The thesis deals with the optimization of terrestrial laser scanning procedures in terms of increasing the accuracy and reliability of the results and reducing the time and cost of the procedure. The main intent of the dissertation is to calibrate the laser scanner and upgrade it with machine learning procedures. In order to carry out the calibration, we need to establish the procedure for precisely determining the target centers from the scans. We propose a robust high precision process with which we also veri�ed the accuracy of the scanner. Two calibration �elds were established and within them we determined the positions of the points with the classical geodetic methodology with the highest achievable accuracy. We developed self-calibration software with an original method for de�ning geodetic datum. Two terrestrial laser scanners were calibrated on the calibration �elds. The self-calibration results were used to determine the additional systematic errors of the measurements at which we used the classical analysis with curve adjustment as well as machine learning. Two practical examples of the application of the terrestrial laser scanning for tasks that require high accuracy measurements and results are shown. In the thermal power plant Brestanica we determined the inclination of its high chimneys. On the high barrier on the Drava River we tried to detect changes in the shape or position of the over ow wall with the use of an innovative method of statistically testing the changes of the plane parameters.