Scholarly article on topic 'Thin Flexible Sheet Handling Using Robotic Hand Equipped with Three-axis Tactile Sensors'

Thin Flexible Sheet Handling Using Robotic Hand Equipped with Three-axis Tactile Sensors Academic research paper on "Economics and business"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Computer Science
OECD Field of science
Keywords
{"Tactile sensor" / Three-axis / "Thin sheet" / Flexible / Pinch / "Sliding fingers" / Handling / "Robotic hand"}

Abstract of research paper on Economics and business, author of scientific article — K. Sugiman, M.A.M. Jusoh, M. Ohka, H. Yussof, S.C. Abdullah

Abstract Since a robotic hand equipped with three-axis tactile sensors can detect not only grasped force caused by pinching motion but also slippage generated by sliding fingers, it can be applied to various tasks such as cap twisting, exploring object shape and detecting hardness. Although thin sheet handling is considered one of the more difficult tasks in robotics because of object flexibility and thinness, the versatility of the hand seems to accomplish this task. In this paper, we evaluate the sensing ability of this sheet thickness to obtain basic data for handling and turning thin flexible sheets. In this experiment, we used 1,000-yen bills as specimens because they have precise and stable size and mechanical properties. Experimental results showed that the robotic hand could not distinguish sheet numbers through a pinching grip causing normal force but could distinguish them through sliding fingers causing tangential force.

Academic research paper on topic "Thin Flexible Sheet Handling Using Robotic Hand Equipped with Three-axis Tactile Sensors"

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Procedía Computer Science 76 (2015) 155 - 160

2015 IEEE International Symposium on Robotics and Intelligent Sensors (IRIS 2015)

Thin Flexible Sheet Handling Using Robotic Hand Equipped with

Three-axis Tactile Sensors

K. Sugiman*3, M. A. M. Jusoha, M. Ohkaa, H. Yussofb, S. C. Abdullahb

aGraduate School of Information Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan bCenter for Humanoid Robots and Bio-Sensing (HuRoBs), Faculty of Mechanical Engineering, Universiti Teknologi MARA, Malaysia

Abstract

Since a robotic hand equipped with three-axis tactile sensors can detect not only grasped force caused by pinching motion but also slippage generated by sliding fingers, it can be applied to various tasks such as cap twisting, exploring object shape and detecting hardness. Although thin sheet handling is considered one of the more difficult tasks in robotics because of object flexibility and thinness, the versatility of the hand seems to accomplish this task. In this paper, we evaluate the sensing ability of this sheet thickness to obtain basic data for handling and turning thin flexible sheets. In this experiment, we used 1,000-yen bills as specimens because they have precise and stable size and mechanical properties. Experimental results showed that the robotic hand could not distinguish sheet numbers through a pinching grip causing normal force but could distinguish them through sliding fingers causing tangential force.

© 2015PublishedbyElsevierB.V. Thisis anopenaccess article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-reviewunderresponsibility of organizing committee of the 2015 IEEE International Symposium on Robotics and Intelligent Sensors(IRIS2015)

Keywords: Tactile sensor; Three-axis; Thin sheet; Flexible; Pinch; Sliding fingers; Handling; Robotic hand

1. Introduction

Recently, the need for robots has expanded to the home and medical fields in part because of the decreasing birthrate and aging population in Japan. In its comprehensive strategy on science, technology and innovation1, the

* Corresponding author. Tel.: +81-52-789-4861; fax:+81-52-789-4800. E-mail a^rasx-ohka^s.nagoya-u.ac.jp

1877-0509 © 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of organizing committee of the 2015 IEEE International Symposium on Robotics and Intelligent Sensors (IRIS 2015) doi: 10.1016/j.procs.2015.12.332

Japanese government said that it expects that advanced robots will be applied to the home and medical fields for the care of elderly and disabled persons. To fulfill this purpose, the robots should be able to perform several tasks related to providing meals and going to and from the toilet and bed. To perform these tasks, the robots should possess tactile sensation, which is essential information for handling unknown objects of varying size, hardness and surface texture.2 Since membrane or film handling is important skill for the robot, there are some research works related to the skill such as membrane peeling3 and picking-up thin object4. Although counting thin-sheet number is one of essential skills for the membrane or film handling, it is not always progressed.

On the other hand, since a robotic hand equipped with three-axis tactile sensors5"12 can detect not only grasped force caused by pinching motion but also slippage force by sliding fingers, it can be applied to various tasks such as cap twisting, exploring object shape and detecting hardness. Although thin sheet handling considered one of the more difficult tasks in robotics because of their flexibility and thinness, the versatility of the hand seems to accomplish this task.

In this paper, we evaluate the sensing ability of sheet thickness to obtain basic data for thin sheet handling through pinching and sliding two fingers. In this experiment, we use 1,000-yen bills as specimens because they have precise and stable size and mechanical properties.

