KSEF RDE: Efficient Routing with Multiple Vehicles for Agricultural Area Coverage Tasks

Nearly every agricultural operation from planting a field with a tractor to imaging a crop from the air involves a vehicle performing some type of action while traveling a path that will result in the coverage of a given area. As the demand for field work capacity grows ever greater but concerns over soil compaction limit vehicle sizes, the coordinated use of multiple vehicles has become important. Although using multiple vehicles makes sense, the issue of efficient scheduling and division of work quickly arises. The development of algorithm to do so is complicated by the nearly limitless variety of field shapes and sizes and the changing importance of various constraints and effectiveness measures based on field operation and weather conditions. In this project we will investigate the application of different algorithms to solve the issue of area coverage in agriculture. A basic heuristic algorithm based on Clark-Wright will be compared with a more powerful, but computationally expensive meta-heuristic algorithm based on Tabu Search. The first stage of this investigation will be Monte Carlo computer simulations of the algorithms in hundreds of field shapes, varying numbers of vehicles, varying starting positions and differing operating conditions. The computer simulations will assume massless vehicles capable of holonomic steering. The second stage of this project will compare the results of the simulation to actual real world systems. Both aerial and ground vehicles will be used in this stage. A representative subset of field shapes will be used in this testing. Aerial testing will be accomplished using UAVs. The ground testing will be performed using autonomous model vehicles. Besides verification, the results will be used to determine how the various modelling simplifications contributed to errors in the model and to what extent the computer simulation models need to be adjusted.