Improved AI Planning for Cooperating Teams of Humans and Robots

Each human usually plans his or her future activities in advance, trying to select those actions and to sort them in such away that as many goals as possible are accomplished. The real world is dynamic and some elements are characterized by a high degree of uncertainty. Nonetheless, the humans manage to adapt their plan fast and easy to the changes and problems that occur, being able to recover or reorganize the activities such that the goals can be further reached.Considering the same requirements for adaptability, the automated task planning for systems with autonomous robots was introduced. However, the existing task planning approaches do not satisfy well enough the requirements from use-cases with cooperating humans and robots, where beside the adaptability, the computation time and the modelling possibilities are of high importance.Considering exactly the requirements from such scenarios,in this work, the automated task planning is integrated in a Three-Level Planning strategy. This strategy has enriched modelling capabilities and ensures a high adaptability to unforeseen situations. Moreover, together with the Replanning approach, with its parallel planning and dispatching features,it results in qualitative plans that can be sent for execution in a short period of time. The implemented methods were validated in a realistic simulation of an industrial environment.