Accessible interfaces for robot assistants

Abstract

Currently, high-level task control of robots is generally performed by using a graphical interface on a desktop or laptop computer. This type of mediated interaction is not natural, and can be problematic and cumbersome for persons with certain types of motor disabilities, and for people interacting with the robot when there are no computer displays present. In this work, we present a framework which enables the removal of such obvious intermediary devices and allows users to assign tasks to robots using interfaces embedded directly in the world, by projecting these interfaces directly onto surfaces and objects. We describe the implementation of the projected interface framework, and give several examples of tasks which can be performed with such an interface.