Human-in-the-loop optimization of the control of a bionic knee prosthesis for Sit-To-Stance tasks

Transfemoral amputations lead to many difficulties in carrying out daily life activities. This is due to the lack of two joints, knee and ankle, and the associated musculature. Passive prostheses are not able to compensate the lack of energy caused by the missing limb. For this reason, to fulfill actions like Sit-to-Stand, people must employ movement strategies which however result in high asymmetry and therefore in possible complications in the remaining musculature and structure of the healthy limb. To counteract the performance deficiency of traditional passive prostheses, active prostheses have been designed to provide the patient a modulated external torque. They require an effective and reliable controller to regulate the joint power delivery in motion. The control methods currently used involve a small set of steady-state locomotion tasks using finite state machines to manage instantaneous transitions between them. Nevertheless, human locomotion is far from steady state. This study aims to implement a new control method for an existing active knee prosthesis prototype, namely the Human-in-the-Loop optimization method. The final goal is to find the correct torque pattern which is able to maximize the symmetry in ground reaction force between the prosthetic and the healthy limbs during the Sit-to-Stand and Stand-to-Sit actions. To assess the efficiency of the controller, Sit-to-Stand and Stand-to-Sit tests will be performed using the active prosthesis. The method will optimize the torque pattern test by test until the optimal one is identified. Furthermore, tests will also be carried out using two different types of passive prostheses to highlight the difference in asymmetry obtained between them and the active prosthesis.