Abstract

Weed management is one of the most urgent and unresolved challenges in specialty crops across most regions of the US. For crops like horseradish, largely cultivated in Illinois and Wisconsin, the weed issue is magnified by the crop’s long growing season and sensitivity to chemical applications. Farmers in these two states account for more than 80% of the US horseradish market. However, current weed control practices rely on expensive manual labor, which is in short supply, or repeated application of unlabeled herbicides that degrade soil and promote herbicide-resistant weed populations. A site-specific, environmentally responsible weeding solution is needed that meets the practical realities of horseradish growers. To address this problem, a mechanical weeding system was developed and deployed to identify and remove weeds in horseradish fields using a mechanical actuation system. An object detection model was developed to distinguish weeds from horseradish with an accuracy of more than 90%. This study builds on that prior work by developing a low-cost, modular, laser-based weed management system as an alternative, non-contact weed control strategy. The proposed system integrates real-time computer vision, object tracking, and precision laser targeting to selectively suppress weeds without disturbing the soil or crop plants. A compact laser module mounted on a motorized linear stage enables accurate positioning along the crop row, while inference and motion control run on an embedded GPU platform. By activating the laser only when a weed is precisely aligned with the target zone, the system minimizes energy consumption and improves operational safety. The modular design allows the system to be retrofitted onto existing cultivation platforms and adapted to different specialty crops with minimal hardware changes. This research demonstrates the potential of combining low-cost hardware with AI-based perception to provide an environmentally responsible, scalable, and economically viable alternative to conventional weed control in specialty crop production.

Keywords

Specialty crops, Horseradish, Herbicides, Mechanical weeding system, Precision agriculture, Laser weeding, Object detection model

DOI

10.5703/1288284318732

Share

COinS
 

Development of a low-cost modular laser weed management for specialty crops​

Weed management is one of the most urgent and unresolved challenges in specialty crops across most regions of the US. For crops like horseradish, largely cultivated in Illinois and Wisconsin, the weed issue is magnified by the crop’s long growing season and sensitivity to chemical applications. Farmers in these two states account for more than 80% of the US horseradish market. However, current weed control practices rely on expensive manual labor, which is in short supply, or repeated application of unlabeled herbicides that degrade soil and promote herbicide-resistant weed populations. A site-specific, environmentally responsible weeding solution is needed that meets the practical realities of horseradish growers. To address this problem, a mechanical weeding system was developed and deployed to identify and remove weeds in horseradish fields using a mechanical actuation system. An object detection model was developed to distinguish weeds from horseradish with an accuracy of more than 90%. This study builds on that prior work by developing a low-cost, modular, laser-based weed management system as an alternative, non-contact weed control strategy. The proposed system integrates real-time computer vision, object tracking, and precision laser targeting to selectively suppress weeds without disturbing the soil or crop plants. A compact laser module mounted on a motorized linear stage enables accurate positioning along the crop row, while inference and motion control run on an embedded GPU platform. By activating the laser only when a weed is precisely aligned with the target zone, the system minimizes energy consumption and improves operational safety. The modular design allows the system to be retrofitted onto existing cultivation platforms and adapted to different specialty crops with minimal hardware changes. This research demonstrates the potential of combining low-cost hardware with AI-based perception to provide an environmentally responsible, scalable, and economically viable alternative to conventional weed control in specialty crop production.