Nowhere do robots promise to enhance the quality of life of humans as much as in the area of rehabilitation. Robotic technology has the potential to assist physically handicapped and elderly people, especially in the mobility for their day-to-day life. For a traditional wheelchair, an individual must navigate through close quarters in his/her home, which, like many homes, was not designed with handicapped people in mind, by carefully and slowly manipulating the wheelchair through a somewhat rudimentary joystick interface. Controlling his/her movements in this fashion may not only be frustrating and time-consuming, but may even be impossible for individuals who have only limited control of their arm, such as elderly persons who experience periodic tremors in their hands, or stroke victims who have been left with only partial dexterity and movement in their arm. On top of that, very similar maneuvering tasks are typically part of the individual's daily routine, for example, a trip from the bedroom to the bathroom which is notoriously difficult to successfully navigate.
Our goal in this project is to develop a smart wheelchair with a human friendly interface for high-level control, and apply human skill models to significantly enhance the usability and functionality of traditional electric wheelchairs. The research issues in this project include (1) what type, how many, and where to locate sensors on the wheelchair so that models of human skill can be reliably represented and learned; (2) how to effectively model human control strategies with very limited data; (3) how to partition human skill module into series of submodule and assemble a skill module for a more difficult maneuver; (4) how to improve human skill models and filter out the deleterious human control signals due to partial dexterity impairment or manual jerk; and (5) how to design user-friendly interfaces for higher-level control of smart wheelchairs.
With the rapidly shrinking costs of modern computing and sensing, we can expect that the costs of outfitting a traditional wheelchair with the required on-board computer and sensors, can be kept at a minimum. In exchange, a disabled person may significantly improve his/her day-to-day life, as he/she teachs the wheelchair over time to perform maneuvers autonomously and with rider comfort in mind.