This paper addresses the complex issue of classifying mode of operation (active, idle, stationary unloading, on-the-go unloading, turning) and coordinating agricultural machinery. Agricultural machinery operators must operate within a limited time window to optimize operational efficiency and reduce costs.

Existing algorithms for classifying machinery operating modes often rely on heuristic methods. Examples include rules conditioned on machine speed, bearing angle and operational time windows. These rules may be specific to machinery operation. In this work, the authors discuss the research gap in utilizing machine learning methods and evaluate their suitability to classify machinery operating modes. A fully automatic algorithm is presented, which has been trained, validated, and evaluated on three datasets featuring different agricultural machinery and harvesting methods. The primary challenge of imbalanced datasets for less common operation modes is addressed through data balancing techniques. The study contributes to the field by exploring and implementing machine learning methods for efficient classification of agricultural machinery operation.