Evaluation of Work Efficiency of Robot Weeding using Multiplied Dominance Ratio

Abstract

We compared weed vegetation and ground workability after autonomous robotic weeding and hand weeding on experimental plots at Tokyo University of Agriculture and Technology, Japan, over a four-month period (July to November 2019). The index used for the volume of weeds was the multiplied dominance ratio (MDR) (m³ m^-2), calculated as the product of the plant coverage ratio (m^-2 m^-2) and plant height (m). Weeds were classified by growth form: erect, branched, procumbent, tussock, rosette, and liane. After weeding, weed vegetation and ground workability were evaluated, and the difference in weed biomass between before and after weeding was quantified for each plot. Furthermore, the difference in the degree of weediness between summer and fall for a plot with various types of vegetation was quantified using the MDR. The types of vegetation and their growth amounts depended on the weeding method. Observation of the autonomous robotic weeding revealed that it was more effective for erect form weeds than branched form weeds and that it suppressed the growth of weeds. There was a positive correlation between the total MDR for each plot and the duration of robotic weeding. Furthermore, the affection to the root operating time could be quantified using the MDR of each growth form as explanatory variables. These results demonstrate that the MDR is useful for evaluating robotic weeding in a field with various forms of weeds.