Autonomous Farming: Modeling and Control of Agricultural Machinery in a Unified Framework

Currently, there are significant challenges faced by the farming industry, not least of which are a reduction in the available labour workforce, and a more 'corporate' style of farming. Such factors demand an increase in farming efficiency and productivity. This paper looks forward to the not too distant future, where the realisation of autonomous farming will aid in the farming communities surviving as well as competing in the global market. In this work, the autonomous farm is seen as a complex system-of-systems, where there is necessarily a seamless integration of requirements, bringing together the areas of robotics for autonomous farming, and precision agriculture (PA), which deals with issues of agronomy. In essence, agricultural robotics uses on-farm sensing and control to actuate autonomous farm machinery with the aim of satisfying agronomy-based objectives. We initially describe a system-of-systems architecture, or unified framework, where a vital building block is the existence of two data sets used as links, or communication, between the various sub-systems. These data sets include a precision farming data set (PFDS) formed off-line before crop cultivation, containing spatially precise navigation data for any and all autonomous machinery, and a precision agriculture data set (PADS), which is a continually evolving entity consisting more of agronomy data in relation to the crop. Secondly, research undertaken in autonomous farm machinery is highlighted, where we present a foundation for the autonomous and robust control of articulated farm vehicles for real-time trajectory tracking in the presence of uncertain conditions. Preliminary results are shown, highlighting the autonomous control of vehicles for the operations of crop seeding and non-herbicidal weeding.