A Comparison of Two Control-Display Gain Measures for Head-Controlled Computer Input Devices
Schaab, John A., Radwin, Robert G., Vanderheiden, Gregg C., Hansen, Per Krogh, Human Factors
Head-controlled computer input devices, or headpointers, allow computer operators to maneuver a cursor in a graphical user interface using head movements instead of the hand movements that are usually needed for mice and trackballs. Head-controlled interfaces provide an alternative method of accessing a computer and have been beneficial for some individuals with impaired upper-extremity control caused by spinal cord injury, degenerative muscular conditions, or cerebral palsy. The focus of this study is to examine characteristics of head motor control in order to provide information that will enhance performance of head-controlled computer input devices.
The linear relationship between movement time and the index of difficulty (ID) as described by Fitts' law has proven to be a robust predictor of various components of the human motor system (Fitts, 1954; Fitts & Peterson, 1964). Langolf, Chaffin, and Foulke (1976) showed Fitts' law to be a good predictor of movement time for the finger, wrist, and arm. Fitts' law has also been shown to be a reliable predictor of movement time for computer-related tasks involving a variety of hand-operated input devices (Card, English, & Burr, 1978; Epps, 1986). Research on head-controlled computer interfaces and the human movement characteristics associated with them has received limited attention.
Jagacinski and Monk (1985) were among the first to report on a head-controlled computer interface. Their study used a helmet-mounted sight as a computer input device and found Fitts' law to adequately describe head movement. The results of their study indicated that movement time was slightly faster for head movements along the vertical and horizontal axes than for those along diagonal axes.
A discrete target acquisition task was developed by Radwin, Vanderheiden, and Lin (1990) to measure and evaluate the performance of various alternative computer input devices. The results of their study supported Jagacinski and Monk's findings that Fitts' law aptly described head movement in computer-related tasks and that performance for diagonal head movements was slower than for movements along the vertical or horizontal axis.
Andres and Hartung (1989) showed that head movement could be described using Fitts' law for a tapping task involving a chin stylus. Spitz (1990) used a prototype head-controlled computer input device to study users' ability to control cursor location by tilting their head. The results of that study also supported a Fitts' law relationship and showed tilting movements of the head to be less efficient than extension/flexion movements at longer movement amplitudes.
Selection of a control-display gain setting for a particular input device has been an important issue for designers of computer interfaces. Chapanis and Kinkade (1972) indicated the need to empirically determine optimum gain for an input device, given the complexities involved within each system. The optimization of control-display gain has been the focus of several studies (Arnaut & Greenstein, 1986; Gibbs, 1962; Jenkins & Connor, 1949; Lin, Radwin, & Vanderheiden, 1992).
Typically the difference between hand- and head-controlled interfaces is that hand-controlled interfaces involve linear hand movement while the head remains stationary, whereas head-controlled interfaces involve head movements relative to a stationary display. As a result, the description of the control-display gain relationship for a head-controlled input device is different from that of traditional movements involving the upper extremities. An exception to this was Gibbs' (1962) study of joystick interfaces, in which control-display gain was defined as the ratio of angular movement of the display's cursor subtended at the user's eyes to corresponding angular movement of the joystick. Similarly, Jagacinski and Monk (1985), and later Lin et al. (1992), defined control-display gain for head-controlled interfaces as the ratio of the angle subtended by the displacement of the cursor to the viewing position and the corresponding angle of head extension/flexion or rotation. …