Filippo Cavallo (Università di Firenze)

Moderator: Andrea Orlandini (CNR-ISTC)

Panelists: Irene Aprile (Fondazione Don Gnocchi), Giuseppe Turchetti (Scuola Superiore Sant'Anna), Manuela Chessa (Università di Genova), Marco Controzzi (Scuola Superiore Sant'Anna), Gionata Salvietti (Università di Siena), Antonio Sgorbissa (Università di Genova)

Organizers: Simone Traverso (Istituto Italiano di Tecnologia), Nicolò Boccardo (Istituto Italiano di Tecnologia), Matteo Laffranchi (Istituto Italiano di Tecnologia)

Website: https://simonetraverso2.wixsite.com/website

Motivation: Lower-limb exosuits are emerging as a new generation of wearable robotic systems capable of providing assistance while preserving the user's natural biomechanics and promoting active participation during movement. Compared with conventional rigid exoskeletons, lightweight wearable technologies offer improved comfort, usability, reduced metabolic burden, and long-term integration into activities of daily living. Recent advances in clinical experience, compliant actuation, cable-driven transmissions, adaptive control algorithms, machine learning, wearable sensing, and combined multimodal approaches based on Functional Electrical Stimulation (FES) are opening new perspectives for personalized assistance, rehabilitation, and long-term support in real-world environments. Bringing together these complementary technologies represents a key step toward the development of more versatile-effective, and clinically translatable wearable robotic systems.

Goals: This workshop aims to bring together researchers from robotics, biomechanics and control engineering to discuss the latest developments and future challenges in lightweight wearable robotics.

The workshop will:

• present the complementary clinical roles of rigid exoskeletons and lightweight exosuits
• present the state of the art in lightweight lower-limb exosuits mechatronics;
• explore multimodal control architectures for gait assistance;
• present adaptive oscillators and machine-learning techniques for gait phase estimation and intention detection;
• discuss the integration of Functional Electrical Stimulation with robotic assistance;
• promote collaboration between academia, clinicians and industry.

Main Topics:

• Clinical experience with wearable robots
• Lightweight lower-limb exosuits
• Cable-driven wearable robots
• Adaptive gait assistance
• High-level control architectures
• Adaptive oscillators
• Machine learning for locomotion
• Functional Electrical Stimulation

Organizers: Donato Caramia (Politecnico di Bari), Enrico Donato (Scuola Superiore Sant'Anna), Egidio Falotico (Scuola Superiore Sant'Anna), Paolo Roberto Massenio (Politecnico di Bari), David Naso (Politecnico di Bari)

Website: https://www.santannapisa.it/en/institute/biorobotics/brair-lab

Robot manipulation is undergoing a paradigm shift: Vision-Language-Action models, world models, and generative policies now manipulate novel objects from natural-language instructions with striking generality. Yet a gap remains between impressive demonstrations and the reliability real deployment demands. The central question is no longer what robots can manipulate, but whether we can trust them to do it reliably. This workshop explores what makes learned manipulation policies trustworthy across three axes. Robustness in physical interaction, from fragile and deformable objects to long-horizon assembly without error accumulation. Verification methodologies, such as digital twins and reproducible benchmarking, needed to validate policies before deployment. And recovery: detecting and responding to failures when execution deviates from plan, particularly through human-in-the-loop execution and online replanning. By bringing together the Robotics, Artificial Intelligence, and Industrial Automation communities, the workshop aims to frame trustworthy manipulation as a measurable engineering property and chart a shared roadmap from research to deployment.

Organizers: Giulia Bodo (Istituto Italiano di Tecnologia), Nicolò Boccardo (Istituto Italiano di Tecnologia), Matteo Laffranchi (Istituto Italiano di Tecnologia)

Website: https://giuliabodo.wixsite.com/irim-robotic-rehab

This workshop explores the multidisciplinary challenges involved in translating robotic upper- and lower-limb exoskeletons from laboratory research to clinical rehabilitation. Through invited talks by engineers and clinicians, topics including mechatronic design, control strategies, clinical validation, rehabilitation protocols, and user-centered design will be addressed. Interactive discussions will foster collaboration and identify future directions in rehabilitation robotics.

Organizer: Emanuele Menegatti (Università di Padova)

Website: https://www.dei.unipd.it/~emg/Workshop-Assistive-Robotics-2026/workshop_assistive_robotics_2026_2.html

Assistive robotics, devices, and technologies are playing a critical role in addressing the needs of an aging population, particularly in promoting independence, enhancing mobility, and reducing risks associated with falls. Among these challenges, fall prevention remains a major concern for the elderly, where fall-related injuries are a leading cause of reduced mobility and loss of autonomy. The psychological impact of falls often leads to a fear of falling, which can confine individuals to their homes and contribute to musculoskeletal decline due to prolonged inactivity.

