Roberto Meattini is a Junior Assistant Professor at the Department of Electrical, Electronic and Information Engineering “Guglielmo Marconi” (DEI) at the University of Bologna, where he also earned his PhD in Biomedical, Electrical and System Engineering. His research focuses on robotic manipulation, telemanipulation, human-robot interaction, electromyographic (EMG) interfaces, and collaborative robotics. He has been involved in European research projects such as REMODEL, IntelliMan, and SUPERHUMAN, with a focus on intuitive robot control, human-in-the-loop systems, and assistive robotics. His work aims to enable more natural and advanced forms of human-robot collaboration.

Fundamentals of Myoelectric Interfaces for Human-In-The-Loop Robotic Manipulation

This lecture presents fundamental concepts that can be exploited for myoelectric interfaces in human-in-the-loop robotic manipulation. We follow a formal understanding of surface electromyographic (sEMG) signals and their relationship to muscle activation, leveraging the notion of myoprocessor for estimating motor intent from muscle electrical activity. The session aims to provide an insight into how humans and robotic grasping devices can be connected through sEMG signals, with a focus on both more established control techniques and broader research perspectives.

Giovanni Berselli (Member, IEEE) is currently a Full Professor and the Chair of Design Methods for Industrial Engineering with the University of Genova, Genoa, Italy, where he coordinates the Ph.D. degree in mechanical, energy and management engineering. He is also an Affiliated Researcher with the Advanced Robotics Department, Italian Institute of Technology (IIT), Genova. He held Visiting Professor positions at Technical University of Munich, Monash University, University of Navarra, University of Twente, German Aerospace Agency – DLR, Harvard Medical School and Massachusetts General Hospital, and Karlsruhe Institute of Technology. His research interests include the design, modeling, and experimental evaluation of robot hands and grippers, compliant mechanisms and soft actuators for safe human–robot interaction, and energy-aware industrial robotics.,Dr. Berselli is a Fellow of the American Society of Mechanical Engineers (ASME), the Chair of the ASME Italy Section, and the past Chair of the ASME Technical Committee on Modeling, Dynamics, and Control of Adaptive Systems.

Computer-Aided Methods for Designing Soft Robotic Hands – Materials and Manufacturing

The main class of compliant mechanisms currently employable for realizing better-behaved robotic hands for prosthetic usage are introduced. The lecture will focus on theoretical and numerical methods for the optimal sizing of the hand mechanical structure and for the selection of the actuation systems. Subsequently, we shall focus on Computer-Aided Design and Engineering tools (both commercial and open source), that can help the hand designer during the conception of novel device, the material selection and technologies employed in the manufacturing phase. Some well-known devices will be compared, with the aim of highlighting advantages, disadvantages and possible directions of improvement. Particular focus will be directed towards a comparative evaluation of under-actuated devices and highly dexterous hands, tackled from the mechatronic designer viewpoint.

Ada Fort earned her degree in Electronic Engineering from the University of Florence in 1989 and obtained her Ph.D. in 1992. She is currently a Full Professor in the scientific-disciplinary sector ING-INF/07 – Electrical and Electronic Measurements at the Department of Information Engineering and Mathematical Sciences (DIISM), University of Siena. In addition to teaching core courses in her field, she also lectures in the specialized course “Sensors and Microsystems”, which focuses on modern sensor technologies and their integration into measurement and monitoring systems. Her research activities are centered on the development of sensor-based measurement systems. In particular, she has worked extensively on the development, characterization, modeling, and interfacing of chemical sensors, the creation of measurement chains based on physical and chemical sensors, as well as the design of innovative sensing technologies and sensor networks within the Internet of Things (IoT) framework.

She coordinates research projects exploring the application of sensors and sensor networks in a variety of domains, including IoT-based systems. She is actively involved in Italy’s National Recovery and Resilience Plan (PNRR), particularly in the Fit4MedRob project, where she contributes to a sub-project focused on the development of custom sensing systems for soft robotics. She also participates in initiatives in the biomedical field (Tuscany Health Ecosystem) and in smart agriculture (CN Agritech) contributing with the study and development of ad hoc sensing solutions.

Prof. Fort also leads the Laboratory of Electronics and Electronic Measurements, where researchers in electronics and measurement science collaborate on the design and development of sensing systems and instrumentation, fostering interdisciplinary innovation and technological advancement.

Her scientific output includes over 340 publications.

Designing and Evaluating Sensors for Reliable Human-Robot Interaction

This course addresses the role of sensing in supervised autonomy, focusing on human-robot interaction (HRI) contexts. The session opens with a brief overview of the fundamental characteristics of a sensing device related to the different possible applications in the SAHRI context, and of a measurement chain — from the physical sensor to the analog front-end (AFE), signal acquisition, and data processing.

Each block in this chain is explored in terms of its technical contribution and its limitations. The course highlights how the performance of a sensing system is not solely defined by the sensor element itself, but also by how it is mechanically integrated into the system, how its signals are conditioned, and how the resulting data is interpreted.

