Encouraging individuals to adopt healthier behaviors, routines, and ways of living often requires a structured process built on shared goals and cooperation—this is the essence of a Behavior Change Intervention (BCI). A variety of technological tools can be used to deliver BCIs, ranging from mobile applications to embodied artificial agents like, e.g., robots; however, comparative analyses of how users perceive the usability of these different systems are still lacking. This paper presents a preliminary investigation into users’ preferences regarding different technologies as tools for promoting behavioral change. We evaluate the perceived potential of these technologies, with particular focus on usability and its relationship to users’ familiarity with them.

This paper introduces a novel approach to providing continuous, personalized, and context-aware assistance through a Migratable Artificial Intelligence (mAI) system. The proposed idea is to enable a single intelligent agent to migrate across heterogeneous embodiments, such as social robots, tablets, and smart TVs, while preserving its identity, memory, and cognitive state. Designed in close collaboration with clinical professionals, the system will aim to address key needs within a long-stay rehabilitation unit, such as maintaining patient engagement during unstructured moments, supporting therapy adherence, and reducing the burden on clinical staff. A user-centered approach is introduced to conceive specific user requirements for a clinical scenario. The mAI system architecture relies on a shared ontological representation and a modular reasoning structure, allowing the intelligent agent to deliver personalised assistance across different stages and environments of care. The architecture is then applied in a simplified version to the selected case study. The resulting system also includes real-time integration of physiological data from wearable sensors and supports adaptive interventions through cognitive and physical training modules. A structured clinical study is identified to evaluate the viability of the system and its impact on motivation, activity, and emotional well-being.

This study investigates the performance of upper-limb kinematic reconstruction obtained using a low-cost, skeleton-based markerless motion capture system with RGB cameras, compared to a laboratory-grade markerless reference system. Kinematic data were collected from healthy participants performing selected upper-limb tasks commonly adopted in musculoskeletal rehabilitation protocols. Two camera configurations, i.e., frontal and lateral, were analyzed to evaluate their influence on upper-limb kinematic estimation.Preliminary results indicate that reconstruction accuracy is task- and viewpoint-dependent. For movements primarily occurring in the sagittal plane, the lateral configuration showed improved agreement with the reference system (Mean Absolute Error (MAE) equal to 2.7° in the best case), whereas for smalleramplitude or multi-planar movements the frontal configuration provided more consistent estimates (MAE ranging between 4°– 16° depending on the task and variable).Overall, the observed errors suggest the potential feasibility of low-cost RGB-based setups for upper-limb functional assessment in controlled environments. These findings highlight the importance of camera viewpoint selection and provide preliminary indications for optimizing acquisition setups in clinical and rehabilitation applications.

Cognitive assessment and training are fundamental tools for supporting healthy aging and monitoring cognitive decline in older adults. Socially assistive robots are being investigated in this context because of their potential to provide multimodal, engaging, and personalized interactions during cognitive interventions. However, current evidence remains fragmented regarding the robotic embodiments, interaction strategies, and monitoring approaches used for cognitive assessment and training, as well as their ability to produce robust and long-term clinical benefits. To address this gap, this systematic review analyzes 37 studies identified through searches of Scopus, PubMed, and IEEE Xplore, following a PRISMA-based methodology. The review examines the main technological and methodological characteristics of the proposed systems, including robot embodiment, interaction capabilities, cognitive intervention paradigms, and the integration of multimodal monitoring technologies. The results highlight the predominance of humanoid robots and multimodal interaction strategies, together with an emerging use of adaptive and AI-based approaches. However, the available evidence is characterized by methodological heterogeneity, limited longitudinal validation, and a lack of standardized and ecologically valid evaluation protocols. Although the reviewed studies report encouraging findings concerning feasibility, engagement, acceptance, and preliminary cognitive outcomes, the evidence remains insufficient to identify the most effective robotic embodiment or confirm long-term clinical benefits. Larger comparative and longitudinal studies are therefore required.

