DESIGN AND IMPLEMENTATION OF AN INTELLIGENT ROBOTIC SYSTEM USING DEEP LEARNING MODEL FOR REAL-TIME PIPELINE GAS LEAKAGE DETECTION

Pipeline gas leakage posessafety and environmental risks in the oil and gas industry which necessitates the adoption of efficient detection methods. This study presents the design and implementation of an intelligent robotic system using YOLOv8 model for real-time pipeline gas leakage detection. In the study, a comprehensive dataset of 8,881 annotated images representing different leak scenarios was used for the training and validation of the proposed model which is further enhanced with Spatial Pyramid Pooling Fast (SPPF) for improved feature extraction. The system was further integrated into an autonomous unmanned vehicle platform and evaluated through extensive simulation and experimental tests. The implementation results of the study demonstrated that it achieved high detection accuracy, with a precision of 83%, recall of 79%, and mean average precision (mAP) of 0.91, alongside rapid inference speeds enabling near real-time operation. From the results, it can be ascertained that the proposed system outperformed conventional detection methods, offering a reliable and scalable solution for enhanced pipeline monitoring.