CMUSV Develops Human – Robot Dialogue Infrastructure
Students at Carnegie Mellon University, Silicon Valley have developed an infrastructure for Human - Robot Dialogue on Robot Operating System. The infrastructure fills in the gaps for a much needed, ready to use and standardized functionalities, which can be used by robot developers to quickly build the robots.
The system has been developed with a generic speech recognition dialogue service as the first offering. The service allows developers to integrate speech recognition based dialogue without writing significant amount of code. The system generates a set of well-defined messages that contain information relevant to speech and is independent of the underlying speech recognition engine. The dialogue system includes a state management infrastructure, which allows a state manager application to dynamically update the language model and acoustic model used by the speech recognition service. The speech recognition service will be provided on Robot Operating System platform, as an open source, free to use package.
The infrastructure is developed by a Vishva Gaurav Sinha, for the client Honda Research Institute (USA), and Professor Ian Lane. This product was developed in twelve weeks as part of the summer practicum course conducted at Carnegie Mellon University.