Missions

Mainly research and development

Activités

1. Design of high precision robotic motion emulator set-up
The current set-up for the knee motion emulator and training consists of two robotic arms (Franka-Emika). These two robots were chosen solely on the basis of their ability at CNRS for other research purposes. Also their relative positioning and their initial configurations were chosen ad-hoc by error and trial. We will reconsider this solution by screening all possible robotic arms that can serve the purpose and considering further extensions of our solution (i.e., beyond the knee) to serve other medical use-cases (e.g., hip, elbow…). Also an utmost calibration set-up will be devised to warrant the highest possible precision of the measurement outcome. We will use tools from robotic kinematics calibration, design optimization and manipulability to devise the best configuration set-up for all possible motions, emulating those of the implants once implanted. Passive tool changers will be used to plug on the robot any implant whose location inside the human body is reproduced by its integration in 3D printed organs (3D printing is made by Pro3D of the University of Montpellier (https://centre-pro3d.fr/)). Concerning the robotic systems, French or EU solutions will be considered at first priority. The outcome of this task is the robotic setup for data training that can adapt to several implants considered in this project.
2. Implementation of the motion emulator and its user interface
This task offers a user-friendly interface to be used by none-robotic experts. Our current set-up uses mc_rtc to program the two robotic arms we possess to emulate the motion of the knee. Any new robotic system can be integrated seamlessly using mc_rtc. The power of mc_rtc lies in its ability to adjust and adapt to any application and, as a framework, it can integrate other software modules such as the AI for learning. We will integrate the explainable AI module such that only one software serves the purpose of the entire system. The software components developed in ROLKNEEMATICS will be open source and made available to the community. The outcome of this task is an enhanced software framework and package serving the purpose of user-friendly robot programming to emulate motions of implants when implanted (even the non plausible ones), and to train the explainable AI with ground truth kinematics of the robotic system.
More details can be provided in interviews and found on the project website: https://sites.google.com/view/rolkneematics/home

Compétences

PhD or Master in Robotics

Contexte de travail

As part of the state-funded robotics challenge program, we have developed a turnkey robotic solution for the learning phase of a device for measuring the real-time and three-dimensional kinematics of a prosthetic knee. This device comes in the form of a knee brace developed by the partner BoneTag https://bonetag.eu/ This knee brace can be worn by any patient implanted with a complete or partial knee prosthesis, or with metal parts (e.g. braces, screws...) following knee repair surgery. The knee brace can be used outside the hospital environment, and the movement data can be transmitted (via the internet or a secure line) to the treating physicians for prevention and post-operative diagnosis.
For more than a year, LIRMM (www.lirmm.fr) and BoneTag have been working to validate the learning of kinematics using two robotic arms for other research subjects. Each robot carries a part of the implanted knee at its endpoint (i.e., one of the two robots carries the femoral part and the other the tibial part). The idea is that the two robots, synchronized, generate a multitude of precise pre-programmed movements that emulate the movements of the implanted knee (including unlikely situations) within the knee brace. The latter emits sensor signals according to the position and spatial orientation of the two parts of the knee. Thanks to the kinematics of the two robots and precise 3D models of the offsets at the end points, the position and orientation of any prosthesis, pin or screw are precisely calculated. The sensor signals and kinematic calculations, carried out using the encoders and geometric parameters of the two robots, are the data of an AI that learns the exact kinematics of prostheses, pins or screws. These can then be implanted on any patient. The knee brace will then be able to reconstruct the kinematics of the implanted patient's knee in real time. Proofs of concept have already been conducted and an advanced version of the smart knee brace has been validated (this is a world first).

Le poste se situe dans un secteur relevant de la protection du potentiel scientifique et technique (PPST), et nécessite donc, conformément à la réglementation, que votre arrivée soit autorisée par l'autorité compétente du MESR.

Contraintes et risques

NA

Informations complémentaires

Part of the ROLKNEEMATICS Robotics Challenge program of the ANR