Study On Robotic Tactile Sensing System Based On Electronic Skin

The paper aims to develop tactile sensing for robots based on electronic skin(e-skin)and improves the robotic sensing in dark environment.The tactile sensing enables robots to sense objects in complicated environment,which improves robotic grasp and manipulation.Firstly,the substrates with different properties were used to improve the tactile sensing of e-skin.The substrate influence on e-skin is studied from thickness,elastic modulus and location of substrate.And the hyperelasticity theory is used to analyze the influence mechanism.Then,crosstalk among sensing elements introduced by substrate is experimentally investigated.Based on the tactile improvement of e-skin,a motion sensing algorithm is designed.The algorithm senses translation by the image centroid identification,and senses the rotation by the image registration.Additionally,image morphology,bilinear interpolation and log-polar transformation are used to remove noises,enhance resolution and improve efficiency.Moreover,a high-precision experimental platform was built to verify the applicability and accuracy of the algorithm.Motion sensing enables robots to sense the objects in complicated environment.To improve the sensing,a material recognition method mimicking human touch is designed.In the method,tactile data is collected by short-duration slide action.Then,statistics and wavelet transform are used to extract features from tactile data,the features are used to train several classifiers after principal component analysis.The classifier with the best performance was selected to recognize materials.Moreover,the method robustness under different interactive forces is experimentally verified,and the performance is compared with human touch and existing works.Tactile sensing based on e-skin enables robots finish tasks like humans,which extends the robotic applications and has important significance.