Optimization of Animated Facial Feature Appearance for VALERIE

The Union College Computer Science Department's social robot, VALERIE, was designed to be just that: a social robot. Whether it is guiding a tour around campus or helping a lost individual find where they are supposed to be going, VALERIE is meant to interact with people and provide them with information. However, many factors limiting VALERIE's effectiveness exist, including its appearance. The face that appears on VALERIE's display should be inviting and easily approachable. In practice, though, users may have found VALERIE's face to be off putting or unapproachable. Finding the "correct" face is not as simple as it may seem. Applying findings from existing research on facial features and human attractiveness could lead to an "uncanny valley" effect, undermining VALERIE's effectiveness. As for animated face optimization, previous findings are either inapplicable due to VALERIE's design or rely on hardware that VALERIE does not have. So, how do we create a face that users find approachable? We ask people to make it for us. By completely tearing down and building back up the code that runs VALERIE's face, we can transform it from an immutable display into a completely editable face with intuitive slider controls. After prompting users with their task, they will create their personal optimal VALERIE face, which will then be stored alongside every other user created face. Analysis of these results will provide the framework for design choices not just for VALERIE, but for all similar social robots with displays.