Exploiting the adaptability of the humanoid robot iCub to personalize a physical human-robot interaction

During our daily routines, we regularly participate in intricate, adaptive, and personalized interactions with our partner, whose sensory and motor skills may diverge from our own. Conforming to our partners' perceptions involves grasping what they will experience and adjusting our actions accordingly. Adaptation is an essential competence in natural (biological) cognitive agents, observable at both behavioral and physiological levels. Therefore, adaptability stands out as a crucial requirement in developing an artificial cognitive agent, empowering it to seamlessly integrate into new environments, navigate alterations in its surroundings, and establish the groundwork for a sophisticated, human-like interaction with other agents. This thesis proposes to exploit the adaptability of the humanoid robot iCub to personalize a physical human-robot interaction. The physical interaction involves a joint action in which a participant and iCub cut together a soft object (e.g., soap bar, wax, etc.) using a flexible steel wire. They hold one end of the wire each and coordinate to optimize the result by exchanging forces through the wire. The initial scenario involves a passive fixed robot behavior, where his only task is following a predefined trajectory to cut the soft object. With these conditions, a dataset of 30 participants was already collected. The follow-up experiment aims to use the position and velocity perceived by iCub as real-time feedback to personalize the robot’s behavior during the interaction. The robot should adapt to the participant’s force strategy to optimize the effectiveness of the cutting task and to improve the interaction and coordination with the human partner. The evaluation of outcomes in the domain of physical human-robot interaction underscores the pivotal role of the robot's adaptive behavior in shaping the diverse perceptions of human partners and subsequently leading to different collaboration strategies.