Fitts' Law -- Virtual Reality (VR) Experiment: Evaluating Display and Input Devices with Fitts' Law

CS 5724: Models and Theories of Human-Computer Interactions
Fall 1996

Summary || Equipment || Experimental Design || Method || Results || Conclusion

Summary

A laboratory experiment was conducted using Fitts' law to evaluate performance of VR hardware and input devices. A VR4 Helmet, a 17 inch CRT monitor, a space mouse, and a regular computer mouse were tested with various parameters of classic Fitts' law experiments.

Equipment

The space mouse used was a Logitech Magellan 3D Controller six degree of freedom pointing device, shown if Figure 1. The space mouse is puck shaped with slight displacement in the six degrees providing the intended movement. For the tasks in this experiment, the participant was only able to traverse forward, backward, left and right. All other degrees of movement will be turned off.

picture of space mouse used in experiment

Figure 1. Logitech Magellan 3D Controller.

The regular computer mouse was a typical 3 button hand held computer mouse.

The helmet-mounted display (HMD) to display the task environment was the Virtual Research Systems, Inc. VR4 shown in Figure 2. The HMD has two 2.7" color LCDs with approximately 40 degrees full overlap field of view. The VR4 was specifically intended for high performance immersive applications.

picture of VR HMD used in experiment

Figure 2. Virtual Research Systems, Inc. VR4 HMD.

The desktop display used was a typical 17 inch CRT monitor.

The software used to model the Fitts' paradigm in this experiment was developed by Scott MacKenzie and others at the University of Guelph. See the Generalized Fitts' Law Model Builder for more information.

Experimental Design

The four conditions tested included:

  1. VR4 HMD with space mouse
  2. VR4 HMD with regular mouse
  3. 17 in. CRT with space mouse
  4. 17 in. CRT with regular mouse
In each of the four conditions participants completed 256 Fitts trials, varying angle of target, width of target, and distance of target from starting point. Each of the individual Fitts' trials was completed four times in each condition. 16 participants participated in one of the four conditions resulting in four participants per condition.

Method

16 Participants were taken from a graduate course on HCI Theories and Models.

Participants were randomly assigned to conditions. Upon assignment to conditions participants received instrustions on how to use the equipment in their condition and were give approxiamately 20 practice trials. The Fitts' software then randomly presented 256 trials for the session varying amplitude, widht, and angle (theta).

Results

A mixed factors ANOVA was used to analyze the data.

Each of the four main conditions were significantly different from one another, with the HMD/space mouse condition having the slowest movement times, then the CRT/space mouse, then the HMD/regular mouse, and the fastest times with the CRT/regular mouse.

For VR equipment, the results showed that input device was significant (p = .0100) with the space mouse being slower than the regular mouse. The display type did not show significantly different movement times.

For Fitts law validation, the results showed that width of target was significant (p = .0001) with larger widths being acquired faster than smaller widths, distance of target was significant (p = .0001) with shorter distances being acquired faster than longer distances, and theta was significant (p = .0191) with targets at 90 degrees being acquired slower than targets at either 0 or 180 degrees.

Significant interactions include amplitude * input device (p = .0090), width * input device (p = .0044), and amplitude * width (p = .0160). Graphs

Conclusions

The experiment replicated established findings that show width and amplitude affect time to acquire targets in a computer environment. Theta also has been shown in the past to influence target acquisition, as found in this experiment. Fitts (1954) originally described arm motions towards the body (flexor) as being faster to acquire targets than those away from the body (extensor). The 90 degree condition required the extensor motion of the arm to acquire targets in all conditions, and showed slower acquisition times than some other angles.

The VR equipment showed interesting results. The input device was highly significant with the space mouse being much worse than the regular mouse. Experience with the space mouse vs. a regular computer mouse is a possible explanation for slower times. The space mouse also may be harder to manipulate. Participants did receive instruction and brief practice before the trials began, but by no means enough to qualify for 'expert' performance. The experimenters hypothesized the HMD would yield slower acquisition times than the CRT display. The data showed a trend for this, but was not close to significant (p = .2721). The conclusion is the HMD did not worsen performance, participants readily adapted to its display of the information.

While this experiment used a 2-dimensional task, a modified version of the Fitts' Law could be used to evaluate 3D VEs for ballistic target acquisition performance.

Back to Fitts' Law Main Page
Back to CS5724 -- Models of HCI Home Page
By Fitts' Law Group
Last Update: 11/125/96