Elsevier

Biosystems Engineering

Volume 146, June 2016, Pages 203-215
Biosystems Engineering

Special Issue: Robotic Agriculture
Research Paper
Selective spraying of grapevines for disease control using a modular agricultural robot

https://doi.org/10.1016/j.biosystemseng.2015.12.004Get rights and content

Highlights

  • A totally automatic selective spraying system for diseases in specialty crops.

  • A modular agricultural robot was configured for precision-spraying applications.

  • Multispectral imaging disease detection system integrated with manipulator control.

  • Selective spraying experiments conducted on powdery mildew disease in grapevine.

  • Pesticide use significantly reduced compared to conventional homogeneous spraying.

Due to their recognised role in causing environmental pressures, the need to reduce production costs and public concerns over the healthfulness of fresh products and food, reducing pesticide use in agriculture is a major objective. In current farming practice, pesticides are typically applied uniformly across fields, despite many pests and diseases exhibiting uneven spatial distributions and evolving around discrete foci. This is the fundamental rationale for implementing the selective targeting of pesticide applications such that pesticides are deposited only where and when they are needed and at the correct dose. This approach is explored using the example of powdery mildew on grape vines controlled by means of a modular agricultural robot developed within the EU-project CROPS. The CROPS manipulator was configured to six degrees of freedom and equipped with a new precision-spraying end-effector with an integrated disease-sensing system based on R-G-NIR multispectral imaging. The robotic system was tested on four different replicates of grapevine canopy plots (5 m in length × 1.8 m in height) prepared in a greenhouse setup by aligning potted plants exhibiting different levels of disease. The results indicate that the robot was able to automatically detect and spray from 85% to 100% of the diseased area within the canopy and to reduce the pesticide use from 65% to 85% when compared to a conventional homogeneous spraying of the canopy. This work, to the best of our knowledge, is the first using a totally automatic selective system for spraying of diseases in specialty crops.

Introduction

In recent decades, the reduction of pesticide use in agriculture has been a major objective of EU policy; it was one of the strategic themes of the 6th Environment Action Programme and the topic of a framework directive on sustainable pesticide use (2009/128/EC). Indeed, pesticides are recognised to play a major role in environmental pressure (Sabatier et al., 2014, Stehle and Schulz, 2015), agricultural production costs and public concerns about the healthfulness and wholesomeness of fresh products (Abdollahi et al., 2004, Burns et al., 2013, Rauh et al., 2012).

The objective of reducing pesticide use has been tackled through different and complementary approaches, including selection of resistant varieties, crop management techniques, crop scouting practices, application of biocides and beneficial organisms, and regular maintenance and optimal setting of spraying equipment.

In current farming practice, pesticides are typically applied uniformly to fields. However, several pests and diseases exhibit an uneven spatial distribution, with typical patch structures evolving around discrete foci (localised areas exhibiting symptoms), especially during early stages of development (Everhart et al., 2013, Spósito et al., 2008, Waggoner and Aylor, 2000).

This is the fundamental rationale for implementing selective spraying capability by means of highly automated equipment or robots. Such systems would enable the selective targeting of pesticide application only where and when it is needed, with the aim of controlling the initial foci and preventing the infection establishment and its epidemic spread to the whole field (West et al., 2003).

This approach has been explored within the EU-funded project CROPS (www.crops-robots.eu), which is aimed at developing, optimising and demonstrating a highly modular and reconfigurable robotic system for accomplishing multiple agricultural operations, including selective spraying, ripeness monitoring and selective harvesting. This system is also able to work on different specialty crops, such as grapes, sweet peppers and apples (Baur et al., 2012, Bontsema et al., 2014, Schütz et al., 2014).

The approach adopted in CROPS was clearly different from previous research on robotic agriculture, which typically relies on adaptation of non-modular, heavy standard industrial manipulators (e.g. Baeten et al., 2008, Katupitiya et al., 2005) or focuses on specific types of produce and operational tasks (e.g. Bac et al., 2014, Guo et al., 2010, Hayashi et al., 2010, Van Henten et al., 2003). A few examples of multipurpose agricultural robotic systems have nevertheless been developed and tested especially for greenhouse operations (Belforte et al., 2006, Hayashi et al., 2008, Mandow et al., 1996).

Over the past two decades, the idea of automated selective spraying (or spot spraying) has been introduced and investigated for herbicide applications (Felton and McCloy, 1992, Paice et al., 1996, Slaughter et al., 1999), and this research has led to the development of some examples of currently available commercial equipment.

Although the concept was extended to crop-disease management (Larbi et al., 2013, Li et al., 2009, Moshou et al., 2011, West et al., 2003), automated, selective spraying for diseases has not yet been developed. The reason is mainly that there have been only limited advances in automated detection systems for disease symptoms. Even if this is currently a blossoming field of research, a huge potential for improvement remains.

