Tactile sensors can be used to build human-machine interfaces, for instance in isometric joysticks or handlebars. When used as input sensor device for control, questions arise related to the contact with the human, which involve ergonomic aspects. This paper focuses on the example application of driving a powered wheelchair as attendant. Since other proposals use force and torque sensors as control input variables, this paper explores the relationship between these variables and others obtained from the tactile sensor. For this purpose, a handlebar is instrumented with tactile sensors and a 6-axis force torque sensor. Several experiments are carried out with this handlebar mounted on a wheelchair and also fixed to a table. It is seen that it is possible to obtain variables well correlated with those provided by force and torque sensors. However, it is necessary to contemplate the influence of issues such as the gripping force of the human hand on the sensor or the different kinds of grasps due to different physical constitutions of humans and to the inherent random nature of the grasp. Moreover, it is seen that a first step is necessary where the contact with the hands has to stabilize, and its characteristics and settle time are obtained.
Attendant joysticks of powered wheelchairs are devices oriented to help caregivers. Diseases and disabilities such as dementia, spinal cord injuries or blindness make the user unable to drive the chair by his or her own. However, this device is not intuitive to use, especially for old people. Proper processing of the information provided by two tactile sensors in the handlebar achieves control signals that allow an easy and intuitive driving. This is done in this paper, where the performance of this approach is evaluated in comparison with that of the joystick by means of objective measurements as well as questionnaires to obtain the subjective perception of the participants in the experiments. The results show a better performance of the handlebar in terms of error in following a trajectory, collisions with the surrounding furniture, and user feeling related to ease of use, comfort, required training, usefulness, safety and fatigue.
Voltage-to-frequency converters (VFCs) do typically use relaxation oscillators to generate an output signal whose frequency depends linearly on an input voltage. VFCs are data converters, that is, they are designed to codify the input voltage by means of the output frequency. Hence, they have strong requirements regarding linearity, dynamic range, and stability. On the other hand, compared with other converters, they have the advantage of low cost, particularly for large resolutions; 13-bit resolution is typical, and more than 20 bits are feasible. However, in such a case several milliseconds may be necessary to complete the conversion – the smaller the resolution, the shorter the conversion time. The serial frequency-modulated output provided by VFCs is particularly well suited for telemetry, to send the outcome of a measurement by wire, radio, or fiber optic links. After the frequency-modulated signal is received, it can be easily codified into a digital word, or converted back into a voltage by a frequency-to-voltage converter. This type of output simplifies galvanic isolation through the use of either an optocoupler or a magnetic coupler. This article deals first with the concepts behind voltage-to-frequency conversion, and then describes the main types of VFCs on the market, their limitations, and main applications. Frequency-to-voltage converters are also briefly described in this article.
Dexterity in robotic hands is not enough to cover the demands for more flexibility in the manufacturing process. Specifically, processes that involve cooperation or learning from humans, slippage detection and tactile feedback to improve grasp stability when different objects are manipulated, object exploration and classification, quality inspection, welding, etc. The sense of touch is the key to the capabilities of human hands, so tactile sensors must be incorporated into robotic hands to improve them. However, tactile sensing technology is not yet mature enough to be an effective tool. One reason for this limited performance is the need of acquisition and processing of large amounts of data in the short time required to perform a fluent manipulation. Smart tactile sensors can help in this goal if they are able to carry out local acquisition and processing of tactile data and transmit only relevant information through buses that guarantee bounded latency. Field-programmable gate arrays (FPGAs) are devices which are especially suitable for this task due to their ability for parallel processing. This paper presents the realization of electronics for a tactile sensor suite based on FPGAs that implements a direct interface with the raw sensor and serial communication between the fingertips and palm.
The typical layout in a piezoresistive tactile sensor arranges individual sensors to form an array with M rows and N columns. While this layout reduces the wiring involved, it does not allow the values of the sensor resistors to be measured individually due to the appearance of crosstalk caused by the nonidealities of the array reading circuits. In this paper, two reading methods that minimize errors resulting from this phenomenon are assessed by designing an electronic system for array reading, and the results are compared to those obtained using the traditional method, obviating the nonidealities of the reading circuit. The different models were compared by testing the system with an array of discrete resistors. The system was later connected to a tactile sensor with 8 × 7 taxels.
Tactile sensors are arrays of force sensing units called taxels and equilibration is necessary to normalize their response. This is commonly done by applying uniform pressure while the sensor lies on a flat surface. However, the surface of the destination system often has a different shape, for instance in robotic hands. Therefore, a device able to exert uniform pressure on curved surfaces could be helpful in this task. This paper explores the use of a custom pressurized chamber, large enough to host the whole system. The chamber has been built and the approach has been tested using a commercial sensor.
