Me
I am a PhD student at Inria of University of Bordeaux since October 2021. In 2021 I graduated as an engineer in cognitics at the Ecole Nationale Supérieure de Cognitique. In 2017, I had a bachelor degree in Pure Mathematics.
Through different internships and work positions, I worked in artificial intelligence, linguistics, physiology, transfer knowledge and software development.
Thesis
Musculoskeletal models and data-driven learning for personalized human force capacities evaluation.
Keywords: Musculoskeletal models
Abstract: The project of the Auctus team is to design the collaborative robotics cells of the future to lower the hardship at work. There are currently important methodological issues concerning the design of the robotic system from the analysis of the task and the capacities of the operator. Simple tools such as the RULA scores are used to assess the body posture and find musculoskeletal disorders (MSD) issues. This is however not sufficient and does not provide cueson the operators force capacities to design collaborative robotic system.
Biomechanical models would help but they are generally complex and biased by individual variability. One important biomechanical characteristic to consider is the ability of individuals to generate forces appropriate to the task at hand without exceeding their physiological limits. To date, two approaches are proposed to evaluate force capacities: one based on musculoskeletal modelling and the other one relying on experimental data obtained from experiments. None of them is still satisfactory due to the important inter-subject variability.
In this framework, the aim of this thesis is to combine the advantages of both paradigms. Using force data obtained during the interaction of a human subject with a robot, the objective is to personalize the parameters of an upper-limb musculoskeletal model by means of machine learning techniques.
Thesis supervisors: This thesis will be supervised by Jean-Marc Salotti, who has an expertise in ergonomics and machine learning, with the help of Nasser Rezzoug (University of Toulon, currently with a Delegation Inria position in the Auctus team), who has an expertise in biomechanics. We think that developing this type of approach combining modelling and machine learning has the potential to better address other issues in the context of collaborative robotics.
This is the list of my publications: