Simulation studies on hybrid neuroprosthesis control strategies for gait at low speeds

Lower-limb rehabilitation programs are beneficial in restoring function for people with neurological disorders. Combining techniques and equipment, such as active orthoses and functional electrical stimulation (FES), promises to accelerate the therapy outcome while simultaneously reducing therapists’ physical burden. However, there are still challenges related to inadequate stimulation response characteristics and the trade-off between weight, size, and active orthoses performance. Moreover, it is often unachievable to obtain specific metrics in experimental environments, and it seems there is still no consolidated simulation environment for these applications. Here we explore an open-source simulation framework to investigate further FES and active orthoses controlling knee motion of a detailed musculoskeletal model. We modeled, implemented, and compared tracking errors and correlations of four FES controllers for two slow speeds (0.1 m/s and 0.3 m/s), showing a higher correlation with the reference for the higher speed (0.89) than for the lower speed (0.73). Further, we observed that the simulation environment presented a similar behavior compared to experiments, as higher errors during the knee flexion and hip extension. These results are compatible with the expected behavior from real experiments, showing that muscle-drive simulations may generate a wealth of data and elucidate the principles that govern muscle coordination to improve rehabilitation outcomes.