Nonlinear theoretical control models applied to human reaching movements

Details

Supervisors

Crevecoeur, Frédéric ; Bianchin, Gianluca

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

This work consists in the investigation on how neural activities associated with motor control can govern the highly nonlinear biomechanics of upper limb. This work addresses the many unresolved challenges from previous attempts to design a nonlinear framework for motor control, aiming to accurately consider nonlinear biomechanics while reproducing key features of motor control. To achieve this, I designed a nonlinear controller based on a Feedback Linearization framework to manage the nonlinear dynamics of upper limb movements in a planar context. By implementing the Feedback Linearization technique, the nonlinear system is transformed into a linear equivalent, enabling the application of the best linear motor control model: the LQG controller. The command for the linear system is then transformed back into a command for the nonlinear plant through an appropriate nonlinear change of variables. The proposed control system has been validated through various applications, demonstrating its ability to reproduce important features of motor control. Firstly, the model is applied to scenarios where task parameters change during movement. This shows an online adaptability to dynamic conditions, modeling experimentally observed mechanisms of motor control. Secondly, the model is extended to reproduce an online adaptation mechanism towards unpredicted environmental perturbations. To address these dynamic changes, the framework is supplemented with a nonlinear adaptive controller, assessing its quality as a motor control candidate. Additionally, the impact of an incorrect gravitational effect on reaching movements is studied, showing that this controller is a valuable tool to investigate and refine hypotheses about complex motor behaviors. These applications are complemented by qualitative and quantitative comparisons with experimental data.