This paper describe the study for assessing the accuracy of positioning and orientation for a developed visual positioning system, based on the already known methods. The main restrictions for the use of vision systems in the automatic control cycle, including with an increase in the number of observed objects, are revealed. The accuracy of positioning was estimated on the basis of AR tags and color blobs. The authors presented its hybrid marker for speed detection comparable with known analogues. A prototype of the developed mobile robot and the principle of automatic control for motion without collisions in the environment with moving objects are demonstrated. A full cycle of the developed control system with estimates of time delays at each stage is described. As a result, the most acceptable configuration of hardware and software for the task was selected.