Investigating Human-Human Approach and Hand-Over

Abstract

Humans interact safely, effortlessly, and intuitively with each other. An efficient robot assistant should thus be able to interact in the same way. This requires not only that the robot can react appropriately to human behaviour, but also that robotic behaviour can be understood intuitively by the human partners. The latter can be achieved by the robot mimicking certain aspects of human behaviour so that the human partner can more easily infer the intentions of the robot. Here we investigate a simple interaction scenario, approach and hand-over, to gain better understanding of the behavioural patterns in human-human interactions. In our experiment, one human subject, holding an object, approached another subject with the goal to hand over the object. Head and object positions were measured with a motion tracking device to analyse the behaviour of the approaching human. Interaction indicated by lifting the object in order to prepare for hand-over started approximately 1.2 s before the actual hand-over. Interpersonal distance varied considerably between subjects with an average of 1.16 m. To test whether the behavioural patterns observed depended on two humans being present, we replaced in a second experiment the receiving subject with a table. We found that the behaviour of the transferring subject was very similar in both scenarios. Thus, the presence of the receiving subject plays a minor role in determining parameters such as start of interaction or interaction distance. We aim to implement and test the parameters derived experimentally in a robotic assistant to improve and facilitate human-robot interaction.

Towards Seamless Human-Robot Handovers

“…We observed very consistent behavior patterns in givers that replicate the conclusion in (Basili et al, 2009), that handovers have consistency. The results of the study show three main activities that happen during human-human handovers: 1) carrying, 2) coordinating, and 3) object transfer.…”

“…(Huber, Rickert, Knoll, Brandt, & Glasauer, 2008) analyzed the efficiency of handovers in terms of the durations of three phases during a handover, and compares human-human handovers with robot-human handovers. (Basili, Huber, Brandt, Hirche, & Glasauer, 2009) analyzed human approach and handover and observed a preparatory movement of lifting the object before the handover, which might play an important role in signaling the timing of the handover.…”

Inferring the goal of an approaching agent: A human-robot study

“…Gaze provides the information needed for performing a task and is therefore directed to the relevant objects or goals before the action itself takes place. In a previous investigation on human-human interaction, we could show that an approach followed by a handover is executed in smoothly by blending the two components into each other rather than executing them serially [20]. This efficiency can be attributed to the ability to mutually predict the actions of each other with high accuracy and thus to be able to rely on these predictions.…”

B-PaDY: robot co-worker in a bumper assembly line

“…Then they applied the desired handover timing, robot configuration, and the approach based on the analysis results to the handover task between the worker and the robot. Basili et al [17] focused on the transfer part in the handover between the humans. They performed a numerical analysis of the handover including the distance and velocity.…”