designing robust architectures to support intelligent environments, developing new sensing techniques for accurate and reliable environmental data acquisition, studying fundamental algorithms for activity recognition and data processing. In a second phase, we have assisted an expansion of applications, services and interdisciplinarity. In particular, articles have focused on implementing intelligent environments in homes to improve quality of life and support home healthcare, enhancing user interfaces to interact with intelligent environments more naturally and intuitively, increasing attention to personal data protection and the security of intelligent systems. In the last years, we have registered a lot of research finalized to the integration of artificial intelligence and machine learning techniques such as, using machine learning algorithms to enhance the predictive and adaptive capabilities of intelligent environments, applying AI to detect and prevent abnormal behaviours and security threats, developing predictive models for preventive maintenance and resource management, and, more recently, research has focused on creating autonomous environments and collaboration among intelligent agents making environments capable of self-organizing and self-managing with minimal or no human interaction. We can state that over the past ten years, the Journal of Reliable Intelligent Environments has witnessed a significant shift in the focus of published research. From initial technical foundations, the focus has moved towards a more application-oriented and interdisciplinary approach, with increasing emphasis on artificial intelligence and collaboration among intelligent agents. This reflects not only technological advancements, but also the evolving needs and expectations of society towards intelligent environments.

At our Inaugural Issue [1] of the journal we provided an area update and highlighted some of the challenges ahead. We made an overview of challenges in relation to system creation, the challenge of assessing correctness, and the system operational viability:

Creation:

we highlighted the difficulties of representing such complex systems of systems, the additional complexity that many of them are increasingly related to human safety, health and well-being, therefore making the lack of support to build such systems even more important. This is an aspect of the area which has not improved and the state of the art does not differ substantially from what it was then and it still requires much more work.

Correctness:

Some progress has been made in some directions of establishing system correctness, for example, on simulation, testing and validation. However in some other directions actually the situation has worsened given with much work being done on techniques which yield some results but use "opaque" algorithms which make difficult to assess the input–output correlation and to obtain explanations on the system behaviour (especially important in safety critical systems, such as those used in health).

Operationalization:

Being this area closer to technology and deployment in society it is more largely influenced by the private sector, still there is increasing involvement from academia as the sector is pressed into more explicit impact in society. There has been many more deployments of Intelligent Environments around the world and lessons have been learnt in many directions. Some progress has been noted on interoperability, reliability of networks and on security. Still deployment of such complex systems remains complex and with a diversity of systems in different societal niches, it is challenging to find recipes which can be applied across sectors.

So it is clear there is still significant work to be done. Emphasis in the last decade has been on quantitative methods, and speed, which have benefits, however emphasis on other aspects such as quality, transparency and accountability will become crucial for society acceptance.

2 On this Anniversary Issue

This Anniversary Issue of our journal has six invited articles which range from wide to deep analysis of specific areas of active research.

Development methodologies for IoT-based systems: challenges and research directions by Miguel J. Hornos and Mario Quinde, provides an extensive review of the ongoing effrots to distill a methodology for IoT-based systems development. This is a very important step for the maturity of the area however due to the inherent complexity of these systems there is not yet a standard and safe way to create them and what is most abundantly are isolated examples on how a system was achieved. The article considers challenges in finding a methodology and suggests open research directions which may help in finding a methodology that is at the same time easier to use and reassuring.

The future of human and animal digital health platforms by Patrick-Benjamin Bök and Daniela Micucci, provides a study on the complexities and challenges inherent to applications of IoT within medicine and health in general, including Electronic Health (eHealth) and the more recent Internet of Medical Things (IoMT). The authors consider the state of the art and propose a new framework to integrate these resources as well as to make it more flexible so that it can be extensible to the health of animals, given the importance they have for society at large.

Surveying neuro-symbolic approaches for reliable artificial intelligence of things by Zhen Lu, Imran Afridi, Hong Jin Kang, Ivan Ruchkin, and Xi Zheng, centres on Artificial Intelligence Internet of Things (AIoT), Machine Learning, with focus on Deep Learning, and alternatives, such as the neuro-symbolic approach, which can be considered to complement different methodologies and combine strengths and desirable features from different approaches.

Towards intelligent environments: human sensing through point cloud by Hirozumi Yamaguchi, Hamada Rizk, Tatsuya Amano, Akihito Hiromori, Riki Ukyo, Shota Yamada, and Masakazu Ohno, focuses on highly specialized sensing and tracking technologies which provides help in the challenging problem of multi-person tracking which in turn supports activity recognition and more personalized services delivery. The authors also illustrate how these technologies can be deployed in applications to smart buildings, shopping malls, and elderly care, with emphasis on their potential to enhance safety, convenience, and understanding of human behavior within intelligent environments.

A survey on graph neural network-based NextPOI recommendation for smart cities by Jian Yu, Lucas Guo, Jiayu Zhang, and Guiling Wang, considers the challenge of enhancing user experiences in relation to systems which provide recommendations based on “ ‘next’ Points of Interest”, location recommendations influenced not only by present context as it is more prevalent, but also by considering past choices to predict the next choices by the user. These systems considers spatio-temporal forecasted preferences, which makes the recommendation potentially more useful. Naturally this more ambitious service is also more complex to achieve and it opens up various challenges. This article focuses on Graph Neural Networks as a possible tool to approach this challenge, it provides an overview of their use and highlight the issues still to be improved to make it a safe solution.

Fuzzy cognitive network process for software reliability and quality measurement: comparisons with fuzzy analytic hierarchy process by Kevin Kam Fung Yuen, presents the Fuzzy Cognitive Network Process (FCNP) as an alternative applied to software reliability and quality measurement. One application based Fuzzy AHP is revisited to demonstrate the feasibility and usability of the FCNP as an alternative.

3 Closing words

We hope these articles stimulate the community to further improvements in this area and perhaps to collaborations between the participating teams so that complementary solutions can be used in a combined way to tackle more complex problems.