Abstract
In this paper, an active tactile sensor system for measurement of Young’s modulus and viscosity is developed. For the active sensing, the sensor unit with the 8 bimorph piezoelectric array is bonded to the speaker that is driven by a square wave. The transient response from the sensor signal of each bimorph piezo cell is processed to reduce the noise. From the output, the first peak and the ratio of first peak to time of first peak are extracted as the indexes for the measurement of Young’s modulus and viscosity, respectively. It is confirmed that the first peak has a clear correlation with the Young’s modulus, and that the ratio of first peak to time of first peak has a definite correlation with the viscosity. Further, the initial contact force affects the sensitivities of Young’s modulus and viscosity.
Keywords
Introduction
In the medical field, palpation by fingers and hand is frequently used since it can obtain easily multiple information such as roughness, softness, and viscosity, and son on, and the skilled palpation diagnosis is high-performance. On the other hand, minimally invasive procedures such as da Vinci surgery, laparoscopic surgery, and endoscopic surgery are widely used. In these surgeries, the operator feels only the force when the surgical instrument hits or does not hit the living body, and there is no system to convey the multiple information described above to an operator. The multiple information is effective to distinguish the illness and to achieve the safe surgery.
There are some studies of the tactile sensor to calculate Young’s modulus of the measured object from change the natural frequency when the vibrator contacts with the object [1], and to measure hardness using a pressure-sensitive element [2], and so on. However, these sensor systems can measure only single information.
Rederman and Krathuky have reported the various manipulation of hand and finger affects that we human can obtain multiple tactile information [3]. Based on these ideas, we have developed an active tactile sensor system that can measure surface roughness and Young’s modulus. For the measurement of roughness, the sensor is stroked over the surface of the object [4], and for the measurement of Young’s modulus the sensor is momentarily pushed into the object [5], while the sensor unit is common.
Biological tissue is a viscoelastic body with both elasticity and viscosity. However, many studies for measuring the hardness of biological soft tissues have quantified and evaluated only elastic properties such as Young’s modulus and stiffness. Viscosity change of biological tissue arises from viscosity change of body fluid in biological tissue, and viscosity change of body fluid is considered to be an important diagnostic index [6]. In other words, by obtaining both elastic and viscous characteristic information, it is possible to evaluate biological soft tissue from various angles, which is considered to be useful for detecting diseases.

Sensor unit.

Sensor unit attached to the speaker.

Experimental setup.
This paper is a study on the development of an active tactile sensor system for measuring Young’s modulus and viscosity of samples, by using the developed sensor unit. For the active sensing, the sensor unit with bimorph array is bonded to a speaker that is driven by input square wave voltage. 5 kinds of samples with various Young’s modulus and viscosity are prepared, and the sample is placed under the sensor unit during the measurement. The obtained transient waveform of the bimorph cell is processed to reduce the power noise and the outputs of center and edge parts are collected. From the signal output after the signal processing, the two indexes for evaluation of Young’s modulus and viscosity are extracted. Comparison of the two indexes and the Young’s modulus and viscosity are performed and the each sensitivity is investigated by changing the initial contact force.
The sensor unit consist of eight bimorph piezoelectric elements as shown Fig. 1 [5]. The sensor unit is bonded a speaker as shown Fig. 2. Figure 3 shows the sensor experimental setup. The speaker is driven by inducing the electric signal 7Vp-p 1Hz. The transient output of bimorph piezoelectric element is stored into a digital oscilloscope and used as the sensor output after signal processing. In the measurement, there are 3 kinds of initial contact force, 0.5 N, 0.7 N, and 0.9 N, adjusted by weight. And the initial force is checked by the electronic scale under the sample. The measurement is performed five times for each sample while changing the initial contact force.
Measurement sample
Measurement sample are 5 kinds. The sample for measurement is a cylindrical hole made of ABS, 85 mm in diameter and 19 mm in depth, filled with polyurethane resin or silicone rubber as shown in Fig. 4. Table 1 shows the characteristics of sample.

Measurement sample.

