YOLOX-based blue laser weeding robot in corn field

A novel mechanical-laser collaborative intra-row weeding prototype: structural design and optimization, weeding knife simulation and laser weeding experiment

Introduction

The competition between intra-row weeds and cultivated vegetables for nutrients is a major contributor for crop yield reduction. Compared with manual weeding, intelligent robots can improve the efficiency of weeding operations.

Methods

This study proposed a novel mechanical-laser collaborative intra-row weeding device structure. A slider-crank mechanism size optimization algorithm was proposed, and the correctness of the algorithm was verified by ADMAS software. Finally, the crank and link lengths were determined to be 87 mm and 135 mm, respectively. The resistance of triangular weeding knives with different penetration angles and edge angles in the soil was simulated and analyzed using EDEM software. The simulation results show that the triangular weeding knife with a soil penetration angle of 0 ° and an edge angle of 30 ° encountered the least resistance. In addition, weed control experiments with different powers and lasers were conducted using 200 W NIR and 200 W blue lasers. The experimental results show that the time it took for a 50 W blue laser and a 100 W NIR laser to remove small weeds was approximately between 0.3 and 0.4 s, and the time it took for a 50 W blue laser to remove larger weeds was approximately between 0.5 and 0.6 s. The time it took for 75 W and 50 W NIR lasers to remove weeds was more than 1 s.

Results

Based on the above research results, a prototype of a mechanical-laser collaborative intra-row weeding device was successfully built.

Discussion

This study provides a new idea for the field of intelligent weeding. The simulation and experimental results can provide a reference for the research and development of mechanical weeding and laser weeding equipment.

Sensing and Perception in Robotic Weeding: Innovations and Limitations for Digital Agriculture

The challenges and drawbacks of manual weeding and herbicide usage, such as inefficiency, high costs, time-consuming tasks, and environmental pollution, have led to a shift in the agricultural industry toward digital agriculture. The utilization of advanced robotic technologies in the process of weeding serves as prominent and symbolic proof of innovations under the umbrella of digital agriculture. Typically, robotic weeding consists of three primary phases: sensing, thinking, and acting. Among these stages, sensing has considerable significance, which has resulted in the development of sophisticated sensing technology. The present study specifically examines a variety of image-based sensing systems, such as RGB, NIR, spectral, and thermal cameras. Furthermore, it discusses non-imaging systems, including lasers, seed mapping, LIDAR, ToF, and ultrasonic systems. Regarding the benefits, we can highlight the reduced expenses and zero water and soil pollution. As for the obstacles, we can point out the significant initial investment, limited precision, unfavorable environmental circumstances, as well as the scarcity of professionals and subject knowledge. This study intends to address the advantages and challenges associated with each of these sensing technologies. Moreover, the technical remarks and solutions explored in this investigation provide a straightforward framework for future studies by both scholars and administrators in the context of robotic weeding.

Improving the maize crop row navigation line recognition method of YOLOX

The accurate identification of maize crop row navigation lines is crucial for the navigation of intelligent weeding machinery, yet it faces significant challenges due to lighting variations and complex environments. This study proposes an optimized version of the YOLOX-Tiny single-stage detection network model for accurately identifying maize crop row navigation lines. It incorporates adaptive illumination adjustment and multi-scale prediction to enhance dense target detection. Visual attention mechanisms, including Efficient Channel Attention and Cooperative Attention modules, are introduced to better extract maize features. A Fast Spatial Pyramid Pooling module is incorporated to improve target localization accuracy. The Coordinate Intersection over Union loss function is used to further enhance detection accuracy. Experimental results demonstrate that the improved YOLOX-Tiny model achieves an average precision of 92.2 %, with a detection time of 15.6 milliseconds. This represents a 16.4 % improvement over the original model while maintaining high accuracy. The proposed model has a reduced size of 18.6 MB, representing a 7.1 % reduction. It also incorporates the least squares method for accurately fitting crop rows. The model showcases efficiency in processing large amounts of data, achieving a comprehensive fitting time of 42 milliseconds and an average angular error of 0.59°. The improved YOLOX-Tiny model offers substantial support for the navigation of intelligent weeding machinery in practical applications, contributing to increased agricultural productivity and reduced usage of chemical herbicides.