Resistive sensor arrays are formed by a large number of individual sensors which are distributed in different ways. This paper proposes a direct connection between an FPGA and a resistive array distributed in M rows and N columns, without the need of analog-to-digital converters to obtain resistance values in the sensor and where the conditioning circuit is reduced to the use of a capacitor in each of the columns of the matrix. The circuit allows parallel measurements of the N resistors which form each of the rows of the array, eliminating the resistive crosstalk which is typical of these circuits. This is achieved by an addressing technique which does not require external elements to the FPGA. Although the typical resistive crosstalk between resistors which are measured simultaneously is eliminated, other elements that have an impact on the measurement of discharge times appear in the proposed architecture and, therefore, affect the uncertainty in resistance value measurements; these elements need to be studied. Finally, the performance of different calibration techniques is assessed experimentally on a discrete resistor array, obtaining for a new model of calibration, a maximum relative error of 0.066% in a range of resistor values which correspond to a tactile sensor.
One of the most suitable ways of distributing a resistive sensor array for reading is an array with M rows and N columns. This allows reduced wiring and a certain degree of parallelism in the implementation, although it also introduces crosstalk effects. Several types of circuits can carry out the analogue-digital conversion of this type of sensors. This article focuses on the use of operational amplifiers with capacitive feedback and FPGAs for this task. Specifically, modifications of a previously reported circuit are proposed to reduce the errors due to the non-idealities of the amplifiers and the I/O drivers of the FPGA. Moreover, calibration algorithms are derived from the analysis of the proposed circuitry to reduce the crosstalk error and improve the accuracy. Finally, the performances of the proposals is evaluated experimentally on an array of resistors and for different ranges.
Direct sensor–digital device interfaces measure time dependent variables of simple circuits to implement analog-to-digital conversion. Field Programmable Gate Arrays (FPGAs) are devices whose hardware can be reconfigured to work in parallel. They usually do not have analog-to-digital converters, but have many general purpose I/O pins. Therefore, direct sensor-FPGA connection is a good choice in complex systems with many sensors because several capture modules can be implemented to perform parallel analog data acquisition. The possibility to work in parallel and with high frequency clock signals improves the bandwidth compared to sequential devices such as conventional microcontrollers. The price to pay is usually the resolution of measurements. This paper proposes capture modules implemented in an FPGA which are able to perform smart acquisition that filter noise and achieve high precision. A calibration technique is also proposed to improve accuracy. Resolutions of 12 effective number of bits are obtained for the reading of resistors in the range of an example piezoresistive tactile sensor.
A New Model Based on Adaptation of the External Loop to Compensate the Hysteresis of Tactile SensorsPublished: 15 October 2015 by MDPI in Sensors
This paper presents a novel method to compensate for hysteresis nonlinearities observed in the response of a tactile sensor. The External Loop Adaptation Method (ELAM) performs a piecewise linear mapping of the experimentally measured external curves of the hysteresis loop to obtain all possible internal cycles. The optimal division of the input interval where the curve is approximated is provided by the error minimization algorithm. This process is carried out off line and provides parameters to compute the split point in real time. A different linear transformation is then performed at the left and right of this point and a more precise fitting is achieved. The models obtained with the ELAM method are compared with those obtained from three other approaches. The results show that the ELAM method achieves a more accurate fitting. Moreover, the involved mathematical operations are simpler and therefore easier to implement in devices such as Field Programmable Gate Array (FPGAs) for real time applications. Furthermore, the method needs to identify fewer parameters and requires no previous selection process of operators or functions. Finally, the method can be applied to other sensors or actuators with complex hysteresis loop shapes.
This paper shows realizations of a piezoresistive tactile sensor with a low cost screen-printing technology. A few samples were fabricated for different materials used as insulator between the conductive layers and as top layer or cover. Both can be used to tune the sensitivity of the sensor. However, a large influence is also observed of the roughness at the contact interface on the sensitivity and linearity of the output, as well as on mismatching between the outputs from different taxels. The roughness at the contact interface is behind the transduction principle of the sensor, but it also limits its performance if the wavelength of the roughness is comparable or even longer than the size of the contacts. The paper shows experimental results that confirm this relationship and discusses its consequences in sensor response related to the materials chosen for the insulator and the cover. Moreover, simulations with FEA tools and with simple models are used to support the discussions and conclusions obtained from the experimental data. This provides insights into the sensor behaviour that are shared by other sensors based on the same principle.
