The regulations to certify modern road cars require plenty of chassis dynamometer tests. In an attempt to optimize the reliance on human operation more and more automation processes are implemented, such as driving robots and on-board software control. These techniques can lack the special characteristics that define a human driving behaviour. To combat this challenge, an algorithm has been created that takes a given driving schedule and converts it into a human-like driving schedule. This human-like driving schedule is then used to calculate a human-like driving coefficient, formally named “rsmooth”. To achieve such an algorithm, first some statistical characteristics must be found that define the human-like driving behaviour. Part of the characteristics are breaking transient, acceleration transient and overlapping as well as reaction and correction times. With these characteristics an evaluation has been done to a transformative function which generates the human-like driving schedule. Finally, a database based on real human driver input been used to confirm the quality of human-like driving coefficients and compare them with the results of a driving robot.