Vous êtes sur la page 1sur 49

Autonomous Systems Lab Prof.

Roland Siegwart

Studies on Mechatronics

Force Sensing Technologies


Spring Term 2010

Supervised by: Christoph Hrzeler u Janosch Nikolic

Author: Matthias Fssler a

Contents
Abstract List of Figures List of Tables 1 Introduction 1.1 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Structure of the Report . . . . . . . . . . . . . . . . . . . . . . . . . 2 Tactile Sensing Technologies 2.1 Mechanical Tactile Sensors . . . . . . . . . . . . . 2.1.1 Whiskers and Antennae . . . . . . . . . . . 2.1.2 Mechanical Displacement . . . . . . . . . . 2.1.3 Pneumatic Touch Sensor and Foil Switches 2.1.4 Digital Tactile Sensor Array . . . . . . . . . 2.2 Capacitive Force Sensors . . . . . . . . . . . . . . . 2.3 Strain Gauges . . . . . . . . . . . . . . . . . . . . . 2.3.1 Metal Strain Gauges . . . . . . . . . . . . . 2.3.2 Semiconductor Strain Gauges . . . . . . . . 2.4 Piezoresistive Force Sensors . . . . . . . . . . . . . 2.4.1 Conductive Elastomers . . . . . . . . . . . . 2.4.2 Carbon Felt and Carbon Fibers . . . . . . . 2.5 Piezoelectric Force Sensors . . . . . . . . . . . . . . 2.6 Pyroelectric Force Sensors . . . . . . . . . . . . . . 2.7 Optical Force Sensors . . . . . . . . . . . . . . . . 2.7.1 Frustrated Internal Reection . . . . . . . . 2.7.2 Opto-Mechanical . . . . . . . . . . . . . . . 2.7.3 Fiber-Optic . . . . . . . . . . . . . . . . . . 2.7.4 Photoelasticity . . . . . . . . . . . . . . . . 2.7.5 Tracking of Optical Markers . . . . . . . . . 2.8 Magnetic Force Sensors . . . . . . . . . . . . . . . 2.8.1 Hall Eect and Magnetoresistance . . . . . 2.8.2 Magnetoelastic . . . . . . . . . . . . . . . . 2.9 Ultrasonic Force Sensors . . . . . . . . . . . . . . . 2.10 Electrochemical Force Sensors . . . . . . . . . . . . 3 Evaluation of Force Sensing Technologies 4 Additional Sensor Features 4.1 Multi-Axes Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Slip Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Measuring Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . i iii v vi 1 1 1 3 3 3 4 4 5 5 7 7 8 8 8 9 10 10 10 10 11 12 14 15 16 16 17 18 18 21 25 25 25 25

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

5 Force Sensors on the Market 5.1 Load Cells and Single Force Sensing Elements . . . . . . . . . . . . . 5.2 Description of the Parameters . . . . . . . . . . . . . . . . . . . . . . 5.3 Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Applications of Force Sensors 6.1 Surgical Applications . . . . . . . . . 6.2 Rehabilitation and service Robotics . 6.3 Agriculture and Food Processing . . 6.4 Specic Applications of force sensors 7 Conclusions Bibliography

27 27 28 29 35 35 35 36 36 37 39

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

ii

Abstract
This report contains an overview over the fundamental force sensing technologies with examples of manufactured sensors. These technologies are evaluated in terms of their applicability for docking maneuvers with unmanned helicopters. Force sensors that are available on the market are listed with their specications and evaluated likewise. Furthermore, some additional features of certain sensing technologies are given and the eld of applications of force sensors is described briey. In a conclusion, the suited technologies and sensors are summarized and a recommendation for a force sensor to use on a helicopter is given.

iii

iv

List of Figures
2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16 2.17 2.18 2.19 2.20 2.21 2.22 5.1 5.2 5.3 Whisker contact sensor by Clements and Rahn [15] . . . . . . . . . . Mechanical transducer with a linear potentiometer [2] . . . . . . . . Cross-section view of the pneumatic touch sensor [1] . . . . . . . . . Digital tactile Array sensor [1] . . . . . . . . . . . . . . . . . . . . . Capacitive touch sensor [1] . . . . . . . . . . . . . . . . . . . . . . . Typical metallic strain gauge pattern [36] . . . . . . . . . . . . . . . Piezoresistance using a separator [1] . . . . . . . . . . . . . . . . . . Carbon felt tactile sensor [1] . . . . . . . . . . . . . . . . . . . . . . . Tactile sensor based on frustrated internal reection [1] . . . . . . . Detecting shear forces with a microlever [1] . . . . . . . . . . . . . . An opto-mechanical array touch sensor [1] . . . . . . . . . . . . . . . Optical ber sensor based on varying coupling between crossed bers [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Light radiation due to microbending [1] . . . . . . . . . . . . . . . . Tactile sensor based on a deformable elastic reective surface and ber-optic technology [1] . . . . . . . . . . . . . . . . . . . . . . . . . Measuring stresses using photoelasticity [1] . . . . . . . . . . . . . . Photoelastic sensor [1] . . . . . . . . . . . . . . . . . . . . . . . . . . Hall eect [35] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Magnetoresistive sensor using current-carrying wires [1] . . . . . . . Magnetoresistive sensor using magnetic dipoles [1] . . . . . . . . . . Changes in ux distribution caused by applied force [1] . . . . . . . . Tactile sensor using ultrasonic pulses to detect elastic skin thickness [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Schematic diagram of the streaming potential sensor [1] . . . . . . . ATI Nano17 6-axis force and torque transducer . . . . . . . . . . . . A Honeywell FSS1500NSB force sensor . . . . . . . . . . . . . . . . . FlexiForce foil sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 4 5 6 7 7 9 9 11 11 12 12 13 14 15 15 16 17 17 18 18 19 27 28 28

List of Tables
3.1 3.2 3.3 General advantages and disadvantages of dierent sensor technologies 22 General advantages and disadvantages of dierent sensor technologies (continued) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Sensor technologies categorized with respect to the desired sensor properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Dierent force sensing elements with their specications . . . . . . . 30 Dierent force sensing elements with their specications (continued) 31 Some examples of special force sensors that are available on the market 32

5.1 5.2 5.3

vi

Chapter 1

Introduction
1.1 Goal

The goal of this report is to collect information about existing technologies and products for force sensing devices. The existing technologies and the available sensors on the market should be described and evaluated. The evaluation refers to the suitability of these technologies and particular sensors for docking maneuvres with an unmanned rotorcraft. Preferred attributes of the force sensor are low weight, small size, robust especially in the sense of shock resistance, low price and a simple construction in case it has to be customized.

