A non-array customizable tactile sensor based on spraying process

Tactile perception is very important for the physical interaction between robots and the external environments. Based on the theory of non-uniform gradient potential distribution (GPD), a novel non-array flexible touch-sensing sensor is proposed in this paper. It can detect whether there is a touch, and further to know where the touch took place. The sensor uses an indium-tin oxide polyethylene terephthalate (ITO-PET) film as signal generation layer. The electric fields in two different directions are constructed by time-sharing in the ITO-PET conductive plane, and the potential distribution over the ITO-PET film has the characteristics of uniqueness and determinacy. The touching position can be worked out by detecting the potentials of the touching point in two directions, respectively. In order to establish the relation between the position and the potential value effectively, quickly and accurately, the radical basis function (RBF) neural network is used to construct the electric field model and calculate the touching position coordinate. A detection circuit was designed to detect the touching position of the sensor sample. The simulation and experimental results show that the detection principle proposed in this paper is feasible, and the real-time detection of single point touching position can be realized. The touch-sensing sensor is simple in structure, thin in thickness and soft in flexibility. It can cover the surface of the robot and obtain external touching information in real time.

This paper proposes two decoupling methods for a flexible tactile sensor, improved back propagation neural network (BPNN) and radical basis function neural network (RBFNN). In the numerical experiments, the number of hidden layer nodes of the BPNN is optimized and k-fold-cross-validation (k-CV) method is also applied to construct the dataset. Information of the tactile sensor array at different scales is also used to construct the BPNN. RBFNN is applied to approach the nonlinear relationship between the deformation and the three-dimensional force of the tactile sensor numerical model built through finite element analysis. The decoupling results show that the RBFNN with high nonlinear approximation ability has good performance in decoupling three-dimensional force and satisfies both the decoupling accuracy and real-time requirements of the tactile sensor. Different white Gaussian noises (WGN) are added into the ideal model of the flexible tactile sensors. Then the modified RBFNN is applied to approximate and decouple the mapping relationship between row-column resistance with WGNs and the three-dimensional deformation. Numerical experiments demonstrate that the improved RBFNN doesn't rely on the mathematical model of the system and has good anti-noise ability and robustness.

For human-robot interactive systems, physical touch identification is essential for security and intelligence enhancement. Flexible tactile sensor with a simple structure and minimal peripheral electric connections will benefit various applications where the contact information is crucial. However, it is still hard to balance the full flexibility, large area, minimal electrical connection and easy fabrication in existed tactile sensing systems. In this letter, a novel positioning method based on a flexible capacitive tactile sensor is introduced. The tactile sensor contains two layers of screen-printed flexible electrodes and an ionic gel membrane in the middle to form electrostatic double layer capacitors (EDLCs). Only four points of the top layer electrode are used in the measurement to identify the touch position by measuring the voltage variation in real-time. Taking advantages of the electric potential re-distribution induced by the EDLC variation from the touch pressure, a calculation method for touch position recognition is tested based on experimental data. The developed method has no dependency on complex fabrication, redundant electric circuit or sensing unit arrays, which owns a promising future in various application fields, such as intelligent robotics, biomedical devices, and wearable equipment.

Reservoir sensitivity prediction is an important basis for designing reservoir protection program scientifically and exploiting oil and gas resources efficiently. Researchers have long endeavored to establish a method to predict reservoir sensitivity, but all of the methods have some limitations. Radial basis function (RBF) neural network, which provided a powerful technique to model non-linear mapping and the learning algorithm for RBF neural networks, corresponds to the solution of a linear problem, therefore it is unnecessary to establish an accurate model or organize rules in large number, and it enjoys the advantages such as simple network structure, fast convergence rate, and strong approximation ability, etc. However, different radial basis function has different non-linear mapping ability, and different data require different radial basis functions. Nowadays, the choice of radial basis function in the network is based on experience or test result only, which exerts a great adverse impact on the network performance. In this study, a new RBF neural network with trainable radial basis function was proposed by the linear combination of common radial basis functions. The input parameters of the network were the influence factors of reservoir sensitivity such as porosity and permeability, etc. The output parameter was the corresponding sensitivity index. The network was trained and tested with the data collected from our own experiments. The results showed that the new RBF neural network is effective and improved, of which the accuracy is obviously higher than the network with single radial basis function for the prediction of reservoir sensitivity.

Early warning of whether an enterprise will be faced with human resource crisis is a new hotspot in the study of enterprise crisis. This study contributes to early warning of enterprise human resource crisis by proposing an integrated model of Rough Set (RS) and Radial Basis Function (RBF) neural network, which overcomes the shortcomings of long training time and complex network structure in the traditional neural network methods. The proposed model fully exerts the advantages of the two methods of RS and RBF neural network. By means of RS for attribute reduction, the input data are reduced but still reflects the main information of the original data. And RBF neural network has simple network structure, strong nonlinear approximation ability, and fast convergence speed. First, this study sets up the enterprise human resource crisis early-warning index system. Second, 55 training samples are trained to construct the human resource crisis early-warning model, and 5 testing samples are used to test the forecasting effect of the model. Finally, this study compares the performance of RS–RBF neural network to those of Back Propagation (BP) neural network and RBF neural network and RS-BP neural network. The model comparison results show that the proposed model simplifies the structure of the neural network, speeds up the learning speed of the network, and improves forecasting efficiency and accuracy, which can give early warning of enterprise human resource crisis more effectively.

