Selective spraying of grapevine’s diseases by a modular agricultural robot

In current viticulture protection of grapevine is obtained with uniform distribution of fungicides, typically repeated according a regular calendar. This continuous protection approach can easily result in ten to fifteen treatments per season in vineyards of several wine-producing regions. Primary infections exhibit nevertheless discrete foci, with uneven spatial distribution. Hence it can be argued that detection of symptoms at early disease stages and their targeted treatment would reduce the spread of the infection to wider patches in the vineyard, while enabling reduced use of pesticides. Within the UE-project CROPS, a modular and multifunctional agricultural robot system for specialty crops is being developed and one of the tasks that has to accomplished is selective spraying of diseases. The robotic system setup integrates a six degrees of freedom manipulator, an optical sensor system and a precision spraying actuator. After a brief description of the requirements of the system, this contribution gives a detailed description of its components and discusses the results obtained in first experiments. As case study we consider here the automatic detection and selective spraying of grapevine canopy areas exhibiting symptoms of powdery mildew (Erysiphe necator), one of the major diseases for this crop. Based on optical sensing feedback, the precision spraying actuator is positioned by the manipulator to selectively and accurately apply pesticides solely to infected areas. Disease foci identification and localization is based on on-the-go processing of images sensed by a multispectral camera inspecting the vertical structure of the grapevine canopy. At the end of the manipulator arm is located the precision spraying actuator, constituted by an axial fan with a flow straightener and an axially mounted spraying nozzle. The sprayer can deliver an air-carrier flow with an adjustable velocity, producing a circular spraying pattern of a constant diameter of 0.15 m over a wide range of spraying distances. A first experiment was conducted in an experimental greenhouse, where vineyard canopy conditions were recreated by aligning plants of grapevine grown in pots. Within the recreated canopy, diseased plants with different levels of disease symptoms were used as targets of automated selective spraying performed by the agricultural robot. The results of these experiments are discussed in view of a possible intelligent, close precision crop protection framework.