1.2

Structure of the Report

This report is organized as follows: Chapter 2 introduces the dierent force sensor technologies. It explains all the fundamental transduction methods for force sensing. In Chapter 3, this technologies are evaluated and their advantages and disadvantages are registered. Additional features of sensors that some technologies provide besides sensing of normal forces are explained in Chapter 4. Common force sensors that are available on the market are presented in Chapter 5 together with their properties as far as known. Chapter 6 presents the main elds where force sensors are applied as well as some specic applications of commercial force sensors. In the last Chapter, conclusions, based on the evaluation of sensor technologies and sensors on the market, are drawn regarding docking maneuvers with unmanned rotorcrafts.

Chapter 1. Introduction

Chapter 2

Tactile Sensing Technologies


This chapter gives an overview and a description of the function principle of the known force sensor technologies. A concrete example with a gure for illustration is given to the described technologies. The technologies presented here are for the most part fundamental transduction methods presented by Russell [1] and Nicholls [2] as well as from chapter 19 of the Handbook of Robotics [7]. The books of Russell and Nicholls were published in the year 1990 and 1992 respectively. But according to newer literature by Lee and Nicholls [6], most of the possible forms of physical transduction methods have now been explored and there seems little scope for new fundamental transducers. So the fundamental transduction methods are still the same. The current research oerings are mainly concerned with novel packagings, better designs, improved engineering and more complete analysis.

2.1
2.1.1

Mechanical Tactile Sensors


Whiskers and Antennae

Whisker or antenna sensors are in essence a hybrid of proprioceptive and tactile information. The main components of the earliest sensors of this type are a base angle sensor and a tip contact sensor to explore the environment. This simple assembly provides information whether contact occurs and if so, also the contact location. For many animals, whiskers or antennae provide an extremely accurate combination of contact sensing and proprioceptive information. Examples of such sensors are reported by Kaneko [14] or Clements and Rahn [15].

Figure 2.1: Whisker contact sensor by Clements and Rahn [15] 3

Chapter 2. Tactile Sensing Technologies

2.1.2

Mechanical Displacement

This kind of force sensors rely on a mechanical displacement caused by an applied force. The simplest example is a spring loaded switch giving on-o contact readings. A linear potentiometer provides a graded scale of deection, and the output can be considered in terms of either force or linear displacement (see Figure 2.2). A common example of mechanical displacement in simple touch sensors is the movement of a linear probe. For example, Presern et al. [16] designed a three degrees of freedom probe for arc welding applications where it is intended for seam tracking.

Figure 2.2: Mechanical transducer with a linear potentiometer [2]

2.1.3

Pneumatic Touch Sensor and Foil Switches

A switch is one of the simplest touch-actuated devices for detecting the presence of an object. A very simple and small form of a switch are foil layers as they are used in cheap calculator keyboards. These switches use two foil layers with a conductor and one layer to separate them. If enough force is applied the two external layers are pressed together and contact between the two conductors is produced. Pneumatic touch sensors are more complex than foil layers but work in a similar way. Usually, pneumatic touch sensors are built by a shallow spherical dome made of thin sheet metal. When a critical value of force is applied the dome collapses with a click, and later returns to its original shape as the force is removed. As the dome collapses, electrical contact with an electrode on the inside of the dome is produced. The sensor sensitivity varies with dierent geometries and material properties of the dome as well as with its pressure inside. Figure 2.3 shows a crosssection view through part of a pneumatic sensor array by Garrison and Wang [11]. The sensitivity of this sensor can be varied by changing the pressurized uid inside the domes. The single sensing elements are spaced 2.54mm apart.

2.2. Capacitive Force Sensors

Figure 2.3: Cross-section view of the pneumatic touch sensor [1]

2.1.4

Digital Tactile Sensor Array

A simple switch indicates only whether or not the applied force exceeds a set threshold. With a closely grouped array of switches, each having a dierent threshold, the magnitude of an applied force can be estimated. For example, this can be implemented with a conductive, elastic material pressed against a V-shaped notch as illustrated in Figure 2.4. More pressure is required to force the elastomer into the notch as the width of the notch narrows. Commonly, a row of aluminium pads deposited along the bottom of the notch forms one electrode and a sheet of conductive elastomer makes the other electrode. Like this a small array of switches, with each switch in the array having a dierent pressure threshold, can be produced. Linear, logarithmic, or exponential response to pressure can be obtained by varying the shape of the notch. An advantage of this sensor design is that no analog-to-digital converter is necessary. A disadvantage of this sensor is that the fragile silicon chip is close to the point of contact with external objects. A tactile sensor array based on this idea has been constructed by Raibert [12] where a sheet of elastic material is pressed against a round hole.

2.2

Capacitive Force Sensors

Over small distances capacitance can be used to measure the separation between two conductive plates. In principle, capacitance can be used to measure both shear and normal forces. Shear forces can alter the area of overlap between two plates and normal forces can aect the plate separation. However, it is dicult to separate the two eects when trying to measure both at the same time. An example of an capacitive sensor array constructed by Siegel et al. [13] is shown in Figure 2.5. This sensor uses a set of row and column electrodes which are spatially separated by a dielectric. A multiplexing scheme allows the capacitance at the cross point of any row and column electrode to be measured and hence the deection at that point is determined. Capacitive sensor arrays can be molded and provide very accurate measurements of skin deection. The properties of the sensor, in terms of hysteresis, creep, memory, non-linearity, etc., are governed by the elastic material between the capacitor plates.

Chapter 2. Tactile Sensing Technologies

Figure 2.4: Digital tactile Array sensor [1]

2.3. Strain Gauges

Figure 2.5: Capacitive touch sensor [1]

2.3

Strain Gauges

Strain gauges are a common tool to measure forces and are often applied in commercial force sensors. So even if semiconductor strain gauges belong to the category of piezoresistive force sensors they are described in this separate section. For measurements of small strain, semiconductor strain gauges are often preferred over foil gauges. Still both types of strain gauges are presented and their dierences and advantages are explained.

2.3.1

Metal Strain Gauges

A strain gauge consists of an insulating exible backing which supports a metallic foil pattern. A typical pattern is shown in Figure 2.6 which measures strain in longitudinal direction. The strain gauge is attached to the object by a suitable adhesive. When the conductive metallic foil is stretched within the limits of its elasticity, it will become narrower and longer, changes that increase its electrical resistance. The pattern with several paths in parallel increases the eect of a change in the total resistance. From the electrical resistance of the strain gauge, the amount of applied stress may be inferred typically using a Wheatstone bridge. By using the material properties of the object which the gauge is attached to, also the applied force can be calculated.

