Learning Human-like Hand Reaching for Human-Robot Handshaking

Abstract

One of the first and foremost non-verbal interactions that humans perform is a handshake. It has an impact on first impressions as touch can convey complex emotions. This makes handshaking an important skill for the repertoire of a social robot. In this paper, we present a novel framework for learning reaching behaviours for humanrobot handshaking behaviours for humanoid robots solely using third-person human-human interaction data. This is especially useful for non-backdrivable robots that cannot be taught by demonstrations via kinesthetic teaching. Our approach can be easily executed on different humanoid robots. This removes the need for re-training, which is especially tedious when training with human-interaction partners. We show this by applying the learnt behaviours on two different humanoid robots with similar degrees of freedom but different shapes and control limits.

Nonverbal social behavior generation for social robots using end-to-end learning

“…However, considering that the network learned the data seven times, this is not a large value. Moreover, the errors in the positions of the head, left hand, and right hand were 5.9 cm, 7.0 cm, and 7.1 cm, respectively, which are reasonable for social interactions (Prasad et al, 2021).…”

“…Similarly, Jonell et al (2019) presented a probabilistic, generative deep-learning architecture enabling robots to learn nonverbal behaviors from YouTube videos. Expanding on this, Prasad et al (2021) established a framework to learn handshaking behaviors solely from third-person human-human interaction data.…”

“…Although these studies have provided significant contributions to social robot behavior generation, many of these studies predominantly focus on manipulation or navigation tasks, or co-speech gestures. Several investigations have tackled nonverbal social behavior generation in robots, however, most targeted the learning of a singular social behavior (Prasad et al, 2021) or produced invalid pose sequences during long-term behavior generation (Ko et al, 2020). In this paper, we propose a Seq2Seq-based (Sutskever et al, 2014) neural network architecture capable of learning multiple nonverbal social behaviors.…”