Design and Control of a Paediatric Robotic Lower-Limb Exoskeleton

Lower-limb exoskeletons enhance motor function in patients, benefiting both clinical rehab and daily activities. Nevertheless, paediatric exoskeletons remain largely underdeveloped. To address this gap, this thesis presents a new robotic lower-limb exoskeleton (LLE) specifically tailored for children, considering the unilateral and bilateral design configuration. Utilizing anthropometric data from the target demographic, the LLE has a size-adjustable design to accommodate children aged 8 to 12. The design incorporates six active joints at the hip and knee, actuated using Brushless DC motors in conjunction with Harmonic Drive gears. A human-LLE system is designed in SolidWorks™ and imported into Simscape™ as a virtual prototype. Considering the system's complexity, particularly in the human-exoskeleton interaction, a unique approach is employed. This approach simplifies the virtual prototype for each leg within Simscape™ into a Simplified Model using Simscape Multibody™, following multiple layers of reduction. This thesis conducts a rigorous analysis of forward and inverse kinematics applied to this Simplified Model. While the Denavit-Hartenberg (D-H) convention is used for the forward kinematic analysis, an algebraic-geometric method is employed to solve the inverse kinematics of the model. This kinematic analysis is then used to develop a dynamic model of the Simplified Model within the Simulink® environment, using a MATLAB Function. This Simulink Model, representing the virtual prototype, aids in designing the PD controller with gravity compensator for the unilateral exoskeleton and computed torque control (CTC) for the bilateral exoskeleton. The control regime is then applied to the virtual prototype to track a standard gait trajectory and manage the highly unstable dynamics of the human-LLE system. The minimal discrepancies observed in the angular positions between the Simulink Model and the virtual prototype while following the desired trajectory provide strong validation for the accuracy of the derived dynamic model. Furthermore, the appropriate control signals and reasonable interaction forces between the human and the LLE effectively validate the controller's performance in the presence of disturbances.