Fast ℒaasso- 𝓜pc Using Extended Kalman Filter for Robot-Assisted Rehabilitation: with Optimal Impedance

Elderly people may lose the ability to walk normally as their muscles weaken, or it may be difficult for them to maintain balance while walking. In addition to aging, nerve damage such as stroke, trauma, infectious diseases, accidents, etc. may cause loss of balance in walking and weakening of muscles. Wearable robots are crucial for helping patients with lower limb diseases, particularly those with trouble walking since their numbers are rising. These robots assist patients in walking, provide comfort, and aid in recuperation. In this study, the model predictive control based on the Lasso regression theory (Lasso-MPC) and the extended Kalman filter (EKF) was used to make a novel controller that helps the patient walk by adjusting the impedance so that, in addition to regular walking, the patient has to put out the most effort when walking. In order to evaluate and the effectiveness of the proposed method, first, using \textit{OpenSIM} software, the required torque data was determined for healthy, disabled and sick people. Then, using these obtained data, the model fitting parameters were determined. At the end, experiments including a healthy person and a modular lower limb exoskeleton were performed. The findings show that the proposed method successfully estimates the patient’s torque and correctly adjusts the robot’s assistance level to the user’s behavior, thereby maximizing his activity during treatment.