Stride-wise Adaptive Assistance Strategy for Ankle Exoskeleton Under Varying Walking Conditions

Efficient and effective personalized assistance strategies are crucial for enhancing exoskeleton performance under varying walking conditions. We proposed a novel real-time adaptive assistance strategy to generate personalized and stride-wise customized ankle exoskeleton assistance profiles that adjusted to diverse and varying human locomotion demands. This approach tuned the assistance magnitude and timing online, starting from a profile pre-optimized during medium constant-speed walking, based on real-time ankle momentum estimation. It eliminated re-optimization processes when gait changes and solved assistance customization during transitional gaits. We recruited ten participants and validated the performance of our approach in two testing scenarios: a single-gait walking condition other than the pre-optimized one, and a varying-gait walking condition, which included multiple distinct steady-state gaits and their transition states. For the single-gait (high-speed walking) case, the stride-wise adaptive assistance of a unilateral ankle exoskeleton reduced muscle activity by 35.9 ± 16.8% compared to no assistance, demonstrating a level of assistance comparable to the state-of-the-art approach (human-in-the-loop optimization), but with improved time efficiency. For the varying-gait case, the stride-wise adaptive assistance reduced muscle activity by 28.4 ± 15.4% compared to no assistance, and by 28.1 ± 15.9% during their transition states. These results demonstrated the efficiency and effectiveness of the proposed stride-wise adaptive assistance strategy in assistance personalization and customization under multiple, unknown, un-optimized, changing conditions, as well as during transitional gaits. This approach has the potential to significantly enhance the performance of real-life exoskeletons.