| Functional electrical stimulation (FES) is a means to restore arm movement to paralyzed people due to high spinal cord injuries. To restore basic daily functions such as eating or interacting with the environment, the FES-controller has to account for the stability of the arm. In traditional control, system stability is typically ensured by applying feedback control; however, characteristics in FES systems, such as low stimulation rate and time delay, limit the ability to apply feedback FES control. Nevertheless, human arms possess feedforward mechanisms to modulate the arm stiffness, thus improving the stability in unstable interactions. The goal of this thesis is to develop feedforward FES control methods to improve the stability of the arm. The necessity of explicitly considering of limb stability in FES controller was first assessed. A computational model was developed to make the assessment of the stability of FES-controlled limbs by predicting the limb stiffness under FES control. Postural maintenance tasks were simulated at postures over the reachable workspace, and the results indicate that only marginal stability was achieved by a typical feedforward FES controller that did not consider limb stability. These results suggest that the limb stability should be considered in the formulation of FES controllers. Subsequently, FES controllers incorporating natural stabilizing mechanisms were formulated and validated with simulated tasks. In one of them, the controller predicted human behaviors during unstable interaction tasks; in the other, the controller ensured arm stability in an unstable tool-usage task.;However, these investigations were made with computational models, which might have only limited ability to predict the behaviors of specific FES subjects. Experimental approaches were therefore developed to identify subject-specific models and to formulate empirical model-based FES controllers. With the identified model of FES force generation, the performance of a feedforward FES force controller was evaluated. Over a wide range of endpoint force targets, the controller could achieve the accuracy of 89% of the maximum force produced by muscle stimulation, suggesting that the controller could be used to restore versatile force tasks useful in daily life. Furthermore, by incorporating the identified model of FES stiffness, an advanced feedforward FES controller was formulated and shown to provide significant stiffness modulations independent from the endpoint force it generated. These results have supported the potential for using feedforward FES control to modulate arm stiffness and improve arm stability, which holds promise for restoring more natural activities to people with spinal cord injuries. |
