Design And Implementation Of Human Gait Prediction Algorithms Based On Attitude Sensors

Lower extremity exoskeletonsare a class of wearable robotic systems that augmentshuman endurance and strength.Exoskeletons provide pilots the ability to carry heavy loads. It will be widely used in the future.The wearer’s gait would have changed when exoskeleton’s actuators move to target positions,so exoskeleton’s gait lags behind its pilot’s. As a result, the exoskeleton would disturb its pilot’s walking. The reference information of exoskeleton controller should be ahead of pilot’s gait data so that the exoskeleton would make as little interference as possible in pilot’s movement. Therefore, it is necessary to predict the gait data obtained by exoskeleton on line.A human gait obtaining system is designed to collect and predict its wearer’s gait data.Firstly, LMS adaptive filter, RLS adaptive filter,SSA and nonlinear time series analysis based on Takens embedding theorem are simulated in MATLAB using human gait data sampled by a motion capture system. The result reveals that LMS adaptive filter and the Takens algorithm have better effect than the other two in predicting human’s gait data.Then,software for the hardware circuit based on ARMis completed. Embedding software for the circuit is designed to calculate human gait data. LMS adaptive filter and the Takens algorithm are implemented on the hardware to predict gait data on line. The experiment result indicates that the Takens algorithm has a better achievement while predicting gait data on line.Lastly,the human gait obtaining system is applied to the exoskeleton. The gait obtaining system does very well in experiments.It not only provides the controller of exoskeleton with reliable reference information, but settles the problem of time delay between exoskeleton and its pilot as well.