Research On Control Gesture Recognition Technology Of A Small Ground Unmanned Platform

Gesture recognition is an important part of human-computer interaction,and its interaction is natural and convenient.It has a wide range of applications in different fields such as industrial control,medical rehabilitation,and home entertainment.The application of gesture recognition technology to the military field is undoubtedly a key research direction.Traditional control methods for unmanned combat platforms mostly use remote controls such as handles and keyboards,which have poor human-computer interaction and are unnatural.This paper studies the method of controlling ground unmanned platform based on hand gestures and military dynamic gestures.The main research contents are as follows:First of all,the research on hand gesture recognition methods for controlling small ground unmanned combat platforms was carried out.The paper used the MYO armband to obtain the EMG signal,used the moving average method to determine the action interval after filtered the signal,and extracted the mean absolute value(MAV),mean absolute value slope(MAVS),wavelength(WL)and variance(VAR)features to form the feature set of this paper,used support vector machine(SVM)and BP neural network realize 7 kinds of gesture instruction classification.The results showed that the classification effect of SVM was better,the overall recognition rate could reach 92.29%,and SVM was selected as the final classifier.Compared with the traditional Hudgins feature set,the feature set of this paper had a better classification effect.Aiming at the problem of low recognition rate of similar gestures in recognition,angular velocity signal was introduced,and the recognition method of angular velocity and EMG signal feature fusion was used.The overall recognition rate of hand gestures reached 97.85%,an increase of 5.56%.Secondly,the research on military gesture recognition for controlling small ground unmanned combat platforms was carried out.A military gesture is composed of several hand-shaped sub-gestures and posture sub-gestures.The paper used the above-mentioned EMG recognition method to classify 7 military hand-shaped sub-gestures,and used MYO to obtain acceleration signals,extracted the acceleration action interval value as features,and used SVM to classify 7 seed gestures.When the hand-shaped sub-gestures and posture sub-gestures were output in a predetermined time sequence,they were recognized as military dynamic gestures,and a military gesture recognition system was established.The results showed that the recognition rate of 14 kinds of military dynamic gestures could reach 92.3%,and the gesture recognition rate was high.Finally,7 military gestures with a small number of sub-gestures and high recognition rate were selected for the control of small ground unmanned combat platforms.Finally,a gesture control experiment system for a small ground unmanned platform was built and the experiment was verified.The thesis built the control system of armband,PC and unmanned platform,used Bluetooth to transmit the gesture signal obtained by armband to PC,and used WIFI to transmit the signal command to the small ground unmanned platform after gesture recognition on PC.The unmanned platform performed corresponding actions in accordance with the instructions,verified the feasibility of hand gestures and military gestures to control a small ground unmanned combat platform.