Approximately 2.2 million people in the United States depend on a wheelchair to assist with their mobility. Often times, the wheelchair user can maneuver around using a conventional joystick. Visually impaired or wheelchair patients with restricted hand mobility, such as stroke, arthritis, limb injury, Parkinson’s, cerebral palsy or multiple sclerosis, prevent them from using traditional joystick controls. The resulting mobility limitations force these patients to rely on caretakers to perform everyday tasks. This minimizes the independence of the wheelchair user. Modern day speech recognition systems can be used to enhance user experiences when using electronic devices. By expanding the motorized wheelchair control interface to include the detection of user speech commands, the independence is given back to the mobility impaired. A speech recognition interface was developed for a smart wheelchair. By integrating navigation commands with a map of the wheelchair’s surroundings, the wheelchair interface is more natural and intuitive to use. Complex speech patterns are interpreted for users to command the smart wheelchair to navigate to specified locations within the map. Pocketsphinx, a speech toolkit, is used to interpret the vocal commands. A language model and dictionary were generated based on a set of possible commands and locations supplied to the speech recognition interface. The commands fall under the categories of speed, directional, or destination commands. Speed commands modify the relative speed of the wheelchair. Directional commands modify the relative direction of the wheelchair. Destination commands require a known location on a map to navigate to. The completion of the speech input processer and the connection between wheelchair components via the Robot Operating System make map navigation possible.
Library of Congress Subject Headings
Wheelchairs--Control; Automatic speech recognition; Speech processing systems; Human-computer interaction
Computer Engineering (MS)
Department, Program, or Center
Computer Engineering (KGCOE)
Echefu, Samuel, "Towards Natural Human Control and Navigation of Autonomous Wheelchairs" (2016). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus