Research And Implementation Of Intelligent Drumming Robot For Education And Entertainment

With the continuous development of robot technology and the strong support of the government,the robot industry has ushered in the spring of development in China,and various robots have appeared in people's field of vision.Among them,playing robot as a new development direction in the field of robot research,its research results have attracted much attention.At present,most of the domestically developed playing robots focus on commercial performance functions,lacking corresponding science education and art teaching functions.This paper uses the jazz drum in percussion as an example to design and develop two sets of drumming robots.One of them is a basic version,named "Xiaohan",which aims to educate primary and secondary school students about robot science.It can not only realize the single robot drum performance function through human-computer interaction,but also complete the independent performance function by recording the user's input action.The other is a smart version,named "Xiaozhi",which is more intelligent and can be used not only for robot group entertainment,but also for drum teaching.The main contributions of this paper include the following three parts:(1)This paper researches and designs a "Xiaohan" drumming robot with low cost,fun,competitive,and certain space.The robot can be widely used in science education for primary and secondary school students.The system uses ATmega328 P as the controller.It has four different working modes: parameter adjustment,human-computer interaction,recording and playback.The working status of each mode will be prompted by the corresponding digital tube.In human-computer interaction mode,the robot completes the performance by detecting the user's input in real time.The adjustable parameter mode and the recording mode complete the power-off storage of the parameter information and the action sequence by capturing the user's input.The play mode is to reproduce the sequence of actions that have been successfullly recorded,thereby completing the autonomous performance of the robot.The system has been successfully applied in the 7th Guangdong Creative Robot Competition.(2)This paper researches and designs a "Xiaozhi" drumming robot that can be used for jazz drum teaching,and provides a new way of learning for the teaching of jazz drums.At the same time,learners can also perform jazz drum performances with robots.In this way,they can not only be used to cultivate learners' interest in jazz learning,but also enhance their actual stage performance.The system software framework is implemented using the C/S architecture.The server is mainly used to identify and classify drums of jazz drum audio and sends the processing information to the robot client.The robot client uses two CPUs.The Raspberry Pi is used as the main controller and is responsible for network communication and audio playback.The auxiliary controller uses STM32,which is responsible for the control of the drumming robot,the prompting of the drum information,and the capture of the tapping motion.In addition,in order to facilitate the user to click on the song,a song client is also designed,which is mainly used to send the server the name of the drum music to be performed or learned next.In addition,in order to allow users to order songs,a song client was also designed.This client is used to send the server a name for the jazz drum that will be performed or learned.(3)This paper researches and implements the drum detection.In this paper,the rhythm extraction of jazz drum audio files is studied firstly.An adaptive spectral differential amplitude algorithm based on threshold segmentation is proposed.Compared to other methods,this method is more accurate when extracting rhythm points.Secondly,through the feature extraction and analysis of the jazz drum audio sample files,five basic features are selected for the jazz drum classification,and the validity of eigenvalue selection is verified by experiments.Finally,for drum classification,this paper constructs an SVM multi-classifier model suitable for the jazz drum music classification,designs the experiment to compare the classification accuracy of the model with other multi-classifier models,and verifies the feasibility of the method.