Timeline-based Planning and Scheduling (P&S) allows to deal with temporal uncertainty and unpredictable behaviors. Flexible temporal plans are constituted by tasks whose duration cannot be exactly foreseen in advance. Such uncertainty entails a dynamic controllability issue for robust execution. We present a software tool for dynamic controllable execution of timeline-based plans leveraging recent results gathered from the integration of P&S and Model Checking techniques. The tool is deployed in a timeline-based planning system to control plan execution while guaranteeing dynamic controllability and robust plan execution. To demonstrate its viability, the tool was tested against several real-world scenarios considering an increasing execution complexity.

Robotic systems are increasingly utilized in rehabilitation contexts, offering tailored support for physical and cognitive functions. Enabling these systems to autonomously personalize treatments entails a critical role for automated planning and scheduling methodologies. However, a comprehensive review to map the control architectures of robotic rehabilitation systems that leverage automated planning for long-term customization is lacking, addressing a key gap in the literature. To bridge such a gap, this review systematically analyzed 26 studies retrieved from scientific databases after a careful selection performed following PRISMA guidelines. The search focused on works employing automated planning in robot-aided rehabilitation across physical and cognitive domains, as well as multi-user robotic environments. Papers were selected based on predefined criteria, including relevance to automated planning, peer-reviewed publication, and a focus on robotic systems in rehabilitation. The discussion highlights the primary similarities and differences in control architectures, adaptation strategies, and validation approaches among the reviewed systems. This review emphasizes the potential of automated planning to enhance personalization and adaptability in robotic rehabilitation, offering a foundation for more efficient and patient-centered therapeutic solutions.

Recognizing affective states from physiological signals is essential for enabling emotion-aware systems, particularly in human-robot interaction. This paper presents a hybrid deep learning framework for multimodal emotion recognition that integrates deep feature extraction with handcrafted physiological descriptors. The system processes electrocardiogram, photoplethysmogram, and galvanic skin response signals to predict arousal and valence in a continuous regression setting. To our aim, we evaluate two fusion strategies - feature-level and decision-level fusion - using two public affective datasets (AMIGOS and DEAP). Features extracted from each modality via a shared one-dimensional convolutional neural network and signal-specific physiological metrics are either concatenated (feature-level fusion) or separately modeled and combined at the prediction level (decision-level fusion). A broad set of machine learning regressors, including boosting methods and tree ensembles, is explored. Experiments were conducted with a leave-one-subject-out cross-validation protocol to assess generalization across users. Results show that feature-level fusion generally outperforms decision-level fusion, achieving the best root mean square error of 0.089 for arousal and 0.053 for valence. Statistical analyses confirm the significance of these differences, particularly favoring adaptive boosting and random forest under feature fusion. The proposed architecture offers a robust and interpretable solution for physiological emotion recognition and provides a solid foundation for real-time applications in emotion-aware social robotics and human-centered adaptive systems.

Behavior change intervention (BCI) can be seen as a collaborative and goal-oriented process that aims to help people make positive adjustments to their behavior, habits, and lifestyle. This paper is aimed at identifying a set of cognitive features necessary to support and stimulate behavior changes through social robots with particular attention on the components required for change state detection and the semantic reasoning for personalizing the intervention.

In the context of socially assistive robotics, there is a growing need for interaction strategies that can adapt to users' emotional states in real time, as fixed or generic communication styles often fail to sustain user engagement or meet individual motivational needs, especially in long-term human-robot interaction. To address this challenge, this paper presents a novel framework for adaptive interaction style modulation in socially assistive agents, combining large language models (LLMs) with reinforcement learning based on real-time emotion recognition. The proposed architecture leverages multimodal sensing to monitor the user's affective state and dynamically selects among predefined communicative styles using Thompson Sampling. At each interaction turn, the user's emotional feedback is converted into a scalar reward, allowing the system to reinforce styles that yield more positive affective outcomes. Style conditioning is operationalized through prompting strategies that guide the LLM to generate responses aligned with the selected tone. A preliminary evaluation using VADER sentiment analysis demonstrates that stylistic prompts successfully induce measurable differences in sentiment polarity, neutrality, and verbosity. These findings suggest the viability of our approach to style-aware dialogue generation and support its potential for long-term adaptation in personalized human-agent interaction.

