Doctoral Research Seminar on "On the identification of industrial robots: from a standard approach to most sophisticated techniques"
The next Doctoral Research Seminar is given by Dr. Alexandre Janot (French national aerospace research center, ONERA) on "On the identification of industrial robots: from a standard approach to most sophisticated techniques"
Abstract: Nowadays, the democratization of model-based control strategies in robotics raises new challenges to researchers. Since the reliability of such control paradigms strongly depends on models' consistency and precision, accurate knowledge of system parameters turns out to be of paramount importance. This trend naturally led to an intensification of research efforts in the field of parameters identification over the past two decades. In the case of rigid robots, dynamic parameter identification algorithms are usually based on a rigorous analysis of joint motion derivatives and torques over a predefined time horizon. The robot is tracking a trajectory that "excites'' the different components of its dynamic model and the residuals in terms of joint motions and torque are used to refine iteratively the parameters estimates. The most common approach for offline dynamic parameters identification is the Inverse Dynamic Identification Model with Least-Squares estimation (IDIM-LS). Both efficient and easy to implement, IDIM-LS and its variants, however, lacks noise immunity and exhibit a strong dependence on the conditioning of the robot's trajectory. Over the past two decades, much research effort has focused on solving these issues. This presentation will actually focus on the instrumental variable (IV) identification technique.
Short bio: Dr. Alexandre Janot is a research engineer, specialized in robotics and system identification at ONERA, the French national aerospace research center.
4 December 2019 at 16:30h, ICS Karlstraße 45, 2. Floor, room 2026