I. Introduction
Single-board computers are gaining increasing popularity due to their size, with the result that they are widely used in robotics. But for the short form factor, you have to pay weak technical characteristics. Vision-based navigation is one of the crucial tasks that robotics address. Many factors are imposing practical limitations on a robot’s ability to see, learn and explore the environment. For this reason, navigation in unknown or partially unknown environments remains one of the biggest challenges in today's autonomous mobile robots implemented on single board computers [1]. This task requires a sufficiently large number of capacities, and to solve it in a single-board computer we should use all possible potential. To date, there are various ways of improving computer performance for a particular task. One of the possible solutions is to utilize of real-time (RT) systems, where compliance with specific time limits is very important, but, along with this, the overall system performance may decrease. Thus, the developer is faced with the hard choices of using real-time systems to dealing with performance or accuracy in image recognition and vision-based navigation. Objectives of this research are to study and use a real-time operating system (RTOS), Linux and their patches on a Raspberry Pi single board computer and to perform its further implementation in a small autonomous mobile robot equipped with the camera module and IR sensor.