Kaspar Causally Explains

Abstract

The Kaspar robot has been used with great success to work as an education and social mediator with children with autism spectrum disorder. Enabling the robot to automatically generate causal explanations is key to enrich the interaction scenarios for children and promote trust in the robot. We present a theory of causal explanation to be embedded in Kaspar. Based on this theory, we build a causal model and an analysis method to calculate causal explanations. We implement our method in Java with inputs provided by a human operator. This model automatically generates the causal explanation that are then spoken by Kaspar. We validate our explanations for user satisfaction in an empirical evaluation.