Design of Multifunctional Seedbed Planting Robot Based on MobileNetV2-SSD

Abstract

As a new generation of intelligent agricultural machinery, agricultural robots have become a research hotspot in agricultural equipment technology. Deep learning and Internet of Things technologies are gradually being applied to agricultural robots to optimize the production process and improve production efficiency. Currently, most of the execution systems and sensors of agricultural robots are installed on movable platforms. This arrangement makes the function of the robot relatively single, the manufacturing cost is high, and it is difficult to promote. For large-scale production facilities, there is currently a lack of modular, intelligent, and networked robotic production equipment. In response to this problem, this paper designs a modularized CNC seedbed planting management robot. For large-scale production facilities, there is currently a lack of modular, intelligent, and networked robots. In response to this problem, this paper designs a seedbed planting management robot with a structure similar to a gantry milling machine. The robot realizes a variety of basic functions such as seeding, irrigation, weeding, and light supplementation, and realizes the Internet of Things and autonomous operation functions based on the embedded system. The robot realizes basic functions such as seeding, irrigation, weeding, and light supplementation, and realizes the Internet of Things and autonomous operation functions based on the embedded system. In terms of weed recognition algorithm, through the application of MobileNetV2-SSD network and color threshold segmentation algorithm, the accuracy of robot weed recognition reaches 94%. The effectiveness of the robot as a production unit of a large- scale agricultural production facility is proved, and an open- source design scheme is given at the end of the article.