Research Of Multi-sensory Myoelectric Prosthetic Hand With Hardness And Thermal Conductivity

The purpose of the research of prosthetic hand is to improve the self-care ability of the disabled and promote the development of medical benefits. At the same time, the artificial prosthetic hand is cross-disciplinary fields of robotics, biomedical engineering, electronics information and has become research focus of related engineering researchers. The upper limb is very complicated and has rich sensory feedback. However, the current commercial electric prosthetic hand controlled by surface electromyography (SEMG) signals, does not have information on the perception feedback of grasping objects. In order to make prosthetic hands act like human hands as much as possible and improve the biomimetic and intelligence, it is necessary to add multi-sensor perception system on electric prosthetic hand system.Multi-sensor perception is to access the multi-faceted and multi-angle physical properties of the grasping object, such as temperature, hardness, thermal conductivity, which provides guidance for follow-up actions. SEMG of the stump is used by electric prosthetic hand to achieve natural control, involving the SEMG signal processing, motion recognition and motion control technology. Combination of multi-sensor perception and SEMG electric prosthetic hand will make prosthetic hand more bionic and improve intelligence of the prosthetic hand. Therefore, this article take theoretical and experimental exploration on hardness testing and thermal conductivity measurements of the grasping objects, SEMG signal filtering, feature extraction and pattern recognition, and other aspects. The paper makes following work and innovations:(1) Documents of related technologies at home and abroad are reviewed. Hardness sensors, thermal conductivity sensors and research background, status and significance of SEMG prosthetic hand are briefly generalized.(2) Based on the designed robotic tactile sensor, design a hardness sensor suitable for electric prosthetic hand and its signal processing circuit. A simple and practical method of testing hardness of grasping objects is proposed. That is, with the same clamping speed and time, cumulative value of the pressure signal is treated as a the quantitative expression. Design a thermal conductivity sensor that can be installed on the fingers of prosthetic hands and its corresponding signal processing circuits. Because the response time is very long, the change in signal amplitude with units of time is adopted as a standard to judge the thermal conductivity of the object.(3) The wavelet domain hidden Markov model (HMM) is used for surface EMG signals filtering for the first time. Compared with the hard or soft thresholding method, the reconstruction wavelet coefficients in this paper are determined by Bayesian posterior estimates. The method doesn't need pre-set parameters, which improves the adaptability. Experimental results show that the method has better filtering effect and noise rejection ratio is increased, while the edge characteristics of surface EMG signals are retained well, which is benefit for subsequent pattern recognition. The optimal wavelet packet is used to extract the wavelet coefficients of the surface EMG signal. The square of these coefficients as fourteen-dimensional feature vector are input to support vector machine (SVM) classifier. Experimental results show that the feature vector composed by the optimal wavelet packet and support vector machine algorithm can distinguish four motion of wrist extension, wrist flexion, exhibition boxing, fist.(4) The hardness sensor and thermal conductivity sensor are used in prosthetic hands. And some typical objects are classified by information fusion. The research result provides technical routes and methods for development of SEMG electric prosthetic hands with multi-feelings.