The last ten years have led to significant advances in action planning. However, the impact of these advances has been rather limited in the area of manipulating spatial entities as it is necessary in a robotics context. In project R7-[PlanSpace], we address this shortcoming by integrating symbolic and spatial planning. In particular, we account for cases where spatial planning influences the symbolic planning process due to execution failures and optimization criteria.

Many interactions between objects and the world cannot be handled on a purely geometric level. Therefore, we integrate physical properties in the planner in addition to a geometric representation.