Active learning based on computer vision and human–robot interaction for the user profiling and behavior personalization of an autonomous social robot. (January 2023)

• Record Type: Journal Article
• Authors: Maroto-Gómez, Marcos; Marqués-Villaroya, Sara; Castillo, José Carlos; Castro-González, Álvaro; Malfaz, María
• Is Part Of: Engineering applications of artificial intelligence. Volume 117:Part B(2023)
• Publication Date: 2023-01
• Subjects: Active learning -- Human–robot interaction -- Social robots -- User profiling -- User recognition; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2022.105631
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Social robots coexist with humans in situations where they have to exhibit proper communication skills. Since users may have different features and communicative procedures, personalizing human–robot interactions is essential for the success of these interactions. This manuscript presents Active Learning based on computer vision and human–robot interaction for user recognition and profiling to personalize robot behavior. The system identifies people using Intel-face-detection-retail-004 and FaceNet for face recognition and obtains users' information through interaction. The system aims to improve human–robot interaction by (i) using online learning to allow the robot to identify the users and (ii) retrieving users' information to fill out their profiles and adapt the robot's behavior. Since user information is necessary for adapting the robot for each interaction, we hypothesized that users would consider creating their profile by interacting with the robot more entertaining and easier than taking a survey. We validated our hypothesis with three scenarios: the participants completed their profiles using an online survey, by interacting with a dull robot, or with a cheerful robot. The results show that participants gave the cheerful robot a higher usability score ( 82 . 14 / 100 points), and they were more entertained while creating their profiles with the cheerful robot than in the other scenarios. Statistically significant differences in the usability were found between the scenarios using the robot and the scenario that involved the online survey. Finally, we show two scenarios in which the robot interacts with a known user and an unknown user to demonstrate how it adapts to the situation.