Go to previous article
Go to next article
Return to 2000 Table of Contents
Edmund F. LoPresti, Jennifer Angelo, David M. Brienza,
Jonathan Sakai, and Lars Gilbertson
Departments of Bioengineering and Rehabilitation Science and Technology,
University of Pittsburgh
Pittsburgh, PA 15260 USA
Many people are unable to operate a standard computer mouse
due to disabilities affecting their hands or arms. One
alternative is a head control interface, which translates head
movements into cursor movements. Head controls can allow many
people with disabilities to gain access to computers for
vocational, educational, or recreational purposes. Head
controls also have potential applications for people without
disabilities as a method for hands-free computing. Providing
alternatives to hand-operated controls can reduce the incidence
of repetitive strain injuries of the wrists, among other
benefits. Although the use of head controls by people with
disabilities is common, little research has been done on the
ergonomics of how people use head controls to operate
computers. Studies of human-computer interaction have focused
on the use of standard mice, keyboards, and other hand-operated
controls (Keates and Robinson, 1997). Research into the
ergonomics of head movements will contribute to the design of
more efficient head controls.
Research is particularly needed regarding the influence of neck movement limitations. One such limitation is reduced active neck range of motion. Active neck range of motion refers to the number of degrees through which a person can move his or her head (Figure 1). This neck range of motion may be reduced as a result of a person's disability or a secondary condition (Stanger et al. 1994). If neck range of motion is limited, a person’s ability to operate a computer with head movements may be reduced.
Figure 1: Range of motion for neck rotations: A. Flexion and Extension B. Lateral Bending. C. Axial Rotation (supine position). D. Axial Rotation (standing position). (Thibodeau and Patton 1996 p. 298) This study investigates the ergonomics of head control use by observing users with and without disabilities, and with different ranges of neck movement. Performance with head controls was evaluated for people without disabilities, to obtain general information about how people use these controls. Data was then collected for people with cervical spinal cord injury, with different ranges of neck movement. This data was compared to data for subjects without disabilities, in order to study the effect of neck range of motion limitations on the use of head controls.
Methods Fifteen subjects without physical disabilities and three subjects with cervical spinal cord injury participated in this study. Each subject attended two sessions, scheduled between 3 and 14 days apart. During the first session, a magnetic tracking/virtual reality based system (DeFrate et al. 1999) was used to measure neck movements in three dimensions. Magnetic sensors (Flock of Birds, Ascension Technology, Burlington, VT) tracked movements of the subject's head relative to the torso. Subjects received visual feedback concerning their movements through virtual reality glasses (Virtual i-O (TM) Personal Viewing Glasses, Virtual i-O, Seattle, WA). The glasses display a virtual environment in which the subject is seated within a large wire-mesh sphere. Cross-hairs in the foreground of the display move against the wire-mesh background based on the movements of the subject's head relative to his or her torso. As the subject performs head movements, these movements are reflected by the motion of the cross-hairs along the background. Each subject received a neck range of motion evaluation using this system. The evaluation included three trials each of flexion/extension, axial rotation, and lateral bending.
During a second session, each subject performed two computer exercises using head controls (HeadMaster (TM), Prentke Romich Company, Wooster, OH). In a tracking task, the subject attempted to keep the cursor on top of a circular target as the target moved across the screen. Target movement patterns included vertical, horizontal, and diagonal paths. The second exercise was an icon selection task. At the beginning of the exercise, a circle appears at the center of the screen. The subject attempts to hold the cursor in this circle. A target symbol then appears elsewhere on the computer screen. The subject attempts to move the cursor to this target, and select it by holding the cursor within the target (dwelling) for 0.5 seconds. The subject then returns the cursor to the circle at the center of the screen. Targets appeared at one of three distances from the center of the screen and in any of eight directions from the center of the screen, for a total of 24 possible required movements.
Each subject performed four repetitions of the tracking task, with eight targets presented in each repetition. All but one subject performed 16 repetitions of the icon selection task, with 24 targets presented in each repetition. One subject with spinal cord injury only completed six repetitions of the icon selection task due to fatigue.