Lorenzo Masia (Technische Universität München): Embodied Intelligence for Human Augmentation and Collaborative Robotics

Organizers: Francesco Scotto di Luzio (Università Campus Bio-Medico di Roma), Marco Rosa (PAL Robotics), George Moustris (National Technical University of Athens)

Website: https://scottodiluzio.github.io/hri-workshop-irim2026/

Human-Robot Interaction (HRI) is becoming one of the defining research challenges for the next generation of intelligent robotic systems. As robots progressively move beyond structured industrial environments into healthcare, rehabilitation, public spaces, homes, and collaborative workplaces, their ability to interact naturally, safely, and effectively with humans is no longer an optional capability but a fundamental requirement. Achieving this vision demands robots that can perceive complex environments, understand human intentions, communicate through multiple modalities, and adapt their behavior in dynamic and uncertain scenarios. Recent advances in multimodal perception, cognitive robotics, generative AI, foundation models, and embodied intelligence are rapidly transforming the capabilities of interactive robotic systems. However, translating these advances into reliable, scalable, and deployable solutions remains a major scientific and engineering challenge. Intelligent HRI systems require the seamless integration of heterogeneous components, including perception, language understanding, planning, reasoning, control, and learning, within software architectures that are modular, reproducible, maintainable, and suitable for long-term deployment. This workshop places Human-Robot Interaction at its core while focusing on the software infrastructures that enable the transition from research prototypes to real-world robotic applications. In this context, ROS 2 is considered the enabling middleware that supports interoperability, modularity, and reproducibility across diverse robotic platforms and application domains. A distinctive feature of the workshop is its emphasis on experiences emerging from national and European collaborative research projects, including Horizon Europe and other large-scale initiatives. These projects demonstrate how common software frameworks, open-source ecosystems, and shared engineering practices facilitate collaboration among multidisciplinary teams, accelerate technology transfer, and promote the development of interoperable robotic systems capable of operating in real-world environments.

Topics of interest include, but are not limited to:

• Software architectures and middleware for Human-Robot Interaction
• ROS 2 for intelligent robotic systems
• Human-aware perception and multimodal sensing
• Natural language interaction and conversational robotics
• Integration of Generative AI and Foundation Models
• Cognitive architectures, planning, and decision making
• Physical and social Human-Robot Interaction
• Healthcare, rehabilitation, and assistive robotics
• Open-source robotics software and reproducible research
• Experiences from Horizon Europe and collaborative research projects
• From laboratory validation to real-world robotic deployment

Objectives

• To discuss emerging scientific and technological challenges in developing intelligent Human-Robot Interaction systems capable of operating in complex real-world environments.
• To present state-of-the-art software architectures, AI technologies, and engineering methodologies that support the development and deployment of interactive robotic systems.
• To showcase experiences from collaborative research projects, emphasizing reusable software solutions, interoperability, and technology transfer across application domains.

Organizers: Lucia Pallottino (Università di Pisa), Cristian Secchi (Università di Modena e Reggio Emilia)

Website: https://sites.google.com/view/i-rim26-multi-robot-workshop/speakers

The workshop focuses on the deployment of multi-robot systems in real-world industrial and logistics scenarios.

While autonomous mobile robots, automated guided vehicles, and heterogeneous robotic fleets are increasingly adopted in warehouses, factories, and other operational environments, several open challenges still limit large-scale deployment. These include fleet coordination, interoperability among robots from different vendors, standardization, safety, reliability, trustworthiness of AI-enabled solutions, and the transition toward progressively automated industrial environments.

The goal of the workshop is to create a focused discussion between academia and industry, combining scientific perspectives with concrete deployment experience. Particular attention will be devoted to interoperability and fleet orchestration, which are becoming central issues as industrial facilities increasingly rely on robots, software platforms, and automation technologies from multiple providers.

Organizers: Christian Tamantini (Istituto di Scienze e Tecnologie della Cognizione, Consiglio Nazionale delle Ricerche), Marta Gherardini (Scuola Superiore Sant'Anna), Chiara Livolsi (Scuola Superiore Sant'Anna), Andrea Tigrini (Università Politecnica delle Marche)

Website: https://fit4medrob-workshop.github.io/fit4med_irim_ws_2026/index.html

Fit4MedRob is a major PNC-funded initiative at the intersection of robotics, bioengineering, rehabilitation medicine, and clinical innovation, bringing together an extensive national network of clinical centres and bioengineering research groups across Italy. Through its three core missions: Clinical Translation and Innovation, Biorobotic Platforms and Allied Digital and Next-Generation Components, the Italian research community involved in the Initiative has made massive efforts to address the key challenges behind the clinical applicability of modern AI-driven and robot-assisted rehabilitation treatments. As the project approaches its final stages, there is the need to share the Italian ecosystem's contributions with the broader European and international communities. Crucially, this workshop aims to highlight the advanced competencies developed by early-career researchers, who are poised to drive future innovation and ensure the sustainability of these technologies beyond the project's lifespan. The primary objective is to provide a concise yet comprehensive overview of the project's outcomes. The session will bridge the gap between high-level strategic visions, presented by senior researchers who orchestrated the three missions, and the tangible, cutting-edge advancements delivered by young researchers.