To illustrate these topics, the course will focus on two categories of widely used sensors: e.g. traditional force or strain sensors (including capacitive variants), and MEMS sensors such as accelerometers and gyroscopes. For each, a critical analysis will be offered, addressing real-world challenges such as mechanical interface issues, signal noise, analog front-end design, calibration complexity, and long-term reliability.

The session will conclude with a concrete case study in soft robotics, e.g. the integration of a sensorized joint, to highlight the interplay between mechanical design, sensing performance, and signal interpretation. This example will serve as a synthesis of the earlier concepts, providing a realistic scenario in which the discussed trade-offs and design decisions become evident.

The course emphasizes a critical, system-level approach to sensor integration, encouraging participants to look beyond datasheets and consider the entire sensing architecture.

(Senior Member, IEEE) received the B.Sc., M.Sc., and Ph.D. degrees in Electronics Engineering from the University of Naples Federico II, Naples, Italy, in 2004, 2007, and 2011, respectively. Since 2022, he has been an Associate Professor with the Department of Electrical Engineering and Information Technology at the same university. His research interests span electrothermal characterization, modeling, and simulation of semiconductor power devices, as well as the design of embedded systems for the Internet of Things (IoT) and wearable health monitoring applications. In recent years, his work has increasingly focused on the integration of emerging sensor technologies with AI methodologies, particularly leveraging Edge AI and Deep Edge AI approaches to enable intelligent, real-time, and resource-efficient processing in embedded platforms. He has co-authored over 158 publications in international journals and conference proceedings and actively contributes to interdisciplinary research at the intersection of electronics, sensors, and artificial intelligence.

Sensors and AI at the Edge: Enabling Supervised Autonomy through Embedded Intelligence

The integration of sensors and artificial intelligence (AI) lies at the core of the evolving paradigm of supervised autonomy in human–robot interaction. This lecture explores how advances in sensing technologies, when coupled with emerging AI methodologies—particularly deep edge AI—can enhance autonomy while preserving human oversight and interpretability.
We begin with a general overview of sensor systems, focusing on their roles as the perceptual front-end of autonomous agents. This sets the stage for discussing how data from heterogeneous sensors (vision, tactile, inertial, etc.) can be fused and processed through AI pipelines. The lecture then delves into the distinctions and synergies between cloud-based and edge-based AI approaches, highlighting how deep edge AI enables low-latency, energy-efficient, and privacy-preserving computation directly on embedded systems.
We further analyze how deep AI techniques—including convolutional and transformer-based models—are being adapted for constrained sensor platforms via deep edge AI, allowing for real-time, in-situ decision-making. Throughout the session, the implications of these technologies, especially deep edge AI, for supervised autonomy are critically examined.
To ground these concepts, we present concrete examples from robotics and wearable systems, illustrating how embedded deep edge AI supports collaborative autonomy, task delegation, and shared control. By the end of the lecture, participants will have gained a comprehensive understanding of how sensor-driven deep edge AI architectures can be designed to support ethically and functionally robust human–robot partnerships under the umbrella of supervised autonomy.

Dr Sylvain Calinon is a Senior Research Scientist at the Idiap Research Institute and a Lecturer at the Ecole Polytechnique Fédérale de Lausanne (EPFL). He heads the Robot Learning & Interaction group at Idiap, with expertise in human-robot collaboration, robot learning from demonstration, geometric representations and optimal control. The approaches developed in his group can be applied to a wide range of applications requiring manipulation skills, with robots that are either close to us (assistive and industrial robots), parts of us (prosthetics and exoskeletons), or far away from us (shared control and teleoperation).
Website: https://calinon.ch

Frugal learning of manipulation skills
Despite significant advances in AI, robots still struggle with tasks involving physical interaction. Robots can easily beat humans at board games such as Chess or Go but are incapable of skillfully moving the game pieces by themselves (the part of the task that humans subconsciously succeed in). Learning manipulation skills is both hard and fascinating because the movements and behaviors to acquire are tightly connected to our physical world and to embodied forms of intelligence.

I will present an overview of representations and learning approaches to help robots acquire manipulation skills by imitation and self-refinement. I will present the advantages of targeting a frugal learning approach, where the term “frugality” has two goals: 1) learning manipulation skills from only few demonstrations or exploration trials; and 2) learning only the components of the skill that really need to be learned.

Toward this goal, I will emphasize the roles of geometry, manifolds, implicit shape representations and distance fields as inductive biases to facilitate human-guided manipulation skill acquisition. For the generation of trajectories and feedback controllers, I will discuss how the underlying cost functions should take into account variations, coordination and task prioritization, where various forms of movement primitives can contribute to the optimization process. I will also show how ergodic control can provide a mathematical framework to generate exploration and coverage movement behaviors, which can be exploited by robots as a way to cope with uncertainty in sensing, proprioception and motor control.