This paper applied an enhanced particle swarm optimization (PSO) technique with levy flight algorithm for training the radial basis function (RBF) neural network to forecast the data of runoff in the Houzhaihe River basin, a typical karst area in Guizhou Province, southwest China. The karst aquifer system is a highly nonlinear and complex system due to its unique aqueous medium, the complexity of its hydrogeological conditions makes the traditional hydrological model research results unsatisfactory, and the establishment of a physical distributed model based on hydrological mechanism requires a large number of hydrogeological parameters, which are often unavailable in karst areas. Radial Basis Function (RBF) neural network has been widely used in various fields because of its simple structure, high-speed calculation and ability to approximate any nonlinear function. Based on the RBF neural network, this paper established a time series prediction model for typical karst regions. In order to improve the performance of RBF network model, we applied an enhanced particle swarm optimization with levy flight in this research. The results show the proposed enhanced RBF model performs much better than the one without improvement by the levy flight.

Electronic skin is an important means through which robots can obtain external information. A novel flexible tactile sensor capable of simultaneously detecting the contact position and force was proposed in this paper. The tactile sensor had a three-layer structure. The upper layer was a specially designed conductive film based on indium-tin oxide polyethylene terephthalate (ITO-PET), which could be used for detecting contact position. The intermediate layer was a piezoresistive film used as the force-sensitive element. The lower layer was made of fully conductive material such as aluminum foil and was used only for signal output. In order to solve the inconsistencies and nonlinearity of the piezoresistive properties for large areas, a Radial Basis Function (RBF) neural network was used. This includes input, hidden, and output layers. The input layer has three nodes representing position coordinates, X, Y, and resistor, R. The output layer has one node representing force, F. A sensor sample was fabricated and experiments of contact position and force detection were performed on the sample. The results showed that the principal function of the tactile sensor was feasible. The sensor sample exhibited good consistency and linearity. The tactile sensor has only five lead wires and can provide the information support necessary for safe human—computer interactions.

This paper presents a novel approach of single neuron PID control for position sensorless switched reluctance motors (SRM) based on radial basis function (RBF) neural network. In the proposed RBF neural network, there is no hidden units at the beginning, and during the process of learning, they are increased or decreased according to an adaptive algorithm. So the RBF neural network is built with a much simpler and tighter structure to form an efficient nonlinear map, and then it facilitates the elimination of the position sensors. Moreover, the paper uses single neuron to construct the adaptive controller of SRM, which has the advantages of simple structure, adaptability and robustness. A RBF network is built to identify the system on-line, and then constructs the on-line reference model, implements self-learning of controller parameters by single neuron controller, thus achieve on-line regulation of controller parameters. The experimental result shows that the method given in this paper can construct process model through on-line identification and then give gradient information to neuron controller, it can achieve on-line identification and on-line control with high control accuracy and good dynamic characteristics.

A flexible tactile sensor array with 6 × 6 N‐type sensitive elements made of conductive rubber is presented in this paper. The property and principle of the tactile sensor are analyzed in detail. Based on the piezoresistivity of conductive rubber, this paper takes full advantage of the nonlinear approximation ability of the radial basis function neural network (RBFNN) method to approach the high‐dimensional mapping relation between the resistance values of the N‐type sensitive element and the three‐dimensional (3D) force and to accomplish the accurate prediction of the magnitude of 3D force loaded on the sensor. In the prediction process, the k‐means algorithm and recursive least square (RLS) method are used to optimize the RBFNN, and the k‐fold cross‐validation method is conducted to build the training set and testing set to improve the prediction precision of the 3D force. The optimized RBFNN with different spreads is used to verify its influence on the performance of 3D force prediction, and the results indicate that the spread value plays a very important role in the prediction process. Then, sliding window technology is introduced to build the RBFNN model. Experimental results show that setting the size of the sliding window appropriately can effectively reduce the prediction error of the 3D force exerted on the sensor and improve the performance of the RBFNN predictor, which means that the sliding window technology is very feasible and valid in 3D force prediction for the flexible tactile sensor. All of the results indicate that the optimized RBFNN with high robustness can be well applied to the 3D force prediction research of the flexible tactile sensor.

