Previous studies from our laboratory have demonstrated that the choice of control algorithm greatly affects how a human adapts to a robotic lower limb exoskeleton. We have shown that proportional myoelectric controllers have distinct advantages over kinematic based controllers in healthy humans. However, many neurologically impaired patients that could benefit from a robotic lower limb exoskeleton do not have sufficient muscle activation patterns for robust myoelectric control. Our long term goal is to develop an artificial neural oscillator that can provide adaptive control for robotic lower limb exoskeletons and other robotic assistive devices for human locomotion. As a first step towards this goal, we used computer simulations to assess the stability and robustness of various neural oscilla-tor algorithms coupled with dynamic systems. Due to the pendular nature of human walking mechanics, our dynamic systems included damped pendulums and a three-dimensional passive dynamic walker. We used a Hopf oscillator as our artificial neural oscillator providing neural control to the actuators. Overall results indicated that the Hopf oscillator could entrain to the resonant movement dynamics of the mechanical system with sufficient proprioceptive feedback. Our future work will examine methods to shape the timing and profile of the Hopf oscillator output to more closely mimic human muscle activation patterns. After coupling the revised Hopf oscillator to more complex dynamic walking models, we plan on testing their implementation on the robotic lower limb exoskeletons in our laboratory.

Категория: Научные статьи | Нет комментариев »

Комментарии