A mobile robot-operator interface based on spoken dialogue

This paper describes an approach to designing a mobile robot-operator interface based on spoken dialogue. This is a research topic wtihin IRST AI project MAIA (Advanced Model of Artificial Intelligence) whose purpose is to approach AI from the integration point of view. This paper argues that speech will be an important capability for robots operating in many different areas. A system for speech control of robot should have both robot behavioural and environmental models together with a dialogue model. This paper describes a probabilistic model for spoken dialogue between an operator and a robot. The model incorporates knowledge about dialogue in order to improve speech recognition by the robot. A prototype of spoken dialogue interface for a mobile robot, navigating inside the authors´ Institute, is also described. The interface is based on a speaker independent continuous speech recognition with a 200 words vocabulary. The recognizer, tested on a sample of 130 sentences in office environment, performed well: 2% semantic errors.<>