Design And Development Of Human-machine Interface Device Based On Wearable A-mode Ultrasound

Most of the traditional ultrasound devices have some limitations such as bulky size,unwearable and high cost,which hinder the application and promotion of ultrasound-based human-machine interface(HMI)in the fields of prosthesis control,AR/VR human-computer interaction and so on.Therefore,this thesis aims to design a HMI device based on wearable A-mode ultrasound,which includes the design of hardware system,embedded software,host-computer software and on-chip integration of gesture classification algorithm.Due to the problem that the existing ultrasound devices are not suitable for HMI application,this thesis gives priority to the design and miniaturization of the hardware system of the A-mode ultrasound device,which is used to generate,receive and process the A-mode ultrasound signals from forearm muscles.Then,the embedded software was designed and developed for the hardware system according to the circuit composition of the A-mode ultrasound device.Through appropriate configuration for the registers of each functional chip,the device can complete the functions of high voltage excitation for A-mode ultrasound signals,echo receiving and processing,echo sampling and transmission.Besides,a human-computer interaction software was also developed for displaying the ultrasound echo signals.After that,the gesture classification performance of three types of ultrasound features were compared and analyzed,in the case of repeated donning and circumferential transducer shift.The most robust ultrasound feature was applied in the on-chip integration of A-mode ultrasound based pattern recognition algorithm,which includes the process of ultrasound feature extraction,classifier training,online recognition and on-chip memory management.Finally,the developed A-mode ultrasound HMI device was able to classify dexterous hand gesture on embedded chip and achieve desirable performance in prosthetic hand control.