Robot Monitoring and Controlling Soybean Field Soil Condition Based On K-Nearest Neighbor Algorithm and Message Queuing Telemetry Transport Protocol

Abstract

Soybean production is decreasing every year. The level of soybean production is strongly influenced by soil moisture. The problem is that farmers let soybeans grow without adequate maintenance, including without checking the soil moisture. Therefore, an autonomous robot is built that could replace the role of farmers in caring for soybeans. This robot is built to monitor the conditions of the soybean field and classify the image of soybean field soil using the K-Nearest Neighbor algorithm. The results of soil classification are used to control the watering node for watering plants. This robot uses the Internet of Things concept with the MQTT protocol integrated with ThingsBoard as a display of monitoring information. The robot is built based on the Raspberry Pi 3 Model B+. In this research, with the KNN algorithm, the robot can classify soil moisture accurately and adequately, where it obtained 83.3% accuracy, 90% recall, 81.8% precision, and 85.7% F1 score. The watering node also performed well with a 94.4% success rate. In addition, soybeans in a field with the robot have better growth than soybeans in a field without robot. That is evidenced by the average plant height and the number of leaves in the field with the robot is better than those in the field without robot, that is 17.28 cm and 9 leaves compared to 15.72 cm and 8 leaves. However, plants without robot have a better stem diameter than those in a field with the robot, which is 2.8 mm compared to 2.74 mm.