Skip to Main Content
Action planning is important for autonomous systems since the capacity to find a solution for a particular problem determines the degree of system autonomy. When designing an action planning system for real world domains, hierarchical systems offer a promising way to split complex planning goals in clearly-arranged sub-goals. Although dividing problems into chunks depicts a performance adequate approach for planning, considerations how plan fragments are saved in memory must be made. Memory demands grow with the abstraction level of the planning system. A decision unit for autonomous agents forms the foundation for this work. An action planning system that was designed to fit the constraints of a bionically inspired information representation module is presented. Finally, the system is integrated into autonomous agents and evaluated within an artificial life simulation.