Investigation of Eye and Head Controlled Cursor Positioning Techniques

摘要

The project objective was to investigate the feasibility of using eye point-of-gaze and head control of a display cursor in place of, or to supplement, manual control for cursor positioning tasks. Of particular concern was the problem of positioning a cursor with respect to targets that may need to be small with respect to eye tracker accuracy and precision. The approach to this problem was to incorporate a second control modality, other than eye movement, for closed loop error correction (i.e. fine tuning). A second objective was to investigate the applicability of Fitts' law and related movement time prediction models to eye control. Several techniques were subjectively tested which allow users to 'close the loop' and fine tune point-of-gaze controlled cursor positions. All but one of these techniques were rejected on the basis of preliminary subjective tests. A technique that allowed the user to switch from point-of-gaze control to a low gain head position control when near the target seemed very effective during informal, subjective tests. This technique was used in a set of formal experiments, along with pure point-of-gaze control, pure head position control, and standard mouse control of display cursor position. The latter two control techniques have been studied fairly extensively by others, and data using these techniques help relate current results to previous work. All four of the above control techniques were formally tested using two serial, cursor positioning tasks. One of the tasks included a search component, and was designed to represent a realistic computer interface task. The other did not have a search component, and was designed to facilitate analysis of motion time with Fitts' law and related models.