Autopilot control unmanned aerial vehicle system for sewage defect detection using deep learning

• DOI: 10.1002/eng2.12852
• Publication Date: 2024-01-30T05:40:30Z
• Authors: Binay Kumar Pandey; Digvijay Pandey; S. K. Sahani
• Keywords: Sewerage; Autopilot

Abstract This work proposes the use of an unmanned aerial vehicle (UAV) with autopilot to identify defects present in municipal sewerage pipes. The framework also includes effective control mechanism that can direct flight path a UAV within sewer line. Both these breakthroughs have been addressed throughout this work. UAV's camera proved useful sewage inspection, providing important contextual data helped analyze line's internal condition. A plethora information for understanding inner functioning and extracting interior visual details be obtained from camera‐recorded imagery if defect is present. In case inspections, nevertheless, impact false negative significantly higher than positive. One trickiest parts procedure identifying defective pipelines negatives. order get rid outcome or positive outcome, guided image filter (GIF) implemented proposed method during pre‐processing stage. Afterwards, algorithms Gabor transform (GT) stroke width (SWT) were used obtain features UAV‐captured surveillance image. camera's then classified as “defective” “not defective” using by Weighted Naive Bayes Classifier (WNBC). Next, images lines captured are analyzed speed‐up robust (SURF) deep learning different types defects. As result, methodology achieved more favorable outcomes prior existing approaches terms following metrics: mean PSNR (71.854), MSE (0.0618), RMSE (0.2485), SSIM (98.71%), accuracy (98.372), specificity (97.837%), precision (93.296%), recall (94.255%), F1‐score (93.773%), processing time (35.43 min).