Results Subjects without Disabilities Fitts' Law has been a valuable tool in studying human-computer interaction (Olson and Olson 1990). Fitts' Law describes a relationship between movement time and target size and distance:
T = A + Blog2(2D/W) (1) where T = movement time, D = distance from starting position to center of target, W = target width, A is a constant term, and the coefficient B is the Fitts’ Law slope (Fitts and Peterson, 1964). Fitts' Law has been used to describe the speed with which people are able to use hand-operated controls such as the mouse or keyboard. Previous research indicates a Fitts’ Law relationship may apply to head movements as well (Jagacinski and Monk 1995, Lin et al. 1992). For this study, regression analysis of the icon selection data for subjects without disabilities indicates a Fitts' Law-type relationship between movement time and target distance:
T = 0.369 + 0.300log2(2D/1.5) (2) Because only one icon size was used, target width W=1.5 cm for all targets. Therefore the data can only indicate a Fitts’ Law-type relationship between movement time and distance, but not between movement time and target size. For each subject, half the subject's data was used to calculate this equation, and the remaining half of the data was used to test the equation's accuracy. The mean absolute-value error between the equation's predictions and the actual data was between 19.90% and 32.79%, with a 27.24% mean error across subjects.
One possible source of error for this model could be the effect of movement direction on movement time. Paired t-tests for the difference in mean movement times for each subject indicated that vertical movements were faster than horizontal (p=0.014) or diagonal movements (p=0.0009).
The effect of movement direction on cursor deviation was also considered. For the tracking task, cursor deviation is used to describe the difference between the path taken by the target symbol and the actual path along which the subject moved the cursor. Specifically, "cursor deviation" is used to refer to the root-mean-square difference between the actual path taken by the cursor and the straight-line path taken by the targets. A paired t-test indicates that cursor deviation was higher for diagonal targets compared to horizontal and vertical targets (p=0.0031).
Subjects with Disabilities Data for three subjects with cervical spinal cord injury was compared to data for the 15 subjects without disabilities. Active neck range of motion data is summarized in Table 1. Subject E01 has neck ranges of motion similar to the subjects without disabilities. Subject E03 has somewhat reduced neck range of motion, while subject E02 has severely reduced neck range of motion. Subjects' injury levels are shown in Table 1. Also, subject E01 has a spinal fusion between the C5 and C7 vertebrae and subject E03 has stabilizing rods between the C3 and T2 vertebrae.
Subject Injury Level Flexion/ Extension ROM Total Axial Rotation ROM Total Lateral Bending ROM C01-C15 None 118.42 ± 15.00 137.87 ± 13.60 87.35 ± 10.65 E01 C7 103.31 ± 5.55 136.90 ± 2.57 90.06 ± 4.74 E02 C1 2.72 ± 0.28 7.64 ± 0.07 6.70 ± 0.61 E03 C5 71.48 ± 2.07 98.41 ± 1.00 37.06 ± 2.68
Table 1: Range of Motion (ROM) in degrees. Mean (± st. dev.) across subjects without disabilities (C01-C15). Mean (± st. dev.) across trials for three subjects with cervical spinal cord injury (E01-E03). Measures of these subjects’ performance on the computer exercises are summarized in Table 2. The measures shown are:
Icon Selection Time: time elapsed between when a target icon appeared on the screen and when it was successfully selected by the subject, not including the 500 ms dwell time.
Error for Fitts’ Law Model: mean absolute-value error between observed icon selection times and the selection times predicted by the Fitts Law equation derived from data for able-bodied subjects (Equation 2). Icon Selection Accuracy: the number of icons successfully selected, divided by the number of icons presented as targets.
Percent Distance Subject Followed Moving Target: For the tracking task, the straight-line distance the cursor moved across the computer screen, as a percentage of the straight-line distance traveled by the target. Used as a measure of the portion of the screen accessible to the subject. Data greater than 100% indicate that the subject moved the cursor beyond the farthest position of the target.
Subject Icon Selection Time (seconds) Error for Fitts’ Law Model Icon Selection Accuracy Percent Distance Subject Followed Moving Target C01-C15 1.18 ± 0.12 27.24% ± 3.59% 99.93% ± 0.14% 99.77% ± 3.14% E01 1.91 ± 0.98 34.98% 97.92% 100.10% ± 8.13% E02 5.01 ± 2.46 72.61% 16.07% 49.76% ± 31.30% E03 1.02 ± 0.40 36.06% 100% 98.16% ± 4.66% Table 2: Results of Computer Exercise. Mean (± st. dev.) across subjects for subjects without disabilities (C01-C15). Mean (± st. dev.) across trials for three subjects with cervical spinal cord injury (E01-E03).
