Intelligent and Autonomous Robot Systems

Research goals


Robotics that Assists and Simplifies

IRAS develops technologies for effective, robust and safe robot-environment and robot-human interaction. The integration of sensorial information into behavioral models and controls of robot systems enables the intuitive, robust and flexible use of robots in context-aware applications and a natural, transparent and furthermore safe workspace sharing.

Although robots high flexibility, in terms of motions and programmability, would make them very valuable for SMEs, industrial robots have been mainly developed for large enterprises. SMEs can be supported increasing robots flexibility and ease-of-use, but they still only embody knowledge about their nominal task, with the possible inclusion of sensing as explicitly instructed by a human operator (who has knowledge of the purpose of a particular motion).
The implementation of more advanced sensing and perception, combined with robust behaviour with respect to unforeseen changes, implies extensive engineering and system integration using current flexible solutions. This, however, is neither desirable (the SME would become too dependent on external competence) nor affordable (for reasons of competitiveness) for a SME. Indeed, a SME, typically, does not have a system integrator who can carry out the required day-to-day changes. This implies cognitive capabilities to be embodied in robot systems. However, minimizing risks and maximizing flexibility, human intervention should be allowed whenever necessary.
As a new vision, IRAS proposes a new work paradigm, where humans and robotic systems work cooperatively in all phases of the robot life-cycle, using information, interaction devices, information services and other resources in order to manufacture products for internal or external customers, in which humans and robots can together manage SME manufacturing uncertainties, and in which humans and robotic systems are cooperatively able to learn from each other and to learn from past uncertainty-handling strategies. Progresses towards the realization of the above mentioned vision require the achievements of the following research goal: development of a new kind of industrial robot systems, capable of responding cooperatively (human-robot and robot-robot), safely, intelligently and largely autonomously to gaps in their knowledge and to situations or contexts that have not been specified in their design (that is, they must be adaptable and robust).
The collaboration with humans and/or among robots is the core of the IRAS research path and the main activities span from efficient and safe human-robot workspace sharing to multi-robot motion synchronization, to meet specific process requirements (collision avoidance, optimal cooperation or force exchange), from robustly safe to stable physical interaction with humans and dynamic environments (including other robots). Furthermore, IRAS is involved in the research area of the autonomous environment perception: advanced industrial perception systems, integration of visual, motion and contact sensing, robust cognition-based algorithms capable of managing reduced and disturbed perception situations. Since endowing robotic systems with novel functionalities to adapt themselves to unstructured environments is strictly connected to the control architecture and to dynamic programming processing, IRAS is investigating innovative control architectures and algorithms and intuitive multimodal programming methodologies and devices.

Manual Guidance Device conceived by IRAS within the framework of SMErobotTM project and then finalized as a COMAU SpA product within framework of the "Methods and Devices for intuitive robot programming" industrial Contract. EP2194434; JP2010149273; US2010145520; Finalist of the 2009 EURON/EUnited Robotics Technology Transfer Award

Industrial process analysis and decomposition, in terms of symbolic actions and goals, advanced sensor-based programming and control are the main activities in order to perform autonomously adaptable industrial applications. Furthermore, in order to cover all the required competences, IRAS is deeply involved in the concurrent real-time control area: soft real-time operations (e.g., reasoning, evaluation of affordance, actions selection and execution, adaptation, smooth action transitions, hierarchical and service-oriented control algorithms executions) as well as hardreal-time control (servo control algorithms, reflexive actions, smooth control adaptation and switching).

Interaction control strategies for automatic assembly processes characterized by geometries, materials and tolerances that are not known a priori

The above mentioned research goals clarify why robot-assisted neuromotor rehabilitation for flexible and advanced therapies cannot be any more considered as a utopian expectation.
Since 2008 IRAS is working on the development of different robotic rehabilitation devices and systems for neuromotor rehabilitation of upper and lower extremities, as well as the investigation and optimisation of advanced motion control to enable patientrobot interaction in various application scenarios to prove, and improve, the therapeutic efficacy of novel rehabilitative treatments in the field of neuromotor rehabilitation. In order to tackle with this aspect, the development of sensor-fusion-based control techniques and the development of innovative safety solutions to make patient-robot interaction reliable are key factors to achieve breakthrough solutions and reduce uncertainties and difficulties in defining therapies customized to actual patient's inabilities.

