Thesis: Modelling and control of robotic flexible endoscopes using machine learning

PhD defense: Rafael ALELUIA PORTO

Title: Modelling and control of robotic flexible endoscopes using machine learning

Team: AVR

Abstract: In the context of minimally invasive surgeries, flexible endoscopes and cable-driven flexible instruments are essential tools. In robotic systems, cable transmission allows to control distal effectors inside the patient from external motors located at the proximal side. However, non-linearities are introduced due to the interaction between the cables and the sheaths inside the endoscopic tools shafts. A common effect that can be observed is the presence of a hysteresis characteristic between the position of the cables at the proximal side and the distal tip position. In this thesis, a novel position inverse kinematic model that can take into account complex non-linearities is proposed for both coupled and uncoupled flexible systems, as well as an extension for any kind of robotic system. The core concept of the developed techniques is to combine classic kinematic modeling with supervised machine learning techniques. This allows for a precise, yet efficient, inverse kinematic model that can handle hysteresis effects. The positioning accuracy is improved compared to other learning-based approaches with a much faster training phase as has been shown on experiments on the STRAS robotic platform.

This defense will be held in English and will take place on Thursday the 14^th of January 2021 at 9:30am. This defense will be partially virtual and will take place in the presence of the jury members listed below:

• M. Michel DE MATHELIN, Professeur, Université de Strasbourg, Director
• M. Florent NAGEOTTE, Maître de Conférence, Université de Strasbourg, Supervisor
• M. Tanneguy REDARCE, Professeur, INSA Lyon, Reviewer
• M. Brahim TAMADAZTE, Chargé de Recherche, CNRS, FEMTO-ST, Reviewer
• M. Nabil ZEMITI, Maître de Conférence, Université de Montpellier, Examiner
• M. Edouard LAROCHE, Professeur, Université de Strasbourg, Examiner