Recognition of Manhole Cover Status for Refuse Collection Vehicles Utilizing YOLO26-ABF

Refuse collection vehicles are essential for urban sanitation. Adverse conditions, including damaged, displaced, or absent maintenance hole covers on roadways, can result in safety incidents—such as equipment jamming or overturning—during operations, consequently impacting the continuity and safety of cleaning activities. To facilitate real-time, precise identification of maintenance hole cover conditions during cleaning operations, this study proposes a detection method using an enhanced YOLO26 model. Through the development of the YOLO26-ABF detection model, which integrates multiscale feature fusion and multidimensional data augmentation techniques, we attained precise identification of five maintenance hole cover conditions—“good,” “broken,” “circle,” “loose,” and “uncovered”—resulting in a moderate rise in parameter count and computational demand. The experimental findings indicate that this method achieves substantial improvements over the baseline model on a custom dataset. It can precisely identify the condition of maintenance hole covers, providing reliable status awareness for waste collection vehicles and ensuring the safe and stable execution of cleaning activities.