Context-aware human-robot collaboration as a basis for future cognitive factories

Today, manufacturing industry is faced with increasing product variants while product cycles decrease. Since conventional optimization strategies are saturating, future competitiveness of manufacturers will be highly related to their ability and flexibility to adapt to these essential market requirements. The collaboration between human and robot has been announced as a promising approach to solve these challenges, because it teams the strength and the efficiency of robots with the high degree of dexterity and the cognitive capabilities of humans into a flexible overall system. This thesis identifies required functionality and mechanisms for both efficient and natural human-robot collaboration. To give the human an appropriate and efficient collaboration partner, the work of this thesis bases on recent psychological research about cognitive processes of joint-action among humans. These psychological studies interpret, that the success of efficient collaboration depends on the team-partners’ abilities to predict actions and to affect own actions using the prediction. Inspired by these findings, this thesis implements and evaluates the required collaboration mechanisms action observation and action coordination for an assistive robotic system. These two mechanisms imply, that the robotic system is capable of perceiving, gathering and representing contextual information such as the environment, recognizing current actions of the human, and by this producing context-aware dynamic and adaptive actions. These demanded capabilities are conceptually realized in a distributed software framework. Human robot collaboration experiments were performed using the integrated demonstration platform JAHIR, that was co-developed along with this work. The results validate the benefits regarding efficiency, adaptability, and flexibility.