Robotics is moving from industrial applications into personal and office environments, in part due to converging trends of an aging population, a shrinking workforce, and a decrease in production costs. The research field of Human-Robot Interaction (HRI) studies the models and algorithms necessary for robots to interact with non-professional humans in these new environments.However, most HRI models are structured in a discrete turn-taking framework, based on dialog, planning, and state-action models. This usually results in a rigid and delayed stop-and-go interaction, which is unlike the simultaneous, fluent action coordination humans are used to from each other.This research proposes to bridge this gap by evaluating new computational models enabling fluent coordination in human-robot interaction. The goal is to design robots that interact with untrained humans in a more natural and human-acceptable way, facilitating their integration into homes, offices, hospitals, and workshops.The research proposed is based on recent cognitive and neurological research in embodied cognition, which sheds new light on the how humans coordinate their actions in a fluent manner. It proposes to computationally model these findings - in particular those of perceptual emulators and real-time action predictors - and to implement them in a human-robot collaborative setting. It also proposes to conduct human-subject studies evaluating the benefits of these new models, and their acceptance by untrained users.The project's potential is to suggest new research paths in HRI, extend the current knowledge in embodied cognition, and produce practical outcomes for the design of robotic systems for the general population. Due to the recent return of the applicant to the EU zone from the US, this proposal is also crucial for the reintegration of the applicant into the EU zone, as well as for the successful transfer of knowledge and future collaboration with top US research institutes.