音声ブラウザご使用の方向け: SKIP NAVI GOTO NAVI

SINGLE SWITCH MOUSE CONTROL INTERFACE

Andrew J. Moynahan, Richard M. Mahoney

Applied Science and Engineering Laboratories (ASEL)

University of Delaware/Alfred I. duPont Institute

Wilmington, Delaware, USA

ABSTRACT

This paper presents the first results from an experimental single switch direct manipulation interface. This interface is based on goal directed movements. The experiment focuses on the goal of target capture in a graphical user interface. Descriptions of the system and the experiment are also given.

INTRODUCTION

Direct manipulation lies at the heart of a graphical user interface. The importance of direct manipulation and its relationship to GUIs has been previously described[1][2]. The main advantage of direct manipulation is that it allows the user to become involved in carrying out whole actions and tasks rather than having to break tasks down into small, sometimes seemingly unrelated, parts. For example, systems that employ on screen scanning menus may require a user to scan through multiple levels to carry out a given task. The previous example illustrates why current systems that provide single switch access to GUIs are unable to provide all of the benefits of direct manipulation to the user. In an effort to improve single switch access systems to graphical user interfaces (GUIs), a new mouse pointer control scheme is being developed and tested. The concept of goal directed movement lies at the heart of this new control method.

BACKGROUND

A survey has identified a number of features that would make a computer more accessible to individuals with disabilities [3]. One of the features listed in this survey is software access to the operating system. Having access to the operating system not only allows software to interject information, such as simulated keystrokes, but allows a program to find out information about the environment in which it is running. The driving force behind this control scheme comes from the desire to increase a user's control over direct manipulation while using a single switch interface [4]. To achieve this desire, a new control scheme for cursor control was proposed. At the heart of this new cursor control scheme is an intelligent software agent. The job of this intelligent agent is to perform tasks under the supervision of the user.

SYSTEM DESCRIPTION

The developed control method and interface differ from existing single switch GUI interfaces in a number of ways: ù the user is in control of a task, such as moving the mouse pointer to a target, rather than controlling a subtask, such as moving the mouse pointer in a particular direction, ù the task is carried out by an intelligent agent, ù the intelligent agent is knowledgeable about the environment it is working in, ù changing to a new task is initiated by the user and then carried out by the intelligent agent. By using this interface a user supervises the actions of the control agent. To activate the control agent, a user supplies an appropriate `GO' signal. In the case of our experiment, this signal comes in the form of holding down the right mouse button. If the user finds the control agent carrying out the correct task, she/he merely continues to supply the `GO' signal. As long as the system sees a `GO' signal it will continue with what it thinks is the appropriate task. If, however, the currently executing task is not the one desired by the user, he/she can terminate the `GO' signal by releasing the right mouse button. This stops the current task. Pressing the mouse button again causes two things to occur: ù the intelligent agent re-evaluates the situation and attempts to come up with a reasonable choice for the task to perform, ù the intelligent agent begins to carry out the newly identified task. The system uses information about the environment and the user's input to make its decisions about which task to execute. Due to the nature of Microsoft Windows, the environment for which this interface is developed, information about the objects on the screen is available [5].

EXPERIMENT DESCRIPTION

In order to test the ideas behind the proposed single switch mouse control interface (SSMCI), an experiment was designed and carried out. The experiment simulates the task of moving the mouse cursor to a target on the screen. The experiment begins with the presentation of a number of identical looking circular targets. After a short delay, one of the targets changes color, indicating this target as the goal target. Once the goal target is indicated, the user must then move the mouse pointer to the goal target. Moving the mouse pointer is accomplished by one of two control methods: the SSMCI and the arrow keys found on the numeric keypad. Figure 1 is an example of a sample experiment screen. This figure shows a total of four possible targets and a mouse pointer. The number of targets on the screen varies from two to five throughout the experiment. The experiment was designed in such a way to guarantee equal number of trials with two, three, four, and five targets. The positions of the targets are random. While participating in the experiment, each subject saw a total of 200 trials. Of these 200 trials, 100 were conducted using the directional keys while the remaining 100 trials were done using the SSMCI. The trial scenarios seen by a user while using the SSMCI were the same as those seen while using the directional keys. The order of trial presentation, however, was random There were a total of eight subjects in this study. Of those eight, five were able bodied while three had different disabilities: cerebral palsy, C5-6 spinal cord injury, and Arthrogryposis Multiplex Congenita. All of the able bodied test subjects typically use a mouse while interacting with a computer. Those with disabilities, however, all used different methods (i.e. trackball, mouse, and directional arrow keys) for interacting with a computer on a daily basis. Data was collected while the experiment was in progress. The data collected includes:

  • total time required to reach the goal target,
  • total distance travelled by the mouse pointer while reaching the goal target,
  • distance the goal target initially was from the mouse pointers position,
  • total number of key or switch activations made while reaching the goal target,
  • in the case of the SSMCI method, the total number of intermediate targets chosen by the intelligent agent before the goal target was reached.