Influence of Errors in Tactile Sensors on Some High Level Parameters Used for Manipulation with Robotic HandsPublished: 19 August 2015 by MDPI in Sensors
Tactile sensors suffer from many types of interference and errors like crosstalk, non-linearity, drift or hysteresis, therefore calibration should be carried out to compensate for these deviations. However, this procedure is difficult in sensors mounted on artificial hands for robots or prosthetics for instance, where the sensor usually bends to cover a curved surface. Moreover, the calibration procedure should be repeated often because the correction parameters are easily altered by time and surrounding conditions. Furthermore, this intensive and complex calibration could be less determinant, or at least simpler. This is because manipulation algorithms do not commonly use the whole data set from the tactile image, but only a few parameters such as the moments of the tactile image. These parameters could be changed less by common errors and interferences, or at least their variations could be in the order of those caused by accepted limitations, like reduced spatial resolution. This paper shows results from experiments to support this idea. The experiments are carried out with a high performance commercial sensor as well as with a low-cost error-prone sensor built with a common procedure in robotics.
Assistive ambulatory devices are used for gait rehabilitation and assistance. In both cases, their benefit is greater when they are used properly. As for canes, embedded sensors can be used for monitoring purposes. In this paper, a custom tactile handle equipping a cane is described. It is composed of cost-effective commercially available pressure sensors. Experimental results involving 10 subjects show that the developed handle can provide information on the cane orientation as well as on the load applied to it during assisted gait. These data can help monitoring the cane usage and misuses detection.
One of the challenges in today’s mobile robotics is the design of high mobility and maneuverability robots. In this work we present the design and construction of a new concept of a locomotion system for mobile robots. It consists of a hybrid leg-wheel module that can be attached to the main body of a robot in a similar way to a conventional wheel. The mechanical configuration of the driving module is described, emphasizing the characteristics which make it different from other hybrid locomotion systems. A dynamic model that simulates the movement of the module was developed to analyze its behavior and to test different control algorithms that were subsequently implemented on the real module. Finally, we have carried out a series of simple experiments that demonstrate the correct operation of the module on flat ground without obstacles.
This paper introduces a novel device based on a tactile interface to replace the attendant joystick in electric wheelchairs. It can also be used in other vehicles such as shopping trolleys. Its use allows intuitive driving that requires little or no training, so its usability is high. This is achieved by a tactile sensor located on the handlebar of the chair or trolley and the processing of the information provided by it. When the user interacts with the handle of the chair or trolley, he or she exerts a pressure pattern that depends on the intention to accelerate, brake or turn to the left or right. The electronics within the device then perform the signal conditioning and processing of the information received, identifying the intention of the user on the basis of this pattern using an algorithm, and translating it into control signals for the control module of the wheelchair. These signals are equivalent to those provided by a joystick. This proposal aims to help disabled people and their attendees and prolong the personal autonomy in a context of aging populations.
A 13 x 13 square millimetre tri-axial taxel is presented which is suitable for some medical applications, for instance in assistive robotics that involves contact with humans or in prosthetics. Finite Element Analysis is carried out to determine what structure is the best to obtain a uniform distribution of pressure on the sensing areas underneath the structure. This structure has been fabricated in plastic with a 3D printer and a commercial tactile sensor has been used to implement the sensing areas. A three axis linear motorized translation stage with a tri-axial precision force sensor is used to find the parameters of the linear regression model and characterize the proposed taxel. The results are analysed to see to what extent the goal has been reached in this specific implementation. © (2013) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Tactile sensing is a serious issue in assistive robotics or prosthetics. It requires the hands and body of the robot are equipped with many tactile arrays, and a large amount of data has to be acquired and transmitted to a central decision unit. Moreover, this has to be done in a very short time to improve dexterity and performance in service tasks Therefore, the local electronics of a smart tactile patch must be powerful enough to acquire and pre-process data in real time to cope with the limited throughput of the communication buses. This paper presents an architecture based on FPGAs that implements a direct interface with the sensor without analog-to-digital converters and robust serial communications between the electronics in the finger tips, mid-digits and palm.
