Navigating the Dark: Advances in Robotic Night Patrol with D-Block Mask Electric EEL Dense Nested R-CNN for Enhanced Safety

Abstract

Ensuring safety and security during nighttime operations remains a critical challenge for automated surveillance systems, often hindered by limited visibility, sensor noise, and inadequate data processing techniques. To overcome these limitations, an advanced robotic night patrol system is proposed, utilizing the D-block Mask Electric Eel Dense Nested R-CNN (D-Mask-EEDN-R-CNN). This system acquires data from multiple sensors implemented on robotic units, which saves it in a Firebase database instantly. The noise that hinders the correct interpretation of the data is eliminated and data quality improved in the pre-processing stage with the help of Improved Sage-Husa Adaptive Kalman Filter (SHAKF). The data is then passed through feature extraction and, classification by the proposed D-Mask-EEDN-R-CNN network which is fine-tuned by the E2FOP for better performance. This system has been developed in Python and the efficiency of this system, in fact, is very high, it is 99.9% accuracy and 99.8% improvements on average thus yielding better recall results than existing methods. These results highlight the effectiveness of the proposed approach in enhancing the accuracy and reliability of robotic night patrols, providing a robust solution for improving nighttime safety and security.