Human Gait Detection And Motion Prediction Based On Omni-directional Lower Limb Rehabilitation Robot

According to the sixth national census report, People aged above60had reached178million at the end of2010in our country, which made up13.26%of the total population.Elderly people will reach248million and the aging level will also increase to17.17%ofthe total population. In addition, thousands of population is suffering serious illness fromthe traffic accidents every year. The aggravation of aging population and high rate ofdisability and morbidity are diminishing the life level for elderly, putting heavier burden onthe society and restricting the rapid growth of national economy.In order to strengthen body functions for elderly, it is essential to have the properrehabilitation training and obtain gait information from the lower limb, which contributesto reducing the probability of the contingency. The gait detection system based on theomni-directional lower limb rehabilitation robot is proposed to acquire the horizontaldistances between every mark pasted on the lower limb and detection plane and then thetime series model for gait is established on the basis of characteristics of gait series and itsphysical meaning. Moreover, recursive prediction and estimation for gait series at the nextmoment are obtained by combining Kalman filter with time series model, gait pattern isrecognized based on the average deviation between the predictive values and estimatedvalues at the period of time.The gait detection system is fixed on the midpoint of Omni-directional lower limbrehabilitation robot with the inductive elements towards the active area of lower limb. Thehorizontal distances between all marks on the frontal surfaces of lower limb and detectionplane are acquired simultaneously in the process of rehabilitation training. Time seriesmodel is established effectively to reflect the changing rules of gait series according to thecharacteristics of randomness, relevance and periodicity, and then Kalman filter isemployed to predict gait information at the next moment and estimate gait informationaccording to the observed value after modeling gait series. In order to recognize the gaitpattern and take some essential measures to avoid the rehabilitation accident as soon as possible, the reference criterion is developed by calculating the average deviations betweenpredictive values and estimated values at a period of time. The normal gait is determined ifthe average deviation at the same period of time is located in the reference criterion above,otherwise the abnormal gait is determined and start the alarm to call the medical workersfor help.