Figure 2.6: Typical metallic strain gauge pattern [36]

Chapter 2. Tactile Sensing Technologies

2.3.2

Semiconductor Strain Gauges

Additionally to the change of resistance due to the geometrical change of the conductor in metal strain gauges, the resistance of piezoresistive strain gauges also change because of the new mechanical state of stress. This new state of stress changes the relative resistance of the piezoresistive material. So the change of the total resistance consists of a geometric part and a material specic part. The total resistance with respect to the strain is given by R() = with () l() A() This can be transformed into R = (1 + 2 + ) = K R0 (2.5) =
0 (1

()l() A()

(2.1)

+ )

(2.2) (2.3) (2.4)

= l0 (1 + ) = A0 (1 )2

Where R is the change in resistance, R0 is the resistance of the unstressed gauge, is the Poissons ratio of the conductor, is a factor of the change of the relative resistance due to strain and is the strain. The factor (1 + 2 + ) is called the K-factor and is a measure of the sensitivity of the strain gauge. For metallic foil gauges, the K-factor is around 2.1 and is dominated by the deformation. For semiconductor strain gauges the change of the total resistance is based up to 98% on the change of the specic resistance. Semiconductor strain gauges can reach K-factors of up to 150 which allows stier deformable elements of the sensor. This large sensitivity is the major advantage of semiconductor strain gauges. In addition, semiconductor techniques allow much smaller strain gauges than metal strain gauges. The major disadvantage of semiconductor strain gauges is that they are more sensitive to temperature changes and they are not as robust as metal strain gauges. Erler [9] discusses the properties of semiconductor strain gauges in detail in Chapter 6 of his book.

2.4
2.4.1

Piezoresistive Force Sensors


Conductive Elastomers

Conductive elastomers are insulating natural or silicone-based rubbers made conductive by adding particles of conducting or semiconducting materials such as silver or carbon. Most of these forms of conductive rubber show little change in bulk resistance as they are compressed. However, area of contact and hence inverse contact resistance can be made to vary with applied force. An example of a sensor using conductive elastomers described by Hillis [17] is shown in Figure 2.7. The elastomer and the contact pad are separated by a woven mesh of a nylon stocking which gives no contact, hence innite resistance, for zero normal force. At a certain threshold force the conductive elastomer makes contact with the electrode. Additional force increases the area of contact and thus reduces the contact resistance. The sensor of Hillis has an array of 256 tactile sensor elements in the area of 1cm2 and a sensing range of 1-100g.

2.4. Piezoresistive Force Sensors

Figure 2.7: Piezoresistance using a separator [1]

2.4.2

Carbon Felt and Carbon Fibers

Larcombe [28] has described piezoresistive sensors constructed by sandwiching carbon felt and carbon bers between metal electrodes as shown in Figure 2.8. As the load increases, the carbon bers are compacted together, making more electrical contacts and reducing the felt resistance. At loads in excess of 5 kg the area of contact between touching bers starts to increase and this leads to a further reduction in resistance. Carbon ber and carbon felt sensors are rugged and can be shaped. They withstand very high temperatures and considerable overloads. A disadvantage of this sensor is a great deal of electrical noise with a load of less than 10 g. However, these sensors are very robust and well suited for sensing in very inhospitable environments.

Figure 2.8: Carbon felt tactile sensor [1]

Chapter 2. Tactile Sensing Technologies

10

2.5

Piezoelectric Force Sensors

The piezoelectric eect gets crystals of quartz to produce an electrical voltage when pressure is applied to the crystal. The piezoelectric eect only occurs in crystals which do not have a center of symmetry. Depending on the design of the sensor, different modes to load the piezoelectric element can be used: longitudinal, transversal and shear. The voltage generated across a sensing element is proportionally related to the applied pressure. If a load is maintained, then the sensor output decays to zero. Therefore, these sensors are most suited for sensing dynamic forces. One of the most common material used for piezoelectric sensors is a polymer known as polyvinylidene uoride (PVF2 or PVDF). PVF2 has good mechanical properties, is a durable material and shows one of the largest piezoelectric eect. Its exibility, sensitivity, and large electrical output oer many advantages for touch sensors in particular. Piezoelectric materials are also pyroelectric as described in the next section. A problem with materials that are both piezoelectric and pyroelectric is separating the two eects, thus protection from thermal variations may be necessary if pressure variations are important. Dario and Buttazzo [19] have developed a skinlike sensor based on PVF2 lm. This sensor contains two force-sensing layers and has the additional capability of sensing thermal properties. The sensing elements are arranged in a hexagon at 5mm spacing. A similar sensor built by Dario et al. [20] has a dynamic range of 4000:1, where load over 0.01 40N were established.

2.6

Pyroelectric Force Sensors

Correspondingly to the piezoelectric eect, the pyroelectric eect is the generation of a voltage when the sensing element is heated or cooled. As piezoelectric sensors, pyroelectric sensors are inherently dynamic which means that the sensor output decays to zero if the temperature is constant. The above presented PVF2 material is also highly pyroelectric. A pyroelectric sensor consists of a pyroelectric element and a heat source. The heat source causes the device to heat up. When an object touches the sensor surface, heat ows from the sensor into the object ore vice versa if the object is hotter than the sensor. The resulting temperature change can then be measured. The pyroelectric eect is not well suited to measure scaled forces but has to be taken into account when building a piezoelectric sensor (Dario and Buttazzo [19]). In return, a pyroelectric sensor is well suited to detect slip on a surface. If the sensor has no slip, the surface temperature at the point of contact becomes the same as the sensor temperature. As soon as the sensor moves, a temperature change and therefore slip can be detected.

2.7
2.7.1

Optical Force Sensors


Frustrated Internal Reection

A sheet of clear plastic can act as a light guide. Light introduced at one edge will propagate across the sheet by total internal reection and emerge at the opposite edge. Total internal reection occurs when light is propagating in the denser of two media and strikes the interface at an angle larger than a particular critical angle with respect to the normal to the interface. When the surface of the light guide comes into contact with an external object then at that point total internal reection is frustrated and light emerges from the opposite side of the light guide. In practice , a reective rubber is placed on the light guide to protect it and to exclude external light as it is shown in Figure 2.9. If the rubber sheet is molded with a textured surface then an output proportional to the area of contact, and

11

2.7. Optical Force Sensors

hence applied force, can be obtained (Tanie et al. [22]). The light which emerges from the back of the light guide is detected either by an array of photodiodes, solid state optical sensors, or transported away from the sensor by optical bers. This kind of sensor can also be made sensitive to shear forces in the reective rubber material as the microlever proposed by Dario et al. [21] that is illustrated in Figure 2.10.

