Proposal for a Control System of a Weeding Robot Considering Terrain Surface Friction

In Japan, the agricultural sector is facing a serious decline in the workforce due to an aging and shrinking population. Combined with rising temperatures, heatstroke incidents among elderly farmers have become increasingly common, especially during weeding tasks. Weeding is typically performed twice a year, around May and early autumn, when the risk of heat-related illnesses is high. Because the task involves prolonged outdoor labor under intense sunlight, it poses significant health hazards. To address these issues, many research institutions and companies are developing remotely operated and autonomous weeding robots. While remote-controlled models are commercially available, autonomous robots capable of operating on uneven or sloped terrain remain underdeveloped. One of the main challenges is accurate self-localization, which is hindered by sensor noise caused by engine vibrations and rough terrain, as well as slippage in the drive system due to weeds. In this paper, we propose a control system for suppressing lateral slippage of a weeding robot on sloped terrain, incorporating a self-tuning fuzzy PI controller. A barometric sensor correction method is also validated, and experimental results show that the proposed method effectively reduces sensor drift compared to the uncorrected case.