Elena De Momi (Politecnico di Milano)

Demos and exhibition continue. The list of demos is in the Demo + Exhibition session of Friday, October 23.

Organizers: Simona Casini (Università di Pisa), Giulio Pisaneschi (Università di Pisa), Andrea Caiti (Università di Pisa)

Website: https://sites.google.com/view/robotsminds/home-page

What does it mean for a robot to be intelligent? This workshop strives to answer this question from different perspectives. Although nowadays intelligence in robotics is frequently associated with artificial intelligence and learning-based schemes, intelligent behaviours may also emerge as natural consequences of cognitive processes, robust and optimal control strategies, physical embodiment, morphology, and continuous interaction with the environment.

The workshop brings together four nearly complementary viewpoints on robotic intelligence: cognitive intelligence, artificial intelligence, control-theoretic intelligence, and embodied or physical intelligence. Cognitive intelligence frames robotics through the lens of cognitive science, exploring how notions such as perception, representation, reasoning, intentionality, and goal-directed action can contribute to the understanding and design of intelligent robotic systems. Artificial intelligence focuses on the computational and learning mechanisms that support autonomous behaviour. Control-theoretic offers a perspective based on feedback, regulation, robustness, adaptation, and optimality. Embodied or physical intelligence shifts the focus toward the robot’s body, morphology, material properties, and interaction with the environment as sources of intelligent behaviour. By bridging these perspectives, we aim to explore how different notions of intelligence can shape the design, analysis, and understanding of robotic systems, thereby contributing to a unified vision for the machines of the future. With this regard, the workshop will address a central question for robotics and intelligent machines: what makes intelligence robotic? Is it located in its internal representations, learning mechanisms, feedback loops, body, or in the interaction among all these potentially orthogonal dimensions?

Organizers: Sara Federico (Università degli Studi della Campania "Luigi Vanvitelli"), Francesco Cufino (Università di Napoli Federico II), Marco Costanzo (Università degli Studi della Campania "Luigi Vanvitelli"), Mario Selvaggio (Università di Napoli Federico II)

Website: https://sites.google.com/view/learning-or-modeling-irim2026/home-page

Despite the remarkable progress achieved in recent years, dexterous object manipulation remains one of the most challenging problems in robotics.

Indeed, prehensile, non-prehensile, and in-hand manipulation all rely on accurately resolving contact dynamics, frictional effects, and physical interactions, which makes both simulation and control particularly demanding and difficult to apply robustly to real-world scenarios. Model-based approaches offer interpretable, predictable, and well-characterized control laws,but their dependence on accurate physical models limits their reliability in unstructured environments. Learning-based algorithms, on the other hand, have demonstrated strong generalization capabilities in precisely these unstructured settings, at the cost of extensive training (e.g., long GPU-based training for reinforcement learning) or large data requirements (e.g., imitation learning), and often by sacrificing the transparency that characterizes their model-based counterparts. The recent trend shows an increasing shift towards learning-based methods to achieve greater generalization. However, understanding when model-based or learning-based methods are preferable — and how their complementary strengths might be combined — remains an open and active research question.

This workshop will examine two major paradigms for robotic manipulation — advanced learning-based and well-established model-based methods — analyzing their respective strengths, limitations, possible combination of the two, and application domains.

Organizers: Michele Canepa (Istituto Italiano di Tecnologia), Nicolò Boccardo (Istituto Italiano di Tecnologia), Samuele De Giuseppe (Istituto Italiano di Tecnologia)

Website: https://sites.google.com/view/biointerenect-i-rim

Rehabilitation robotics and neurotechnology are rapidly evolving from standalone assistive devices toward integrated ecosystems capable of continuously exchanging information between users and robotic systems. This transition is driven by multimodal bidirectional platforms that combine physiological sensing, artificial intelligence, embedded computing, neural interfaces, and sensory feedback within a common modular architecture. Such platforms enable technologies developed for different applications, including upper- and lower-limb prostheses, wearable robots, functional electrical stimulation, implantable neural interfaces, and neuro-urological systems, to share hardware, software, and control methodologies, accelerating innovation and clinical translation. Bringing together these complementary technologies represents a key step toward the development of reusable, interoperable, and patient-centered rehabilitation solutions.

Chorus Cafè, Via della Conciliazione 4, 00193 Roma

Demos and exhibition continue. The list of demos is in the Demo + Exhibition session of Friday, October 23.