2. Optical Three-axis Tactile Sensor

2.1. Sensing Principle

Since the principle of our optical tactile sensor has been repeatedly explained in several conferences, journals and books,9"12 it is only briefly reviewed in this chapter. In the optical tactile sensor, light reflection and elastomer deformation are used as the basic principle.13 First, we assume a structure composed of a transparent board, a rubber sheet on it and small air gap between them; light is introduced from the end of the board. Since total reflection is caused in the inside of the board, light is not observed from the back side of board. However, if an object is put on the sheet and force is applied to the object, contact between the board and rubber sheet back surface under the object is caused, and the contact is observed from the back side of the board as a bright area because the total reflection is violated on the contact surface.

We have introduced several designs of the three-axis tactile sensor. Our sensors are divided into two groups: one of them is conical and columnar feeler type9"12 and the other is feeler movement type.14 At present, the former is applied to a hand-arm-robot to achieve several tasks such as cap-twisting, object-passing and object-assembling.9"12 The former sensor uses a cylindrical rubber element with a hemisphere and conical projections on the point and bottom, respectively.

To explain three-axis force sensing, we introduce Fig. 1. If a force is applied to the parietal point, normal force is observed by bright spots caused on the contact between the conical tips and the acrylic dome. The normal force component is proportional to the brightness of the spots, while the tangential force component is observed by centroid movement of the bright spots. Since there is almost no interference between brightness and centroid movement, normal and tangential components are independently measured.

Thus, normal and tangential forces are measured from an integrated brightness of the image data G and the centroid movement of the bright area, which is expressed by ux and uy in local coordinates embedded on each

sensing element. We assume that three components of applied force, Fx , Fy, and Fz, are proportional to ux, uy ,

and G .

2.2. Specifications of Ordinal Optical Three-axis Tactile Sensor

Figure 2 shows the design of the ordinal optical three-axis tactile sensor, which has repeatedly appeared in the authors' articles. Since we have recently introduced new designs, we call it the ordinal optical tactile sensor. It is composed of sensing elements, an acrylic dome, optical fiber, a fiber scope, a light source and a CCD camera. Light is directed from the light source through optical fibers to the acrylic dome; the side view attachment is mounted on the fiber scope end to observe the back surface of the acrylic dome.

Although it has some disadvantages such as an invisible zone and unacceptable large force that we are trying to overcome through new sensor designs,11 the ordinal optical three-axis tactile sensor91012 is still used because the new designs still require several tests and improvement for application. In this study, we used the ordinal optical three-axis tactile sensor because the task of sheet handling does not require large force.

The specifications of this sensor are shown in Table 1. Maximum detectable normal and tangential force components are 2 N and 0.15 N, respectively; resolutions of these components are 10 ^N and 35 ^N, respectively. This sensor does not accept large force, whereas it has precise resolution offorce components.

Fig. 1. Principle of three-axis force Table 1 Specifications of ordinal optical three-axis tactile sensor

Item Value

Number of elements 41

Detectable maximum force per element

Fx, Fy 0.15 N

Acceptable maximum force per element

Fx, Fy 2N

Resolution

Fx, Fy 10 ^N

Fz 35 ^N

Acrylic dome

Fig. 2 Design of ordinal optical three-axis tactile sensor

Fig. 3 Hand robot equipped with ordinal optical three-axis tactile sensors

3. Hand Robot Equipped with Ordinal Optical Three-axis Tactile Sensors

In our laboratory, an articulated hand-arm robot is developed to achieve the aforementioned intelligent tasks. Since we used the hand part of the robot in this study, it is shown in Fig. 3. Each finger has three motors (Micro actuator, Yasukawa Co.) and an ordinal optical three-axis tactile sensor. The motor is composed of an AC servo motor, a miniature harmonic drive and an encoder; since it generates 0.7 Nm torque, around 10 - 40 N force can be obtained as maximum force at the fingertip.

The hand is mounted on an articulated robotic arm: the hand-arm robot is shown in Fig. 4. Since the arm robot has only five degrees-of-freedom (DOF), the finger motors installed on the base are used for one wrist DOF. The velocity vector of the fingertip is decided by normal force and its motion mode (pinching or sliding two fingers). In this paper, tactile data obtained from the tactile sensor mounted on Finger 2 are used for data sampling and the grasping force control.

We used three computers for finger control, arm control and tactile sensors. Tactile data and the hand-and-arm status are transmitted through a local area network.