Tactile sensors are basically arrays of force sensors. Most of these force sensors are made of polymers or conductive rubber at lower cost, especially in the case of large area low-medium resolution tactile sensors. The consequence of such a decrease in cost and complexity is a worsening in performance. Hysteresis and drift are two main sources of error. Other tactile sensors do not present such limitations per se, however they are covered by a protective elastic layer in their final location and this covering can also lead to limitations. This paper presents a method to reduce the error caused by hysteresis in tactile sensors. This method is based on the generalized Prandtl–Ishlinskii model that has been applied to characterize hysteresis and saturation nonlinearities in smart actuators. The approximation error depends on several parameters as well as on the envelope functions that are chosen. Different alternatives are explored in the paper. Moreover, the model can also be inverted. This inverse model allows the force values to be obtained from the tactile sensor output while reducing the errors caused by hysteresis. In this paper the results of such an inversion are compared with other alternatives to register the data that do not compensate hysteresis. The average value of the hysteresis error measured in the experimental curve is 7.20% for an input range of 206 kPa, while this error is 1.51% following the compensation procedure. Since the implementation of tactile sensors usually results in the electronics being close to the raw sensor, and this hardware is also commonly based on a microcontroller or even on a FPGA, it is possible to add the algorithms presented in this paper to the set of compensation and calibration procedures to run in the smart sensor.
Back pain is one of the most common health problems in developed societies. Stress and bad habits like having bad postures are often behind, so postural hygiene is highly recommended to prevent it. Systems to measure back posture based on piezoresistive or optical bend sensors, or on inertial sensors have been reported. This paper presents one based on eight accelerometers connected through I2C bus to a small board that provides wireless communication. Data is collected and sent to a PC for analysis and warning. The spine is modelled as a kinematic chain where each segment has an associated accelerometer, and a 3D representation of the spine is obtained for easy interpretation of data.
This paper reports the design of a tactile sensor patch to cover large areas of robots and machines that interact with human beings. Many devices have been proposed to meet such a demand. These realizations are mostly custom-built or developed in the lab. The sensor of this paper is implemented with commercial force sensors. This has the benefit of a more foreseeable response of the sensor if its behavior is understood as the aggregation of readings from all the individual force sensors in the array. A few reported large area tactile sensors are also based on commercial sensors. However, the one in this paper is the first of this kind based on the use of polymeric commercial force sensing resistors (FSR) as unit elements of the array or tactels, which results in a robust sensor. The paper discusses design issues related to some necessary modifications of the force sensor, its assembly in an array, and the signal conditioning. The patch has 16 × 9 force sensors mounted on a flexible printed circuit board with a spatial resolution of 18.5 mm. The force range of a tactel is 6 N and its sensitivity is 0.6 V/N. The array is read at a rate of 78 frames per second. Finally, two simple application examples are also carried out with the sensor mounted on the forearm of a rescue robot that communicates with the sensor through a CAN bus.
Tactile sensors are basically arrays of force sensors. Most of these force sensors are made of polymers or conductive rubbers to lower the cost, especially in the case of large area low-medium resolution tactile sensors. The price to pay for such decrease in the cost and complexity is a worse performance. Hysteresis and drift are the two main sources of error. This paper presents a method to reduce the error caused by hysteresis. This method is based on the generalized Prandtl- Ishlinskii model that has been applied to characterize hysteresis and saturation nonlinearities in smart actuators. The classical Prandtl-Ishlinskii model is not suitable because the lack of symmetry the output curves from the sensor show. Other alternatives like the Preisach model are too complex to implement, especially taking into account that a tactile sensor provides many data to process. The approximation error depends on several parameters as well as on the envelope functions that are chosen. Different alternatives are explored in the paper. Moreover, the model can also be inverted. This inverted model allows obtaining the force values from the tactile sensor output while reducing the errors caused by hysteresis. Since implementations of tactile sensors usually have the electronics close to the raw sensor, and this hardware is also commonly based on a microcontroller or even on a FPGA, it is possible to add the algorithms presented in this paper to the set of compensation and calibration procedures to run in the smart sensor.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Tactile sensors have increasing presence in different applications, especially in assistive robotics or medicine and rehabilitation. They are basically an array of force sensors (tactels) and they are intended to emulate the human skin. Large sensors must be implemented with large area oriented technologies like screen printing. The authors have proposed and made some piezoresistive sensors with this technology. They consist of a few layers of conductive tracks to implement the electrodes and elastomers to insulate them, on a polymer substrate. Another conductive sheet is placed atop the obtained structure. Pressure distribution in the interface between this conductive sheet and the electrodes has a direct impact on the sensor performance. The mechanical behavior of the layered topology with conductive tracks, elastomers and polymers must be studied. For instance, the authors have observed experimentally the existence of pressure thresholds in the response of their sensors. Finite element simulations with COMSOL explain the reason for such thresholds as well as the dependence of the pressure distribution profile on the properties of the materials and the geometry of the tactel. This paper presents results from these simulations and the main conclusions that can be obtained from them related to the design of the sensor.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Tactile sensors are basically arrays of force sensors that are intended to emulate the skin in applications such as assistive robotics. Local electronics are usually implemented to reduce errors and interference caused by long wires. Realizations based on standard microcontrollers, Programmable Systems on Chip (PSoCs) and Field Programmable Gate Arrays (FPGAs) have been proposed by the authors for the case of piezoresistive tactile sensors. The solution employing FPGAs is especially relevant since their performance is closer to that of Application Specific Integrated Circuits (ASICs) than that of the other devices. This paper presents an implementation of such an idea for a specific sensor. For the purpose of comparison, the circuitry based on the other devices is also made for the same sensor. This paper discusses the implementation issues, provides details regarding the design of the hardware based on the three devices and compares them.