A PID control combined with Radial Basis Function (RBF) neural network was proposed for course control of ship steering. The traditional PID control has a wide application on ship steering fields because of its simple structure and obvious effect. But uncertainty of ship models and the disturbance of real-world environments reduce its control precision. Neural network has better control effect for such non-linear external disturbance system. Through the ability of the neural network to approximate arbitrary nonlinear to adjusting the PID three parameters in real time to achieve optimum PID control, to overcome the impact due to model uncertainty and disturbance, to achieve the purpose of automatic tracking of ship course. Through the Matlab Simulink environment simulation to verify the control effect of the traditional PID control and the RBF neural network PID control under the sinusoidal reference signal. The comparison of simulation results that PID control that adding neural network can track the reference signal more effectively, so it can achieve more precise control of the ship steering.

In this article, the wing structural damage is identified and located by using modal analysis and Radial Basis Function (RBF) neural network. The finite element model of an aircraft wing is set up which is used for model analysis. The number of network centers is increased gradually which can ensure that the network has a simplest structure; RBF center is determined by K-means clustering algorithm which can improve the representative of each center and improve the training accuracy; the network weights is determined using the concept of pseudo inverse matrix and inverse matrix, which can shorten the training period and improve training efficiency. The computer simulation result shows that this damage identification method has high identification accuracy. The relative error is 1.422%, and the absolute error is 31.28mm. Comparing with the analyzing spar and skin individually, this method has a more spreading value.

Flexible tactile sensors with simple structures, minimal peripheral electric connections and straightforward data processing will benefit the human-machine interactions in which the contact information is crucial. However, it is still hard to balance the easy fabrication, full flexibility, large measurement area and minimal electric connection in existed tactile sensing systems. This study introduces an innovative positioning method based on a flexible supercapacitive tactile sensor. A 100 mm×100 mm prototype sensor, which contains two layers of flexible electrodes and a layer of ionic gel membrane in the middle to construct electrostatic double-layer capacitors (EDLCs), is developed to study the underlying physical principles. Following the established method, minimized electric connections are needed to achieve the mm-scale touch position/movement trace identification. Under the touch pressure, the formation of supercapacitors around the touch area leads to re-distribution of electric potential within the sensor. The electrical voltage variation is gauged at four points, and the data are calculated to estimate the touch positions following a two-step protocol. The developed method demonstrates high accuracy of position identification (around 5% in the 100 mm×100 mm flexible sensor), superior anti-disturbance capability (more than ∼104 variation of capacitance) and fast response (∼ms level). At the same time, it has no dependency on complex fabrication, redundant electric circuit or sensing unit arrays. These promising characteristics can benefit various application fields, such as intelligent robotics, biomedical devices and wearable equipment.

The performance assessment of hydraulic servo systems has attracted an increasing amount of attention in recent years. However, only a few studies have focused on practical approaches in this field. A performance assessment method based on radial basis function (RBF) neural network and Mahalanobis distance (MD) is proposed in this study; the method is quantized by the performance confidence value (CV). An observer model based on RBF neural network is designed to calculate the residual error between the actual and estimated outputs. The root mean square (RMS), peak value, and average absolute value are then extracted as the features of residual error, which serve as the coordinates of the feature points. Lastly, the MD between the most recent feature point and the constructed Mahalanobis space is calculated. The condition of the system is assessed by normalizing MD into a CV. The proposed method is proven to be effective by a simulation model in which leakage faults are injected. Experimental results show that the proposed method can assess the performance of hydraulic servo systems effectively.

3D double-vision inspection is very necessary. It has a larger field of view, and can solve the problem of “blind area” for 3D measurement, as proposed by 3D single-vision inspection. At the beginning of this paper, the principle of structured-light based 3D vision inspection is introduced. Then, a method of gaining calibration points for 3D double-vision inspection system is proposed in detail. In order to gain calibration points with high precision, a double-directional photoelectric aiming device is designed as well, and a method for compensating the position-setting error of the aiming device is described. The coordinates of all calibration points are precisely unified in a world coordinate system. The application of RBF (radial basis function) neural network in establishing the inspection model of structured-light based 3D vision is described in detail. Finally, with the use of the calibration points, the inspection model of 3D double-vision based on RBF neural network is successfully established. The model’s training accuracy is 0.078 mm, and the testing accuracy is 0.084 mm.

With the continuous improvement of robot intelligence, machine tactile and machine vision technology have been greatly developed, but most tactile sensors can not meet the requirements of tactile sensing applications. Therefore, piezoresistive effect and force information detection principle based on new sensitive materials are proposed, and a flexible force sensitive tactile sensor with three-dimensional array structure is designed. The mathematical description of calculating the three-dimensional force is obtained through analysis, and the finite element simulation of the array structure of the multi-dimensional flexible tactile sensor is carried out. The three-dimensional displacement changes of the coordinates of the central node under different three-dimensional forces are obtained, and the output resistance of the force-sensitive conductive rubber array unit before and after the three-dimensional force action is calculated. The changes provide an effective basis for verifying the experimental results and new structure design. The test results show that the proposed sensor can measure normal force with high accuracy, good repeatability, and can detect sliding phenomena. In addition, the proposed sensor has simple structure, low cost, and is easy to mass production and application.