Design On Self-propelled Robot For Greenhouse Based On Machine Vision

Spraying pesticides is an effective method to control pests and weeds in greenhouse production.Traditional artificial spraying methods have the low utilization rate of pesticides,not only deposited in the soil,but also spreading in the environment.Pesticides that evaporate into the air can harm human health.Therefore,it is particularly important to use unmanned pesticide application in the production process.This paper developed an unmanned machine vision autonomous navigation pesticide application system suitable for greenhouse production.The main research contents are as follows:(1)A method for image processing of field walking roads was proposed,and a path recognition system was designed.Aiming at the problem that the navigation path is affected by the change of light,the appropriate color space is selected for the acquired image,the selection of the the clustering center and the clustering number of K-means segmentation algorithm were optimized.Through the graying processing,filtering processing,segmentation,edge processing,and linear fitting,the navigation centerline was finally determined.The tests showed that the system can effectively adapt to different lighting conditions.The average time for extracting the crop row centre line was 12.36ms.(2)A four-wheel mobile robotic application platform was built.According to the environment and requirements of the greenhouse application,the mechanical structure of the whole machine was designed and hardware selection and layout were completed.The hardware design of the driving and steering part of the mobile platform was clarified.The selection and design of image acquisition module,application module,and controller module were completed.(3)Self-propelled control system and application control system were designed.The self-propelled system used fuzzy control algorithm.When the position of the center of the center of the walking path deviates from the theoretical center position,fuzzy control was used to adjust the front wheel turning angle in real time,so that the body can advance autonomously along the crop row centre line.The application system used fuzzy turning PID control algorithm.The pressure sensor detected the water pipeline pressure in real time.By comparing the measured pressure value with the set value,the pressure deviation and the rate of change of the deviation.These were the input quantity and the duty cycle,PWM wave was the final output.Through regulating the PWM duty cycle,the pressure of the water pump was changed and the pressure regulation of the pesticide application system was realized.(4)Greenhouse tests were conducted on the self-propelled system and the spray system.Self-propelled system test results showed that the system can effectively adapt to different lighting conditions,and can accurately track the path under different light conditions and different speeds.The maximum lateral deviation was less than 5cm.The spray system test results showed that the upper,middle and lower layer coverage of plant leaf blade were 63.26%,50.89%and 75.82%,respectively,and the average number of droplets per unit area were 55,42,78/cm2,which could meet the requirement of the controlled application of mobile robots to control diseases and pests in greenhouse.