The field of action planning is concerned with the automatic generation of action sequences (plans) which guide the planning agent from a given initial situation to a desired goal situation. One commonly distinguishes between domain-independent and domain-specific planning systems. Domain-independent planners solve planning tasks given only a declarative description of the initial situation, possible actions, and goals. Domain-specific planners additionally require domain-specific control knowledge provided by experts, which reduces the space of possible action sequences to be considered and hence simplifies finding solutions. Domain-independent systems are more flexible than domain-specific ones, but the latter are considerably more efficient (given suitable expert knowledge) and hence indispensable in many applications. The objective of this project is to reduce the gap in scalability between domain-independent and domain-specific planning systems by devoloping automated procedures for deriving control knowledge and using it in the context of a domain-independent planning system. For this purpose existing ideas for deriving certain kinds of control knowledge (invariants, landmarks, macros) shall be integrated, generalized, and exploited within a domain-independent planning system by combining them with heuristic search algorithms.