Development of a Generalized Voice-Controlled Human-Robot Interface: One Automatic Speech Recognition System for All Robots

摘要

Automatic speech recognition (ASR) technology has been used by researchers to develop voice-controlled robots. However, the usual pattern is to develop an ASR interface that works specifically with one particular robot. Generalized ASRs, ones that can communicate with multiple robots, are needed. A generalized ASR allows an operator to control several robots that are operated by different programming languages at the same time via a single interface. This paper describes the development and an initial user study of a generalized ASR interface using a commercially available ASR software. The user study provided initial validation of the interface's functionality using a variety of commands. Results indicated a reasonably high input flexibility; the interface successfully recognized 85% of 1200 spoken commands. It also demonstrated output flexibility; the generalized ASR fed the robot a sequence of keystrokes that formed any syntax required to trigger its task programs. Implications for human-robot interaction are discussed.