Sorghum halepense control in soybean with ATAR smart sprayer system

Abstract

The ATAR smart sprayer is an innovative system created by the startup company ATAR AGTEK d.o.o., which enables precise identification and treatment of weeds and can be easily integrated on any type of sprayer. The system consists of cameras that record of the field in front of the sprayer during movement, and the images are forwarded to an advanced artificial intelligence chip that analyzes them in real time and activates the necessary nozzles in order to target and effectively treat only detected weeds. In 2023 and 2024, the precision of this system in recognition and effectiveness of conrol S. halepense from rhizomes in the soybean crop was tested in field trials. Fields with a very uneven distribution of S. halepense plants and where it was estimated that its control was justified were selected for the tests. In both trials, a clethodim based product (144 g a.i. ha^-1) was applied broadcast on total area of plots, and only targeted where the presence of this weed was detected. Herbicide was applied post-emergence at stages 13-31 of BBCH scale of S. halepense and at stages 12-13 BBCH scale of soybean. In 2023, an trial was performed on large plot without repetition, while the trial in 2024 was set up according to a randomized block design with three repetitions. To confirm the effective recognition and treatment of weeds in each basic plot, 5 water-sensitive papers placed horizontaly next to the weed plants and SnapCard application was used to analyze % coverage. The most important parameter was the efficiency, that was the proportion of the number of controled weed plants 15 days after herbicide application, which was evaluated on a sample of 400 m^-2 in the middle of each basic plot. Also, for each basic plot, a map of the treated parts of the plots was created and the percentage of the treated area was calculated. Analysis of water-sensitive papers confirmed that weeds were fully recognized and treated using this innovative approach. The created treatment maps confirmed that in 2023 only 19% of the total area was treated, while in 2024 that area ranged from 29-75%. These differences depended on the number and spatial distribution of weeds, but also on the setting of the required area for treatment in front of and behind the detected weeds in order to achieve maximum reliability of the system. By analyzing the number of controled and uncontroled (newly sprouted) plants from the rhizome 15 days after the application of the herbicide in the trial in 2023, the effectiveness of S. halepense control was 92.5%. In 2024 a high and approximate efficiency was achieved with all four tested treatments, i.e. at speeds of 6 and 8 km h^-1 when treating the entire surface was 91.2 and 94.2%, and with spot spraying 90.8 and 93.5% at the same speeds. The obtained results confirmed that with this system is possible to achieve reliable control of S. halepense with big savings of herbicides and thus a reduced risk for non-target organisms and the environment.