Implementation of a Speech Enabled Virtual Reality Training Tool for Human-Robot Interaction

This paper presents the outline and implementation of a virtual reality-based training tool for interacting with the Julius Open-Source Voice Recognition Software [1]. This implementation builds on our previously designed tool for training police officers the command words associated with an integration of Julius with a robotic system with which they trained. By expanding the design into virtual reality, it was possible to increase realism of training as well as rapidly prototype and test new interfaces to support individuals' interactions with the voice recognition system. A simulated wearable display was provided to allow officers to look up the commands they needed and to provide feedback on what the voice recognition system detected. Additional functionality was incorporated to support new interaction methods such as push to talk and push to confirm.