Alen DéIDIÆ -  Department of Agricultural Engineering and Physics, Wageningen Agricultural University (WAU), Institute of Agricultural and Environmental Engineering (IMAG-DLO), Wageningen, The Netherlands, Dairy Science Department, Faculty of Agriculture, Univeristy of Zagreb Svetoöimunska 25, 10000 Zagreb, Croatia

Jasmina LUKA» HAVRANEK - Dairy Science Department, Faculty of Agriculture, Univeristy of Zagreb Svetoöimunska 25, 10000 Zagreb, Croatia

For the planning of the barn layout, cow traffic and facility locations (such as: cubicles, forage lane, etc.), the farmer has to know the milking robot utilization of his production herd. Therefore, prediction of the milking robot utilization has to be done. The milking robot utilization depends on the cowís visiting pattern and capacity of the milking robot. The models used for prediction were generalized multiple regression models. Behavioural data were obtained by video observations and electronic measurements. For eleven behavioural variables used in the model from all three experiments, only two (number of cows and sum of milk yields per hour in kilograms) were statistically significant (p ≤ 0.05) and measurable on a commercial farm. A part from the milking capacity, forage feeding routine influenced utilization of the robot. Combined cow traffic used in experiments appeared to be feasible.