Discussion The results of this study provide information about the effects that movement distance and direction have on speed and efficiency. Understanding these effects can contribute to the design of better user interfaces for head control users, as well as head controls themselves. First, the results indicate that head movements can be described by a Fitts' Law equation. This equation could then be used to predict how quickly head control users could work with a particular computer software product, based on the cursor movements required to by the software. Different software, or different arrangements of objects on the screen, could then be compared in order to select the arrangement that will allow the user to work most quickly. The results also indicate that vertical cursor movements can be performed more quickly than horizontal or diagonal movements, and that people have more difficulty following a diagonal line using head movements compared to a horizontal or vertical line. These effects of movement direction could be taken into account in the design of software for head-control users, and in the arrangement of icons on the computer screen.
In addition to guiding the design and selection of computer software, these results could be used in the design of head control hardware. Information about the effects of movement distance and direction on user's performance could allow manufacturers to design head controls that accommodate a user's movement patterns.
In addition to obtaining general ergonomic information about head controls, it is important to consider the effect of disabilities on the effectiveness of a head control system. Subject E02, who has severely limited neck range of motion, had low accuracy in the icon selection task. He also had difficulty following the tracking targets across the screen, on average only reaching halfway across the screen. Subject E01 also had reduced performance with the head controls, despite having normal neck ranges of motion. This subject could move the cursor across the screen, as shown by his performance in the tracking task and his high accuracy in the icon selection task. However, his accuracy in the icon selection task was lower than that achieved by the control subjects, and the times needed to acquire icons were longer. This may be due to neck stiffness or some other symptom not considered in this study. Subject E03, on the other hand, despite somewhat reduced neck range of motion, was able to perform the tracking and icon selection exercises with performance comparable to the subjects without disabilities.
The results from subjects with disabilities also shows the limitations of human-computer models, such as the Fitts' Law equation described above. This Fitts' Law model was developed based on the data for subjects without disability. Based on the results shown in Table 2, it does not seem to be as successful at predicting the performance of subjects with disabilities compared to subjects without disabilities. Ideally, models of human-computer interaction would be developed to account for disabilities, and thus be applicable for all computer users.
Conclusion A study was conducted to analyze head movements in the context of two computer exercises: an icon selection task and a tracking task. The results indicate relationships between movement distance and movement time; movement direction and movement time; and movement direction and cursor deviation. The results also show that disabilities such as cervical spinal cord injury can have an impact on the ergonomics of head controls. In further research, data will be collected for 12 more individuals with disabilities, allowing further analysis of the effect of disabilities on the subjects' performance with head controls. Based on these data, models of the human-computer interface will be developed which address the use of head controls by people with and without disabilities. These results will also be used in the development of head control software that can adapt to a user's available neck range of motion.
Acknowledgments This research was funded by the Microsoft Corporation and the Whitaker Foundation.
References DeFrate, L.E., Sakai, J.L., Gilbertson, L.G., Moon, S-H., Nishida, K., Donaldson, W.F. III, Kang, J.D., and Woo, S. L-Y. 1999. Magnetic Tracking/Virtual Reality Based System for Comprehensive Kinematic Assessment of the Cervical Spine. ASME Summer Bioengineering Conference. 1999.
Fitts, P.M. and Peterson, J.R. (1964). Information Capacity of Discrete Motor Responses. Journal of Experimental Psychology. 67(2):103-112. Jagacinski, R.J. and Monk, D.L. 1985. Fitts’ Law in Two Dimensions with Hand and Head Movements. Journal of Motor Behavior. 17(1):77-95. Keates, S. and Robinson, P. 1997. User Performance Modeling and Cognitive Load. Proceedings of the RESNA 1997 Annual Conference. 342-344. Lin, M., Radwin, R.G., and Vanderheiden, G.C. 1992. Gain Effects on Performance Using a Head-Controlled Computer Input Device. Ergonomics. 35(2):159-175.
Olson, J.R. and Olson, G.M. 1990. The Growth of Cognitive Modeling in Human-Computer Interaction Since GOMS. Human-Computer Interaction. 5:221-265.
Stanger, C., Phalangas, A., and Cawley, M. 1994. Range of Head Motion and Force of High Cervical Spinal Cord Injured Individuals for the Design of Test-bed Robotic System. Proceedings of the 1994 ICORR Conference. 37-40. Thibodeau, G.A. and Patton, K.T. 1996. Anatomy and Physiology Third Edition. Mosby, St. Louis.
Go to previous article
Go to next article
Return to 2000 Table of Contents
Return to Table of Proceedings