Research On Flexible Tactile Sensor Array And The Inverse Solution Of Contact Information

It has always been the ambition to produce a multi-finger dexterous artificial hand just like human hands in the fields of the humanoid robot and the intelligent prosthesis.And the intelligent artificial hand largely depends on the high-performance tactile sensing to obtain the interaction information with the outside world whenever it needs to complete some complicated or precise operations.At present,the tactile sensing technology has achieved great development in terms of basic sensing performance and sensor unit flexibility.However,there still remains some technical problems such as the lack of flexible skin layer packaging in most of tactile sensors,the difficulty in integrating both the static and dynamic measurement abilities with stable performance,and the contact information recognition relying too much on the high-density sensor array but still having detection blind zones.Thus,based on the combination of theoretical analysis and experimental methods,this dissertation researches on the contact mechanics of the flexible skin layer of the tactile sensor array,the design and fabrication of the novel tactile sensor with solid-liquid composite structure,and the inverse solution of contact information.The main contents of this dissertation are briefly stated as follows:1.The mechanical conduction properties of the flexible skin layer for the contact force on the tactile sensor array is studied.According to the basic mechanical properties of human skin and the performance requirements of the contact interaction,the mechanical constitutive relation of the bionic skin material is determined.The contact mathematical model and the finite element simulation model of the flexible skin layer are respectively established to analyze the conduction law of the flexible skin layer for the contact force,that is,it has both the attenuation effect and the coupling effect on mechanical stimulus signals.The basic mechanical properties of the PDMS bionic skin materials are tested,and the loading platform with the pressure sensor array is established to further verify the force conduction law of the flexible skin layer by experiments.2.The structural design and fabrication process of the solid-liquid composite tactile sensor arrays are studied.Aiming at the problem of contact force attenuation caused by flexible skin layer and the brittleness of silicon sensing unit,a solid-liquid composite tactile sensing structure is proposed.Then,the mechanical model of spring system and the finite element simulation model are established to analyze the fluid pressure conductivity in the flexible cavity.The fabrication process and packaging process of the tactile sensor array are studied,and 3×3 sensing array samples are successfully prepared.The experimental platforms are established to test the sensing performances of the prepared tactile sensor array samples,which shows that the tactile sensor array meets the basic performance requirements of tactile sensing:The sensor unit's force range is 0?1.6 N,the minimum force measurement ability is 5 mN,and the dynamic response frequency range is 5?400 Hz.3.An inverse solution of contact force for flexible tactile sensor arrays is studied.Aiming at the lack of spatial resolution caused by the coupling effect of the flexible skin layer and the sparseness of the sensor array itself,an inverse solution of contact force based on correlation analysis is proposed.The applicable principle of the method is analyzed based on the contact model in elastic half space,and the specific algorithm is compiled.A high-precision automatic loading system is built to carry out the inverse solution experiments using the prepared 3X3 solid-liquid composite tactile sensor array,which shows the average position error of the inverse solution is 0.46mm,and the average amplitude error is 0.043N.4.An object curvature recognition method for flexible tactile sensor arrays is studied.Based on the Hertzian contact theory,the mapping relationship between the stress change in the flexible skin layer and the radius of curvature of the object is verified.Then,the neural networks for the surface type recognition and curvature radius prediction are established respectively.The experimental samples with plane,convex spherical surface and convex cylindrical surface and radius of curvature between 1-30 mm are prepared.The contact loading experiments are implemented to obtain sample data for the neural network training and validation.The test results show that the established neural network model can eliminate the influence of different loading positions and contact directions,and effectively predict the surface curvature radius with the average prediction error of 1.87mm,and for 81.8%of tests,the prediction error is less than 3mm.