Abstract:With increasing global environmental and economic pressures on agriculture, traditional chemical and physical weed control methods face significant challenges, such as environmental pollution and inefficiency in operations. Intelligent mechanical weeding technology has emerged as a sustainable alternative, effectively addressing these challenges. This review examined the research progress on intelligent mechanical weeding machines designed specifically for dryland environments, focusing on their working principles, key technologies, practical applications, and development status both domestically and internationally. Intelligent weeding machines increasingly utilized high-precision image recognition and advanced deep learning algorithms to achieve accurate weed identification and precise positioning. These systems used mechanical arms or other units to perform efficient and targeted weeding operations, enhancing crop yield and reducing reliance on chemical herbicides while providing substantial environmental and economic benefits. However, challenges such as variable field conditions, high equipment costs, and technical limitations hindered widespread adoption. This review also explored essential technologies in dryland intelligent mechanical weeding, including crop row detection, seedling-avoidance control mechanisms, and multi-sensor integration, emphasizing the importance of improving real-time processing and precision in weeding operations. Future directions included multi-sensor fusion, modular design, and adaptations for various crop environments to enhance the practicality and adoption of intelligent weeding technologies in agriculture.