Research On Electrostatic Tactile Adaptive Rendering Method Based On Meta-texture

Human-Computer Interaction(HCI)is the field of research on the way people and computers communicate.The existing physical technology of human-computer interaction basically relies on three human senses: vision,hearing,and touch.The tactile perception is the most difficult to construct and the most demanding to achieve conditions.Interactions based on this have always been the most difficult to implement in interactive interfaces.As a relatively novel tactile presentation method in the field of humancomputer interaction,electrostatic tactile sensation generates a weak electrostatic force between the finger and the electrode by applying periodic signals to the electrode plate to change the friction force.Its feedback is highly real-time,the information transmitted is relatively rich,and it also has a high degree of flexibility.In this article,we first propose a new tactile rendering method based on electrostatic force.Taking the relatively lossy pattern of human tactile perception on a small scale of time as a starting point,a new concept of meta texture is proposed,which decomposes virtual textures into several single frequency waveforms on a time scale.Texture irregularity is achieved by changing the frequency from meta texture to meta texture.In this paper,multiple sets of pre-experiments were conducted to determine the optimal signal parameters for the meta texture.The inertial measurement unit is used to measure the triaxial acceleration and angular velocity of the finger across the surface of the electrostatic force screen.The final dataset is obtained through filtering,remining,and processing processes such as de trending the data in the dataset,and applying the DTW algorithm to measure.Then,an inverse dynamic model based on convolutional neural networks from signal to meta texture sequences is constructed to describe the nonlinear relationship between its output and internal input.In another work in this article,we propose an adaptive improved algorithm.Firstly,we construct a force velocity data grid by collecting experimental data under different velocities and normal forces.Secondly,the effects of different velocities and normal forces on the collected data in the time and frequency domains were explored.Finally,for arbitrary conditions,different operations such as resampling or linear interpolation are performed based on their adjacent nodes in the grid.Ensure that under the same texture,regardless of the conditions under which the collected data are processed,they have relatively similar characteristics in the time and frequency domain distribution of the signal.We conducted subjective psychophysical experiments and objective error analysis on the results of this method,and the results confirmed the feasibility and effectiveness of our adaptive algorithm.In addition,we have also reconstructed a network model based on CNN and LSTM to fully handle the temporal correlation of one-dimensional data.From the performance of the network model,the proportion of completely correct prediction of the sequence by the network constructed in this part of work is 84.25%,which can better reconstruct the real texture.In terms of subjective perception,the data generated after adjustment by our adaptive algorithm has subjectively achieved the desired effect.Users cannot subjectively distinguish signals of the same material collected under different experimental conditions.When freely exploring textures adjusted by our adaptive algorithm,the proportion of identifying differences in textures is only 9.75%.At the same time,after quantitative analysis of the adjusted signal in the time and frequency domain,it was found that the error value of the data after adaptive processing decreased by 65%.Integrating the above multiple evaluation dimensions has verified the feasibility and correctness of our adaptive tactile rendering scheme based on meta-texture.