Performance analysis of an off-line BMI for the control of a robotic digit

Sofiane Ouanezar (1, 2) , Selim Eskiizmirliler (1) , Marc Maier (1)

1 CESEM - UMR 8194 - Centre d'étude de la SensoriMotricité
2 TSI - Département Traitement du Signal et des Images

Various recent brain-machine interface (BMI) applications have emphasized the potential of using cortical signals for the kinematic control of artificial devices such as robotic arms or legs [1], [2]. However, it is less clear whether hand and digit movements can be controlled in the same way in order to potentially restore manual dexterity. Recently, we have provided a proof-of-concept that an anthropomorphic robot finger with 4 degrees of freedom (DoF) can be controlled off-line by intracortical signals recorded in the behaving monkey [3]. Here we report the performance analysis based on the same data and robot, the statistical results including aspects of learning/training, prediction, as well as control.

Mots clés: BMI and signal processing; Information coding; Neural interfaces and softwares; Neural fields and attractor networks

Domaines: Neurosciences [q-bio.NC]; Neurosciences

Sofiane Ouanezar, Selim Eskiizmirliler, Marc Maier. Performance analysis of an off-line BMI for the control of a robotic digit. Cinquième conférence plénière française de Neurosciences Computationnelles, "Neurocomp'10", Aug 2010, Lyon, France. ⟨hal-00553416⟩