Search

Our research effort takes inspiration from human social learning mechanisms to focus on situations in which an expert guides a learner through explanations. The proposed approach incorporates explanations into maximum likelihood inverse reinforcement learning. We computationally evaluate explanations against other teaching signals (reward, demonstration and explanation) in three navigational scenarios. The generated explanations are also evaluated in a user study with 150 participants. The user study investigates participants' preferences between the different types of teaching signals and the impact of contextual situations, i.e., distance from the task's goal, on their preferences. Our simulations' results show that explanations lead to better performance compared to reward and demonstration signals, and that explanations are preferred by human teachers in situations where the goal is far from the learner.

International audience; The issue of how to make embodied agents explainable has experienced a surge of interest over the last three years, and, there are many terms that refer to this concept, e.g., transparency or legibility. One reason for this high variance in terminology is the unique array of social cues that embodied agents can access in contrast to that accessed by non-embodied agents. Another reason is that different authors use these terms in different ways. Hence, we review the existing literature on explainability and organize it by (1) providing an overview of existing definitions, (2) showing how explainability is implemented and how it exploits different social cues, and (3) showing how the impact of explainability is measured. Additionally, we present a list of open questions and challenges that highlight areas that require further investigation by the community. This provides the interested reader with an overview of the current state-of-the-art.

We present the design and evaluation process of a robot aimed at stimulating creativity in humans in a collaborative drawing-oriented task. Collaboration towards a common goal is known to be one of the most influential factors for creativity nourishing. The social robot acted as a collaborative peer in a drawing activity, taking turns with the human, to complete their drawing. In each turn, both "artists" (the robot and the human) draw something new until they achieved their final drawing. The goal was to study if a robot, in comparison with a tablet, can stimulate creativity. A total of 28 participants were distributed across robot and tablet conditions. Creativity scores for each drawing were calculated. Contrary to what was expected, the creativity score did not present a statistically significant main effect across conditions. We discuss our results under a design perspective, addressing the role of the robot’s embodiment and presence, as well as the purpose of use for robots in society and what implications there is for design.