Precision Agriculture Applied to Harvesting Operations through the Exploitation of Numerical Simulation

When it comes to harvesting operations, precision agriculture needs to consider both combine harvester technology and the precise execution of the process to eliminate harvest losses and minimize out-of-work time. This work aims to propose a complete control framework defined by a two-layer-based algorithm and a simulation environment suitable for quantitative harvest loss, time, and consumption analyses. In detail, the path-planning layer shows suitable harvesting techniques considering field boundaries and irregularities, while the path-tracking layer presents a vision-guided Stanley Lateral Controller. In order to validate the developed control framework, challenging driving scenarios were created using IPG-CarMaker software to emulate wheat harvesting operations. Results showed the effectiveness of the designed controller to follow the reference trajectory under regular field conditions with zero harvest waste and minimum out-of-work time. Whereas, in presence of harsh road irregularities, the reference trajectory should be re-planned by either selecting an alternative harvesting method or overlapping the harvester header by some distance to avoid missing crops. Quantitative and qualitative comparisons between the two harvesting techniques as well as a relationship between the level of irregularities and the required overlap will be presented. Eventually, a Driver-in-the-loop (DIL) framework is proposed as a methodology to compare human and autonomous driving.