The SHAARE associate team is created in 2024 between the IRiS lab at KAIST and the Auctus team at Inria, to share our complementary methodological orientations in haptic shared control. Together, we aim at developing shared-control approaches that, either, better guide the human through adaptive haptic guidance, or adjust the robot behavior according to the human gestures.
The IRiS lab develops virtual-fixture feedback, generated from a task description given by the user. Our partner also studies machine learning strategies to transfer skills from the human to the robot in haptic teleoperation.
The Auctus team focuses on the formulation of shared controllers under safety, human-centered, and task/environement constraints. They build upon model-based optimization problems to both generate the force feedback from virtual guidance and the robot command from the joint human inputs and robot skills.
These different approaches should together form a comprehensive shared haptic framework, where the human is guided toward optimal gestures and helped by the robot taking more autonomy on the task. This collaboration specifically aims at augmenting the robot assistive behavior in haptic teleoperation by simultaneously improving haptic guidance and transferring expert skills to the robot. The works focuse on:
- Generating optimal virtual fixtures from the task semantics, user preferences, and robot/environment constraints;
- Developing robust and adaptive robot assistive skills, modeled through task-oriented force-motion control primitives;
- Transferring expert skills from the human operator to the robot during the interaction.
People involved
at Inria
- Alexis Boulay (Auctus)
- Vincent Padois (Auctus)
- David Daney (Auctus)
- Margot Vulliez (Auctus, Investigator)
at KAIST
- Kwang-Hyun Lee (IRiS)
- Donghyeon Kim (IRiS)
- Seong-Su Park (IRiS)
- Jee-Hwan Ryu (IRiS, Investigator)