Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
COORDINATION OF HARVESTING AND TRANSPORT UNITS FOR AREA COVERAGE
Document Type and Number:
WIPO Patent Application WO/2018/185522
Kind Code:
A1
Abstract:
Methods are presented for the coordination of harvesting and transport units for area coverage, comprising at least a working area, a harvesting unit, a transport unit and a depot for the storage of the harvested good. One of the methods includes grouping of all available harvesting units; assigning groups of transport units to said groups of harvesting units; assigning an area coverage path plan to all harvesting units; each transport unit during each transport tour serving sequentially all harvesting units of its assigned group of harvesting units, whereby a transport tour of a transport unit comprises leaving the depot with an empty storage tank and returning with a storage tank filled up to its capacity limit; thereby sequentially transfering harvested crop from the storage tanks of said group of harvesting units to the storage tank of said transport unit while all harvesting units of said group remain at zero velocity during said transfer process until the storage tanks of all harvesting units of said group are emptied; thereby enabling cyclic scheduling of transport units.

Inventors:
GRAF PLESSEN MOGENS (CH)
Application Number:
PCT/IB2017/051899
Publication Date:
October 11, 2018
Filing Date:
April 04, 2017
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
GRAF PLESSEN MOGENS (CH)
International Classes:
G06Q50/02; G06Q10/04
Foreign References:
DE102004027242A12005-12-22
US20070135190A12007-06-14
EP0821296A21998-01-28
US20120310691A12012-12-06
Other References:
D. BERTSEKAS: "Dynamic programming and optimal control.", vol. 1, 1995, BELMONT, MA: ATHENA SCIENTIFIC
Download PDF:
Claims:
Claims

1. A method for the coordination of harvesting and transport units for area coverage, wherein the method comprises

- receiving working area defining data including at least the working area boundary coordinates;

- determining an area coverage path plan for a given set of available harvesting units that have a storage tank of limited capacity for the storage of harvested crop and an unloading auger for the transfer of harvested crop to a transport unit;

- grouping of said set of available harvesting units according to a heuristic criterion;

- assigning a number of transport units to each of said group of harvesting units, whereby said transport units have a storage tank of limited capacity for the storage of harvested crop;

- determining a cyclic scheduling plan for each of said transport units whereby each transport unit is:

- shuttling between its assigned group of harvesting units for loading of harvested crop and a stationary or mobile depot for unloading of harvested crop;

- serving all harvesting units of its assigned group of harvesting units on each of the cyclically scheduled paths between loading and unloading locations;

- thereby being filled up to its storage capacity limit on each of the cyclically scheduled paths between loading and unloading locations;

and whereby each harvesting unit is

- driving in close formation with all harvesting units of its assigned group;

- stopping upon reaching a specified storage tank fill level which is less than its capacity limit and which is assigned depending on the maximal storage capacity of the transport unit next assigned to said group of harvesting units for the transfer of harvested crop;

- transfering harvested crop to said transport unit while being stopped;

- remaining stopped until all harvesting units of said group of harvesting units having transfered their harvested crop to said transport unit; before continuing harvesting and following the cyclic scheduling plan until said working area is entirely harvested;

- applying said cyclic scheduling plan in real-time operation during harvest. 2. The method of Claim 1, whereby the specified storage tank fill level assigned to each harvesting unit of each group of harvesting units is such that the sum of said storage tank fill levels of all harvesting units of the same group of harvesting units is equal to the storage tank capacity limit of the transport unit cyclically scheduled to next transfer harvested crop from said group of harvesting units. 3. The method of Claim 1, whereby the area coverage path plan for said set of harvesting units is determined independently from the cyclic scheduling of transport units, and whereby it is distinguished between two general types of area coverage patterns comprising:

- eroded headland traversals only whereby said headlands are generated by making a copy of the area-contour and geometrically translating this copy in parallel to the interior of the working area, thereby obtaining an eroded (mathematical morphological operation) area being smaller than said area, and whereby these headlands are selected to permit coverage of the entire working area during harvest;

- at least one eroded headland traversal and additional traversals along straight lanes that are assigned to the working area for area coverage during harvest. 4. The method of Claim 1, whereby the first headland traversal, which is common to both two general types of area coverage patterns according to Claim 3, requiring a specific maneuver of all harvesting units of the highest prioritized group of harvesting units, involving backwards driving for a short spacing while maintaining the priority ordering of corresponding harvesting units, and to be performed in order to make space for serving transport units to align for crop transfer. 5. The method of Claim 1 for an area coverage pattern involving straight lanes according to Claim 3, whereby transitions between any two straight lanes are performed in a manner to maintain initially assigned priority orderings of harvesting units within each group of harvesting units, to maintain a constant turning spacing for each said harvesting unit and to ensure each harvesting unit being enabled to visually align and harvest along an edge indicating the border between already harvested and not yet harvested

6. The method of Claim 1 and for an area coverage pattern involving straight lanes according to Claim 3, whereby requiring a specific maneuver of all harvesting units of a group of harvesting units when harvesting along a straight lane on which the constraint operation of unloading augers of said group of harvesting units requiring such a maneuver in order to make space for a serving transport unit to align for crop transfer without the repression of not yet harvested crop, and whereby said maneuver must be performed because of the group of harvesting units reaching its assigned target storage tank fill levels according to Claim 1, and whereby said maneuver characteristically requiring backwards driving for a short spacing while maintaining the priority ordering of corresponding harvesting units.