Noise reduction methods. (a) Raw waveform bimorph sensor (b) generated noise signal by using the raw wave form of second half part, and (c) the signal by subtracting the noise wave (b) from raw wave (a).
Characteristics of samples
By the speaker vibration, the sensor unit is momentarily pushed into the object and the transient output of bimorph sensor is obtained. The raw output from the each bimorph piezoelectric element array contains a power noise as shown in Fig. 5(a). In order to reduce the power noise signal, the power noise waveform is generated by using the second half part of raw waveform in Fig. 5(a). The generated noise is presented in Fig. 5(b). For the next process the output signal shown in Fig. 5(c) is obtained by subtracting the noise component (Fig. 5(b)) from the raw waveform (Fig. 5(a)).
Furthermore, the output is averaged by the number of measurements, and the center elements output are obtained by adding the outputs Ch4 and Ch5, and the edge elements output are obtained by adding Ch1 and Ch8. In this study, in order to investigate the shear effect [7], the both center and edge elements outputs are investigated.
Figure 6 is the typical sensor output of center and edge parts when initial contact force changes. Figure 6(a) and (b) are the sensor output at the case of initial contact force 0.5 N, and Fig. 6(c) and (d) are the case the force are 0.9 N. Figure 6(a) and (c) are the sensor output of center parts and Fig. 6(b) and (d) are the sensor output of edge parts.
Comparing the outputs, the initial contact force greatly affects the sensor output, it can be seen that the attenuations of the sensor output are fast when the initial contact force are 0.9 N. And it is found that the difference of sensor output between the center and edge parts is not significant. This is considered to be caused by the shape of the sample of this study.

Typical sensor output. (a) Output of center part for initial contact force 0.5 N, (b) output of edge part for initial contact force 0.5 N, (c) output of center part for initial contact force 0.9 N, and (d) output of edge part for initial contact force 0.9 N.
Two indexes for evaluation of Young’s modulus and the viscosity are extracted. The magnitude of the first peak of output Vp is extracted as the first index, since the first peak of a transient output is proportional to the generated force by the sample with the vibration and the force is proportional to the Young’s modulus.
For the second index of the evaluation of the viscosity, we focus on the rise characteristics of the sensor output. For the evaluation of the rise characteristics, the derivative of output at time t = 0 is considered to be suitable. However, the process to obtain the index is demanded to be simple. Therefore the ratio of the first peak of output Vp to the time of the first peak tp Vp/tp is extracted. The ratio Vp/tp can be used as the second index, because the relationship between the derivative of output at time t = 0 and the ratio Vp/tp is checked and it is confirmed the relation is linear and there is a one-to-one correspondence. The first index, Vp and the second index Vp/tp are obtained each contact condition and each sample.
Results and discussion
The relationships between the extracted indexesVp and Vp/tp and the Young’s modulus and viscosity listed in Table 1 are investigated. The relationship between the Young’s modulus and Vp is shown in Fig. 7, and the relationship between viscosity and Vp/tp is shown Fig. 8.
Figure 7(a) and (b) show the relationships between the Vp and Young’s modulus when the initial contact force is 0.5 N, Fig. 7(c) and (d) show that when the force is 0.9N. Figure 7(a) and (c) and Fig. 7(b) and (d) show that of center and edge elements, respectively. It is found that the first index Vp increase with increase of the Young’s modulus of the sample and that the distribution of Vp is divided according to the material of sample when the initial contact force is small.
Figure 8(a) and (b) show the relationships between the Vp/tp and the viscosity when the initial contact force is 0.5 N, Fig. 8(c) and (d) show that when the force is 0.9 N. Figure 8(a) and (c) and Fig. 8(b) and (d) show that of center elements and edge elements, respectively. It is found that the second parameter Vp/tp increase in proportional to the viscosity of the samples.

Relationship between the Young’s modulus of sample and the 1st parameter Vp. (a) Center parts when the initial contact force is 0.5 N, (b) edge parts for 0.5 N, (c) center parts for 0.9 N, and (d) edge parts for 0.9 N.

Relationship between the viscosity of sample and the 2nd parameter Vp/tp. (a) Center parts when the initial contact force is 0.5 N, (b) edge parts for 0.5 N, (c) center parts for 0.9 N, and (d) edge parts for 0.9 N.
Coefficient of determination between Vp and Young’s modulus
Coefficient of determination between Vp/tp and viscosity
In order to investigate the relationship in detail, the coefficients of determination for each initial contact force between Vp and Young’s modulus and between Vp/tp and viscosity are calculated. Table 2 and Table 3 show the calculated coefficients between Young’s modulus and Vp, and between the viscosity and Vp/tp, respectively. From Table 2 and Table 3, the initial contact force greatly affects the relationship, it is found that the larger the initial contact force, the stronger the relationship between Vp and Young’s modulus of the sample, and that the smaller the initial contact force, the stronger the relationship between Vp/tp and the viscosity of the sample.
The active tactile sensor system was designed using bimorph array as a sensory receptor. The measurement for Young’s modulus and viscosity was investigated. Through the signal processing the two indexes were proposed. The first index was the first peak of the output and the second one was the ratio of the first peak to the time of the first peak. The measurement of 5 kinds of sample was undertaken, and the comparisons were made between the two indexes and the specificity of the sample. The results are summarized as follows.
(1) The first peak of the output is proportional to the Young’s modulus, and the sensitivity is high when the initial contact force is large.
(2) The ratio of the first peak to the time of the first peak has a clear correlation with viscosity, and the sensitivity is high when the initial contact force is small.
In the future, the miniaturization is necessary to use in the medical field.
