About the Lab
J-LAIR is a cooperation of multiple research groups: Robotics and Mechatronics, Biomechatronics and Rehabilitation Technology, Precision Engineering and Human Media Interaction. In different projects, other research groups also join the lab to jointly work on new solutions.
The ultimate objective of the J-LAIR is to realize stand-in robotic assistance. We consider any daily life setting with all its uncertainties and variations, be it in healthcare, in industry, or at home.
We put a robotic system in any such environment, and we enable the system to assist in the following two ways:
- by specifically designing system aspects (hardware, software) based on prior knowledge of the situation, and
- by teaching the system on the spot how to (better) assist in any given task. Strong design will ease the required learning, whereas learning will allow flexibility towards problems that are challenging to design beforehand. With this we aim to tackle problems arising from staff shortages and workload.
RESEARCH LINES
To achieve the above, we envision the following research lines:
LEARNING FROM HUMAN FEEDBACK FOR SHARED AUTONOMY How can we transform human input and feedback into meaningful teaching signals for a robotic system? Various paradigms will be considered, including learning from demonstration in which a task demonstration by a human is collected through tele-manipulation and then used in offline learning algorithms, as well as online learning in which a robotic system continuously adjusts to a human in the control loop. We consider a spectrum from full human control through tele-manipulation, to shared control, to full robot autonomy, and aim to continuously shift towards more robot autonomy by learning from humans.
CO-DESIGN PRINCIPLES FOR TRANSPARENT INTERACTIVE SYSTEMS Robot systems for interactive tasks are often an assembly of off-the-shelf components, possibly leading to suboptimal performance. This research line proposes an integrated systematic design approach for high fidelity robot-human and robot-environment interaction, based on application specific requirements. Examples include, highly force transparent robot grippers and responsive master-slave devices. The focus lies on leveraging enabling technologies in mechanisms design, sensor design, controls and modeling.
MODELING OF ROBOT-ENVIRONMENT INTERACTIONS To enable robots to learn over the full control spectrum, from full human control to full robot autonomy, a robotic system must understand its environment and the physical and social interactions with it. This requires models that capture the task at hand, the expected effects of environmental interactions, and the intention and expected (re)actions of human operators. This includes providing safety in robot-environment interaction scenarios for planned or unplanned interactions between robot and environment, trading risks and performance.
EMBEDDING IN REAL-WORLD SCENARIOS This line investigates the effect of design choices on stakeholders, to allow informed design. The design choices made in the other lines of research impact the way in which the system will be embedded in real-world scenarios and influence the experience of- and perception by stakeholders.
PROJECTS
Exploring the Future of Care with Robotic Avatars At the Joint Lab for Interactive Robotics, we are currently exploring how Robotic Avatars can make a meaningful difference in healthcare. These robots can help ease professionals' workloads and help people stay connected with loved ones.
In collaboration with a care organization, two enthusiastic bachelor students are diving into what it takes to successfully introduce these robotic avatars into real-life care settings. One student looks at the bigger picture (laws, regulations, and funding) while the other interviews the people who matter most: clients, family members, and care professionals.
Their research is part of our ongoing effort to understand how robotic innovations fit into the real world. By learning from stakeholders, we can design robots that genuinely support and adapt to the needs of people.
Contact
