Design of an unmanned boat for waste collection and an automatic sorting device based on image recognition

This paper presents an intelligent control system that leverages image recognition technology to automate target detection and classification processes. The proposed system employs a multi-device collaborative framework integrating unmanned aerial vehicles (UAVs), unmanned vessels, and robotic arm to realize a unified operational workflow. The YOLOv5 target detection algorithm is adopted and optimized through model pruning and quantization for deployment on the Shannon Orange Pi embedded platform, significantly enhancing accuracy and real-time performance for small target detection. The UAV achieves target localization using a Global Positioning System (GPS) and LoRa modules, while the unmanned vessel is equipped with multi-sensor arrays to facilitate environmental sensing and target retrieval, and the robotic arm integrates the OpenMV platform to perform automatic classification. Experimental results demonstrate that the optimized YOLOv5 model on the Shannon Orange Pi platform improves inference speed by a factor of three compared to the original model while reducing the model size to 40% of its initial volume. Furthermore, the multi-device collaborative approach achieves a low communication latency of approximately 120 milliseconds through the LoRa protocol. This paper proposes a reusable technical framework for real-time target detection, lightweight deployment on resource-constrained platforms, and collaborative control of heterogeneous devices in edge computing scenarios. Validation experiments confirm that the system attains enhanced detection efficiency on embedded platforms, with stable multi-device collaboration latency maintained within 120 milliseconds, thereby providing a valuable technical reference for intelligent control system design in industrial inspection and smart agriculture.