Abstract
In recent years, computer vision has emerged as a key technology for autonomous mobile robots and transportation systems, enabling their ability to perceive the environment and make real-time decisions. One of the most critical and widely demanded tasks is object detection, which is essential for navigation, collision avoidance, and the execution of mission-specific operations. This paper examines experimental studies of computer vision algorithms used for object detection in mobile robots. The focus is on the analysis and comparison of modern methods, including classical image processing algorithms and deep learning-based neural network approaches. The experimental study was conducted in a pipeline system to detect in-pipe defects, which requires high accuracy and robustness of algorithms given the limited computing resources of mobile platforms. Experiments were conducted using various object detector architectures (YOLO, SSD, Faster R-CNN) and lightweight models, such as MobileNet, designed for resource-constrained environments. Algorithm performance was evaluated based on accuracy and processing speed. The results of the study enable the identification of optimal solutions for integrating computer vision systems into mobile robots, ensuring a balance between detection quality and operational efficiency. Furthermore, the study explores the potential adaptation of these architectures for specialized scenarios, such as pipeline inspection, where detecting complex-shaped defects within confined spaces is required.