Periodic Reporting for period 1 - HRI-CoDeOp (Collaborative, Decision making, and Operational Shared-Control Framework for Human-Robot Interaction)
Periodo di rendicontazione: 2021-11-09 al 2023-11-08
The development of collaborative robots (cobots) has enabled humans and robots to share workspace and interact. The benefit of human–robot interaction (HRI) lies in the largely complementary capabilities of humans and robots; while humans still outperform robots at reasoning, (re-)planning and adapting in unstructured environments, robots excel at performing structured tasks repetitively and precisely. One of the main challenges in HRI is to enable robots to be collaborative partners or assistants to humans, since it requires modelling of human decision-making behaviour and a task-relevant and human-centric control sharing between collaborative partners. In the context of physical HRI, there exists no shared-control framework that enables interaction between the partners on cognitive (decision-making) and operational (physical) levels simultaneously. Therefore, this project aims at making scientific and technological advances towards effective, intuitive, and safe human–robot collaboration with the novel idea of a human-centric hierarchical shared-control framework. The shared-control framework will enable HRI in many application domains ranging from industrial, service, medical, to exploration of dangerous or inaccessible environments. In particular, the Covid-19 pandemic has revealed a pressing need for robot assistants in hospitals because of a shortage of clinical professionals and the risk of infection spreading to hospital staff who interact with infected patients.
The goal of this research is highly relevant for the advancement of collaborative robotics in Europe, especially in its most demanding and unexplored area of physical human–robot collaboration. The results of this research address broadly the following challenges and priorities at a global/European level:
health care – robots that can physically interact with humans are highly needed in the cases of infectious diseases so that other humans are protected from exposure, and in medical interventions, e.g. teloperated surgeries, assisting the elderly in having an active life, or in rehabilitation of disabilities and injuries,
obtaining natural resources – deploying robots in mines, where human safety cannot be guaranteed, is possible within the scope of this research by teleoperating robots through haptic devices,
collaborative manufacturing – cognitive and physical support of humans is necessary in industrial settings, e.g. for flexible assembly, and
AI in robotics – AI is especially beneficial for data-driven modelling and control when humans are included in the loop, since physical models of human behaviour are difficult to obtain.
The overall aim of this project is to develop a holistic, human-centric shared-control framework for human-robot collaboration that will consider the interaction on: (i) the cognitive/decision-making level for a dynamic shared control and conflict resolution, and (ii) the physical/operational level with human-in- the-loop and safety guarantees.
Modeling human decision-making behavior with an online inverse optimal control approach proved to be general enough in cases of human-robot interaction where the robot is passive and human-human interaction in cooperative tasks, implying that it can be suitable for modeling human decision-making behavior in interactive tasks with an active robot as well. Shared-control approach for human-robot collaboration through haptics that considers cognitive/decision-making aspect of the interaction together with lower-level physical (operational) control is necessary for scenarios with multiple tasks that are executed by the human operator and the robot autonomy.
The goal of this research is highly relevant for the advancement of collaborative robotics in Europe, especially in its most demanding and unexplored area of physical human–robot collaboration. The results of this research address broadly the following challenges and priorities at a global/European level:
health care – robots that can physically interact with humans are highly needed in the cases of infectious diseases so that other humans are protected from exposure, and in medical interventions, e.g. teloperated surgeries, assisting the elderly in having an active life, or in rehabilitation of disabilities and injuries,
obtaining natural resources – deploying robots in mines, where human safety cannot be guaranteed, is possible within the scope of this research by teleoperating robots through haptic devices,
collaborative manufacturing – cognitive and physical support of humans is necessary in industrial settings, e.g. for flexible assembly, and
AI in robotics – AI is especially beneficial for data-driven modelling and control when humans are included in the loop, since physical models of human behaviour are difficult to obtain.
The overall aim of this project is to develop a holistic, human-centric shared-control framework for human-robot collaboration that will consider the interaction on: (i) the cognitive/decision-making level for a dynamic shared control and conflict resolution, and (ii) the physical/operational level with human-in- the-loop and safety guarantees.
Modeling human decision-making behavior with an online inverse optimal control approach proved to be general enough in cases of human-robot interaction where the robot is passive and human-human interaction in cooperative tasks, implying that it can be suitable for modeling human decision-making behavior in interactive tasks with an active robot as well. Shared-control approach for human-robot collaboration through haptics that considers cognitive/decision-making aspect of the interaction together with lower-level physical (operational) control is necessary for scenarios with multiple tasks that are executed by the human operator and the robot autonomy.
The researcher gained deeper understanding of how humans make decisions and how they transfer them to actions in the real world. An inverse optimal control (IOC) method to model human decision-making behavior for continuous, physically coupled HRI tasks has been proposed. The IOC approach proved to be general enough to model human-robot interaction where the robot is passive and a human-human interaction in cooperative tasks, implying that it can be suitable for modeling human decision-making behavior in interactive tasks with an active robot.
A shared-control approach that enables decoupling of multiple simultaneous subtasks, has been proposed and implemented. Safety of the proposed approach has been insured with formal passivity guarantees.
A shared-control approach that enables decoupling of multiple simultaneous subtasks, has been proposed and implemented. Safety of the proposed approach has been insured with formal passivity guarantees.
The IOC approach proved to be general enough to model human-robot interaction where the robot is passive and a human-human interaction in cooperative tasks, implying that it can be suitable for modeling human decision-making behavior in interactive tasks with an active robot.
The shared-control approach enables HRI in teleoperation tasks, so that control over one subtask can be transferred from human to robot autonomy, which increases flexibility of shared-control approach and expands state of the art.
The work carried out during this project addressed multiple industrial and societal needs at regional level. The results of this research were motivated by societal challenges, health, ageing population, environment, and security, according to the EU Commission policy on Robotics. The fellowship strengthened international research and innovation cooperation which is one of the H2020 work program priorities in order to strengthen scientific leadership and industrial competitiveness in the EU. With this actionthe UN’s Sustainable Development Goals of Industry, Innovation and Infrastructure (Goal 9), and Partnerships for the Goals (Goal 17) were addressed.
A journal publication which includes the conducted literature review and a discussion on potential future directions in this field, has been drafted and is in preparation for the Proceedings of the IEEE, a leading journal focusing on reviewing technical developments in the field of electrical and computer engineering. Below is the current title and the list of authors of the planned journal publication:
MUSIC, S. & Hirche, S. (under development). Human-centric control sharing for physical human-robot interaction, Targeting the Proceedings of the IEEE journal.
The shared-control approach enables HRI in teleoperation tasks, so that control over one subtask can be transferred from human to robot autonomy, which increases flexibility of shared-control approach and expands state of the art.
The work carried out during this project addressed multiple industrial and societal needs at regional level. The results of this research were motivated by societal challenges, health, ageing population, environment, and security, according to the EU Commission policy on Robotics. The fellowship strengthened international research and innovation cooperation which is one of the H2020 work program priorities in order to strengthen scientific leadership and industrial competitiveness in the EU. With this actionthe UN’s Sustainable Development Goals of Industry, Innovation and Infrastructure (Goal 9), and Partnerships for the Goals (Goal 17) were addressed.
A journal publication which includes the conducted literature review and a discussion on potential future directions in this field, has been drafted and is in preparation for the Proceedings of the IEEE, a leading journal focusing on reviewing technical developments in the field of electrical and computer engineering. Below is the current title and the list of authors of the planned journal publication:
MUSIC, S. & Hirche, S. (under development). Human-centric control sharing for physical human-robot interaction, Targeting the Proceedings of the IEEE journal.