DC15: User-centred model predictive control approach for robotic assistance
Project: User-centred model predictive control approach for robotic assistance (WP4)
Host institution: WR (Italy)
Supervisor: F. Salsedo
Co-supervisor(s): A. Filippeschi (SSSA), Prof. H. Hallez (KU Leuven)
Objectives:
Control algorithms for exo-based assistance can highly benefit from an accurate and updated identification of human parameters and behavior. The objective of this DC is to devise and test new online calibration/identification methods for human-robot systems and a model predictive controller that accounts for the user’s physical characteristics and behavior to provide usercentered, personalized assistance.
Expected Results:
Develop an online identification method for human-robot systems that informs the wearer of the segment to be excited and a subject-specific torque model predictive control approach for wearable robots, utilizing a minimal number of sensors.
Planned secondment(s):
- Sant’Anna (3 months, M13-M15): Validation of online identification method for human-robot-specific calibration method with A. Filip-peschi (KPI: joint journal paper) (KPI: joint journal paper)
- KU Leuven (3 months, M29-M31): sensor fucsion algorithm for online identification method, with H. Hallez (KPI: joint conference paper)
Enrolment in Doctoral degree: Doctoral School of Sant’Anna School of Advanced Studies (Italy)
Required profile: Applicants should hold an MS degree in Engineering, preferably in Robotics, or Computer Science, and have a solid background in control and/or optimization. Good communication skills in English are mandatory.
Desirable skills/interests: Applicants should have a genuine interest in human-robot interaction research, and be willing to merge an analytical approach to control with experimental activities to test their algorithms on exoskeletons with humans in the loop. Programming skills (C/C++, Matlab/Simulink) are fundamental to successfully achieving the objectives.