Home | Business News | Browse by Publication | H | Human Factors

The form of augmented force-feedback fields and the efficiency and satisfaction in computer-aided pointing tasks.

Article Excerpt
INTRODUCTION

The development of graphical user interfaces (GUIs) with mouse control has been an important step toward making direct manipulation of objects on a computer screen possible (Shneiderman, 1983). Direct manipulation of objects has greatly improved the effectiveness, efficiency, and user satisfaction in many computer tasks. Control of such direct manipulations is normally mediated by the visual system; the user observes both the object in question and a cursor that is under mouse control and tries to coordinate hand and arm movements to obtain some specific graphical effect (e.g., the cursor moving toward and landing on a target object).

When people manipulate objects in the real world, their tactual and kinesthetic senses provide important feedback, in addition to their visual system. It is therefore not surprising that quite a few attempts have been made to add force feedback to the control device in GUI-mediated operations. Engel, Goossens, and Haakma (1994) presented a force-feedback trackball that significantly shortened the completion time of moving a cursor through a maze. Using the same control device, Keyson (1996) showed that the time needed to move a cursor to a target area could be shortened by an average of 12% if, through force feedback, the target area was made to feel like a ball rolling into a hole. Akamatsu and Sato (1994) developed a haptic mouse with computer-controlled friction and a vibrotactile warning signal applied when a target was reached. This device also reduced target acquisition times by a significant amount.

Application of force feedback does not have to be limited to the immediate target area. Just as McGuffin and Balakrishnan (2002) successfully enhanced visual targets as they were being approached, augmented force feedback can be applied to a control device when the cursor is still outside the target area. The applied force might allow the user to move with less accurate manual control, relying on the force feedback to home in on a target. Goossens (1992), who also used the previously mentioned force-feedback trackball, found a 22% reduction in movement time when the "hole-shape" force was extended by a distance of 45 pixels in every direction beyond the target border. This reduction was only 12% when the force was not spatially augmented (Keyson, 1996). Hasser, Goldenberg, Martin, and Rosenberg (1998) used a force-feedback (FEELit) mouse to show that movement times to targets could be reduced by up to 61% if a force with a gradual onset, a constant segment, and a gradual offset was used extending to 30 pixels outside the target area. In addition, they found that on average, 26% fewer acquisition errors were made. Dennerlein and Yang (2001), in a similar experiment, found an acquisition time reduction of 25%.

There is also evidence that the shape of the force-feedback function has a large influence on the "feel" and may therefore have consequences for the operation's efficiency and user satisfaction. There are many choices of force-field models, and there is no a priori "optimal" force field that will obviously yield the best results. The reason most researchers tend to choose a gravity model is probably that people tend to be naturally familiar with these field types. For instance, the force on a ball rolling down into a bowl-shaped hole is probably more familiar than the force exerted on a charged particle in an electric field. Leenders (1994) studied target acquisition efficiency and user preferences for hole-like gravity force-feedback fields that were either flat bottomed or bowl shaped and could have maximum force values of 0, 100, 225, and 350 millinewtons (mN). He found that in general, the duration of moving the cursor to a visually specified target was affected very little by force feedback in a stand-alone task but was significantly reduced if participants had to perform other simultaneous tasks. Flat-bottom and bowl shapes were equally effective, but participants showed a clear preference for bowl-shaped holes. Although generally higher force levels led to shorter movement times, the 225 mN maximum force function was the most preferred. The experiments by Hasser et al. (1998) and Dennerlein and Yang (2001) also appear to confirm that the shape of a force-feedback function can have considerable effect on operation efficiency and user satisfaction.

Therefore, the two questions addressed in this study are (a) is there an optimum force-field shape for circular force fields with respect to operation efficiency and user satisfaction, and (b) what are the constraints on the magnitude of such force functions?

PILOT EXPERIMENT

The choice for a gravitational force field was made partly because of the participants' assumed natural familiarity with such fields and partly because of the availability of software tools to create such fields (Keyson & Tang, 1995). Within this class of fields and general hole-type models, however, there still are many possible hole shapes to choose from. A first restriction imposed was to use only holes with a circular perimeter centered at the target. The set of potential shapes can be further reduced by imposing three additional constraints on the hole's cross-sectional shape, based on the observation that a force applied over a certain period of time or distance generally has an onset, a steady-state portion, and an offset. The first constraint imposed is that in all three stages the force is directed toward the target. The second constraint is that the steady-state part is...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!