Sensor technologies for crop diseases have been extensively reviewed recently by Sankaran, Mishra, Ehsani, and Davis (2010), while a more focused discussion of the applications of proximal optical sensing for disease detection in arable crops can be found in West et al. (2003) and Mahlein, Oerke, Steiner, and Dehne (2012); a similar discussion for specialty crops can be found in Lee et al. (2010).

Among other case studies, grapevine is a perfect candidate crop to explore the concept of selective and targeted spraying of initial disease foci. Indeed, in current practice in viticulture, pesticide spraying is applied uniformly through the vineyard using a continuous protection approach throughout the growing season. For some of the most advanced wine-producing regions worldwide, this results in ten to fifteen or more applications per season, often conducted at high volume rates (typically 1000 l ha−1 or greater). A successful implementation of a timely detection system and the selective spraying of disease foci may have a dramatic impact on the amount of pesticide necessary to prevent an infection's establishment and its epidemic spread to the vineyard.

In this research we investigated the possibility of automatically detecting the symptoms of powdery mildew, a major fungal grapevine disease, and of selectively spraying the diseased canopy areas by means of the CROPS modular robot, equipped with a precision-spraying end-effector. The evaluations were based on deposition of spray on targets, with the implicit assumption that the treatment has contact action on the disease. The evaluation of the biological efficacy for different modes of action of the protection treatment is beyond the goal of this particular work.

The paper describes the components and the architecture of the robotic system, and details the methods and results of the experiments conducted in a demonstrating yet realistic crop scenario.

To the best of our knowledge, this is the first test conducted on totally automatic, selective spraying of diseases in specialty crops.

Section snippets

The robot system

The robot system used for the experiments was based on a CROPS manipulator configured to six degrees of freedom (DoFs); a precision-spraying end-effector and pesticide liquid-circuit; and a disease-sensing system. All the components were integrated thanks to an electronics and communication framework architecture.

Plant material preparation and canopy setup

In 2013, multiple sessions of greenhouse experiments were conducted on grapevine canopy with localised symptoms (i.e., disease foci) of powdery mildew (Erysiphe necator) to test the concept of automated, selective spraying to prevent crop diseases.

A relevant feature of powdery mildew is that the fungus colonises the adaxial (i.e. upper face) tissues of grapevine leaves by developing greyish branching filaments that possibly can expand to other green tissues, as young shoots and bunches. This

Results and discussion

The robotic system was tested on four different dates on four different grapevine canopy preparations which exhibited varying powdery mildew disease symptom levels and spreading densities within the foliage. The operative results obtained with robotic selective spraying of disease symptoms were quantitatively assessed through: a) the sensitivity of the selective treatment, i.e. the capability of covering real targets (fraction of disease area which was sprayed by the robot); b) the specificity

Conclusions

This study investigated the feasibility of automatically detecting the symptoms of powdery mildew, a major fungal grapevine disease, and of selectively spraying the diseased canopy areas.

To this aim, the CROPS modular agricultural robot was equipped with a precision-spraying end-effector and configured for this application. A disease-sensing system based on multispectral imaging and associated detection algorithm was integrated with real-time manipulator control in a ROS based communication

Acknowledgements

The CROPS project (GA-246252) was funded by the European Commission under the 7th Framework Programme within the theme “Automation and robotics for sustainable crop and forestry management”.

References (41)

  • M. Abdollahi et al.

    Pesticides and oxidative stress: a review

    Medical Science Monitor

    (2004)
  • C.W. Bac et al.

    Harvesting robots for high-value crops: state-of-the-art review and challenges ahead

    Journal of Field Robotics

    (2014)
  • J. Baeten et al.

    Autonomous fruit picking machine: a robotic apple harvester

    Tracts in Advanced Robotics

    (2008)
  • J. Baur et al.

    Design and development of a redundant modular multipurpose agricultural manipulator

    IEEE/ASME Advanced Intelligent Mechatronics (AIM)

    (2012)
  • M.C. Bélanger et al.

    Detection of powdery mildew in grapevine using remotely sensed UV-induced fluorescence

    International Journal of Remote Sensing

    (2008)
  • J. Bontsema et al.

    CROPS: high tech agricultural robots

  • C.J. Burns et al.

    Pesticide exposure and neurodevelopmental outcomes: review of the epidemiologic and animal studies

    Journal of Toxicology and Environmental Health – Part B: Critical Reviews

    (2013)
  • A. Calcante et al.

    Evaluation of “ground sensing” optical sensors for diagnosis of Plasmopara viticola on vines

    Spanish Journal of Agricultural Research

    (2012)
  • S.E. Everhart et al.

    Spatio-temporal patterns of pre-harvest brown rot epidemics within individual peach tree canopies

    European Journal of Plant Pathology

    (2013)
  • W.L. Felton et al.

    Spot spraying

    Agricultural Engineering

    (1992)
  • Cited by (0)

    View full text