Artikel CC BY-SA 4.0
referiert
Veröffentlicht

From traditional weed mapping to an autonomous robot: developments and results from Hungary

Zugehörigkeit
Széchenyi István University, H-9200-Mosonmagyaróvár, Vár tér. 2., Hungary
Reisinger, Péter;
Zugehörigkeit
Széchenyi István University, H-9200-Mosonmagyaróvár, Vár tér. 2., Hungary
Borsiczky, István

We are developing precision weed control technology in Hungary since 2000. From the beginning, for financial reasons, we focused our attention on the map-based technology. The post-emergence weed control method developed for winter wheat begins with weed mapping, continues with post-processing of information and ends with creation of application maps with GIS software. In the first five years, we have developed the weed mapping methodology. Based on experiments and calculations, we have found the solution in a Hungarian method based on weed coverage. The field is divided in 0.5-hectare quadrats and the weed coverage of each quadrat is estimated. The information is georeferenced with DGPS coordinates. The processing algorithm considers the coverage limits for all weed species with focus on the presence of dangerous weeds, selects the cells where chemical weed control is mandatory and the cells with low risk, where the absence of chemical protection can be allowed. At the beginning, we have used a conventional, tank mix sprayer, later we have switched to a twin tank, direct injection machine. From 2008 to 2016 the technology was tested on 1237 hectares, 38 fields, 2459 quadrates in total. We can report 51% average herbicide saving for the nine-year interval. The advantages of the developed method are costeffectiveness and safe weed detection. The disadvantage is the high expenditure of time by the weed expert because the whole field must be covered. As a further development, we have created an autonomous weed mapper in 2016. The result is a significant increase of image samples. The used office environment improves the accuracy of image processing, the identification of species and coverage estimation.

Vorschau

Zitieren

Zitierform:
Zitierform konnte nicht geladen werden.

Zugriffsstatistik

Gesamt:
Volltextzugriffe:
Metadatenansicht:
12 Monate:
Volltextzugriffe:
Metadatenansicht:

Rechte

Nutzung und Vervielfältigung: