European Journal of Sustainable Agroecosystems

Background: Weed management poses a major agronomic challenge in row crop production throughout the world, requiring a considerable amount of labor and chemicals. The shortage of workers in agriculture combined with the growing environmental awareness about herbicides makes it essential to explore new mechanical weed control methods.
Objective: The research evaluated how well the autonomous field robot equipped with advanced sensor fusion systems, artificial intelligence-based crop-weed identification technology, and automation in mechanical weeding was able to manage weeds in maize and soybeans.
Methods: Field trials were completed over three growing seasons (for the years 2022-2024) in three geographical locations. The vehicle with RTK-GNSS navigation, LiDAR and RGB cameras, and a machine learning algorithm was used to conduct inter-row and intra-row mechanical weed management. The performance of the robot was compared to the performance of hand weeding and conventional mechanical weeding with the help of randomized complete block design with six replications.
Results: The autonomous robotic system achieved a weed control efficiency of 87.3 % and a crop injury of only 2.1 % which was compared to 94.2 % of weed control efficiency and 3.8 % of crop injury for manual weeding, along with 76.5% of weed control efficiency and 8.3% of crop injury for conventional methods of mechanical weeding. The navigation precision stood at 3.2 cm in average with the application of RTK-GNSS correction. The operation of the robot resulted in a labor requirement diminished by 78%. Thus, the performance in terms of yield was equal to the results of manual weed control (9.2 t/ha for robots as compared to 9.4 t/ha for manual weed control; p > 0.05). The power consumption averaged 2.4 kWh/hectare while the operating costs amounted to about $87/hectare, which would lead to 34 % of the savings margin when considering the paid cost over ten years of operation.
Importance: Autonomous field robots can significantly contribute to making weed control sustainable and within the reach of labor-intensive methods applicable in row crops.