Hybrid neuroprostheses are promising gait restoration devices that combine the benefits of FES and powered exoskeletons. However, controlling these devices can be complicated due to numerous inherent challenges in the system such as actuator redundancy, mixed actuator dynamics, EMD, and muscle fatigue. In this project, solutions to these challenges were explored from a human motor control perspective. The human motor control system is able to achieve fluid and coordinated gait despite high degree of freedom and multiple muscles. It is hypothesized that the human central nervous system (CNS) activates multiple muscle fibers in groups or patterns called muscle synergies, or motor primitives, to efficiently perform movements such as reaching and posture control. The benefit of synergies is their ability to transform higher dimensional and complex systems into lower dimensional and simpler systems that are easier to control. In this project, two methods of generating time-invariant synergies were derived. In the first method PCA was used to extract a synergy basis from precomputed optimal trajectory and input data. These artificial synergies were numerically simulated and it was shown that the adaptive synergy-based controller when combined with a feedback controller can achieve quasi-static walking. In the second method, synergies were designed through optimizations to produce a predetermined set of dynamic postures. When these postures are activated, they can be used as a basis for generating the swing phase of a gait. These synergies were incorporated in adaptive control laws to overcome model uncertainties and information loss due to synergy decomposition. EMD during FES and mixed actuator dynamics due to the combined use of FES and electric motor can result in system instability or uncoordinated responses. These issues were addressed by using a DSC framework and an EMD compensation term. Finally, to compensate for the diminishing control effectiveness due to the FES-induced muscle fatigue, a model-based fatigue estimate was used as a scaling factor in the feedforward path. These control methods were tested on an able-bodied subject and a person with an incomplete SCI. The experimental results show that the synergy-based controller achieves good tracking results, and importantly can compensate for actuator redundancy, mixed actuator dynamics, EMD, and muscle fatigue. These results demonstrated that a hybrid walking neuroprosthesis can be achieved. Particularly, it was shown that the multiple lower-limb muscles can be stimulated via FES, along with the use of a powered exoskeleton, to reproduce quasi-static walking movements. Potentially, the combined use of FES and the powered exoskeleton can lead to a light-weight walking rehabilitation technology for individuals with mobility disorders. This technology will enhance their muscle health by actively exercising the muscles, integrate them in the society by increasing their mobility, and overall improve the quality of life.

Last Modified: 07/03/2018
Modified by: Nitin Sharma