Application Research Of Gesture Recognition Based On Hexapod Robot

Since its birth,hexapod robot has been widely used in many fields with its unique characteristics such as high carrying capacity,terrain adaptability and high flexibility.It has been extended from the traditional robot field to the intelligent robot field.At the same time,it has been continuously expanded in new fields such as star survey and seabed exploration.So far,the hexapod robot has been widely studied by many researchers in the field of robotics in the world,which makes the structure design,control system and kinematics of the hexapod robot further developed;on the other hand,many robot researchers are committed to mining the intelligent performance of the hexapod robot,and most of the research focuses on the optimization of robot control algorithm and mechanical structure And innovation,gradually improve the environmental adaptability and intelligence of hexapod robot.However,in some special environments,due to local adaptability and other reasons,the hexapod robot can not play the optimal performance,which needs to be effectively solved through human intervention.In addition,the existing human-computer interaction generally depends on simple devices such as keyboard,mouse,remote control,touch screen,etc.,which has some shortcomings.Therefore,it is necessary to constantly update the human-computer interaction mode of hexapod robot,so that the non-contact human-computer interaction technology is gradually emerging and widely used in the field of robot control,in order to more tap the potential of hexapod robot.As an indispensable part of human-computer interaction,gesture recognition has been widely studied and applied in education and training,smart home,virtual technology,disability assistance,and autonomous driving.It has a large impact on the performance of humancomputer interaction.Including the nature and flexibility of human-computer interaction.In the current gesture recognition methods,researchers generally simplify the gesture background according to the experimental environment,and use the relevant algorithms to segment and extract the target gesture under a single background.The target gestures finally extracted will be systematically analyzed by common methods.Due to the simplification of the specific environment and background,the above experimental results are often limited In real life,the environment in which gestures are located is usually more complicated,such as: the brightness and darkness of light and angle changes,too many gestures in the environment,different distances between gestures and the device being interacted with,etc.The innovation and improvement of algorithm is a necessary way to realize gesture recognition in the context of complex environments.The improvement and application of gesture recognition efficiency will contribute to future human-computer interaction that is more in line with human habits.Therefore,based on the hexapod robot using gesture recognition,this paper studies and develops the application of gesture recognition,mainly using simple and complex background gesture data,using different methods to carry out gesture recognition experiments.Through the description of the hexapod robot with gesture interaction,the different ways of the hexapod robot to perform actions in response to different gestures are formulated.The purpose of the hexapod robot gesture recognition and human-computer interaction is achieved.The hexapod robot realizes gesture recognition and can perform corresponding action tasks through the recognition results,which lays the foundation for the design and implementation of the hexapod robot gesture interaction.The gesture interaction system based on hexapod robot is composed of the lower computer which is responsible for controlling the hexapod robot and the upper computer which is responsible for image information processing.Among them,the lower computer is stm32f103zet6 as the main controller.The function of the lower computer is to send the robot's own state to the upper computer and control the corresponding action of the hexapod robot by receiving the command from the upper computer.The upper computer is composed of bramble pie 3b,camera V2 and PC,which is responsible for image information processing and communication with the lower computer.In the field of target detection,YOLOv3 is one of the most representative algorithms,with high detection accuracy and good timeliness.Based on the above reasons,this paper proposes a new hexapod robot interaction mode—a human-computer interaction mode based on hexapod robot.The hexapod robot is based on the Sublime Text3 development environment,the open source computer vision library Opencv3.0 class library,and the deep learning YOLOv3 algorithm.The Python2.7 language is used for host computer development,and the Keil4 development environment is used for stm32 for lower computer development.Human gestures directly interact with the hexapod robot,enabling the hexapod robot to perform five different actions(forward,backward,left,right,and gait switching).By using different methods for different gesture data sets,and comparing and analyzing the experimental results,the superiority of the yolov3 algorithm is verified.Finally,under the yolov3 method,the average accuracy of indoor and outdoor gesture recognition on the test set can reach 95.08% and 89.00% respectively.Compared with traditional algorithms based on manual feature extraction and other machine learning-based algorithms,it has higher detection accuracy and stronger robustness.The gait and movement of the hexapod robot can be controlled through gesture recognition results,so that the hexapod robot has better human-computer interaction mode,environmental adaptability,and good passing performance.