© TU Chemnitz

HAnD-AI

Subproject A1: From Action Perception to Joint Actions: Learning from Joint Handover Actions of Human Dyads for Robotic Actions and Human-Robot Interactions

Advancing control mechanisms for coordinated joint actions between humans and robotic agents is one of the overarching goals of this CRC. Joint actions of two individuals are extremely difficult to process as each partner’s movement is controlled separately and the actions need to be directed toward a common coordinated movement output. Human dyads are, however, often found to perform joint actions in a highly coordinated manner. Examples can be found in many sports (e.g., partner dance or tandem cycling), but also during everyday collaborative activities (e.g., handing over or conjoint lifting of objects). In joint handover tasks, both the deliverer and receiver need to coordinate their actions with respect to each other.
A smooth handover action requires

  • both partners to meet at the same position to exchange the object;
  • the deliverer to grasp the object long enough so it does not drop, but to release it early enough to ensure a smooth transfer, and
  • the receiver to plan his grasp force in advance tight enough so that the object does not drop and light enough so that fragile objects are not squashed.

Joint handover tasks occur during a variety of everyday activities and inclusion of robotic devices in joint handover tasks can be useful in many situations (e.g., assistive robots in primary care settings). Observation and understanding of the partner’s intentions and actions is therefore crucial for successful completion of a joint action. The main goal of this project is to understand how observation and anticipation of a partner’s movement facilitate joint actions in the context of handover tasks in human dyads (with special focus on elderly individuals, who may present motor difficulties). Results obtained in the planned project will provide essential information for the development of neuro-computational models for the control of robots during joint actions in human-robot dyads. Data Aquisition finished in Juli 2023.

For further Informations to SFB Hybrid Societies click here
This project is suspported by the DFG

Researchteam:
Prof. Dr. Claudia Voelcker-Rehage
PD Dr. Dieter Kutz
Dr. Julian Rudisch
Lena Kopnarski

Contact:
Lena Kopnarski

Sponsor: Deutsche Foschungsgemeinschaft