Elsevier

Biosystems Engineering

Volume 110, Issue 3, November 2011, Pages 247-252
Biosystems Engineering

Research Paper
Design and testing of an intra-row mechanical weeding machine for corn

https://doi.org/10.1016/j.biosystemseng.2011.07.007Get rights and content

As an alternative to chemical weed control, mechanical weed control between crop rows can be achieved using standard tools such as field cultivators. This paper addresses the related problem of achieving mechanical intra-row weed control in maize. The object was to non-specifically remove weed plants within the row by enabling dual tine carriers to engage the soil whilst circumventing the maize stalks. The maize stalks were distinguished from the weeds and maize leaves by utilising 1) the typical vertical quasi-cylindrical stalk of the maize plant, 2) the limited range of maize stalk diameters, and 3) by assuming constant plant spacing.

To assess the performance of the machine, a video was taken during field plot experiments. This allowed determination of the number of plants that were “fatally damaged” after inadvertently being pushed over by the implement. This was assumed to cause the plant to die, or “minimally damaged” where the implement merely touched the plant, when the plant was assumed to survive. Experiments were carried out using three arrangements being 1) three rows without weeds, 2) three rows with broadleaf weeds (simulated by planting soybean) and 3) three rows with grassy weeds. The percentage of plants that were fatally damaged was 8.8%, 23.7%, and 23.7% in cases 1, 2, 3 respectively. In addition, the percentage of plants that were minimally damaged was 17.6%, 20%, and 25.9% in cases 1, 2,3 respectively.

Graphical abstract

Highlights

► Intra-row weeding can be achieved using four laser beams as maize stalk detectors. ► A lateral quasi-sinusoidal motion allowed circumventing maize stalks in rows. ► In the absence of weeds, 8.8% of the plants were fatally damaged. ► Under heavy weed cover, up to 23.7% of plants were fatally damaged.

Introduction

Two recent trends in agriculture have encouraged the development of alternative weed control mechanisms to chemical application: Firstly, several weed species have shown resistance to glyphosate. Despite previous claims that target-site resistance to glyphosate will not occur (Bradshaw, Padgette, Kimball, & Wells, 1997), biotypes of goosegrass (Eleusineindica) (Baerson et al., 2002, Lee and Ngim, 2000) and horseweed (Conyzacanadensis) (VanGessel, 2001) have been reported with target-site resistance to glyphosate. Rigid ryegrass (Loliumrigidum) and Italian ryegrass (L. multiflorum) biotypes also have been identified with glyphosate resistance, although in these cases resistance occurred by unknown mechanisms (Powles et al., 1998, Heap, 2011). Also, Palmer amaranth (Amaranthuspalmeri) biotypes in Georgia have been identified that demonstrate resistance to glyphosate (Culpepper et al. 2006). As a close relative of waterhemp, Palmer amaranth is considered the most aggressive and competitive member of the pigweed complex. The term glyphosate-resistant now also has been attached to waterhemp.

The second trend is the growth of organic farming. In the US state of Wisconsin, for instance, organic farming has increased more than 90% over the past five years (NASS, 2007). It is therefore important to develop techniques and practices to increase its productivity. In particular, the availability and cost of labour for weed control are limiting its progress, and therefore the development of suitable mechanised weeding methods is an imperative.

In earlier research, Wisserodt et al., (1999) developed a ‘cycloid hoe’ which consists of rotary disc-mounted tines that could be extended and retracted resulting in a combined cycloidal motion. This ingenious mechanism did however require precise lateral placement of the implement above the row, which was later addressed in research by Nørremark, Griepentrog, Nielsen, and Søgaard (2008). Earlier research by Bontsema, Grift, and Pleijsier (1991) used infrared sensors to identify the location of weeds within a row in sugar beet. One of the challenges in this research was that the weed and sugar beet plants had similar morphological features. In maize the problem is simpler. Maize has a vertically oriented cylindrical stalk of considerable size that can be more easily distinguished from common broadleaf and grassy weeds that grow nearby since they typically do not exhibit this feature. The method, as developed here, is non-weed specific. It is based on determining the location of the maize stalks, whilst considering every other plant in the row as a weed. The challenge now lies in producing a machine that can mechanically remove the weeds between standing plants in the row, whilst causing minimal damage to the crop.

The objective of this research was therefore to develop a system for identifying maize stalk locations, allowing a soil engaging tool to mechanically remove weeds in the row, whilst circumventing the maize plants.

Section snippets

Materials and methods

The mechanical weeding machine as developed consisted of three subsystems: stalk sensing, a control algorithm and a mechanical weeding mechanism. These three segments will be discussed along with the experimental methodology.

Results and discussion

How effective the implement was in weeding could not be judged, since the tines did not contact the soil. Instead, the accuracy of the implement in circumventing the maize stalks under various weed cover scenarios was assessed. A video of the end effector movement around the plants showed that the path through which the end effectors moved was close to the theoretical path as shown in Fig. 4.

Two criteria for determining the effectiveness of the weeding system were set. Firstly, when the end

Conclusions

A mechanical weed control machine containing a sensing arrangement, control algorithm and dual mechanical end effectors was successfully developed and tested. The overall functionality of the machine was proven, but the percentage of fatally damaged plants was 8.8% in the absence of weeds and reaching 23.7% in heavy weed infested areas with hundreds of weeds per m2. More work is needed to identify the causes of misjudging the locations of the corn plants. In addition, for the machine to be

References (11)

  • M. Nørremark et al.

    The development and assessment of the accuracy of an autonomous GPS-based system for intra-row mechanical weed control in row crops

    Biosystems Engineering

    (2008)
  • S.R. Baerson et al.

    Glyphosate-resistant goosegrass. Identification of a mutation in the target enzyme 5-enolpyruvylshikimate-3-phosphate synthase

    Plant Physiology

    (2002)
  • Bontsema, J., Grift, T. E., & Pleijsier, K. (1991). Mechanical weed control in sugar beet growing: detection of a plant...
  • L.D. Bradshaw et al.

    Perspectives on glyphosate resistance

    Weed Technology

    (1997)
  • S.A. Culpepper et al.

    Glyphosate-resistant Palmer amaranth (Amaranthuspalmeri) confirmed in Georgia

    Weed Science

    (2006)
There are more references available in the full text version of this article.

Cited by (57)

  • Development and evaluation of a seed position mapping system

    2021, Computers and Electronics in Agriculture
  • Sensor-based mechanical weed control: Present state and prospects

    2020, Computers and Electronics in Agriculture
  • Deep localization model for intra-row crop detection in paddy field

    2020, Computers and Electronics in Agriculture
View all citing articles on Scopus
View full text