In order to enable a tight cooperation between humans and robots, there is the need for a perception-based coordination of both partners' actions. The scene is observed and analyzed by a depth camera. A background model is dynamically built up and the foreground objects are automatically segmented. The interaction space is defined through the activity radius of the robot and the interacting person, respectively. For the coordination of the turn-based actions of both partners, an activity model has been developed as a spatial representation of the interaction space. Based on this model, the robot grasps the object lastly "activated" by the person.
Download Factsheet PDF
Integration into Research:
Research on 3D computer vision and modelling of the interaction space is conducted in the collaborative research center 'Alignment in Communication' (CRC 673, projects A4 and C1). Aspects of service robotics are explored within CITEC.
Potential areas of application are in the fields of service robotics and flexible assistance in assembly tasks.