Issue 4 (August)

With the aggravation of urban water pollution, the research and application of water surface garbage cleaning robots have gradually emerged. Among them, the garbage collection mechanism, as a key module, directly affects the operational efficiency and environmental adaptability of the system. To address the complexity of garbage distribution in different water bodies, researchers have designed various types of collection mechanisms. This paper systematically reviews the current mainstream water surface garbage collection mechanisms from three perspectives: structural principles, advantages and disadvantages, and applicable scenarios. The mechanisms discussed include tipping-bucket type, conveyor-belt type, baffle type, robotic-arm sorting type, rotary-gathering type, and vortex type, which are further classified into external and embedded configurations according to their integration modes. Through comparative analysis, the technical features and adaptation trends of each mechanism are summarized. Furthermore, future research directions are proposed in terms of intelligent perception integration, multi-structure collaboration, and green energy-driven solutions, aiming to promote the intelligentization and practical development of cleaning robot systems.

With the advancement of agricultural modernization, the intelligent management of greenhouses has become crucial for improving agricultural productivity. Neural Radiance Fields (NeRF), as an emerging 3D modeling and rendering technology, offers new solutions for environmental monitoring, crop growth modeling, and pest and disease early warning in greenhouses. This paper reviews the principles of NeRF, its 3D reconstruction methods, and its application progress in greenhouses. It also explores the potential of integrating NeRF with digital twin technology and analyzes the current status of plant growth models in greenhouses. The study shows that NeRF has broad prospects for application in agriculture but still faces challenges in data acquisition, model optimization, and computational efficiency. Future research needs to further integrate multi-source data and optimize neural network models to promote the intelligent management of greenhouses.