This paper presents results from many experiments whose purpose is evaluating the effect of the limitations of a tactile sensor on the tactile image as it is seen for control. Specifically, the variations of a few significant parameters of the tactile image are measured when hysteresis, drift and spatial resolution are taken into account.
This paper presents results from a selection of tactile sensors that have been designed and fabricated. These sensors are based on a common approach that consists in placing a sheet of piezoresistive material on the top of a set of electrodes. We use a thin film of conductive polymer as the piezoresistive material. Specifically, a conductive water-based ink of this polymer is deposited by spin-coating on a flexible plastic sheet, giving it a smooth, homogeneous and conducting thin film. The main interest in this procedure is that it is cheap and it allows the fabrication of flexible and low cost tactile sensors. In this work, we present results from sensors made using two technologies. Firstly, we have used a flexible printed circuit board (PCB) technology to fabricate the set of electrodes and addressing tracks. The result is a simple, flexible tactile sensor. In addition to these sensors on PCB, we have proposed, designed and fabricated sensors with screen-printing technology. In this case, the set of electrodes and addressing tracks are made by printing an ink based on silver nanoparticles. The exhaustive characterization provides us insights into the design of these tactile sensors.
This paper shows the design of a tactile sensor intended to cover the forearm of the rescue robot ALACRAN. This robot is able to lift hundreds of kilograms so it has to be carefully designed because it will manipulate human beings that can be hurt. So it has to be aware of being in contact with a human being and how he or she is pressed. Not just a binary output is required because contact is often necessary, for instance when a human is held in the arms of the robot. For this sort of operations a kind of artificial skin must provide information about the contact between the robot and the human. This skin will cover the hands to carry out fine manipulation, but it will also cover large areas like forearms. Some devices have been proposed to face such demand. This paper presents one of these sensors that has been obtained by arranging commercial force sensing resistors. Design issues related to this approach and results are presented.
This paper presents a bioinspired integrated tactile coprocessor that is able to generate a warning in the case of slippage via the data provided by a tactile sensor. Some implementations use different layers of piezoresistive and piezoelectric materials to build upon the raw sensor and obtain the static (pressure) as well as the dynamic (slippage) information. In this paper, a simple raw sensor is used, and a circuitry is implemented, which is able to extract the dynamic information from a single piezoresistive layer. The circuitry was inspired by structures found in human skin and retina, as they are biological systems made up of a dense network of receptors. It is largely based on an artificial retina , which is able to detect motion by using relatively simple spatial temporal dynamics. The circuitry was adapted to respond in the bandwidth of microvibrations produced by early slippage, resembling human skin. Experimental measurements from a chip implemented in a 0.35-mum four-metal two-poly standard CMOS process are presented to show both the performance of the building blocks included in each processing node and the operation of the whole system as a detector of early slippage.
