Tactile sensor hardware based on programmable SoC

Results and discussion: Fig. 2 shows a photograph of a prototype of the tactile sensor based on the architecture of Fig. 1. It is worth highlighting here that the columns in Fig. 1 are depicted from the circuital point of view, but the taxels they contain can be arranged in more than four columns in the sensor. For instance, the taxels in Fig. 2a are distributed to resemble the shape of a fingertip. Fig. 2b (left) shows the tactile image obtained when a handmade bar of rubber is pressed against its surface. Note that the shape of the bar is clearly distinguishable, and the crosstalk is cancelled. Fig. 2b (right) shows the output of the sensor when the electronics does not cancel the crosstalk. The PSoC in Fig. 2a is the CY8C27443 from CYPRESS. Depending on the resolution and dynamic requirements we obtain different solutions with the same hardware. For instance, the array is scanned in 1.37 ms if four 6 bit successive approximation (SAR) analogue-to-digital converters are used, while it is scanned in 7.2 ms with two multiplexed incremental converters of 8 bits and in 31.8 ms if the latter have 9 bits of resolution.
(Vbias ¼ 2.5 V). The latter can be obtained with a voltage regulator. Another alternative is the use of the reference Vbias generated on-chip, although one pin is dedicated to providing access to it and the number of columns is reduced by one (three with the PSoC of this Letter). The array is scanned to read the sensor data. This process sets the driving voltage of a row to zero, while the others are set to a high impedance. It takes as many steps as rows in the array, while the columns are read in parallel. Note that the voltage drop at the resistive paths in the notdriven rows is zero and they do not contribute to the output. RG = ((VS/Vbias) − 1)RS min at Fig. 1, where VS is the voltage supply (5 V) and RSmin is the minimum value of RS determined by the pressure range of our application. Another important equation is Vbias((1/Rbias) + (4/RS min)) ≤ IDRV max to guarantee that the current sunk by the row driver does not exceed the limit imposed by the PSoC (IDRVmax).
Tactile sensor hardware based on programmable SoC
F. Vidal-Verdu´, D. Bravo-Ga´lvez and O. Oballe-Peinado
This reported work focuses on the implementation of data acquisition electronics for a piezoresistive tactile sensor. It commonly involves the use of quite bulky electronics not very suitable to be housed in artificial hands, for instance. A very compact implementation because it is not based on a conventional microcontroller but on a programmable system-on-chip (SoC) is presented. This device implements analogue blocks that are not present in standard microcontrollers, so they do not have to be added as extra integrated circuits on the PCB, and a compact implementation based on a single chip plus a few electronic components is obtained.
a Output with electronics in Fig. 1 b (left) With an electronics that does not cancel interferences (right)
Conclusion: The electronics for a piezoresistive tactile sensor based on a single chip plus a few electronic components has been shown. Crosstalk cancellation and some parallelism in data acquisition are achieved by using the resources for signal conditioning on the PSoC.
A direct way to cancel this crosstalk implements one track per taxel in the array [2]. Another popular method for cancelling the interferences due to these parasitic currents is based on grounding and consists in driving the sensor in a way that the voltage drop throughout the parasitic resistive path is zero [3]. This Letter shows a sensor the circuitry of which is based on a programmable system-on-chip (PSoC). This device has analogue blocks inside. When adequately configured, these blocks allow the implementation of the strategy to cancel the crosstalk. This has been done for the sensor shown in this Letter, where a PSoC is used instead of a standard microcontroller that results in more bulky hardware [4].
Introduction: Tactile sensors are designed to be a sort of artificial skin for robots or machines with smart interfaces with humans [1]. These sensors provide an array of force measurements or pressure maps. Many sensors have been proposed, most of them based on piezoresistive, capacitive or optical principles. Piezoresistive sensors are the most popular because their simple construction results in a robust and relatively cheap sensor. On the other hand, these sensors are quite prone to error, and errors owing to drift, hysteresis, nonlinearity or crosstalk are significant. Crosstalk cancellation is a particularly relevant issue in the design of their data acquisition electronics. Crosstalk is due to parasitic current paths. These paths are commonly associated with the fact that these sensors are implemented with continuous sheets of piezoresistive materials that are shared by taxels. However, they are also caused by paths present in an architecture that addresses the sensor in rows and columns because the metal tracks that connect these rows and columns also form parasitic paths.