Research activities

Synthesis and Optimization of Parallel and Hybrid Mechanisms. In the last decade IRAS has been working on analysis, design and experimental validation of parallel and hybrid robots, developing both novel prototypes and novel frameworks for the analysis, synthesis and optimization of such machines. Results have been applied to the development of several innovative manipulators and machine tools.
Physical Human-Robot-Safe-Interaction (pHRSI). It is a wide research field; IRAS focuses its efforts in: real time algorithm for safe workspace sharing between robot and human beings; soft-robotics (i.e. intrinsically safe mechanical design and control strategies for robots interacting with human); impedance and interaction control; haptic interfaces and devices; safe and multimodal devices for easy and intuitive robot programming; open, modular and reconfigurable pc-based real-time control platforms that allow the use of robots in application where autonomous and adaptable human-machine interaction or machine-environment interaction or machine machine interaction is mandatory; dynamic multi-robot planning and coordination.
Robot-Environment Interaction. A comprehensive theoretical and experimental coverage of interaction control problems, starting from the mathematical modelling of robots interacting with complex dynamic environments and proceeding to various concepts for interaction control design and implementation algorithms at different control layers, is mandatory to achieve a high autonomy of IRs in performing technological application (e.g. deburring, polishing, grinding, guidance, etc.).
3D Vision. Vision technology is the IRAS enabling technology in the field of sensor-based robot programming and control. The acquisition of objects position and shape in loosely structured environments, automatic motion tracking, automatic object recognition and feature extraction, reverse engineering approaches are fundamental to achieve a robust automation of industrial process. IRAS is mainly developing novel 3D Laser-Scanning Technologies and devices focusing on efficient off-line calibration algorithms, rather than on mechanical and/or optical high accuracy, that get an extremely fine model of the whole acquisition workspace while using low-cost mechanics.

Clinical Trial for RehaMSR - Rehabilitation Multisensory Room (RehaMSR) natively designed for upper-limb robot-assisted rehabilitation - developed by IRAS in cooperation with Ospedale Valduce. The system integrates several devices for multi-sensorial feedback. The platform is used by medical personnel to both investigate human neuromotor behaviours and to assess innovative rehabilitation protocols

Robotics for medical applications. IRAS competences and research activities enable a cross fertilization with medical applications field, in particular for robotic rehabilitation and surgery issues. The development of advanced medical-robots involves both integrated mechatronic design and innovative interaction control algorithms, in order to cope with both the complexity of human structure and the necessity to realize robots able to safely and autonomously perform advanced rehabilitation therapies and/or surgical acts.
Beside the activity conceived and promoted within Regional, National, European and International research frameworks, IRAS is strongly committed to promoting the knowledge value chain by contributing actively to create specific conditions for technology transfer and value and competitive advantage creation in National and European firms - often SMEs providing both effective business consulting services and the design and pre-industrial development and testing of highly innovative products.

Main research projects

The European Robot Instiimative for Strengthening the Competitiveness of SMEs in Manufacturing, EU FP6-NMP (2005-2009)
Hyperflexible And Autonomous Industrial Robot System, Regione Lombardia - Cooperazione Scientifica Internazionale (2010-2011)
Smart robotics for high added value footwear industry, EU FP7 (2010-2013)
Active Constraints Technologies for Ill-defined or Volatile Environments, EU FP7-NMP (2011-2015)
Sistemi di produzione flessibili ed eco-efficienti per veicoli su gomma, Ministero per lo Sviluppo Economico (2012-2015)
Robust and FLEXIble CAST iron manufacturing, EU FP7-FoF (2012-2016)
Cluster Fabbrica Intelligente, Cluster Tecnologici Nazionali, Ministero dell'Istruzione, dell'UniversitÓ e della Ricerca (2013-2016)
Recupero post-Ictus: Piattaforma per la RiabilitazionE Neuromotoria orientata al Domicilio, Accordo Quadro di collaborazione tra Regione Lombardia e il Consiglio Nazionale delle Ricerche (CNR) (2013-2016)
Contratti Industriali:
Metodi e device per la programmazione intuitiva di robot, COMAU S.p.A. (2007-2008)
Metodi e device per la programmazione sensor-based di robot, CIA Srl (2008-2010)
Metodi e device per il controllo del processo di taglio laser, Adige S.p.A. (2010-2013)
Metodi e device per la riparazione di grandi strutture in composito, Alenia Aermacchi S.p.A. (2013-2016)

3D Laser Scanner (device and methodology) for Heritage developed by IRAS within the framework of the "Methods and Devices for sensor-based robot programming" industrial contract and of the HyFlexAIS project

key words

Human-Robot Interaction Control; Environment-Robot Interaction Control; Robot Design, Dynamics and Control; Multi Robot Cooperation.


Eng. Ph.D. Lorenzo Molinari Tosatti
IRAS Group responsible
stiima.milano [at]