Researches On Autonomous Robot For Weed Control And Smart Spray System Based On Variable Rate Technology

Agriculture has been developing for thousands of years, from ancient style to modern style with giant machinery, and then to precision agriculture. With the increase of population, we are running out of natural resources. At the same time, the environment deteriorates terribly. Precision agriculture and precision forestry are the best solution to these problems. For more than 20 years, precision agriculture has been studied all over the world and has brought us many positive results, indicating benefit future. In this paper, the recent development of precision agriculture, especially, the VRT and precision chemical application, was discussed.In order to minimize the herbicide dosage and droplet drift, a direct herbicide delivering method was firstly suggested in this research. Herbicide was mopped onto the weeds surface, after the weeds were cut. Chemical experiments both in field and in greenhouse showed that about 90% herbicide could be saved using this method. If the spatial distribution was considered in field, around 96.4% herbicide could be saved.An autonomous robot for weed control was developed in this research, which consisted of a digital camera, a computer, a robotic arm, a walking system, a chemical delivering system, a robot body and etc. After weeds were identified in the field by the machine vision system, the robotic arm could reach out, cut weeds and mop chemical onto their cut surfaces.A robot end-effector with high efficiency and stability was developed. The end-effector cut the weeds and mopped chemical onto the cut surface.Experiments showed that the image processing and robot-controlling program written in VC++ could identify weeds from field background correctly and control the robotic arm precisely.Machine vision and image processing techniques have wide applications in industry, as well as in agriculture. The author developed a smart pesticide spray system for trees based on machine vision and fuzzy control. After a tree canopy and the distance between the tree and the nozzles were measured and integrated, the tree size could be judged by fuzzy reasoning and decision-making. The system could apply pesticide according to different tree sizes with different nozzles and with variable flow rate and nozzle sprayingranges. The system could be mounted on a tractor. Nozzles opened when the system identified trees and closed when there was no tree with the help of the machine vision. In this way, pesticide could be saved considerably compared to traditional spray methods. Environment is the biggest beneficiary.Experiments were done to study the characteristics of PWM nozzles. Results showed that PWM nozzles could alter flow rate while maintaining the same spray pattern. Solenoid valves with faster frequency are needed in future researches.