Physical rehabilitation is essential for restoring functionality and improving the quality of life for individuals affected by neurological or musculoskeletal conditions. Rehabilitation robots emerged as key-enabling technology to deliver intensive treatments and objectively quantify patients' motor performance. In the context of Healthcare 5.0, personalization of the treatment is paramount to improve the effectiveness of the interventions. Personalization can be implemented reactively, by providing real-time physical assistance and feedback, and deliberatively, by planning sessions based on therapeutic goals. Inspired by Kahneman's dual-system theory, this paper proposes a cognitive architecture for a robot-aided rehabilitation platform capable of delivering personalized treatment through deliberative and reactive techniques. The proposed cognitive architecture is described and validated through experimental sessions. Six healthy participants were enrolled in the experiments, simulating a robot-aided rehabilitation session with a TIAGo service robot serving as the physical interface to deliver the planned session. The results highlight that the plans generated according to different clinical objectives elicited distinct physiological responses from the participants, demonstrating the effectiveness of the personalized approach.

The increasing complexity of natural disaster incidents demands innovative technological solutions to support first responders in their efforts. This paper introduces the TRIFFID system, a comprehensive technical framework that integrates unmanned ground and aerial vehicles with advanced artificial intelligence functionalities to enhance disaster response capabilities across wildfires, urban floods, and post-earthquake search and rescue missions. By leveraging state-of-the-art autonomous navigation, semantic perception, and human-robot interaction technologies, TRIFFID provides a sophisticated system composed of the following key components: hybrid robotic platform, centralized ground station, custom communication infrastructure, and smartphone application. The defined research and development activities demonstrate how deep neural networks, knowledge graphs, and multimodal information fusion can enable robots to autonomously navigate and analyze disaster environments, reducing personnel risks and accelerating response times. The proposed system enhances emergency response teams by providing advanced mission planning, safety monitoring, and adaptive task execution capabilities. Moreover, it ensures real-time situational awareness and operational support in complex and risky situations, facilitating rapid and precise information collection and coordinated actions.

Workload estimation is essential for artificial systems designed to assist users across various domains. These systems can provide personalized support by continuously assessing the user state and optimizing intervention strategies. Physiological data acquisition through advanced sensors enables objective and real-time workload estimation, offering a more reliable alternative to self-reported measures. Despite the growing interest in workload estimation, existing literature reviews are often domain-specific or focus on cognitive workload only, without providing a comprehensive analysis of methodologies for estimating both physical and cognitive workload across different applications. To address this gap, this systematic review analyzes 35 studies on multimodal physiological monitoring, examining feature extraction methodologies and supervised learning models used for workload estimation. The review identifies key challenges, including the need for standardized protocols, improved generalization across real-world scenarios, and the integration of adaptive artificial intelligence models. It underscores the role of sensor-based workload estimation in healthcare, rehabilitation, and assistive technologies, positioning it as a fundamental component for developing intelligent, user-centered, and adaptive human-machine interaction systems.

The diffusion of Human-Robot Collaborative cells is prevented by several barriers. Classical control approaches seem not yet fully suitable for facing the variability conveyed by the presence of human operators beside robots. The capabilities of representing heterogeneous knowledge representation and performing abstract reasoning are crucial to enhance the flexibility of control solutions. To this aim, the ontology SOHO (Sharework Ontology for Human-Robot Collaboration) has been specifically designed for representing Human-Robot Collaboration scenarios, following a context-based approach. This work brings several contributions. This paper proposes an extension of SOHO to better characterize behavioral constraints of collaborative tasks. Furthermore, this work shows a knowledge extraction procedure designed to automatize the synthesis of Artificial Intelligence plan-based controllers for realizing flexible coordination of human and robot behaviors in collaborative tasks. The generality of the ontological model and the developed representation capabilities as well as the validity of the synthesized planning domains are evaluated on a number of realistic industrial scenarios where collaborative robots are actually deployed.

