In-field and inter-field path planning for agricultural transport units

Abstract

Path planning in agricultural field operations involving cooperating machines (e.g. combine harvesters and transport units) has to satisfy both the objectives of the individual mobile unit and the team of the cooperating mobile units. Especially, the planning and execution efficiency for transport units can significantly affect the productivity of the whole system. In this paper a path planning method for transport units in agricultural operations involving in-field and inter-field transports was presented. The approach incorporated (1) the optimization criterions of time or traveled distance; (2) the generation of paths for both in-field and between fields movements of the transporting units; and (3) the adoption of restricted movements as imposed by the controlled traffic farming concept. A “Metric Map” is generated involving the creation of a geometric description of the different fields, the followed fieldwork pattern by the harvester, and the road network associated with the coupled operation. The topology of the Metric Map is then represented by a graph on where the single-source path planning problem is solved by implementing the Dijkstra’s algorithm. Based on the results provided by selected scenarios, alteration between optimality criterions provides discrepancy between solutions in the range of 2–10% indicating that identification of the appropriate criterion suited to the specific operational conditions is of significant importance. Furthermore, the low computational requirements of the planer, taking into consideration the realistic demands of the harvesting operation system indicating that it is feasible to use the planner for on-line planning efforts.

Introduction

A path planning problem is the derivation of a pre-determined continuous curve, or path, in a configuration space from the starting configuration of a mobile unit to the end configuration, without having the mobile unit colliding with any obstacles (Choset, 2005). The last decades, because of the introduction of new autonomous and robotic systems there is an increasing demand for path planning approaches and implementations. Implementations of automated path planning include indoor implementations, e.g. for indoor transportation robots in hospitals (Takahashi, Suzuki, Shitamoto, Moriguchi, & Yoshida, 2010), industrial robot manipulation (Ting, Lei, & Jar, 2002), robots moving goods in warehouses (Wurman & Andrea, 2008) and optimal trajectory planning for surface spray painting (Diao, Zeng, & Tam, 2009) and outdoor implementations, e.g., for the task of following roads and parking among other moving cars (Likhachev & Ferguson, 2009), loading/unloading of ship container crane (Huang, Liang, & Yang, 2009), shortest-path databases for parcel delivery systems (Jung, Lee, & Chun, 2006), routing Personal Rapid Transit vehicles in a rail network (Berger et al., 2011) and path planning in emergency logistics (Yuan & Wang, 2009).

In agriculture domain, however, the introduction of automated systems has been slower than in industry because of the lower adoption of new technology of farmers (Sørensen & Bochtis, 2010). In the case of area coverage path planning there have been some development of approaches (Alia et al., 2009, Bochtis and Vougioukas, 2008, Jin and Tang, 2010, Oksanen and Visala, 2009), whereas the development of path planning methods has not been enforced except for some specific custom applications. In (Linker & Blass, 2008) presented a path planner for agricultural units operating in orchards that relies on an A^∗ heuristic search of the vehicle configurations space incorporating vehicle and agricultural specific constraints. The planner computes feasible path plans, which due to the resolution of the search space requires computations times in the range of 1.5–20 s. (Johnson, Naffin, Puhalla, & Wellington, 2009) used two different approaches to path planning for autonomous tractors involved in peat moss harvesting operations, a visibility graph path planner was used in situations where few obstacles were present and accurate end position was needed and a grid based path planner was used in other cases. (Bochtis, Sørensen, & Vougioukas, 2010) presented a path planning approach for supporting units in field operations (i.e. transport units) based on an abstraction of a field as a two-dimensional grid and implanting a breadth-first search algorithm.

Regarding the later application, the path planning in field operations involving cooperating machines (e.g. combine harvesters and transport units) has an extended notion in terms of satisfying both the objectives of the individual mobile unit and the team of the cooperating mobile units. Especially, the planning and execution efficiency for a transport unit can significantly affect the productivity of the whole system. A non-optimal path of a transport unit may cause a high-capacity combine to remain idle, while waiting to unload its fully loaded grain hopper and furthermore, assuming the heavy loads carried by transport units, the planned paths may have a significant impact on soil compaction (Bochtis et al., 2010). Additional, path planning for supporting units in cooperating machinery operations is considered a complex task since it has to take into account in-field attributes (e.g. fieldwork tracks, headland passes) and inter-field configurations (e.g. rural road network, field entrance/s).

In this paper a path planning method for supporting units in agricultural operations involving inter-field transports will be presented. The approach will incorporate (1) the optimization criterions of time or traveled distance; (2) the generation of paths for both in-field and between fields movements of the transporting units; and (3) the adoption of restricted movements as imposed by the controlled traffic farming concept.