Research On Bidirectional Interface And Interactive Control For Anthropomorphic Prosthetic Hand

Although prosthetic hands available now are capable to mimic the human hand interms of appearance and motion modes, the interactive control of artificial hands is farfrom being perfect. Users can neither control the prosthetic hands reliably nor receivefeedback from the prosthetic hands intuitively. To solve the problems, an interactivecontrol strategy is proposed with a novel myoelectric control approach andelectrocutaneous stimulation feedback. The contents of this paper include： themyoelectric control system based on the knowledge of grasp taxonomy and theprinciple of Finite State Machine (SFM), the multi channel biphasic electrocutaneousstimulator based on the technology of Transcuataneous Electrical Nerve Stimulation(TENS) and Direct Digital Synthesizer (DDS), and the stimulation noise abatingmethod based on the algorithm of Least Mean Square adaptive filter.Overviews about newly developed bidirectional interfaces and interactive controlmethods of anthropomorphic prosthetic hands are given in details. Then the overallmethodology is established, which interpret users’ control intention via surfaceElectromyogram (sEMG) signals and transmit the sensor information to users viaelectrocutaneous stimuli.The pattern recognition method, a commonly used method to decode the user’scontrolling intention from sEMG signals, needs a number of active muscles and can beeasily affected by the variations of electrode skin interface and muscle fatigue. Toovercome the shortcomings of the pattern recognition method, a novel control methodbased on the knowledge of grasp taxonomy and the principle of Finite State Machine(SFM) is proposed. The validation experiments showed that, the method can accomplishthe combined position and force control of85%of human hand gestures in activities ofdaily life (ADLs) with one pair of agonist/antagonist muscles. The whole graspingprocess was completed with short time and high success rate.To eliminate the inconvenience caused by the lack of sensory feedback, a novelmulti channel stimulator based on the technology of TENS, DDS and biphasicvoltage controlled current source circuit is proposed. To verify its function, experimentsof different stimulation types and stimulation parameters were performed. To clear upthe noise caused by the TENS, a novel approach based on biphasic stimulation andLMS adaptive filter is proposed. Comparison experiments between single polarstimulation and biphasic stimulation were performed and the results showed thatbiphasic stimulation is capable to concentrate the stimulation current and decrease thepower of stimulation noise. Experiments comparing the correlation between the sEMG signal and the noise signal before and after noise cancellation showed that the noisecancellation algorithm can improve the SNR (Signal to Noise Ratio) effectively.To research on the property of the bidirectional human machine interface proposedin the thesis, a new interactive control strategy is proposed based on user demands ofsensory feedback for grasping operations. Experiments based on the HIT IV hand andand the personal computer were performed to verify the efficiency and reliability theinteractive control strategy. The results of the experiments showed that, with theelectrical stimulation sensory feedback, subjects can control the HIT IV hand to trackthe expectation force with high accuracy and short time. The results also showed that,the electrical stimulation sensory feedback can partly replace the visual sensoryfeedback, and the combination of the electrical stimulation sensory feedback and thevisual sensory feedback can improve the grasp efficiency evidently.