The calibration of modern automotive diesel engines, stored in form of maps in the engine control unit, must fulfill stringent requirements in terms of smoothness, ensuring a subtle transition of control parameters between neighbor operating points. However, this could lead to penalties in emissions or fuel consumption. It is therefore necessary to develop a methodology capable of carrying out the engine calibration task in a quick and automatic way. In this paper, an original fully automated methodology for the generation of smooth calibration maps is proposed. Starting from a population of more than 80 optimized calibrations for 20 engine operating points, generated by means of a genetic algorithm-based multi objective optimizer, a final calibration was then selected in an automated way, on the basis of a trade-off between the performance of the calibration and the smoothness of maps. The results achieved clearly showed that in comparison with traditional methods similar level of smoothness can be achieved while having 5-10% lower NOx and soot emissions with an additional benefit of 1% in fuel consumption. Furthermore, the time required for the calibration task of an automotive diesel engine can be reduced by more than half. (C) 2019 Elsevier Ltd. All rights reserved.