The Hybrid team coordinates an ANR DGA ASTRID project funded to develop biomimetic controls for robotic arm and prosthesis applications. The project has just started for 3 years (2021-2023), and involves the INRIA/IMN team Mnemosyne, the army hospital (HIA) Percy, and the UGECAM rehabilitation center Tour de Gassies.
The goal of the project is to extend a proof of principle showing that movements of 4 distal joints missing in a person amputated at the level of their upper arm, could be successfully reconstructed with an artificial neural network on the basis of natural shoulder movements plus contextual information. Indeed, we recently showed that control subjects could reach close to normal performance at picking and placing bottles of various positions and orientations in a virtual environment based on this biomimetic control (Mick et al., 2021). The current funded project aims at extending this approach to more complex tasks, in virtual reality as well on a robotic platform (Reachy, Pollen Robotics), and testing those control principles on amputees.
The project also plans to integrate those controls with MyoTact, an EMG bracelet for muscle recordings with vibrotactile feedback developed in the Hybrid team, and supported by a ‘Aquitaine Science Transfer’ as well as a ‘CNRS Innovation’.
Mick S, Segas E, Dure L, Halgand C, Benois-Pineau J, Loeb GE, Cattaert D, de Rugy A (2021) Shoulder kinematics plus contextual target information enable control of multiple distal joints of a simulated prosthetic arm and hand. Journal of NeuroEngineering and Rehabilitation.
