National Science Foundation
Grant No. 0413204
Prior research in our lab has found that shear forces are important in haptic exploration. However, in a haptic tactile task, shear forces on the fingertip are often not considered. This work seeks to address this deficiency.|
The Variable Friction Haptic Display (VFHD) displays friction on a planar display. The slipperiness of the surface can be varied under computer control. Changing the friction on the surface indirectly generates shear forces on the fingertip, which in turn creates interesting haptic effects. This work additionally will determine what the effect of varied friction has on a haptic task.
Conceptual graphic of haptic effect
It should be noted that the finger must be actively exploring the surface to receive any stimulation. If the finger is stationary, no shear forces are transmitted to the fingerpad. This means that the device does not work in passive exploration tasks.
The VFHD uses squeeze film bearings to create a reduction in friction. These squeeze bearings act as a cushion of air that lifts and separates the finger from the planar display surface. By varying the thickness of the squeeze bearing, the friction on the surface can be indirectly reduced.
Squeeze film generated by vibrating lower plate
In general squeeze bearings are created when a surface is vibrated in the 20 kHz to 90 kHz range with sufficient amplitude. In the VFHD, the oscillation has approximate amplitude of 10 µm at 50 kHz (ultrasonic).
A good analogy for this effect is an air-hockey table. In an air-hockey table, high pressure is generated by an air compressor and forced through tiny holes in the surface to cause the puck to float and move with reduced friction. In the VFHD, high pressure is produced by the vibration of the surface, and causes your finger to float like the puck in the air-hockey table.
The novel contribution of this work is to incorporate position feedback with varying friction. The display of the VFHD, although composed of tiles, behaves as a monolithic surface; if one of the tiles has a given friction level, all of the tiles have that level of friction. Position feedback is done with a passive, parallel linkage system and encoders. The output friction on the VFHD display is based on the measured finger position and velocity. We assume that only one finger is in contact with the display, and that the entire fingerpad will be subject to the same friction.
The current prototype of the VFHD is composed of 6 tiled actuators, each 0.75 inches X 0.75 inches in a 2 X 3 configuration, yielding a planar display surface of 1.5 inches X 2.25 inches. The parallel linkage system is a modified, passive linkage system based on the McGill pantograph, using 14400 cpr encoders. It is capable of providing tactile sensations to one finger under several different simulated textures.
Several control regimes have been explored to determine the effects of friction on a tactile task. Below is a sampling of representative friction on the surface. Wave your mouse over the blue list item titles below to see a pop-up graph showing the friction on each of the surface types.
Preliminary Human Testing with a Variable Friction Display
1x4 VFHD Prototype
As previously mentioned, forcing the plates to move vertically at this high ultrasonic frequency creates a squeeze film effect on the surface of the plate. When a human subject places his finger on these moving plates its feels as though his finger were gliding on a cushion of air. Therefore by actuating this array of plates at ultrasonic frequencies, the friction force felt by the user is effectively reduced. It is suspected that this squeeze film effect is not a pure on and off system, and that the friction reduction is related to the amplitude of vibration. By recording the subject’s finger position as they interact with the display it is possible to vary the coefficient of friction across the plate. This spatial variation in the coefficient of friction across the plate is perceived by the user as a surface texture. Temporal variation of the coefficient of friction was observed to feel more like vibration rather than a surface texture.
Several spatial texture/coefficient of friction patterns were explored, and a 19 texture library was developed. Human subjects were tested and perceptual data in the form of descriptive words and texture grouping was collected. From this data a similarity matrix of the textures in the library was created and an MDS (Multi-Dimensional Scaling) analysis was performed on the data. The MDS plot shows a few trends, the most salient of which is the trend of low to high spatial frequencies as you move across the plot from left to right. It can be inferred from this data that spatial frequency of the friction variation is very noticeable to human subjects. It can also be noted that asymmetric spatial patterns (i.e. fish scales, the direction of finger velocity dictates the friction pattern felt by the subject) and localized friction patterns are both distinguishable from continuous and discontinuous periodic spatial friction patterns.
MDS map of Perceptual Data
This prototype had several limitations which are planned to be corrected with future research. The range of friction coefficients the display was capable of portraying was not very broad. It is suspected that by replacing the current aluminum display surface plates to a material with much higher coefficient of friction against human skin, such as glass, this limitation can be resolved. The driving circuitry for the display was limited to operate at one frequency per plate during subject exploration. This limitation does not allow for resonant frequency shifts as the dynamics change while subjects are interacting with the device. By developing driving circuitry with resonant frequency tracking this problem can hopefully be eliminated. Lastly this prototype is roughly 1 inch by 3 inches by 4 inches in size (not including driving circuitry). It would be ideal for this technology to be packaged with existing force feedback devices. With this goal it is necessary to develop a display on a much smaller scale. Currently we are in the developement of a display which is actuated by piezoelectric bimorphs opposed to the crystal stacks. It is believed that by moving to the bimorph our display (not including driving circuitry) can be packaged into a 1 inch by 1 inch by 0.08 inch size. With such improvements it is predicted their will an increased library of textures. Further subject testing will be performed with this new display to explore additional texture distinctions and device capabilities.