Crowdsourcing human behaviors for conversational social robot

Robots are becoming more capable and affordable, and interactive humanoid robots such as Pepper are already being deployed in retail shops in Japan. Yet, designing interaction logic and behavior contents can still be a labor-intensive process, heavily dependent on the designer’s ability to imagine a variety of social situations (for example, anticipating all of the questions people may ask the robot) and use their intuition to specify relevant and appropriate behaviors and execution rules for the robot. To address the above problems, we present several approaches found in the field of human-robot interaction (HRI) of acquiring verbal and non-verbal dialog behaviors information by using the “crowd”. The approach has the advantage of producing dialog variation by eliciting language samples from an array of individuals as well as capturing unique style of a certain individual. Finally, we will discuss the how crowdsourcing can be applied to other domain to enable data-driven application.