Learning from HRI Datasets

Context The filed of social human-robot interaction is growing. There are more and more openly available datasets that features social interaction between humans and social interactions between humans and robots. Interpreting the transferability of human-human communition to human-robot communication is crucial in build social human-robot interactions. Credits Severin Lemaignan - PinSoRo Dataset In this project we propose to take a data-driven approach to build predictive model of the social interaction for human-human (HH) and human-robot scenarios.

Perception Load as a Measure of Anthropomorpshim

Context: Research in HRI has been investigating how robot design and in particular humanlikeness can influence the interaction. The uncanny valley illustrate for instance how robots’ appearance influence emotional responses. In this project, we aim to take it a step further and investigate how robots’ appearance can influence cognitive load. Follwing a similar protocol to [1], we will use robot’s pictures as distractors in a perception task. Past research has shown that distractors with human faces impaired more strongly performances in the task when compared to other objects.