Figure 2.9: Tactile sensor based on frustrated internal reection [1]

Figure 2.10: Detecting shear forces with a microlever [1]

2.7.2

Opto-Mechanical

The Lord Corporation produced an opto-mechanical touch sensor as shown in Figure 2.11. It contains a rubber skin with an array of mushroom-shaped projections molded into its surface (Rebman and Morris [23]). The head of the mushroom concentrates the normal force and the stalk acts as an optical shutter to modulate light transmission between a light-emitting diode and a photo detector depending upon normal force. Because the deecting element is made of rubber, the sensor response will be subject to the usual problems of creep, hysteresis, memory, and temperature variation. The Lord LTS-210 for example consists of an array of optomechanical force sensors on a 1.8mm spacing. The Construction of this sensor is quite labor intensive.

Chapter 2. Tactile Sensing Technologies

12

Figure 2.11: An opto-mechanical array touch sensor [1]

2.7.3

Fiber-Optic

Light Coupling Between Adjacent Fibers Light propagates along an optical ber with very little lost due to radiation; however, if the surface of the ber is roughened then at that point light can leave and enter the ber. If two optical bers pass close to each other and both have a roughened surface then light can pass between the bers. The sensor design shown in Figure 2.12 uses D-section cords made of rubber as a deformable member and the light coupling between crossed plastic optical bers to measure the resulting deection. A 4x4 array with 1cm spacing has been reported by Schoenwald et al. [24]. Loads applied normal to the sensor surface compress the D-section elastomer cords moving the bers closer together and thus increases the light coupling. This sensor is exible and can conform to complex curved surfaces. It has the advantage of noise immunity associated with optical transducers and the imperfections introduced by the use of elastomer materials as the deformable member.

Figure 2.12: Optical ber sensor based on varying coupling between crossed bers [1] Bending Losses in Optical Fibers Light propagates through an optical ber by repeated internal reection from the cladding interface. For total internal reection to occur, light must strike this surface at an angle greater than a critical angle with respect to the normal to the surface. If the ber is subjected to a signicant amount of bending over a length comparable to the distance between successive internal reections then the angle of

13

2.7. Optical Force Sensors

incidence can be reduced suciently for the light to leave the core. This eect is illustrated in Figure 2.13. Under these conditions of microbending, the amount of light transmitted by an optical ber is greatly reduced. This eect has been put to use in microbend touch sensors. An experimental 2x2 robot sensor has been reported by Winger and Lee [25]. This sensor is capable of detegting a 5g variation in applied load in its linear region which ranges between 125g and 225g per sensor element. Hysteresis proved to be a large problem and this was thought to be caused by the cladding material.

Figure 2.13: Light radiation due to microbending [1]

Optical Skin Thickness Sensor

Figure 2.14 shows a cross-section of a sensor that determines the thickness of a transparent, deformable elastomer layer by measuring the intensity of light reected back from the far side of the layer. Light is introduced into the sensor via an optical ber. A widening cone of light propagates out through a layer of transparent elastomer and is reected by an outer skin of white elastomer. The reected light is received by a second ber which is viewed by a computer vision system to measure the reected light. When an external force compresses the transparent elastomer this shortens the distance travelled by the light cone, limiting the light dispersion and thus reducing the light gathered by the receiving ber. A sensor described by Schneiter and Sheridan [26] contains 2100 sensitive points per square inch (6.45cm2 ) and exhibits a dynamic range of only 18:1. This sensor is very labor intensive since each optical ber was positioned by hand and it is rather frail since the clear rubber material was found to fatigue after only a few hundred operating cycles.

Chapter 2. Tactile Sensing Technologies

14

Figure 2.14: Tactile sensor based on a deformable elastic reective surface and ber-optic technology [1]

2.7.4

Photoelasticity

Photoelasticity can be used to measure stresses in a sample of optically active material. Consider the experimental set-up shwon in Figure 2.15. (a) Light radiated from a light source contains many waves of diering polarization and amplitude. Both polarization and amplitude vary with time. (b) Polarizer 1 only allows through components of each wave having a particular plane of polarization, and blocks those components which are at right angles. (c) Upon entereing the birefringent material light is split into two components, polarized at right angles. These components are aligned with the planes of maximum and minimum stress in the material. (d) If the two waves emerge from the birengent material with the same relative phase that they had when they entered then the original wave is reconstructed. (e) Polarizer 2 is rotated 90 with respect to polarizer 1 and therefore blocks transmission of the reconstructed wave. If the two waves do not emerge with the same relative phase that they had when they entered the birefringent material then an elliptically polarized wave results. Part of this elliptically polarized wave is passed by polarizer 2. Dark areas, or fringes, where light is not transmitted through the system, are the result of two eects: 1. Isoclinics: Areas where light from polarizer 1 is in line with one of the principal stress axes. In this case the light is not split into two components and therefore emerges unaltered. This eect provides information about the orientation of principal stresses in the birefringent material. 2. Isochromatics: Result when light is split by the birefringent material but emerges with the same phase that it entered. This will be true if the principal stresses in the material are identical or dier by an amount which produces an integral number of phase rotations. An example of such a sensor was developed by Cameron et al. [27] and is shown in Figure 2.16. It measures the forces applied to a sheet of birefringent material. The sheet of birefringent material is illuminated by circularly polarized light to visualize isochromatic fringes. A CCD camera records the resulting stress pattern.

15

2.7. Optical Force Sensors

The camera image can be analysed to determine the distribution of stresses within the birefringent material.

Figure 2.15: Measuring stresses using photoelasticity [1]

Figure 2.16: Photoelastic sensor [1]

2.7.5

Tracking of Optical Markers

This principle is based on the idea of using deformable tactile sensors. It combines optical tracking with models of the sensors skin to predict the sensor skin deformation. The sensor skin consists of markers on the inside which are tracked by a tiny CCD camera. The measured data of the camera is then combined with the mechanical model of the skin to derive its deformation and the applied force on the skin. Ferrier and Brocket [29] implemented a tactile sensor with a CCD camera focused on a 7x7 array of dots that are marked on the inside of a gel-lled silicone ngertip membrane. An algorithm is then used to construct a 13x13 grid over the array of dots. Another interesting tactile sensor uses vision to track an array of spherical markers embedded in a transparent elastomer to infer the stress state of the skin material due to applied force. This sensor is commercialized under the tradename GelForce by the Tachi Lab [37].

