Real-time Hybrid Gait Phase Recognition Algorithm Under Continuous Multimodal Locomotion for Lower Limb Exoskeleton

Nowadays, exoskeletons' ability to operate in complex environments is increasingly important. It is challenging to obtain an accurate gait phase under continuous multimodal locomotion. A hybrid gait phase recognition algorithm is proposed combining the adaptive-oscillator-based algorithm and the model-based algorithm. The modification law is adopted to make two algorithms amend each other reciprocally. The design of the modification law combines the advantages of oscillator-based algorithms and model-based algorithms. The gait phase estimation results are more accurate via the hybrid gait phase prediction algorithm (GPRA). Therefore, appropriate assistance actuation and higher efficiency are obtained under the complicated multimodal locomotion. An experiment was designed to verify the performance of the hybrid GPRA. As a result, the hybrid GPRA can still produce a high-precision gait phase with low latency even when the locomotion mode switches. The hybrid GPRA can provide accurate gait phase and assistance timing under continuous multimodal locomotion.