4. Experimental Results and Discussion

4.1. Experimental Procedure

Using the above hand robot, this experiment was performed as follows. The schematic flowchart of finger motion control is shown in Fig. 5. First, the robot pinches papers at the parietal point of each fingertip as shown in Fig. 6 in the first loop. At the start of this motion, each fingertip speed is set as 0.1 mm/s along xH or —xH direction; when contact force at the point (at element #00 shown in Fig. 7) reaches 1.0 N, the grasping motion is stopped. For the pinching motion test, we obtained data for nine elements (from #00 to #08 shown in Fig. 7) to measure normal force distribution at the fingertip. We expected to observe difference in normal force distribution related to the number of pieces of paper.

For sliding two-finger motion in the second loop, after the hand robot grasps papers with 1 N, Fingers 1 and 2 are linearly moved along yH and —yH directions (the coordinate is shown in Fig. 3) with 0.5 mm/s, respectively. This sliding motion is terminated when slippage on the fingertip is observed; the slippage is defined as a time derivative of tangential force, of which magnitude becomes larger via vibration of slippage.

The ratio of element #00's normal force value to element #01 - #08s' mean normal value is obtained for evaluation of sheet number: the difference between maximum and minimum norm of tangential components of force vector at element #00 for sliding finger motion. Furthermore, in this experiment, we used 1,000-yen bills as specimens because they have precise and stable size and mechanical properties.

4.2. Pinching

When we measure the thickness of a thin sheet, we use a micrometer caliper to pinch them. First, we examine the simplest way to examine the number of pinched sheets as with a caliper. At this time, thejoint data of the articulated finger do not seem suited for thin thickness meaurement because of the small playes being accumulated in the serial links of the finger. From the preceding psychophysical experiment, we introduced a hypothesis that "when judging thin sheet thickness, a human being judges the spring constant of the sheet instead of geometrical thickness."15 For robotics, we assume that the hypothesis is applied to thin sheet thickness.

C 1 st loop start

Arm part

Hand part

Recording data; inversed kinematics to control finger motors

C 1 st loop end C 2nd loop start ^

sliding two-finger mode?_ lYes"_

Sliding velocity along yH, ±vvg=0.5 mm/s | Qlotion is terminated^) Recording data; inversed kinematics to control finger motors

C 2nd loop end Oviotion is terminated^)

Fig. 4. Whole view ofhand-arm robot; arrows show arm part's DOF

Fig. 5 Flowchart for grasping and sliding sheets

If we intend to obtain the spring constant without displacement value, we can estimate hardness from force distribution. When a thicker sheet is grasped, the difference between maximum and minimum normal forces becames smaller because the sheet is not so deformed, and thiner sheet vice versa. Based on this idea, we introduce a new parameter for emulation of the spring constant as follows:

_ (m ean normal force of element #01 to #08)

7 T xiuu.

(Normal force of element #00]

Figure 8 shows the results of 1-sheet to 10-sheet grasping. As shown in Fig. 8, an increase of sheet number increases the R1 value. The value of R1 slightly increases from 1 to 3 sheets, then it is saturated, while standard deviation is very large. Although the sheet number seems to be counted through pinching motion in the range from 1 to 3 sheets, the discrimination ofnumber is not so easy because of small difference in Rl.

Fig. 6 Hand robot grasping sheets

Fig. 7 Configuration of sensing elements

Number of sheets

Number of sheets

Fig. 8. Experimental result for pinching motion Fig. 9 Experimental result for sliding two fingers

(Error bar: SD) (Error bar: SD)

4.3. Sliding Two Fingers

Compared to pinching motion, sliding two fingers seems to be effective because we have experience handling two sheet separation through a similar manner. To evaluate the effect of this motion, we introduce another parameter for easiness ofsliding as follows:

R2 = max^ Fx (t )2 + Fy (t )2 ) - mm(j Fx (t )2 + Fy (t )2 ), (2)

where Fx (/)and Fy (t) are time functions of x- andy-directional force vector components at element #00.

Figure 9 shows the difference in parameter R2 obtained from 1 sheet to 10 sheets (unit of R2 is pixel). As shown in Fig. 9, R2 value is 1.10 for 1 sheet, while it decreases to 0.87 for 2 sheets and keeps a constant value of around

0.83.after that; standard deviation is small enough at a glance. Thus, the result of 1 sheet is completely different from other cases. This is because slippage between sheets is occurring in the case of more than 2 sheets, while stational friction force is generated in the case of 1 sheet. Therefore, the difference between 1 sheet and others can be evaluated through sliding two fingers; the criteria of R2 becomes effective for evaluation ofthe difference.

Furthermore, for distinguish between 2 sheets and over 2 sheets, we will introduce the caliper-like measurement. Since the robot has at least repeatibility of 0.1 mm, thickness of over 2 sheets seems to be measured by pinch distance of the two fingers through revolution angles obtained from joint encoders. In the future, if we use the combination oftactile andjoint angle information, the robot will comprehend sheet number as human beings can.