Chapter 2. Tactile Sensing Technologies

16

2.8

Magnetic Force Sensors

2.8.1

Hall Eect and Magnetoresistance

The Hall eect is closely related to the motor eect observed as a force on a currentcarrying conductor in a magnetic eld. In Figure 2.17, if charge carriers (1) ow through a conductive material (2) and a magnetic ux (4) is established, then they experience a force orthogonal to their ow direction and the magnetic eld direction. This deection of the charge carriers is then producing a resulting Hall potential in direction of the deection which can be measured. Due to this deection of the charge carriers, they take a longer path to travel the length of the conductive material. Eectively the deected particles have a lower mobility and this shows as an increased electrical resistance. This eect is known as magnetoresistance. Both the Hall eect and magnetoresistance can be used to measure magnetic eld intensity. Note that a Hall eect sensor is only sensitive to magnetic elds in one direction while the magnetoresistive eect can be used to detect magnetic eld having any orientation within a plane normal to the current ow. A schematic example of a magnetoresistive sensor is shown in Figure 2.18. The magnetic eld is provided by current-carrying wires within an elastomer. When the elastomer is compressed, a change of the magnetic eld in the magnetoresistive element can be measured. Hackwood et al. [30] developed a magnetoresistive sensor that can measure normal forces, tangential forces and torques. In this sensor design, a sheet of silicone elastomer contains an array of embedded magnetic dipoles (see Figure 2.19). Beneath each dipole four Permalloy magnetoresistive sensors are mounted on a rigid substrate. Magnetic eld strength at each of the four magnetoresistive sensors is used to determine both position and orientation of the magnetic dipole.

Figure 2.17: Hall eect [35]

17

2.8. Magnetic Force Sensors

Figure 2.18: Magnetoresistive sensor using current-carrying wires [1]

Figure 2.19: Magnetoresistive sensor using magnetic dipoles [1]

2.8.2

Magnetoelastic

Magnetoelastic sensors are made from a magnetostrictive material. These materials change their magnetic permeability when they are deformed. The sensor element shown in Figure 2.20 contains two windings arranged at right angles. In the unstressed condition the magnetostrictive material is isotropic and hence there is no ux coupling between the two windings and no output voltage at the secondary winding. When a force is applied to the sensor the permeability of the magnetostrictive material decreases in the vertical and increases in the horizontal direction. This changes the magnetic eld of the primary winding in such a way that ux links the secondary winding and an output voltage is produced in the secondary winding. A sensor array of 16x16 magnetoelastic sensor elements with 2.5mm spacing has been reported by Luo et al. [31]. The sensor array was covered by a sheet of elastomer to provide protection and an improved gripping surface. Good sensitivity and linearity, and low hysteresis are claimed for the sensor but the sensors and their associated circuits are relatively complicated.

Chapter 2. Tactile Sensing Technologies

18

Figure 2.20: Changes in ux distribution caused by applied force [1]

2.9

Ultrasonic Force Sensors

Ultrasound has been in use for a long time to measure the thickness of objects. We can measure the time taken for an ultrasonic pulse to travel through the material, reect o the back surface and return. If the speed of propagation of the ultrasonic wave in the material is known then the material thickness can be calculated. This principle can be used to measure the thickness of a exible elastomer layer at many closely spaced points. Like this the deformation of the elastomer can be measured as it is shown in Figure 2.21 and the pressure applied on its surface can be calculated. Grahn and Astle [32] claim a dynamic range of 2000:1 and a spatial resolution of 0.5mm for their ultrasonic force sensor. If the sensor is immersed in a liquid or touches an object with a similar acoustic impedance to the skin material, it is possible that the ultrasonic pulse propagates past the surface of the skin material without being reected. So attention has to be paid on the material properties of the sensor surface and the objects that should be touched.

Figure 2.21: Tactile sensor using ultrasonic pulses to detect elastic skin thickness [1]

2.10

Electrochemical Force Sensors

Chemical-impregnated gels have been formulated to make them sensitive to deformation. The sensor shown in Figure 2.22 by De Rossi et al. [33] uses a gel that contains an immobile negative charge which is balanced by a mobile positive charge. An ionized gel disk 1cm in diameter and 0.40.5mm thick is made from polyacrylic

19

2.10. Electrochemical Force Sensors

acid and polycinylic acid. When pressure is applied to the gel, positively charged liquid is forced out of the gel and an inhomogeneity of charge is formed which constitutes the streaming potential. This potential is picked up by two thin electrodes. The sensor can detect low-frequency deformations but has no steady state response.

Figure 2.22: Schematic diagram of the streaming potential sensor [1]

Chapter 2. Tactile Sensing Technologies

20

Chapter 3

Evaluation of Force Sensing Technologies


First, a few terms which are needed for the evaluation of the force sensing technologies are explained.

1. Dynamic Range: The ratio of largest to smallest detectable force.

2. Spatial Resolution: The space that one single sensing element takes. So for an Array sensor the spatial resolution gives the amount of single sensing elements that can be placed in a given length or area.

3. Inherently Dynamic: Sensor output decays to zero for constant load.

A very general evaluation of the dierent sensor technologies is given in Table 3.1 and Table 3.2. Due to the fact that most of the technologies are only applied in research it makes no sense to give specic characteristics of the built sensors since they vary very much between dierent sensor designs within the same technology. Some specic examples are given in Chapter 2 to give an impression on the limitations of the corresponding technology. Table 3.3 categorizes the technologies with respect to the desired sensor properties introduced in Section 1.1. In this table ++ denotes very good or well suited and denotes bad or not at all suited. 21

Chapter 3. Evaluation of Force Sensing Technologies

22

Table 3.1: General advantages and disadvantages of dierent sensor technologies

23

Table 3.2: General advantages and disadvantages of dierent sensor technologies (continued)

Chapter 3. Evaluation of Force Sensing Technologies

24

Table 3.3: Sensor technologies categorized with respect to the desired sensor properties

Chapter 4

Additional Sensor Features


4.1 Multi-Axes Sensors

Load cells (see Section 5.1) usually measures forces and torques in multiple axis. But also some of the presented fundamental force sensing technologies are able to measure more than just normal force in one direction. A sensor based on frustrated internal reections for example have been made sensitive to shear forces by Dario et al. [21]. This sensor is described in Section 2.7.1 and illustrated in Figure 2.10. With piezoelectric sensors (see Section 2.5), depending on the design of the sensor, dierent modes to load the piezoelectric element can be used: longitudinal, transversal and shear. A further example is the sensor described in Section 2.8.1 by Hackwood et al. [30] is illustrated in Figure 2.19. This hall eect sensor can measure normal and shear forces as well as torques.

4.2

Slip Detection

Several force sensing technologies can also be used to detect slip. Pyroelectric sensors (see Section 2.6) are well suited to detect slip by measuring a change in the temperature as soon as the sensor starts to move along a touched object. Also carbon felt sensors (see Section 2.4.2) have been used to detect slip by measuring the noise due to the friction at the contact point. A capacitive sensor which is capable of detecting slip is described by Luo [34]. This sensor uses the change in capacitance caused by relative contact movement between sensor and object. The contacting sensor surface comprises a set of parallel rollers. Each roller is a half cylinder of conductive material, and a half cylinder of nonconductive material and therefore act as a variable capacitor. Other approaches have been realized by detecting the sound generated by movement against the sensor surface with piezocrystal microphones. Of course there are many other approaches to detect slip, which are not discussed here.

4.3

Measuring Curvature

Provancher and Cutkosky [10] have built a sensor that can measure the curvature of a touched object. The sensor consists of an array of strain gauges embedded in a compliant membrane. The curvature is estimated with a least squares procedure with the measurements of all the strain gauges. The resulting sensor is inexpensive and robust and can be used for object handling and exploration. The curvature 25

Chapter 4. Additional Sensor Features

26

measurements provide information for manipulation planning and control and provide an estimate of the local object geometry.

Chapter 5

Force Sensors on the Market


5.1 Load Cells and Single Force Sensing Elements

There are two dierent main elds of force sensors on the market. One are so called Load cells (see Figure 5.1) and the other are small single sensing elements. These sensing elements are based on piezoresistive transduction and can again be split up in two categories, namely sensors that use some sort of steel ball to concentrate the force to a silicon sensing element (see Figure 5.2) and sensors that consist of piezoresistive foil layers (see Figure 5.3). Load cells are a mechanical arrangement where the applied forces and torques are measured by semiconductor strain gauges (see Section 2.3.2). Usually 6-axis transducers can be found on the market. Mostly these load cells are designed for large industrial applications and can measure large forces up to hundreds of kilo Newton. Since they are rather large, heavy and expensive they are not considered in this report which focuses on applications on a small helicopter. The smallest available 6-axis transducer is the ATI Nano17 shown in Figure 5.1. Its properties and performances can be found in Table 5.3. With its size of 17mm in diameter and 15mm height and a weight of 9g it would also be suitable for applications on a unmanned helicopter. Nonetheless such a transducer will not be rst choice since it is very expensive and force and torque measurements in all axis together is hardly required for docking maneuvers with a helicopter.

Figure 5.1: ATI Nano17 6-axis force and torque transducer 27

Chapter 5. Force Sensors on the Market

28

Figure 5.2: A Honeywell FSS1500NSB force sensor

Figure 5.3: FlexiForce foil sensor

5.2

Description of the Parameters

In this section the parameters which are used to evaluate the sensors found on the market are explained in case their meaning or denition is not clear. In Tables 5.1 and 5.2 the unit [%Span] can be found for dierent parameters. This denotes the relative error compared to the whole span of the sensor. 1. Range: The range of forces which the sensor is dimensioned for. In Tables 5.1 and 5.2 the forces are given in [N ] but one can often nd the unit [kg] or [kgf ] which means the force corresponding to the declared mass in a 9.81m/s2 gravitational eld. For array sensors the range is given in [P a]. 2. Dynamic Range: The ratio of largest to smallest detectable force.

29

5.3. Sensors

3. Over Force: The maximum force which may safely be applied to the sensor for it to remain in specication once force is returned to the operating force range. 4. Sensitivity: The ratio of output signal change to the corresponding input force change. Sensitivity is determined by computing the ratio of Span to the specied operating force range. 5. Span: The algebraic dierence between output signal measured at the upper and lower limits of the operating force range. Also known as full scale output or simply span. For some sensors only the output resistance dierence is declared. 6. Linearity Error: The maximum deviation of the true response to the best t straight line (BFSL). 7. Repeatability: The maximum dierence between output readings when the same force is applied consecutively, under the same operating conditions, with force approaching from the same direction within the operating force range. 8. Temperature: The temperature range over which the product will produce an output proportional to force but may not remain within the specied performance limits. This is the operating temperature range, the storage temperature range may be dierent. 9. Mechanical Hysteresis: The maximum dierence between output readings when the same force is applied consecutively, under the same operating conditions, with force approaching from opposite directions within the operating force range. 10. Elements: The number of simple sensing elements contained in an array sensor. 11. Resolution: The distance between to simple sensing elements in an array sensor.

5.3

Sensors

This Section presents the sensors that were found on the market at dierent distributors. In Table 5.1 and Table 5.2 standard single element force sensors are presented. Table 5.3 presents some special sensors such as sensor arrays and a 6axis transducer. The introduced properties of each sensor are gathered from data sheets that were found in the internet. Since the quality and contained information of the data sheets varies very much not every property could be evaluated for each of the presented sensors.

Chapter 5. Force Sensors on the Market

30

Table 5.1: Dierent force sensing elements with their specications

31

5.3. Sensors

Table 5.2: Dierent force sensing elements with their specications (continued)

Chapter 5. Force Sensors on the Market

32

Table 5.3: Some examples of special force sensors that are available on the market

33

5.3. Sensors

We can now generally evaluate the three dierent sensor categories on the market in terms of the desired properties described in Section 1.1. 1. Load Cells: As mentioned before, load cells are constructed for large forces and therefore are also very robust. But they have some crucial disadvantages for applications on a small helicopter. They are very expensive, rather large and heavy. 2. Silicon-sensing Elements: These elements are small, light and not too expensive. But they have one major disadvantage, which is their robustness. Their sensing range goes only up to around 15N and their over force is only around 45N which can already be reached when the helicopter bounces against a wall. 3. Foil Sensors: These sensors are very light, small (especially thin) and very cheap. Sensors for forces up to 445N are available which is entirely sucient for applications on a small helicopter. So generally these sensors t the desired properties very well.

Chapter 5. Force Sensors on the Market

34

Chapter 6

Applications of Force Sensors


During the 1980s, most authors predicted that the major application area for tactile sensing would be industrial automation tasks such as robotic assembly. However, the demand for such sensors proved to be low. There are still very few fully developed applications of tactile sensing but this chapter considers current trends. We see three main elds where tactile sensing is likely to play a key role. These are: medical procedures, especially surgery; rehabilitation and service robotics; and agriculture and food processing. In addition, some examples of applications using commercial force sensors are given.

6.1

Surgical Applications

Surgery is perhaps the most exciting and rapidly developing area where tactile sensing is actually of central importance. Minimally Invasive Surgery (MIS) is still a young method but is now routinely used as the preferred choice for many operations. However, despite its advantages, MIS severely reduces the surgeons sensory perception during manipulation. Surgery is essentially a visual and tactile experience and any limitations on the surgeons sensory abilities are most undesirable. For example, in laparoscopy long slender tools are inserted through small puncture openings in the abdominal wall and the surgeon uses a range of tip mounted instruments guided by video feedback images. As the instruments are rigid rods and eectively have xed pivots at the entry points, the available degrees of freedom are restricted. This is one of the main diculties together with lack of depth from 2D vision and the almost complete lack of sense of touch. The reason that tactile sensing is so important in surgery is that soft tissue can only be properly examined and identied by assessing its softness, viscosity and elasticity properties. So it is clear that tactile sensing is greatly needed in this area.

6.2

Rehabilitation and service Robotics

A major concern for the next century is the enormous numerical increase in the elderly population that will generate great economic pressures. This demographic change is well accepted and many governments have initiated programmes of research in health care, hospital services and social support. It is clear that there will be greatly increased demand on these services and researchers are looking for methods of support and assistance that do not involve central services but can be distributed as aids within the home and community. Despite the contrast with industrial robotics, there are some emerging interests in human-robot cooperation within manufacturing. As the next stage in automation, 35

Chapter 6. Applications of Force Sensors

36

Toyota envisage workers and robot machines coexisting in a safe partner relationship. As yet there is little in this area that is new specically concerning tactile sensing. However, as for medicine, we can see many opportunities where sense of touch will be a real need.

6.3

Agriculture and Food Processing

The eld of agriculture and food production is now well automated but, like service robotics, also does not feature many new technological advantages in tactile sensing. Uses of tactile sensing are often mundane, but the importance of this area is its potential for imminent development. Unlike manufacturing automation, the processing of natural produce usually involves high numbers of human operators. This is because of the problems of handling soft, delicate and highly variable items by machine have not been solved at low enough cost. Recently, there has been increased interest in the prospect of reducing human involvement in order to reduce hygiene risks, eliminate human errors and use ecient but more hazardous environments.

6.4

Specic Applications of force sensors

Most of the research of tactile sensing in the eld of robotics is made for producing robotic hands. It deals with the problem of gripping objects and therefore measuring contact forces and contact locations. Usually an array of force sensing elements is placed into the skin of the robotic hand to be able to measure contact over the whole surface of the hand. The goal of most of the technologies presented in Chapter 2 is to build a robotic hand covered by an articial skin. Further applications are in human and robot interaction such as haptic devices used for rehabilitation. Another example are touch screens which are very popular nowadays and are mostly based on a resistive detection of the contact location. Some also use capacitive techniques to detect contact on the surface. The Tachi Lab [37] developed some very interesting applications of force sensing. Their eld of research is virtual reality. They have several projects dealing with haptic interaction between humans and a virtual reality such as the GravityGrabber. They also did research on force sensing technologies and developed for example GelForce, a force sensor which uses tracking of optical markers (see Section 2.7.5) to measure the distribution of both the magnitude and direction of force. The most frequently applied force sensors are strain gauges. They are cheap and can be easily used in a custom application. A lot of larger commercial force sensors are using strain gauges to measure the applied forces. Load cells (see Section 4.1) for instance use several strain gauges to measure forces and torques in multiple directions. Another example of a sensor using strain gauges is explained in Section 4.3. Foil sensors such as produced by FlexiForce are also commonly used in commercial applications. One example are PlayStation controllers which have buttons that do not only have an on-o function but also a scaled input according to the applied force on the button.

Chapter 7

Conclusions
Over the past four decades, tactile sensing has developed into a sophisticated technology. There has been a long-standing and widely held expectation that tactile sensors would have a major impact on industrial robotics and automation. However, this promise has not been realized, and few, if any, tactile sensors can be found in factory based applications [4]. The main elds where tactile sensing is likely to play a key role are medical procedures, especially surgery; rehabilitation and service robotics; and agriculture and food processing. After all the production of new designs and congurations of sensors continues apace. However, most of the possible forms of physical transduction methods have now been explored and there seems little scope for new fundamental transducers. The basic transduction methods are described by Russell [1] and Nicholls [2] and were already developed in 1990. The current research oerings are mainly concerned with novel packagings, better designs, improved engineering and more complete analysis. The technologies that satisfy the requirements of a small, light, robust, simple and cheap force sensing element best, are mechanical displacement sensors, strain gauges and more advanced, a carbon felt and carbon bers sensor. A carbon felt sensor may be considered for sensing in inhospitable environments and if large forces or shocks on the sensor are possible. The problem is that no commercial sensor using this technology is available and the development would be very time-consuming and expensive. Therefore it makes more sense to use strain gauges which are the most commercialised sensors that can measure forces. They are cheap compared to other force sensors and can be attached to a simple customized deformable element, such as a small metallic arc, at the end of the helicopter. Finally the simplest considerable solution is a self-built mechanical displacement sensor maybe combined with some kind of damping element to prevent the helicopter to bounce back from a touched object. Such a sensor could be developed very cheaply and can be adapted quite easily to a given task. The commercial sensors presented in Chapter 5 are nearly all based on piezoresistive technologies. Sensors like those produced by Honeywell will most likely not be suitable for the purpose of docking maneuvers with a helicopter because the maximum force of 40N which these sensors withstand and the actual range of 015N is barely enough. Foil sensors such as FlexiForce have a much larger range of 0440N and are therefore more appropriate. A disadvantage for both of these sensor types can be that the movable part of the sensors can only perform small deections. For instance if the helicopter is ying towards a wall, no damping of the impact can be realized just by the sensor, so one have to consider to build an assembly of such a sensor with some damping element. 37

Chapter 7. Conclusions

38

After all, the two force sensing methods that t best for a helicopter are on the one hand mechanical displacement and on the other hand a customized assembly with a foil sensor. Both t the desired properties introduced in Section 1.1 very well and are easy to build by yourself. Therefore, these two should be considered rst and tested if their measuring performance is suitable.

