| People with severe paralysis typically cannot use traditional human-computer interface devices. Those with limited or no motor control below the neck are able to use buttons or switches pressed with their heads as a way to control wheelchairs or interact with computers. Computer vision-based interaction systems have recently enabled other modes of interaction, such as controlling a mouse pointer by tracking a user's head movements. This dissertation builds upon the success of these systems. Typically, users are required to adapt to the interfaces that they wish to use. We propose interfaces that change and adapt to the user and their individual abilities. This dissertation consists of two parts.;First, we present a framework to adapt software to the needs of individuals with severe motion disabilities who use mouse substitution interfaces. The Hierarchical Adaptive Interface Layout (HAIL) model is a set of specifications for the design of user interface applications that adapt to the user. In HAIL applications, all of the interactive components take place on configurable toolbars along the edge of the screen. We show two HAIL-based applications: a general purpose web browser and a Twitter client.;Second, we study the human-computer interaction of users with severe motor disabilities who control the mouse pointer with a head tracking interface. We describe the difficulties users encounter in having to conform to the motion requirements of their interface system. Our analysis shows that some users are able to move in some directions easier than other directions. We propose several mouse pointer mappings that adapt to a user's movement abilities. These mappings take into account the user's motions in two or three dimensions to move the mouse pointer to the intended location on the screen.;We conducted user studies to evaluate the efficacy of our interfaces. This dissertation presents observations of participants with severe motion impairments, including small children, teenagers, and adults. We also propose the inclusion of social networking features in our interfaces. In particular, we investigate if automatically generated messages, posted by users with severe disabilities, can help enhance users' communication with family and caregivers. |
