Learning Whole-Body Human-Robot Haptic Interaction in Social Contexts
This work addresses the challenge of enabling robots to engage in safe and effective social interactions involving full-body contact, which is incremental as it builds on existing learning-from-demonstration frameworks with specific strategies for data sparsity.
The paper tackles the problem of teaching robots whole-body haptic social interactions, such as hugs, by addressing high dimensionality and sparsity in demonstration data, achieving a model that generalizes well to unseen inputs and partners using 121 sample hugs from 4 participants.
This paper presents a learning-from-demonstration (LfD) framework for teaching human-robot social interactions that involve whole-body haptic interaction, i.e. direct human-robot contact over the full robot body. The performance of existing LfD frameworks suffers in such interactions due to the high dimensionality and spatiotemporal sparsity of the demonstration data. We show that by leveraging this sparsity, we can reduce the data dimensionality without incurring a significant accuracy penalty, and introduce three strategies for doing so. By combining these techniques with an LfD framework for learning multimodal human-robot interactions, we can model the spatiotemporal relationship between the tactile and kinesthetic information during whole-body haptic interactions. Using a teleoperated bimanual robot equipped with 61 force sensors, we experimentally demonstrate that a model trained with 121 sample hugs from 4 participants generalizes well to unseen inputs and human partners.