Many artificial skins for robotics are based on piezoresistive films that cover an array of electrodes. Local preprocessing is a must in these systems to reduce errors and interferences and cope with the large amount of data provided by the sensor. This paper presents circuitry based on an FPGA to implement the interface to the artificial skin. The approach consists of a direct connection. The analog to digital conversion procedure is simple. It consists of measuring the discharging time of a capacitor through the resistance we want to read. This first proposed approach needs isolated tactels, so the raw sensor has to be fabricated in this way. If the tactile array is large, the strategy is not feasible. For instance, up to 288 pins are required to implement the interface with an array of 16x16 tactels. The proposal of this work for this case is to replace passive integrators by active ones. The result is a circuitry that allows the cancellation of interferences due to parasitic resistors and the sharing of the addressing tracks. Moreover, the FPGA allows the processing of data from the tactile sensor at a very high rate. This is because the high number of I/O pins of the device allows the conversion of many channels (in our case one per column) in parallel. The internal processing of the tactile image can also be done in parallel. This means we could be able to respond to very high demanding tasks in terms of dynamic requirements, like slippage detection. This also means we can run complex algorithms at real time, so a smart, programmable and powerful sensor is obtained.© (2009) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
This paper presents results from a few tactile sensors we have designed and fabricated. These sensors are based on a common approach that consists of placing a sheet of piezoresistive material on the top of a set of electrodes. If a force is exerted against the surface of the so obtained sensor, the contact area between the electrodes and the piezoresistive material changes. Therefore, the resistance at the interface changes. This is exploited as transconduction principle to measure forces and build advanced tactile sensors. For this purpose, we use a thin film of conductive polymers as the piezoresistive material. Specifically, a conductive water-based ink of these polymers is deposited by spin coating on a flexible plastic sheet, giving as a result a smooth, homogeneous and conducting thin film on it. The main interest in this procedure is it is cheap and it allows the fabrication of flexible and low cost tactile sensors. In this work we present results from sensors made with two technologies. First, we have used a Printed Circuit Board technology to fabricate the set of electrodes and addressing tracks. Then we have placed the flexible plastic sheet with the conductive polymer film on them to obtain the sensor. The result is a simple, flexible tactile sensor. In addition to these sensors on PCB, we have proposed, designed and fabricated sensors with a screen printing technology. In this case, the set of electrodes and addressing tracks are made by printing an ink based on silver nanoparticles. There is a very interesting difference with the other sensors, that consists of the use of an elastomer as insulation material between conductive layers. Besides of its role as insulator, this elastomer allows the modification of the force versus resistance relationship. It also improves the dynamic response of the sensor because it implements a restoration force that helps the sensor to relax quicker when the force is taken off.© (2009) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
This paper reports on the experience of the 2008 international summer school on mechatronics, jointly organized by the University of M'alaga (Spain) and the Technical University of Dresden (Germany). An important part of the hands-on practice and two student competitions have been based on the LEGO Mindstorms NXT Set. To stimulate lab work with representative general purpose software tools, LabVIEW and the NXT add-ons have been used. The paper proposes basic LabVIEW structures for several LEGO case study practices. A description of the course as well as an assessment on student competences are also included.
Tactile sensors are applied to different areas like robotics, medicine or virtual reality. Many of these sensors are based on piezoresistive films that cover an array of electrodes. This approach is simple and cheap and seems to be able to fit the requirements of applications like manipulation with robotic hands or grippers. Local preprocessing is a must in these systems to reduce errors and interferences and cope with the large amount of data from the sensor. This paper presents a circuitry based on a FPGA to implement the interface to the tactile sensor. The approach consists in a direct connection and can save area and get a more compact solution than other proposals that need more integrated circuits. Results from a first implementation based on a development board are shown to illustrate the feasibility of this strategy.
The interest in tactile sensors is increasing as their use in complex unstructured environments is demanded, like in telepresence, minimal invasive surgery, robotics etc. The array of pressure data provided by these devices can be treated with different image processing algorithms to extract miscellaneous information. However, as in the case of vision chips or artificial retinas, problems arise when the size of the array and the computational complexity increase. Having a look at the skin, the information collected by every mechanoreceptor is not sent to the brain for its processing, but some complex pre-processing is performed to fit the limited throughput of the nervous system. This is specially important for high bandwidth demanding tasks, as the case of slip detection with tactile sensors, which is demanding in computing requirements. Here we show some results from a tactile processor based on circuitry proposed for an artificial retina that has been modified to mimic the way the biological skin works.
Artificial sensitive skins are intended to emulate the human skin to improve the skills of robots and machinery in complex unstructured environments. They are basically smart arrays of pressure sensors. As in the case of artificial retinas, one problem to solve is the management of the huge amount of information that such arrays provide, especially if this information should be used by a central processing unit to implement some control algorithms. An approach to manage such information is to increment the signal processing performed close to the sensor in order to extract the useful information and reduce the errors caused by long wires. This paper proposes the use of voltage to frequency converters to implement a quite straightforward analog to digital conversion as front end interface to digital circuitry in a smart tactile sensor. The circuitry commonly implemented to read out the information from a piezoresistive tactile sensor can be modified to turn it into an array of voltage to frequency converters. This is carried out in this paper, where the feasibility of the idea is shown through simulations and its performance is discussed.© (2007) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
This paper presents an up-to-date survey of graphical tactile displays. These devices provide information through the sense of touch. At best, they should display both text and graphics (text may be considered a type of graphic). Graphs made with shapeable sheets result in bulky items awkward to store and transport; their production is expensive and time-consuming and they deteriorate quickly. Research is ongoing for a refreshable tactile display that acts as an output device for a computer or other information source and can present the information in text and graphics. The work in this field has branched into diverse areas, from physiological studies to technological aspects and challenges. Moreover, interest in these devices is now being shown by other fields such as virtual reality, minimally invasive surgery and teleoperation. It is attracting more and more people, research and money. Many proposals have been put forward, several of them succeeding in the task of presenting tactile information. However, most are research prototypes and very expensive to produce commercially. Thus the goal of an efficient low-cost tactile display for visually-impaired people has not yet been reached.
