Context
A postdoctoral position is opened in collaboration between the AUCTUS team at Inria and the IRiS lab at KAIST, about shared control and skill transfer in haptic teleoperation.
The postdoc contract will have a duration of one year, later extendable. The postdoctoral fellow will be recruited at the Inria center of the University of Bordeaux (AUCTUS team), where he/she will start the research project. A six-month exchange period will be planned at KAIST (IRiS lab) to share the works and progresses.
Research activities
Recent shared-autonomy concepts have been proposed in the literature to transfer part of the task from the human operator to the robotic agent. These approaches range from complementary and predefined subtask allocations to adaptive shared-control methods. Focusing on this second and more flexible paradigm, the postdoctoral project aims at improving shared control in haptic teleoperation, to make the robot gain responsibility on the task, adapt its behavior with respect to the human intent, and ultimately act as an effective collaborator.
According to their background and preferences, the postdoctoral scholar will direct the research works to tackle the two following key challenges:
- Generating robust haptic guidance from generic robot skills (robot control primitives). An important research corpus has shown that adding active constraints to the haptic feedback can either guide the operator toward an optimal gesture during the remote task execution, or convey robot-related or contextual information. However, haptic guidance is commonly based on geometric or motion-related virtual fixtures, highly sensitive to the environment uncertainties. We will build upon robust robot control primitives to generate adaptive and generic haptic guidance. By locally exploiting the robot skills and autonomy, controlling its behavior through sensory-based force-motion task primitives, the guidance can directly feedback the robot adaptive behavior, robust to the environment changes. We will particularly focus on maintaining a safe and stable interaction with this new concept of skill-based adaptive haptic guidance.
- Online skill transfer during teleoperation. Assisting Humans in their activity requires the robot to master elementary gestures or actions, and to accurately control physical interactions with the environment. Despite their sensorimotor ability, robots lack of functional autonomy. This second activity aims at increasing robot manipulation skills by taking advantage of the rich multisensory data available during a task performed by the human expert in haptic teleoperation. The first issue will be to transfer new skills to the robot during the teleoperation from the collected data. It can be either done through segmentation and parameterization methods, to extract basic force-motion patterns and encode new task primitives, or through an online learning by demonstration. We will pay a particular attention to ensure robustness of the new robot skills, through distribution-based task descriptions (Gaussian Mixture Models, Locally Weighted Regression, Hidden Markov Models, etc.).
Skills
The candidate should have graduated with a PhD in robotics. He/she should have solid skills in robotic control, programming (C++, Python), and kinematic/dynamic modeling. Their past works should demonstrate a proper balance between fundamental works and experimental studies. A background in haptics or telerobotics is highly recommended. Some additional experiences in planning or machine learning would be appreciated.
Instruction to apply
To apply at this Inria-KAIST postdoc, please email the following documents:
- A detailed CV with a description of the PhD and a complete list of publications with the two most significant ones highlighted.
- A motivation letter with a description of the candidate interests and planned methodology to tackle the research project.
To the investigators: