Tactile feedback devices
The tactile feedback devices are a type of haptic devices, as haptic includes tactile and force interaction. A bidirectional interaction is perceived between the finger and the haptic device. This means that it is not just an interaction from the user’s finger towards the surface or the screen that can be made, but it is also an interaction from the surface towards the finger. The main objective of the tactile feedback devices is the reproduction of the sense of touch of real surfaces using a programmable device. The ways to touch are related to what we want to feel. There are two types of finger stimulation: local and global stimulations. Each of these stimuli corresponds to different types of tactile feedback devices.
Local stimulation can reproduce a local physical sensation under the finger of the user, differentiated at the level of the fingerprint grooves. To achieve this aim, this family of tactile stimulators is operating as follows: the finger is placed statically on the interface in motion. Typically, the stimulation is based on the programming of a matrix of pins; the normal movement on the surface of these pins is programmed to stimulate the finger when it touches the active surface of the device. In some cases,  the pins are working in a quasi-static mode, rather to simulate shape (Velazquez et al., 2005). In most cases, they vibrate with a frequency less than the mechanoreceptors threshold (400 Hz) typically between 10 and 200 Hz and with a displacement of the order of a few micrometres. The main application for this type of tactile display is the development of the Braille display (Echenique et al., 2010) to help blind people to read texts. Generally, blind readers use the sound output to access to the texts (Coutinho et al., 2012; Supriya and Senthilkumar, 2009). But, the sound output is limited in term of scientific equations, graphic or music. Additionally, (Varao Sousa et al., 2013) demonstrated that the fact of listening to text is fundamentally different from actively reading for visual readers. They report that more active modes of reading (such as reading silently or reading aloud) lead to less mind-wandering and greater comprehension than when participants passively listened to the text being read to them. Recently, researchers have investigated on the development of a low cost refreshable Braille display able to help blind reader to read texts and other materials (Russomanno et al., 2015).
Various techniques exist for powering pins. The electromagnetic technologies are the more common (Benali-Khoudja et al., 2004; Lee et al., 1999; Streque et al., 2010).Using a pneumatic pump for expelling air through microvalves (Asamura et al., 1998) proposed a stimulator with 2 mm of diameter and 0.5 mm of thickness. Piezoelectricity is also a good candidate to actuate tactile displays. One of the most famous piezoelectric device is the OPTACON (Efron, 1977) dedicated to blind people who can perceive on their finger the Braille translation of a camera read text (Nye, 1976). Another type of local stimulation device using piezoelectric technology has been proposed in 2000 by (Hayward and Cruz-Hernandez, 2000). It differs from the preceding as it stretches the skin. Indeed, the pins are actuated in a lateral movement; this device is called the STReSS. Figure 1-2 illustrates the final assembly of 6 * 10 piezoelectric bimorph with a spatial resolution of 1.8 * 1.2 mm
Figure 1-2: STReSS lateral skin tactile display (Hayward and Cruz-Hernandez, 2000).
In general for local stimulation, it must be noted that this technique is effective and does give a good sensation only if the pins’ matrix density is high. The main difficulty is that each pin must be controlled independently, which makes this technique complex in terms of manufacture and expensive in terms of cost.
In contrast to the local stimulation, the global stimulation consists of having the same stimulation on the whole finger. To perceive changes in stimuli, the finger must be in motion on the tactile device, and the physical parameter which may be controlled is the friction between the moving finger and the surface. By modulating the friction according to the finger position, it is possible to produce the illusion of touching various surfaces such as smooth or rough surfaces (Watanabe and Fukui, 1995). The finger position is a highly important parameter because the simulation is based on it. This type of devices is dedicated mainly to improve the tactile touch screens like smartphones, tablets or touch screen pc by adding a tactile feedback. This feedback from the tactile device towards the user’s finger will give more realistic touch conditions. For example, it will be felt a textured image on an LCD screen when entering in contact with the image Figure 1-3.
Stimulators based on the friction modulation
These devices belong to the global stimulation device family. We can distinguish two techniques to control the friction which are Ultrasonic Vibrations and Electrovibration. The Electrovibration generates an attractive force between the finger and a polarized surface. A high voltage must supply the plate which is covered with an insulator to polarize the finger. This electrostatic force attracts the finger and increases the coefficient of friction, which gives a modified sensation from the non-supplied surface (Kaczmarek et al., 2006).
Different devices have been designed based on the electrovibration effect, we can cite for example TeslaTouch illustrated in Figure 1-5 (Bau et al., 2010), 3D rendering texture by Disney research (Kim et al., 2013) or the Senseg device (Wijekoon et al., 2012). It must be noted that no physical movement of the surface is required to create the tactile feedback using this type of devices because the effect is based on electrostatic attraction. On the other hand, the other family of the friction modulation devices, on which this thesis is focused, relies on ultrasonic vibrations, which constitute an alternative technique to modulate the friction. A first physical explanation for the friction reduction using ultrasonic vibration is given by the squeeze film theory (Biet et al., 2007; Watanabe and Fukui, 1995; Winter et al., 2013): ultrasonic vibrations generate an air film between the fingertip and the stimulator which acts as a lubricant between the surface and the finger, and reduces the friction. This air film created by the plate is modulated by the amplitude of the ultrasonic vibrations of the surface. More recently, some studies proposed other causes for friction reduction, based on an intermittent contact mechanical approach (Dai et al., 2012; E. Vezzoli et al., 2015b). For ultrasonic excitation, piezoelectric ceramics are utilized to provide a vibration at the whole system mechanical resonance frequency; this type of actuators is chosen for its high responsivity and its low power consumption. The first device based on this principle was developed by (Watanabe and Fukui, 1995) twenty years ago. Some conditions have to be respected to produce this effect. The first one is that the vibration frequency should be more than 25 kHz and the second condition relies on the vibration amplitude that should be superior to 1 μm peak to peak to generate friction reduction (Biet et al., 2007). A standing wave is created when the plate is excited at a specific mechanical resonance frequency of the vibrating plate. This frequency is defined by its material, its geometry and the number and the position of the piezoelectric ceramics actuators. The resonance frequency can be pre-determined using a FEM analysis. Then a vibrometer is operated to check the simulated results.
Different tactile feedback devices have been proposed and evaluated, we can cite for example (Wiertlewski et al., 2014; Winter et al., 2013). The TPad project from Northwestern university was one of the first proposal, (Winfield et al., 2007) and was based on a small circular piezoelectric bending element. From this first prototype, E. Colgate’s team went on working on the topic and in 2013, the principle has been implemented in a Kindle FireTM (Figure 1-6) tablet computer having a large screen area (88 mm x 152 mm) to increase the range of potential applications. The Stimtac project developed in the L2EP laboratory started in 2004 has proposed multiple tactile feedback devices based on the ultrasonic friction modulation. This project recorded several notable changes from the first prototype, fully glued by piezoelectric actuators (Biet et al., 2007), to an optimization of the actuator number to reduce the power consumption in (Giraud et al., 2010). The finger position sensing has been evolved from the use of a 1D LVDT sensor to a 2D position sensing using four force sensors which can also measure the normal force applied by the user. A transparent tactile stimulator has been also developed using a glass material plate in (Giraud et al., 2012). An optimized tactile device connected to the PC by an USB cable has been developed in (Amberg et al., 2011), a picture of the Stimtac device is illustrated in Figure 1-7.
It may be noted that all these global stimulation devices allow only the tactile feedback on a single finger, as the active surface is in the same state at a given time. To allow multi touch tactile feedback with global stimulation devices, another approach has been studied recently, which aims at creating different touch sensations in the same surface. This method, which is based on ultrasonic vibration using a multimodal approach, consists in exciting several vibration modes at the same time corresponding to different mechanical resonant frequencies (Ghenna et al., 2015). The superimposition of several modes allows to produce different tactile stimulations on several fingers located in different positions. Another approach based on the multi-touch stimulation using electrovibration is reported in (Nakamura and Yamamoto, 2014).
As the two effects, Electrovibration and Ultrasonic Vibrations, modulate the friction in an opposite way (one increases and the other effect decreases it), it seems interesting to couple the two effects together in order to increase the range of the modulated friction. The idea to couple the two effects has been applied and implemented for the first time in the same device by (Frederic Giraud et al., 2013). The two effects, using different principle are compatible. A tribological measurement confirmed that by giving step references of ultrasonic vibration and electrovibration (Figure 1-8), it was possible to increase and decrease the friction. It is also technically possible to produce almost the same modulation level using the two effects.
Dependence of the friction on the sliding parameters
It was observed using the accelerometer that the sliding conditions influence the extracted signal. By measuring the friction when sliding a tool on a surface, it was also observed in several studies that the sliding conditions such as the normal applied force ( ) and the velocity ( ) affect the coefficient of friction noted µ. The variability of µ as a function of the sliding parameters depends highly also on the shape and the material of the slider. Several researches analyse the effect of µ by using different probes or real finger and for different ranges of and . The following Table 1-1 sums up the evolution of the coefficient of friction (COF) depending on normal force and the slider materials.
Previous studies on ultrasonic tactile devices in L2EP
The former tactile stimulation devices developed in the L2EP laboratory are mainly based on the ultrasonic vibration. This vibration with an amplitude of a few micrometres creates the feeling of a smoother surface. By just adjusting its intensity depending on the finger position, it is then possible to recreate the sensation of touching a textured surface. This slippery feeling of the finger when sliding on the surface was first explained by the creation of the squeeze film effect theory which proved that an air film is created between the finger and the surface (Watanabe and Fukui, 1995; Wiesendanger, 2001). However, some recent studies have been focused on the physical explanation of this friction weakening effect (Eric Vezzoli et al., 2015b; Xiaowei Dai et al., 2012). Even if squeeze effect may occur between the vibrating surface and the fingertip, some other phenomena such as intermittent contact together with stochastic adhesion seem to be influent (Ben Messaoud et al., 2015b; Eric Vezzoli et al., 2015b).
Several prototypes of ultrasonic friction modulation devices are developed worldwide (Winfield et al., 2007; Winter et al., 2013), and in L2EP, since 2007, under the name “Stimtac”. The first Stimtac was developed by (Biet, 2007) and made up of copper. This device had the dimension 83 mm × 49 mm and was fully glued in its reverse side by piezoelectric ceramics forming an array of actuators to produce a stationary wave at its resonance frequency. It was designed to stimulate gratings regardless of the power consumption. This first prototype used 1D LVDT sensor to track the finger position, so the simulation was just in one direction. This device was liable to thermal heating due to unoptimized design of the copper plate. An analytical modelling of the air pressure between the finger and a vibrating surface characterized by its roughness, wavelength, vibration amplitude, and resonance frequency was developed. In 2008, to determine the finger position, a custom-made 2D sensor built from two white LEDs with a set of mirrors replaced the LVDT sensor. A Digital Signal Processor was used to estimate the finger position from its shadow according to the (x, y) axis with a frequency of 120 Hz. This principle increased the finger position precision and allowed the 2D stimulation. However, heating problems remained.
The following study (Giraud et al., 2010) which focused on the design optimization of the plate allowed a reduction in the power consumption by 90% without influencing the tactile sensation. This outstanding result was obtained thanks to the accurate choice of the piezoceramics and plate thickness, and by the optimization of the number and location of piezoceramics. Another solution was proposed and implemented to reduce the bulkiness of the Stimtac: the 2D optical sensor was substituted by force sensors. On the other hand, the serial port was replaced by a USB connection to provide read/write communication signals and power supply. Demo applications were implemented so that they detect the finger displacement and adapt the vibration amplitude based on the color of the pointed pixel in order to give the illusion of touching a displayed texture (Amberg et al., 2011). In 2012, a glass material has been utilized as a vibrating plate to make the surface transparent (Giraud et al., 2012). The transparency of the plate allows the device to simulate a more realistic tactile feedback by displaying the simulated surface on a LCD screen under the transparent surface. Force sensors have been employed to estimate the finger force in one direction and the finger position as well.
Table of contents :
LIST OF TABLES
1 STATE OF THE ART
1.1 THE SENSE OF TOUCH
1.2 TACTILE FEEDBACK DEVICES
1.2.1 Local stimulation
1.2.2 Global stimulation
1.3 STIMULATORS BASED ON THE FRICTION MODULATION
1.3.1 Tactile stimulation principle
1.4 SIMULATION OF REAL SURFACES: CASE OF TEXTILE FABRICS
1.4.1 Applications of the texture simulation
1.4.2 Texture feature extraction
1.4.3 Dependence of the friction on the sliding parameters
2 DESIGN AND CONTROL OF THE TACTILE STIMULATION DEVICE
2.1 PREVIOUS STUDIES ON ULTRASONIC TACTILE DEVICES IN L2EP
2.2.DESIGN OF THE TACTILE STIMULATOR
2.2.1 Mechanical structure of the tactile stimulator
2.2.2 Finger position measurement
2.2.3 Control of the tactile plate
2.3 ROBUSTNESS ANALYSIS
2.4 SYSTEM MODELLING
2.4.1 General equations in d-q frame
2.4.2 Equations at resonance
2.4.3. Transfer functions for the control
126.96.36.199 Modelling of the vibration amplitude in the vicinity of the resonance
188.8.131.52 System behaviour to frequency changes
2.5 IDENTIFICATION OF THE MODEL PARAMETERS
2.5.1 Experimental setup description
2.5.2 Identification approach
184.108.40.206 Identification of the transfer function between Wd and V
220.127.116.11 Identification of the transfer function between Wq/Wd and Δω
2.6 CONTROL OF THE VIBRATION AMPLITUDE
2.6.1 Tuning of the controller coefficients
2.6.2 Discretization of the controllers
2.6.3 Results for the two control loops
2.6.4 Psychophysical validation
3 FRICTION AND PERCEPTION
2.1 RELATION BETWEEN FRICTION AND PERCEPTION
3.1.1 State of the art
3.1.2 Which friction parameters influence the perception?
3.2 TACTILE PERCEPTION’S THRESHOLD
3.2.1 Experimental protocol
3.3 FRICTION DISCRIMINATION CRITERION
3.3.1 Experimental protocol and tribological criteria
3.3.3 Criterion choice
3.3.4 Influence of the maximal friction on the friction criterion and on the perception
3.3.5 Influence of the sliding velocity
4 DESIGN AND EVALUATION OF A SMART TACTILE STIMULATOR: SMARTTAC
4.1 WHY A SMART DEVICE?
4.2 DESIGN OF THE SMARTTAC
4.2.1 Device particularity
4.2.2 Vibration amplitude control
4.3 FRICTION CONTROL
4.3.2 Control method
4.3.3 Results of the coefficient of friction control
4.3.4 Friction contrast control
5 REAL SURFACE SIMULATION
5.1 TEXTURE RENDERING METHOD
5.1.1 Friction contrast definitions
5.1.2 Fabrics-Finger friction contrast
5.1.3 SmartTac-Finger friction contrast
5.1.4 Texture rendering strategy
5.1.5 Sum up of the control strategies
5.2 FABRICS MEASUREMENT
5.2.1 Characteristics of the fabrics
5.2.2 Friction measurement
18.104.22.168 Fabrics characterization with a real and an artificial finger
22.214.171.124 Fabrics characterization with a special probe
126.96.36.199 Complete tribological features extraction
5.3 PSYCHOPHYSICAL VALIDATION
5.3.1 Thin texture determination
5.3.2 Validation of the simulated fabrics
188.8.131.52 Experimental protocol
184.108.40.206 Results and discussion
CONCLUSION AND PERSPECTIVES
FUTURE WORK AND PERSPECTIVES