5. Conclusion

Although thin flexible sheet handling is considered one ofthe more difficult tasks for advanced robots because of their flexibility and thinness, the versatility ofthe hand seems to accomplish this task. In this paper, we evaluate the sensing ability ofsheet thickness to obtain basic data for thin sheet handling.

In this experiment, we used 1,000-yen bills as specimens because they have precise and stable size and mechanical properties. We introduced two parameters for pinching motion and sliding two fingers: one of the parameters is for evaluation of sheet pliability; the other is for evaluation of slippage between two sheets. Experimental results show that sheet number discrimination is difficult for pinching motion causing normal force. However, it is possible through sliding fingers causing tangential force.

In future work, we will incorporate a hybrid control mechanism of positioning and force to the hand robot to accomplish turning and counting pages. Furthermore, the developed technology will be applied to handling another material such as aluminum foil and various cables.

References

1. Cabinet Office, Government ofJapan, Comprehensive Strategy on Science, Technology and Innovation; 2013, Retrieved from

http://www8.cao.go.jp/cstp/english/doc/20130607cao_sti_strategy_provisional.pdf

2. M. Ohka, Robotic Tactile Sensors, Wiley Encyclopedia of Computer Science and Engineering (Editor: B. W. Wah); 2009, 4, pp. 2454-2461.

3. M. Balicki, A. Uneri, I. Iordachita, J. Handa, P. Gehlbach and R. Taylor, Micro-force Sensing in Robot Assisted Membrane Peeling for Vitreoretinal Surgery,Med Image Comput Comput Assist Interv., 2010;13(Pt 3):303-10.

4. T. Yoshimi, N. Iwata, M. Mizukawa, Y. Ando, Picking up Operation of Thin Objects by Robot Arm with Two-fingered Parallel Soft Gripper,

2012 IEEE Workshop on Advanced Robotics and its Social Impacts (ARSO), 2012, pp.7-12.

5. Y. Yamada and M. R. Cutkosky, Tactile Sensor with 3-axis Force and Vibration Sensing Function and Its Application to Detect Rotational Slip,

1994 IEEE International Conference on Robotics and Automation; 1994, pp. 3550-3557.

6. K. Kamiyama, K. Vlack, T. Mizota, H. Kajimoto, N. Kawakami, S. Tachi, Vision-Based Sensor for Real-Time Measuring of Surface Traction

Fields, IEEE Computer Graphics and Applications; 2005, pp. 68-75.

7. Y. Tanaka, A. Nakai, E. Iwase, T. Goto, K. Matsumoto, and I. Shimoyama, Triaxial Tactile Sensor Chips with Piezoresistive Cantilevers Mountable on Curved Surface. Proceedings ofthe 4 th Asia Pacific Conference on Transducers and Micro/Nano Technologies; 2008, 1B1-1.

8. M. Hamidullah, M. Cheng, L. S. Lim, C. He, H. Feng, and W. Park, MEMS Tri-axial Tactile Sensor Packaging Using Polymer Flexible Cable

for Sensorised Guide Wire Application, 2011 IEEE 13th Electronics Packaging Technology Conference (EPTC); 2011, pp. 207-210.

9. M. Ohka, J. Takata, H. Kobayashi, H. Suzuki, N. Morisawa, and H. Yussof, Object Exploration and Manipulation Using a Robotic Finger

Equipped with an Optical Three-axis Tactile Sensor, Robotica; 2009, 27-5, pp. 763-770.

10. M. Ohka, S. C. Abdullah, J. Wada and H. B. Yussof, Two-hand-arm Manipulation Based on Tri-axial Tactile Data, International Journal of Social Robotics; 2012, 4-1, pp. 97-105.

11. M. Ohka, Y. Yamamoto, H. Yussof and S. C. Abdullah, All-in-type Optical Three-axis Tactile Sensor, 2014 IEEE International Symposium on Robotics and Manufacturing Automation; 2014, pp. 79-84.

12. M. Ohka, H. Yussof, S. C. Sukarnur, Three-axis Tactile Sensor, Introduction to Modern Robotics II, (Edited by D. Chugo and S. Yokota), Concept Press Ltd.; 2012, pp. 113-126.

13. K. Tanie, K. Komoriya, M. Kaneko, S. Tachi, and A. Fujikawa, A High Resolution Tactile Sensor, Proc. of 4th Int. Conf. on Robot Vision and Sensor Control; 1984, pp. 251-256.

14. M. Ohka, Y. Mitsuya, I. Higashioka and H. Kabeshita, An Experimental Optical Three-axis Tactile Sensor for Micro-Robots, Robotica, 2005, 23-4, pp. 457-465.

15. T. Miyaoka and M. Ohka, Tactile Information Processing Mechanisms to Discriminate Foil Thickness, Proceedings of the Seventeenth Annual Meeting ofthe International Society for Psychophysics, 17; 2001, pp. 529-534.