| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a device and a method for executing a processing of appropriately selecting a user speech to the device from environment where a plurality of user speeches coexist. <P>SOLUTION: A probability value calculation using a Bayesian network is performed by inputting observation information such as image information photographed by a camera in addition to user speech information from the environment including users who make speeches, and the user speech to an information processor is selected. Furthermore, since the information processor is constituted so as to execute grounding processing for understanding a user request by applying Partial Observable Markov Decision Process (POMDP) to the selected user speech, a request to a device is exactly selected to promptly perform a task even under the environment in which the plurality of speeches are mixed. <P>COPYRIGHT: (C)2009,JPO&INPIT |
