UC Santa Cruz

For people with poor upper limb mobility or control, interaction with a computer may be facilitated by adaptive and alternative interfaces. Visual head tracking has proven to be a viable pointing interface, which can be used when use of the mouse or trackpad is challenging. We are interested in new mechanisms to map the user’s head motion to a pointer location on the screen. Towards this goal, we collected a data set of videos of participants as they were moving their head while following the motion of a marker on the screen. This data set could be used to training a machine learning system for pointing interface. We believe that by learning on real examples, this system may provide a more natural and satisfactory interface than current systems based on pre-defined algorithms.