In robot navigation, precise knowledge of the robot’s position and orientation is essential for accurate trajectory tracking, obstacle avoidance, and goal attainment, especially in scenarios where human supervision is limited or absent. This paper describes the different established methods in simultaneous localization and mapping (SLAM) algorithms, such as the most advanced SLAM techniques for extreme environmental conditions, including dynamic objects, illumination and brightness variability. Namely, visual information received from cameras is less susceptible to radio interference and does not depend on any additional device, such as GPS and satellite signals. The SLAM community’s main approaches to solving these problems are introduced. Finally, we consider current research in the field of visual odometry (VO), as well as its practical implementation in robotics.