Abstract: This article presents a method for learning well-coordinated human–robot interaction (HRI) from human–human interactions (HHI). We devise a hybrid approach using hidden Markov models (HMMs) ...