Остання редакція: 2025-04-23
Анотація
This paper presents the development and evaluation of an opto-electronic system for object recognition and identification in industrial robotics. The system utilizes RGB cameras combined with computer vision algorithms (OpenCV) and neural network models (YOLOv5, CNN). The proposed system was integrated with an autonomous robot to test its performance in a simulated industrial environment. The results showed high recognition accuracy (92-96%) with a processing delay of approximately 40 ms, demonstrating robustness in noisy and varying lighting conditions. The system's effectiveness in real-time object recognition highlights its potential for enhancing autonomous industrial robots' performance. Future work will focus on sensor data fusion and further optimization for resource-constrained platforms.
Ключові слова
Посилання
[1] Zhang, Z., & Xu, C. (2021). Real-Time Object Detection for Industrial Applications Using Deep Learning Techniques. International Journal of Advanced Manufacturing Technology, 113(1), 45–55.
– Explores CNN-based object detection with an emphasis on real-time performance on embedded devices.
[2] Tsai, Y.-T., & Lin, C.-C. (2021). Vision-Based Object Recognition System for Industrial Robots Using Deep Learning Techniques. Sensors, 21(5), 1576. https://doi.org/10.3390/s21051576
– Discusses the development of a deep learning-based recognition system for robotic platforms in industrial environments.
[3] Kim, H., Park, J., & Lee, S. (2022). Robust Object Recognition in Factory Environments Using YOLOv5 and Image Preprocessing. IEEE Access, 10, 120434–120445. https://doi.org/10.1109/ACCESS.2022.3215437