The interest in tactile sensors is increasing as their use in complex unstructured environments is demanded, like in telepresence, minimal invasive surgery, robotics etc. The array of pressure data provided by these devices can be treated with different image processing algorithms to extract the required information. However, as in the case of vision chips or artificial retinas, problems arise when the array size and the computation complexity increase. Having a look at the skin, the information collected by every mechanoreceptor is not sent to the brain for its processing, but some complex pre-processing is performed to fit the limited throughput of the nervous system. This is specially important for high bandwidth demanding tasks. Experimental works report that neural response of skin mechanoreceptors encodes the change in local shape from an offset level rather than the absolute force or pressure distributions. Something similar happens in the retina, which implements a spatio-temporal averaging. We propose the same strategy in tactile preprocessing, and we show preliminary results illustrated for the case of slip detection, which is certainly demanding in computing requirements
The interest in tactile sensors is increasing as their use in complex unstructured environments is demanded, like in telepresence, minimal invasive surgery, robotics etc. The matrix of pressure data these devices provide can be managed with many image processing algorithms to extract the required information. However, as in the case of vision chips or artificial retinas, problems arise when the array size and the computation complexity increase. Having a look to the skin, the information collected by every mechanoreceptor is not carried to the brain for its processing, but some complex pre-processing is performed to fit the limited throughput of the nervous system. This is specially important for high bandwidth demanding tasks. Experimental works report that neural response of skin mechanoreceptors encodes the change in local shape from an offset level rather than the absolute force or pressure distributions. This is also the behavior of the retina, which implements a spatio-temporal averaging. We propose the same strategy in tactile preprocessing, and we show preliminary results when it faces the detection of the slip, which involves fast real-time processing.© (2005) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
Voltage‐to‐frequency‐converters (VFCs) do typically employ relaxation oscillators to generate an output signal whose frequency depends linearly on an input voltage. VFCs are data converters; they are designed to codify the input voltage by means of the output frequency. Hence, they have strong requirements regarding linearity, dynamic range, and stability. On the other hand, compared to other converters, they have the advantage of low cost, particularly for large resolutions; 13 bit resolution is typical, and more than 20 bits are feasible. However, in such a case several milliseconds may be necessary to complete the conversion—the smaller the resolution, the shorter the conversion time.The serial frequency modulated output provided by VFCs is particularly well suited for telemetry, to send the outcome of a measurement by wire, radio, or fiberoptic links. After the frequency‐modulated signal is received, it can be easily codified into a digital word, or converted back into a voltage by a frequency‐to‐voltage converter. This type of output simplifies galvanic isolation through the use of either an optocoupler or a magnetic coupler. The serial output is also useful for increasing system compactness if throughput requirements are not high. Although many smart monolithic sensors have serial frequency‐modulated outputs, a VFC is needed in case we have an output voltage like that of a Wheatstone bridge after being amplified by an instrumentation amplifier, or the output of a generating sensor such as a thermocouple.This article deals first with the concepts behind voltage to frequency conversion, and then describes the main types of VFCs on the market, their limitations and main applications. Frequency‐to‐voltage converters are also briefly described.Keywords:data converters;relaxation oscillators;telemetry;sensor signal conditioning;remote sensing;telecontrol;variable-frequency oscillators;analog–digital conversion
This paper presents a mixed-signal neuro-fuzzy controller chip which, in terms of power consumption, input-output delay, and precision, performs as a fully analog implementation. However, it has much larger complexity than its purely analog counterparts. This combination of performance and complexity is achieved through the use of a mixed-signal architecture consisting of a programmable analog core of reduced complexity, and a strategy, and the associated mixed-signal circuitry, to cover the whole input space through the dynamic programming of this core. Since errors and delays are proportional to the reduced number of fuzzy rules included in the analog core, they are much smaller than in the case where the whole rule set is implemented by analog circuitry. Also, the area and the power consumption of the new architecture are smaller than those of its purely analog counterparts simply because most rules are implemented through programming. The paper presents a set of building blocks associated to this architecture, and gives results for an exemplary prototype. This prototype, called multiplexing fuzzy controller (MFCON), has been realized in a CMOS 0.7 μm standard technology. It has two inputs, implements 64 rules, and features 500 ns of input to output delay with 16-mW of power consumption. Results from the chip in a control application with a dc motor are also provided.