Bibliography
[1] R.-A. Russel: Robot Tactile Sensing, Prentice Hall, New York; Sydney, 1990 [2] H.-R. Nicholls: Advanced Tactile Sensing for Robotics, World Scientic Publishing Co., Inc., 1992 [3] H.-R. Nicholls, M.-H. Lee: A Survey of Robot Tactile Sensing Technology, The International Journal of Robotics Research, Vol. 8, No. 3, 1989, pp. 3-30 [4] M.-H. Lee: Tactile Sensing: New Directions, New Challenges, The International Journal of Robotics Research, Vol. 19, No. 7, 2000, pp. 636-643 [5] R.-S. Fearing: Tactile Sensing Mechanisms, The International Journal of Robotics Research, Vol. 9, No. 3, 1990, pp. 3-23 [6] M.-H. Lee, H.-R. Nicholls: Tactile sensing for mechatronicsA state of the art survey, Mechatronics, Vol. 9, No. 1, 1999, pp. 1-31 [7] M.-R. Cutkosky, R.-D. Howe, W.-R. Provancher: Robotics, Springer, 1990, Chap. 19 Handbook of

[8] J. Rausch: Engineering Haptic Devices, Springer, 2009, Chap. 10 [9] W. Erler: Elektrisches Messen nichtelektrischer Grssen mit Halbleitero widerstnden, Verlag Technik, Berlin, 1973 a [10] W.-R. Provancher, M.-R. Cutkosky: Sensing Local Geometry for Dexterous Manipulation, ISER, 2002, pp. 507-516 [11] R.-L. Garrison, S.-S.-m. Wang: Pneumatic touch Sensor, IBM Technical Disclosure Bulletin, Vol. 16, No. 6, 1973, pp. 2037-2040 [12] M.-H. Raibert: An All Digital VLSI Tactile Array Sensor, Proceedings of the IEEE International Conference on Robotics, 1984, pp. 314-319 [13] D.-M. Siegel, et al.: A Capacitive Based Tactile Sensor, SPIE Conference on Intelligent Robots and Computer Vision, Cambridge, Massachusetts, 1985, pp. 153-161 [14] M. Kaneko, N. Ueno, T. Tsuji: Active antenna-basic considerations on the working principle, Proceedings of the IEEE International Conference on Intelligent Robotic Systems, Vol. 3, 1985, pp. 1744-1750 [15] T.-N. Clements, C.-D. Rahn: Three-dimensional contact imaging with an actuated whisker, IEEE International Conference on Intelligent Robotic Systems, 2005, pp. 598-603 [16] S. Presern, F. Dacar, M. Spegel: Design of three active degrees of freedom tactile sensor for industrial arc welding robots, Proceedings of the 4th British Robot Association conference, Brighton, UK, 1981, pp. 207-213 39

Bibliography

40

[17] W.-D. Hillis: Active Touch sensing, MIT AI Memo 629, April 1981 [18] M.-H.-E. Larcombe: Carbon Fibre Tactile Sensors, Proceedings of the First International Conference on Robot Vision and Sensory Controls, Bedford, UK, 1981, pp. 273-276 [19] P. Dario, G. Buttazzo: An Anthropomorphic Robot Finger for Investigating Articial Tactile Perception, International Journal of Robotics Research, Vol. 6, No. 3, 1987, pp. 25-48 [20] P. Dario, D. De Rossi, C. Domenici, R. Francesconi: Ferroelectric polymer tactile sensors with anthropomorphic features, Proceedings of the IEEE International Conference on Robotics, Atlanta, GA, 1984, pp. 332-340 [21] P. Dario, et al.: Advanced Rehabilitative Robots, Proceedings of the International Symposium and Exposition on Robots, Sydney, Australia, 1988, pp. 687-703 [22] K. Tanie, et al.: A High-Resolution Tactile Sensor, Robot Sensors, Vol. 2, 1986, pp. 189-198 [23] J. Rebman, K.-A. Morris: A tactile Sensor with Electro-optical Transduction, Robot Sensors, Vol. 2, 1986, pp. 145-155 [24] J.-S. Schoenwald, et al.: A Novel Fiber Optic Tactile Array Sensor, Proceedings of the IEEE International Conference on Robotics and Automation, Raleigh, North Carolina, 1987, pp. 1792-1797 [25] J. Winger, K.-M. Lee: Experimental Investigation of a Tactile Sensor Based on Bending Losses in Fiber Optics, Proceedings of the IEEE International Conference on Robotics and Automation, Raleigh, North Carolina, 1988, pp. 754759 [26] J.-L. Schneiter, T.-B. Sheridan: An Optical Tactile Sensor for Manipulators, Robotics and Computer-integrated Manufacturing, Vol. 1, No. 1, 1984, pp. 65-71 [27] A. Cameron: Touch and Motion, Proceedings of the IEEE International Conference on Robotics and Automation, Philadelphia, 1988, pp. 1062-1067 [28] M.-H.-E. Larcombe: Carbon Fibre Tactile Sensors, Proceedings of the First International Conference on Robot Vision and Sensory Controls, Bedford, UK, 1981, pp. 273-276 [29] N.-J. Ferrier, R.-W. Brockett: Reconstructing the shape of a deformable membrane from image data, The International Journal of Robotics Research, Vol. 19, No. 9, 2000, pp. 795-816 [30] S. Hackwood: A Torque-Sensitive Tactile Array for Robotics, The International Journal of Robotics Research, Vol. 2, No. 2, 1983, pp. 46-50 [31] R.-C. Luo, et al.: An Imaging Tactile Sensor with Magnetoresistive Transduction, Proceedings of the First International Conference on Intelligent Sensors and Computer Vision, Cambridge, Massachusetts, 1983 [32] A.-R. Grahn, L. Astle: Robotic Ultrasonic Force Sensor Arrays, Proceedings of the Robots 8 Conference, Detroit, Michigan, 1984, pp. 21.1-21.18

41

Bibliography

[33] D. De Rossi, A. Nannini, C. Domenici: Articial Sensing Skin Mimicking Mechanoelectrical Conversion Properties of Human Dermis, IEEE Transactions on Biomedical Engineering, Vol. 35, No. 2, 1988, pp. 83-92 [34] R.-C. Luo: Design and implementation of hand-based tactile sensors for industrial robots., Proceedings of the International Conference on Robotics and Factories of the Future, 1984, pp. 423-432 [35] Wikimedia: Hall Eect, http://commons.wikimedia.org/wiki/File:Hall_ effect_A.png [36] Wikimedia: Strain Gauge Pattern, http://upload.wikimedia.org/ wikipedia/commons/thumb/5/53/Strain_gauge.svg/424px-Strain_ gauge.svg.png [37] Tachi Lab: http://tachilab.org/

Vous aimerez peut-être aussi