Cerca nel sito
DiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporamaDiaporama
Accueil
ICube   >   Agenda : PhD : "Online models learning for adaptive assistance to teleoperated minimally invasive surgery."

PhD : "Online models learning for adaptive assistance to teleoperated minimally invasive surgery."

3 maggio 2023
09h30
IHU, 1 place de l’hôpital 67000, Strasbourg

PhD defence : Thibault Poignonec

Team: RDH

Date & time : Wednesday 3rd May at 9 :30 at IHU of Strasbourg, meeting room Hygie (RDC)

Title : "Online models learning for adaptive assistance to teleoperated minimally invasive surgery."

Abstract : This thesis explores assistive robotic behaviors for the realization of robot-assisted minimally invasive surgery (MIS).

We present novel approaches to perform the online backlash model identification of cable-actuated flexible endoscopes. Backlash in cable transmission degrades open-loop positioning accuracy and increases the cognitive load of the practitioner. However, its compensation requires an accurate model identification, which should ideally be done in-situ, i.e., just before or even during the surgical procedure. We propose several methods that can be applied to different relevant robot architectures and scenarios.

In the second part of this thesis, we investigate the online learning of the task and robot model parameters to continuously improve assistance to the operator. We consider the case of haptic guidance during remote teleoperation, a scenario especially relevant to surgical robotics. In this context, we avoid relying on exteroceptive sensors as the main source of information as they could be limited or intermittently unavailable. Instead, we rely on the presence of the operator to extract the information necessary for learning. The algorithms we propose are evaluated in different simulated and real telerobotic scenarios, demonstrating the applicability of the methods to online registration problems.

Keywords: Medical robotics; Haptic guidance; Registration; Backlash; Online parameters learning.

The jury will be composed of:

  • Arnaud Lelevé (Prof., INSA Lyon) – Rapporteur
  • Paolo Robuffo Giordano (DR CNRS, INRIA Rennes) – Rapporteur
  • Caroline Essert (Prof., Université de Strasbourg) – Examinatrice
  • Phillipe Catin (Prof., Université de Bâle) – Examinateur
  • Nabil Zemiti (MdC, Université de Montpellier) - Encadrant de thèse

À la une

Le dépôt des candidatures pour les postes d’enseignants-chercheur est ouvert. Les offres sont...

Flux RSS

Flux RSS