Using natural language processing (NLP) for designing socially intelligent robots

The design of natural interaction with social robots is highly complex process, given the huge design space of robots in terms of appearance and behavior and the challenges arising when using face detection and speech recognition in the wild. More natural and highly autonomous interaction is necessary to foster trust and engagement and hence establishing a long-term social relationship between users and robots. In this abstract, an adaptive and interactive dialogue system is designed to exchange a chat with a user using personal information stored in his/her user profile. NLP is used to extract user's basic information, hobbies and interests for building a rich user profile. The information from the user profile is used to customize and adapt subsequent dialogues in a way to build trust and initiate comfort between users themselves and the robot. The user profile is continuously updated whenever new information is extracted in subsequent dialogues. Face detection (FD) is used to identify the user. An artificial neural network (ANN) based FD system is used to increase the system's predictability. Failure to recognize a user's face leads to creating a new user profile for the new unidentified user. NLP failure leads to storing the whole sentence and manual fixing thereafter.