Avoiding the Content Treadmill for Robot Personalities

Abstract

Robot personalities are today painstakingly hand-crafted by content creators who specify motions, facial animations, and scripts in order to convey a unifying sense of character via interactions. While user preferences can be incorporated by modifying these behaviors to some degree, such as by inserting a user’s name or preference into a script, by and large new content must be continually created in order to maintain the robot’s ‘freshness.’ This content treadmill represents a long tail of work that diverts resources away from improving the underlying capabilities of the robot. Here we outline a novel method for defining robot personality that allows each robot to have and express a unique personality through varying behaviors that reduces the need for content creation.