Design of Control System for Multifunctional Integrated Agricultural Machine

With the development of agricultural intelligence, smart agricultural robots have played a significant role in increasing production. However, weeds compete with crops for resources, reducing yields and causing economic losses. Mechanical weeding of some root crops is prone to damage, and the widespread use of herbicides can lead to environmental issues. Reducing herbicide use while maintaining production has become a major challenge, and precision weeding is an effective method.This research focuses on the control system of an efficient agricultural weed killer, utilizing the Raspberry Pi platform and innovatively employing the YOLOV5 recognition algorithm to distinguish between crops and weeds. By adjusting the neural network structure of the algorithm and optimizing the training set, the recognition speed and accuracy are enhanced, enabling precise identification and output of weed coordinates to avoid damaging seedlings. The control system employs Bézier curve discretization for motion trajectory, sending instructions to the Delta parallel arm control system through a serial port to drive the end effector for precise weed removal.