Combining symbolic and geometric reasoning in multiagent systems is a challenging task that involves planning, scheduling, and synchronization problems. Existing works overlooked the variability of task duration and geometric feasibility intrinsic to these systems because of the interaction between agents and the environment. We propose a combined task and motion planning approach to optimize the sequencing, assignment, and execution of tasks under temporal and spatial variability. The framework relies on decoupling tasks and actions, where an action is one possible geometric realization of a symbolic task. At the task level, timeline-based planning deals with temporal constraints, duration variability, and synergic assignment of tasks. At the action level, online motion planning plans for the actual movements dealing with environmental changes. We demonstrate the approach's effectiveness in a collaborative manufacturing scenario, in which a robotic arm and a human worker shall assemble a mosaic in the shortest time possible. Compared with existing works, our approach applies to a broader range of applications and reduces the execution time of the process.

Deploying collaborative robots in manufacturing presents diverse challenges. Rapid adaptability to the environment while ensuring user safety and engagement is paramount. Existing human-aware task sequencing solutions often lack explicit risk modeling and management. International standards emphasize severity, exposure, and avoidance as critical risk factors. To enhance intelligent risk awareness control, we propose integrating multiple risk factors into task sequencing models. This forms the basis for a cutting-edge planning framework-backed risk-aware task sequencing system. Our approach's evaluation across various scenarios showcases its efficacy and adaptability to diverse risk levels. Experimental results show a positive equilibrium between productivity and safety, achieving both high throughput and low operator risk.

This work investigate temporal planning to synthesize personalized physical rehabilitation programs. The first contribution of the work concerns the representation of (heterogeneous) clinical and spatial constraints into a planning framework. The second contribution is the integration of numerical and symbolic reasoning to synthesize technically valid and coherent plans with respect to different clinical objectives. The experimental section discusses the developed planner from a technical view, assessing solving and personalization capabilities, and from a clinical view, assessing the efficacy of plans on the involved patients.

Robots acting in real-world environments may interact with humans at different levels of abstraction (e.g., process, task, physical), entailing different control and coordination challenges. When acting in social situations, robots should be able to pursue (joint) goals by behaving according to the context as well as the skills/features of involved humans. Although reliable and effective, standard control techniques may limit the adaptability of robots. Novel control technologies based on Artificial Intelligence can endow robots with the cognitive capabilities needed to achieve a higher level of autonomy in terms of flexibility, reliability, and awareness. In this context, this paper introduces a goal-oriented acting framework based on timeline-based planning and execution. The framework is evaluated on a realistic Human-Robot Collaboration manufacturing scenario. Results show the capability of dealing with the uncontrollable dynamics of humans achieving effective and reliable collaborations.

In the last years, several mobile APPs have been developed within the cultural tourism domain to give new impetus to this sector which is booming both in Italy and worldwide. In the wake of the increasing importance of technologies based on artificial intelligence, even mobile applications for the use of cultural tourism heritage are increasingly taking advantage of these techniques. Machine learning strategies are increasingly used to recommend points of interest and itineraries that are compatible with the user’s preferences, requirements and constraints. The quality and integrity of the data acquired become the starting point for training and implementing AI models. By levering well-structured data, these algorithms can offer valuable insights, personalised recommendations, and enhanced user interaction in the cultural tourism domain. The HerMeS APP that we present in this paper was designed starting from these premises. The application aims to provide a wide range of artificial intelligence-based features to enhance the enjoyment and exploration of cultural heritage, both tangible and intangible.