Information acquisition method for mechanical intra-row weeding robot

In order to study the information acquisition method of intra-row weeding robot for transplanted vegetables, an algorithm was developed based on fast operation speed and most accredited accuracy. Green plants could be segmented from background directly using the advantage of G component in RGB color space. As the crop seedlings grow in order and have different shape and distribution from weeds, individual crop plants could be located in the robot’s moving direction through the histogram analysis of plant-pixels vertical to the crop rows. According to the experiment, when the robot moved with a speed from 0 to 3 km/h, the average error of the locating algorithm was ±5 mm and the average processing time was less than 20 ms. The developed method can meet the technical requirement of the weeding robot, and lay a foundation for the field work of weeding robot.