Embedded Device-Based Oil Spill Measurement System for Inland Waters

Oil pipeline ruptures frequently cause severe terrestrial ecological pollution, with secondary pollution from oil spills into inland waters being particularly prominent. Although the measurement and response systems for ocean oil spills are relatively well developed, it is difficult to directly apply these existing technologies to inland water oil spill incidents. This is mainly due to the fundamental differences between inland water bodies and the marine environment. To address the challenges in this emerging field, a measurement system for inland water oil spills was proposed and designed, which is based on the collaborative operation of a Raspberry Pi and an ARM microcontroller. The system employs a hierarchical "perception-decision layer - interaction-motion layer - control terminal" architecture. The "perception-decision layer" is built on the Raspberry Pi 4B platform. It runs a lightweight YOLOv5n deep learning model, which is optimized with TensorRT to achieve real-time oil film detection at 8 to 15 FPS. The detection results are then transmitted in real time through a heterogeneous communication link. The "interaction-motion layer" is built around a custom PCB main control board powered by an ARM microcontroller. It controls the catamaran's propulsion system and sensor data acquisition through multi-channel pulse-width modulation (PWM). A redundant communication protocol is used to ensure real-time motion control. Testing in both simulated and real-world environments demonstrated 96.6% recognition accuracy in inland water conditions, with the Raspberry Pi maintaining 91.7% model inference precision, verifying the design's reliability and adaptability in practical scenarios. And it provides a new solution for the emerging field of oil spill measurement in inland water bodies.