In recent years control systems have been used to alleviate the task of harvesting machinery operators. Automation allows the operator to spend more time on other tasks such as coordinating transport. Moreover, such control systems guarantee constant performance throughout the day whereas an operator gets tired.

The perfect control system anticipates on the harvest condition, just like an experienced operator would. The operator makes a visual assessment of the condition in terms of amongst others crop density in front of the machine, and acts upon it. At present automatic control systems do not have this information, which makes it hard to compete with experienced operators.

A lot of research has been performed into sensors to measure the harvest condition in front of the machine. In this paper a different approach is taken. The harvest properties in front of the machine are predicted based on earlier measurements near the current location. The technique is based on the assumption that the variation of these properties is continuous. As new measurements become available, the model of the field is updated. .