A flexible assembly cell is considered to be an essential building brick for flexible automation of small to medium sized order-based industries. In order to be able to respond adequately to market changes a production facility that can be reprogrammed or even reconfigured in a very short time is vital. Such a system will have to exhibit some form of intelligent behavior. By intelligent behavior we mean that this system should be able to learn from experience. The experience gained in making subsequent batches of products will be used to increase the efficiency of future batches. Unreliable strategies will be replaced by more reliable ones, actual time data will make it possible to choose a better plan. In this paper a reference model for planning, control and error management for a flexible assembly cell is presented and scheduling and a cost function for the selection of strategies for primitive operations are discussed.