Study On Gait Recognition Of Cross Individuals Based On Electromygrophy Of Human Lower Limbs

Lower limb gait phase recognition is an important and basic part in the research of human motion intention perception.It is also a hot research direction in human-computer interaction and artificial intelligence.The existing gait phase recognition methods based on EMG seldom consider the difference of EMG signals between different individuals,which makes the accuracy of cross user gait recognition can not meet the requirements.In this paper,cross user gait recognition is regarded as a domain adaptive problem in transfer learning,which effectively improves the accuracy of online and offline cross user gait recognition.The main contents of this paper are as follows:Aiming at the problem that the dimension of EMG signal is too high,the unsupervised t-SNE algorithm is used to reduce the dimension and visualize the feature of EMG signal without users.The existence of individual difference of EMG signal was proved,and the separability of gait phase was reasonably explained.The cross user gait recognition problem is regarded as the domain adaptive problem in transfer learning.Based on the basic principle and design process of the classifier difference maximization algorithm,the cross user gait recognition is realized by using this algorithm.The simulation results show n that the method can greatly improve the accuracy of cross user identification.With the decrease of the number of samples,the accuracy of cross use r gait recognition based on classifier difference maximization algorithm will have a significant downward trend,so this method is only suitable for offline recognition.In order to meet the requirements of online recognition,the classifier difference maximization algorithm and adaptive batch normalization method are combined.After giving the principle and design steps of the self-adaptive batch normalization algorithm,the effectiveness of the algorithm was proved by simulation with several evaluation indexes,and the online cross user gait recognition could be realized.In order to further verify the performance of the algorithm,different users were selected to carry out online experiments of cross user gait recognition based on EMG signals.Based on the detailed introduction of the experimental process and equipment,the experimental data shown that the algorithm has excellent performance,which can ensure the accuracy of cross user gait recognition.