Abstract

Physical human–robot interaction (pHRI) is arguably the most efficient and natural communication modality between humans and robots. Across diverse pHRI scenarios—ranging from robot-assisted rehabilitation and collaborative manipulation to shared driving—the key challenge lies in designing robot controllers that enable appropriate and adaptive interaction. In this talk, I will present our approach to this problem. We begin by modeling pHRI and then develop methods that infer the human user’s motor control strategies to adapt the robot’s controller accordingly. Building on this, we employ game-theoretic analysis to study situations in which both the human and the robot simultaneously identify and adapt to each other. Finally, I will highlight open research challenges and share ongoing work in this direction.