Research On Design Of Fingertip Dynamic Tactile Sensor Array And Texture Encoding Method For Intelligent Prosthetic Hand

Tactile sensing and recognition is indispensable for new generation intelligent prosthetic hand.Along with the development of the intelligent prosthetic hand,tactile sensing and recognition technologies have increasingly attracted wide attention.Contact force information from tactile sensors together with pertinent tactile recognition methods plays important role to the improvement of the grasp stability and endowing an intelligent prosthetic hand with environmental perception.Currently,the tactile sensing system of prosthetic hand mostly focused on static force detection and lack of reliable dynamic force sensing capability.This situation does not meet the intelligent prosthetic hand application well especially in grasp events and object features discrimination which more related to dynamic tactile information,such as dynamic contact events perception,slippage detection and texture encoding.A novel flexible tactile sensor array capable of three-axis force detection with good dynamic performance is developed in this paper.The design and fabrication,mechanical modeling and three-axis force decoupling method,and tactile texture encoding have been investigated in detail.This work was supported by the National Basic Research Program(973)of China under Grant 2011CB013303.Firstly,flexible polyvinylidene fluoride(PVDF)is selected as sensing material of the novel tactile sensor array due to its highly reliable piezoelectric effect and good dynamic responsibility to mechanical stimulation.The piezoelectric equation of the PVDF film is briefly introduced,followed by a simplified equation under special application.These analyses provide theoretical basis for following design of the tactile sensor array.Secondly,a novel structure based on patterned electrodes and truncated pyramid bump is proposed,so as to satisfy the fingertip limited space and curved surfaces integration requirements.Surface micromachining compatible process has been developed,which enables tactile array with the possibility of miniaturization and good spatial resolution.Three-axis force decouple method has been successfully developed and dynamic measurement properties of the tactile sensor array was checked.Thirdly,a novel texture encoding method based on a new characteristic variable,which is product of the response time interval between adjacent sensor units and the principal frequency of vibration,is first time proposed.Based on this method,texture exploration can be more anthropomorphic.Extensive validating experiment to the proposed new method has been performed and better reliable texture encoding under varying scanning velocities is obtained.The remainder of the dissertation offers the following:1.Research on a new dynamic three-axis force sensing structure based on patterned electrodes and truncated pyramid bump.A PVDF film layer is sandwiched between top and lower patterned electrodes layers to form four piezoelectric capacitors in one tactile unit.A truncated pyramid bump is fixed on top of the four piezoelectric capacitors to transmit contact forces.The structure of the four piezoelectric capacitors and the truncated pyramid bump can enable dynamic three-axis force measurement.Based on the proposed dynamic three-axis force sensing structure,a 3×2 tactile sensing array has been designed to realize distributed measuring.Mechanical model of the tactile sensing unit is developed to obtain stress distribution on the surface of the four piezoelectric capacitors when normal and shear force components are applied.2.Research on three-axis force decoupling model and experiments validation.Based on the load transfer mechanism of the bump and the arrangement of the four piezoelectric capacitors in one tactile unit,a three-axis force decoupling model is developed by combining responses of the four piezoelectric capacitors.To verify the model,three-axis force measurement experiments are performed.Experiment results show that after decoupling the maximum coupling rate between each two axes is no more than 9.49%;the tactile unit has good linearity and sensitivity,with the non-linearity of only 2.45%,2.37%and 1.74%F.S.and the sensitivity of 14.93,14.92,and 6.62 pC/N in the x-,y-,and z-axes,respectively;and the response frequency band is in the range of 5-400 Hz to both normal and shear load,which satisfies tactile application.3.Research on fabrication and encapsulation process of the flexible dynamic tactile sensor array.To reduce the complexity of the fabrication procedure and enhance alignment accuracy of electrode layers,the top and lower electrode layers are developed directly on the top and lower surfaces of a metallized PVDF film with double-side photolithography,and then encapsulated with the substrate layer and bump layer.With the double-side photolithography,the top and lower electrode layers can be aligned with each other well.Meanwhile,the key process steps of fabrication and encapsulation are carefully analyzed and designed.Finally,a sensor prototype has been presented by designed fabrication and encapsulation method.4.Research on a new texture encoding method based on response time interval between adjacent sensor units and the principal frequency of vibration.To eliminate the impact of scanning velocity on the accuracy of texture distribution,a new texture encoding method based on a novel characteristic variable which is product of the response time interval between adjacent sensor units and the principal frequency of vibration is first time proposed.For the response time interval between adjacent sensor units is varied with scanning velocity but independent of texture features,it can be used to get rid of the item of scanning velocity from the characteristic variable.To verify the method,an experimental setup based on the designed texture samples and the assembled tactile fingertip with the flexible dynamic tactile sensor array is developed.A series of dynamic tests have been carried out on the experiment setup by exerting different velocities ranging from 10 to 150 mm/s on texture samples with different spatial periods ranging from 300 to 1000 ?m,to reveal the relationship among texture spatial period,scanning velocity,response time interval and principal frequency.Experiment results shows that the proposed method is effective and texture encoding can be realized under varying velocities with identification accuracy of 99.93%.