Learning algorithms have become of great interest to be applied not only to neural or hybrid neuro-fuzzy systems, but also as a tool to achieve a fine tuning of analog circuits, whose main drawback is their lack of precision. This paper presents accurate, discrete-time CMOS building blocks to implement learning rules on-chip. Specifically, a voltage mode high precision comparator as well as an absolute value circuit. These blocks, plus multiplexing in time techniques, are used to build a circuit to determine the polarity of the learning increments. Compactness and low power consumption have been considered the main requirements, since they are essential to increase the complexity of the neural systems. An example circuit has been simulated with HSPICE with the parameters of a 1 μm CMOS technology. Statistical variations of technological parameters were considered. The results show that all curves from 30 runs of a Monte Carlo analysis behave as expected, and at least 8 bits of resolution is achieved by the proposed techniques
Analog circuits are natural candidates to design fuzzy chips with optimum speed/power figures for precision up to about 1%. This paper presents a methodology and circuit blocks to realize fuzzy controllers in the form of analog CMOS chips. These chips can be made to adapt their function through electrical control. The proposed design methodology emphasizes modularity and simplicity at the circuit level - prerequisites to increasing processor complexity and operation speed. The paper include measurements from a silicon prototype of a fuzzy controller chip in CMOS 1.5 μm single-poly technology
This paper presents adaptive circuit blocks and related learning algorithms to design neuro/fuzzy inference systems using analog integrated circuits in CMOS, standard VLSI technologies. Proposed circuit building blocks are arranged in a layered architecture composed of five layers: fuzzification, T-norm, normalization, consequent, and output. Inference is performed using Takagi and Sugeno's if-then rules, particularly where the rule's output contain only a constant term-a singleton. The proposed learning scheme uses weight perturbation for the fuzzification layer and outstar for the output layer. A three-input, four-rule controller has been designed for demonstration purposes in a 1.6 μm CMOS single-poly, double-metal technology. Its operation speed is in the range of 5MFlips with systematic errors around 1%
Limits to precision impose limits on the complexity of analog circuits, hence fuzzy analog controllers are usually oriented to fast low-power systems with low medium complexity. This paper presents a strategy to preserve most of the advantages of an analog implementation, while allowing a marked increment in system complexity. Such a strategy consists of implementing a reduced number of rules, those that really determine the output in a lattice controller, which we call the analog core, then this core is dynamically programmed to perform the computation related to a specific rule set. HSPICE simulations from an example controller are shown to illustrate the viability of the proposal
Learning algorithms have become of great interest to be applied not only to neural or hybrid neuro-fuzzy systems, but also as a tool to achieve a fine tuning of analog circuits, whose main drawback is their lack of precision. This paper presents accurate, discrete-time CMOS building blocks to implement learning rules on-chip. Specifically, a voltage mode high precision comparator as well as an absolute value circuit. These blocks, plus multiplexing in time techniques, are used to build a circuit to determine the polarity of the learning increments. An exemplary circuit has been simulated with HSPICE with the parameters of a 1 μm CMOS technology. Statistical variations of technological parameters were considered. The results show that all curves from 30 runs of a Monte Carlo analysis behave as expected, and at least 8 bits of resolution are achieved by the proposed techniques
This paper focus on the design of adaptive mixed-signal fuzzy chips. These chips have parallel architecture and feature electrically-controllable surface maps. The design methodology is based on the use of composite transistors-modular and well suited for design automation. This methodology is supported by dedicated, hardware-compatible learning algorithms that combine weight-perturbation and outstar
Presents a mixed-signal fuzzy controller chip and its application to control of DC motors. The controller is based on a multiplexed architecture presented by the authors (1998), where building blocks are also described. We focus here on showing experimental results from an example implementation of this architecture as well as on illustrating its performance in an application that has been proposed and developed. The presented chip implements 64 rules, much more than the reported pure analog monolithic fuzzy controllers, while preserving most of their advantages. Specifically, the measured input-output delay is around 500 ns for a power consumption of 16 mW and the chip area (without pads) is 2.65 mm2. In the presented application, sensed motor speed and current are the controller input, while it determines the proper duty cycle to a PWM control circuit for the DC-DC converter that powers the motor drive. Experimental results of this application are also presented
The sections in this article are1Introduction2Basic Concepts and Models for Oscillator Design3Converters Based on Multivibrators4Schmitt Trigger Type Converters5Design Issues and Limitations of Converters Based on Multivibrators6Frequency to Voltage Conversion7Applications8Acknowledgment
Fernando Vidal-Verdú participated at conference 4th International Symposium on Sensor Science.