FSC VIM has developed the conceptual basis for creating an automated potato harvester with a digital system for identifying soil lumps and separating them from marketable products. Existing machines carry out the technological process in conditions of high soil moisture, which negatively affects the quality of harvesting as a result of a decrease in the completeness of separation. To determine the optimal values of the developed separating system and recommendations for subsequent changes in the design and technological parameters of the machines, a design diagram of an automated harvester is presented, a schematic diagram of modeling a system for identifying soil lumps in the Matlab/Simulink environment, data selection from field studies of the operation quality of the harvesting machine separating system, as well as the work of the neural network system and the neuron error surface for identifying soil lumps while moving along the separating surface are shown.