This paper presents the previous experience in the licensing process of the examiners at the driving test in the Republic of Serbia, with an analysis of the way in which the examiners assess the polygonal actions. The analysis covers the most common mistakes that have been registered with the examiners as well as in the candidates in the training. The main groups of application of this model are only candidates for the training for passing a driving test. The material is a practical part of taking the driver's exam. Based on the database of errors in the work of the examiners and the candidates themselves, a database of electronic knowledge is modeled - (video material with comments) based on which the training model of the training candidate will be formed, based on frequent and observed errors. Further analysis of the database will be updated continuously in everyday work. Forming a database of errors is the first step in the model. At the same time, the selection of candidates with observed errors in the practical part of the training is formed. The formed target group independently examines the video material (as well as the examiner) evaluates its and the practical content of other candidates. A comparative analysis compares the observed errors of examiners and candidates. The collected data is analyzed and interpreted with the aim of identifying frequent errors both in training and in the practical exam itself. E content of the new training in the second cycle is enriched by documenting (video material with comments on past mistakes). After a short cycle of do-training with the application of the model, the newly acquired knowledge is checked by assessing the knowledge test based on the review of the simulated video with errors. The aim of this paper is to create a model of e-learning based on mistakes, which will annoy the creation of mistakes that a candidate creates in taking the practical part of the driving test. We believe that the application of this model in constant work would create conditions for improving the quality of training, as well as easier learning model with the candidate. The result would be excellent quality training and a quality driver in traffic. [For the full proceedings, see ED621557.]