RESULTS

Once the data was collected, the two methods were compared to judge how well they performed in the task of target capture. Figure 3 shows the data collected concerning the average number of switch hits and key presses required by the able bodied users to reach the goal target. Figure 2 shows the average time required to capture the target by the able bodied subjects. For the data in Figure 2 and Figure 3, a one way ANOVA was performed which resulted in a p << 0.001. In Figure 4 the average number of switch hits and key presses is shown for the test subjects with disabilities. Figure 5 shows the average target capture times for the subjects with disabilities. As in the case for Figures 2 and 3, a one way ANOVA was performed which resulted in a p << 0.01.

Figure 1: Sample experiment screen

Figure 2: Average target capture time for able bodied subjects

Figure 3: Average number of switch/key presses for able bodied subjects

Figure 4: Average number of switch/key presses for subjects with disabilities

Figure 5: Average target capture times for subjects with disabilities discussion The above figures show a performance gain for the SSMCI when compared to the directional keys for the given task. This performance gain may be due in part to a number of factors:

  • the intelligent agent alleviates the user from some of the burdens (mental and physical) associated with the task,
  • by knowing the final destination, the SSMCI can move the mouse pointer at a faster velocity than if it were being controlled by hand.

By knowing the location of the targets, the SSMCI is able to plot a direct path to the target. Using a system that only allows selection of movement directions typically forces a user to move along an indirect path. Knowing the final destination also allows for velocity profiling. The current velocity profile includes both periods of movement at a constant velocity as well as periods of acceleration and deceleration. Since the SSMCI is able to decelerate to a slow speed as it approaches a target, it is able to travel at high speeds for the majority of travel towards the goal target. After completing the experiment, a number of subjects remarked that the SSMCI made the task easier. These subjects felt that acting as a supervisor to the SSMCI was easier than dealing with controlling the cursor movement themselves. Future studies with the SSMCI could look at a number of issues including:

  • improving the intelligence of the agent to make it more robust,
  • looking at how the SSMCI performs in a more complex world,
  • adding features to allow interaction with windows, menus and buttons.

Further performance gains and added functionary may be achieved through improving the intelligent agent. In is current form, it acts as a rule based planner.

REFERENCES

[1] Shneiderman, B., Designing the User Interface: Strategies for Effective Human-Computer Interaction, Addison Wesley, 1992.

[2] Ziegler, J., Fahnrich, K., Direct manipulation. In Martin Helander, editor, Handbook of Human-Computer Interaction, pages 123 - 134. North-Holland, 1990.

[3] Vanderheiden, G., Lee, C., Scadden, L., Features to Increase Accessibility of Computers by Persons With Disabilities: Report From the Industry/Government Task Force., Proceedings of the RESNA 10th. Annual Conference, pages 750 - 752, 1987.

[4] Moynahan, A., Mahoney, R., Single Switch Direct Manipulation, Proceedings of the RESNA `95 Annual Conference, pages 422 - 424. June, 1995.

[5] Petzold, C., Programming Windows 3.1, Microsoft Press, 1992

ACKNOWLEDGEMENTS

This research is supported by the U.S. Department of Education, Grant #'s H129E20006 from the Rehabilitation Services Administration and H133E30013, the Rehabilitation Engineering Research Center on Rehabilitation Robotics, from the National Institute on Disability and Rehabilitation Research (NIDRR). Additional support is provided by Nemours Research Programs.

Andrew Moynahan Applied Science and Engineering Laboratories A.I. duPont Institute 1600 Rockland Road, P.O. Box 269 Wilmington, DE 19899 USA

moynahan@asel.udel.edu

Andrew J. Moynahan, Richard M. Mahoney Applied Science and Engineering Laboratories (ASEL) University of Delaware/Alfred I. duPont Institute Wilmington, Delaware USA

SINGLE SWITCH MOUSE CONTROL INTERFACE SINGLE SWITCH MOUSE CONTROL INTERFACE SINGLE SWITCH MOUSE CONTROL INTERFACE