Embodying and Uncovering Intelligence in Mechanical Systems

What does it mean to embody intelligence in mechanical systems? I will begin this talk by sharing how my group, the PhI-Lab at TU Delft, engages with this question through the creation of physically intelligent robots. We explore how motor intelligence, extracted from data, can be automatically re-targeted to the mechanics of robotic hands, quadrupeds, and continuum soft robots, and how computational strategies can complement this physical intelligence, equipping these systems with a brain that both harnesses and amplifies their capabilities. In the second part of the talk, I will discuss a question that has increasingly guided our recent work: Can nonlinear (physical) dynamics themselves serve as inductive biases for machine learning? I will explain how this idea inspires our investigation of neuromorphic and other unconventional computing architectures, our efforts to learn physics directly from data, and our approaches to representing learned skills. Although these directions are still in their infancy, they promise new ways to generate intelligence that combine effectiveness, data efficiency, and provable guarantees.

Maximilian Mühlbauer received his B.Sc. and M.Sc. degrees in Mechanical Engineering respectively Robotics from the Technical University of Munich (TUM). He is currently a Ph.D. student at the Technical University of Munich in close collaboration with the department of Cognitive Robotics at the Institute of Robotics and Mechatronics at the German Aerospace Center (DLR). His research focuses on learning, adaptation, arbitration and control of probabilistic Virtual Fixtures, virtual force fields that assist a human operator in teleoperation or hands-on robot control, with applications in space, industrial and medical robotics.

Probabilistic Virtual Fixtures: Arbitration, Automation and Stable Control

Virtual Fixtures (VFs) are software-generated guides that assist a human operator in telemanipulation and human-robot interaction tasks by providing force feedback. Often multiple fixtures with different properties and input modalities are required, to optimally support the human in challenging tasks. The uncertainty measure available from a probabilistic fixture formulation, which utilizes Gaussian Mixture Models to capture the variability and uncertainty in human motion and task execution, allows to allocate control authority between human and fixture based on stiffness adaptation as well as to arbitrate between different fixtures. Such formulation furthermore allows to automate VFs while still keeping the human in control. This lecture explores the required control algorithms to implement such behavior and achieve a stable controller. Additionally, the implementation on haptic devices and hands-on robots is covered. The practicability of these approaches is shown on a set of different use cases in space, industrial, and medical applications.

Marilena Vendittelli is Full Professor of Automatic Control at the Department of Computer, Control and Management Engineering (DIAG) of Sapienza University of Rome. From the same University she received the PhD in Systems Engineering in 1997. From 1997 to 1998 she held a post-doctoral position at the LAAS-CNRS in Toulouse (France), supported by two Marie Curie Research Training grants, on the problem of motion planning and control of wheeled vehicles.
From 1998 to 2001 she was a research associate at DIAG where she then held the role of Tenured Researcher (2001–2016). From 2017 to 2019 she was Associate Professor at the Department of Information Engineering, Electronics and Telecommunications, and in 2020 she rejoined DIAG.

Over the years she has been Visiting Scholar at Carnegie Mellon University (2005), the Courant Institute of New York University (2012), the Simons Institute of UC Berkeley (2016).

From January 2010 to December 2013 she served as Associate Editor for IEEE Transactions on Robotics. She was Registration Chair of ICRA 2007, member of the National Organizing Committee of the 18th IFAC World Congress and continues to serve on the program committee of numerous international conferences, including the IEEE International Conference on Robotics and Automation (ICRA) and the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). She was General Chair of I-RIM 3D 2023, the annual scientific event of the Italian Institute of Robotics and Intelligent Machines.

For the period 2021-22, she was appointed member of the Scientific Council of the Institut des Sciences de l’Information et de leurs Interactions (CNRS Sciences Informatiques, CNRS-INS2I). She was part of several doctoral commissions in Italy and abroad, she was a member of the Research Commission for the evaluation of the research projects of Sapienza (2011-2012) and a member of the panels for the evaluation of European projects under the Horizon 2020 programme.

Ensuring Safety in Supervised Autonomy: Control Challenges for Human–Robot Interaction

As robots become increasingly integrated into unstructured and safety-critical environments, ensuring their safe operation in close proximity to humans is paramount. In these scenarios, fully autonomous systems are often insufficient to handle unpredictability, necessitating human oversight and intervention. This lecture explores how safety can be effectively framed and implemented as a core component of supervised autonomy, where control is shared between humans and autonomous systems.

Particular emphasis will be given to physical safety in human–robot interaction (HRI), while also addressing challenges related to cognitive safety (e.g., predictability, workload), and interface-level safety (e.g., transparent feedback, shared intent).
Using examples from domains such as collaborative manufacturing, assistive robotics, and surgical robotics, we will discuss how safety can be embedded in control architectures through formal methods, adaptive strategies, and risk-aware planning.