Hardware-Based Touch Sensing

Hardware-based touch sensing generally involves the construction of a grid or other configuration of sensors which experience a change in resistivity or capacitance when touched by a user’s fin- gers or hands [68, 76, 90, 112, 149, 150]. Compared to camera-based approaches, these techniques require significantly less data to be processed, and the desired properties of touch events are more directly measured rather than inferred from images [68]. However, this often comes at the cost of decreased touch-sensing resolution [61], the inability to project imagery, and increased difficulties when considering surfaces with non-planar or non-parametric geometry.

2.1.1.1 Resistive Sensing

Force-sensitive resistance approaches feature arrays of sensors that experience an increase in con- ductance in response to increasing application of force [48, 68, 77, 98, 124, 147]. Because of this

relationship, they are capable of detecting variations in pressure. Resistive touch interfaces are generally cheaper than other hardware-based methods [45]. Unlike capacitive sensing approaches, they are capable of detecting input from devices like styli—though this limitation may be overcome in capacitive systems through the construction of special input devices (e.g. [116]).

The UnMousePad achieves multi-touch sensing through the use of resistive elements [124]. The authors call their technique Interpolating Force Sensitive Resistance (IFSR), which they claim supports higher resolution than traditional resistive touch sensing approaches that are often limited in their ability to detect touches in locations between the hardware touch sensors. It is capable of detecting a large spectrum of pressure variation. The device is flexible and has been applied to cylindrical surfaces. In addition to detecting finger input, the authors also focused on high- resolution stylus sensing, noting that many other approaches are completely incapable of detecting a stylus unless contact occurs directly on one of the sensing elements.

TactileTape is a pliable resistive touch-sensitive material that can be applied to non-planar sur- faces [73]. Primarily designed to allow for rapid prototyping of simple touch-sensitive objects, it is composed of a resistive and a conductive layer that forms a closed circuit when combined with user touch input. The proposed prototype is only capable of sensing the occurrence—but not location—of touch at some point on the strip. However, as it is made of common materials—a pencil, tin foil, and shelf liner—designers can quickly and easily experiment with applying it to a variety of objects to support simple touch sensing.

2.1.1.2 Capacitive Sensing

Similar to resistive touch sensing, capacitive touch sensing is generally achieved through a grid or other configuration of sensors which experience a change in capacitance when touched by fin- gers or hands [76, 90, 112, 149, 150]. The raw change in capacitance can be used as a measure of

touch pressure [90, 149]. Unlike camera-based approaches, capacitive sensing is not susceptible to interference from ambient lighting [143]. Also, capacitive sensing approaches typically require significantly less bandwidth than image-based ones. However, because the collection of sensors is discrete, by default touch sensing resolution can be low, and often a single touch affects the capacitance of several adjacent sensors. To improve the resolution of detectable touches, such ap- proaches often include an interpolation scheme that considers the capacitance of many neighboring sensors [90, 149]; Westerman suggests that a 4 mm spacing of sensors can achieve a precision of 0.2 mm via interpolation [148], though it is important to note that such tightly packed grids may not always be achievable, especially for non-planar surfaces. Finally, unlike camera-based approaches, capacitive ones depend on specific electrical properties of the human body, often restricting touch input to only fingers [124]. However, it is possible to design special objects, such as styli, that can be tracked by such systems [116].

Using capacitive coupling, DiamondTouch supports touch from multiple users on a tabletop sur- face [43]. Location-specific electric fields are transmitted throughout the table using a series of antennas driven by a transmitter. Each user is capacitively coupled to his or her own receiver in a chair. When touch occurs, a circuit that runs from the transmitter to the table to the appropriate user’s receiver and back to the transmitter is completed. Thus, detected touches can be immediately associated with the corresponding user, facilitating multi-user touch input. Depending on the array of antennas, multiple touches from a single user may be too close together to be distinguished. Interestingly, even conductive objects left on the surface do not complete the entire circuit, and so they do not interfere with DiamondTouch; the authors propose creating special objects that could be used to interact with the table. As an extension, Wu and Balakrishnan explore multi-finger and hand gesture input on DiamondTouch [154].

SmartSkin comprises a grid of capacitive sensors in the form of transmitter and receiver electrodes on a table- or tablet-sized surface [122]. Along with detecting the position of a user’s hand on

the surface, it is also capable of computing the proximity of the hand to the surface. For the table-sized surface, the system is accurate to within 1 cm and is capable of sensing a hand within 5–10 cm of the surface. The authors suggest that SmartSkin is suitable for non-planar physical surfaces, though they do not explore this. Additionally, the authors experimented with applying conductive electrode barcode patterns to objects, which they refer to as “capacitance tags.” As the objects are ungrounded, they are not directly sensed by SmartSkin; when a user touches them, they become grounded and hence can be detected.

Villar et al. created an alternative computer mouse based on capacitive sensing [143]. While some similar approaches have difficulties with detecting multiple touches in close proximity, the Cap Mouse is able to sense the user’s fingertips separately. Internally, the mouse uses capacitive el- ements printed with conductive ink in a 5 mm grid, interpolating a touch’s position based on the specific subset of sensors whose capacitances are affected. Interestingly, the values read from the capacitive sensors are converted into a grayscale image, which is interpreted by the same touch sensing processing pipeline used by two camera-based multi-touch mouses the authors created; however, as the source of this data is a conductive ink grid, we categorize this approach together with other capacitive sensing methods.

2.1.1.3 Other

Gu and Lee presented TouchString, a flexible multi-touch sensor that can be applied to objects of various shapes, including planes and cylinders [59]. Their design is somewhat general, supporting optical, capacitive, or resistive sensors. As examples, they used the TouchString structure with phototransistors to allow for multi-touch support on a cellphone frame, a plane, and a bottle, using form-fitting configurations of sensors. However, they note that the prototype is not flexible enough to be truly general purpose, and they experienced problems when repeatedly folding and unfolding

it. Also, the spatial resolution of TouchString is a limitation, as the sensors are placed around 18 mm apart. The authors considered the addition of an interpolation scheme once they reduce the cell spacing by half.