7. The method of Claim 1 and for an area coverage pattern involving eroded headland lanes according to Claim 3, whereby the field contour is reactangular, thereby permitting a spiral-like area coverage path plan for harvesting units, and involving backwards driving of harvesting units of said groups of harvesting units in order to perform rectangular turns without the repression of crop, and maintaining close formation for each group of harvesting units.

8. The method of Claim 1, whereby any transport unit is serving its assigned group of harvesting units in inverse order of priorities of harvesting units within said group of harvesting units such that the least prioritized and the highest prioritized harvesting unit are served first and last, respectively.

9. The method of Claim 1, whereby the transportation of harvested crop from harvesting units to a stationary depot is partitioned into in-field and out-field transportation, requiring transport units for transportation within and outside the working area, respectively; and whereby the transportation units operating outside the working cyclically scheduled to not constrain the operation of in-field transportation.

10. A method for the coordination of harvesting and transport units for area coverage, wherein the method comprises:

- receiving working area defining data including at least the working area boundary coordinates;

- determining an area coverage path plan for a given set of available harvesting units that have a storage tank of limited capacity for the storage of harvested crop and an unloading auger for the transfer of harvested crop to a transport unit;

- receiving fill levels of storage tanks and position coordinates of harvesting and transport units;

- real-time matching of harvesting and transport units based on a heuristic optimization criterion;

- in-field navigation planning subject to the constraint of transport units not repressing not yet harvested crop area.

- shuttling of transport units between harvesting units for loading of harvested crop and a stationary or mobile depot for unloading of harvested crop.

11. The method of Claim 10, whereby said heuristic optimization criterion is trading- off, firstly, forecasted traveling times of a transport unit to potential harvesting units, and, secondly, forecasted fill levels of the storage tanks of said harvesting units at the forecasted times of arrival.

12. The method of Claim 10, whereby any of the employed transport units is according to an optimization criterion either

- visiting as many harvesting units as required until the strorage tank of said transport unit is filled up to its capacity limit, thereby requiring multiple real-time matchings of said transport unit with suitable additional harvesting units according to Claim 10 and Claim 11 and in- field navigation path planning for the transition between harvesting units, before returning to a mobile or stationary depot for unloading, or

- visiting at least one harvesting unit but returning to a mobile or stationary depot for unloading with a storage tank filled up to a level less than its capacity limit.

13. The method of Claim 12, whereby said optimization criterion is trading-off, firstly, forecasted traveling times of said transport unit to said suitable additional harvesting units, and, secondly, corresponding forecasted fill levels of storage tanks of said harvesting units at the forecasted times of arrival.

14. The method of any one of Claims 1 to 13, whereby transport units are unmanned aerial vehicles or ground vehicles.

15. A device for performing the method according to one of Claims 1 to 9 and Claim 14, wherein the device comprises input means for providing the parameters, means for reading of geographic location measurements returned from a location or positioning sensor, means for reading fill levels of storage tanks and for choosing heuristics and wherein the device comprises:

- display means for displaying data enabling the harvesting units and/or users of the harvesting units to follow a specified area coverage path plan and/or

- output means for delivering data enabling the harvesting units and/or users of the harvesting units to follow the area coverage path plan and/or

- display means for displaying data enabling the harvesting units and/or users of the harvesting units to stop once a specified target storage tank fill level is reached and/or

- output means for delivering data enabling the harvesting units and/or users of the harvesting units to stop once a specified target storage tank fill level is reached and/or

- display means for displaying data enabling the transport units and/or users of the transport units to follow a dynamically specified path to a next assigned location, whereby this location can be an unloading point or a loading point.

- output means for delivering data enabling the transport units and/or users of the transport units to follow a dynamically specified path to a next assigned location, whereby this location can be an unloading point or a loading point.

16. A device for performing the method according to one of Claims 10 to 14, wherein the device comprises input means for providing the parameters, means for reading of geographic location measurements returned from a location or positioning sensor, means for reading fill levels of storage tanks and for choosing heuristics and wherein the device comprises:

- processing means for processing data enabling the solution of the matching problem of assigning transport units to harvesting units for the transfer of harvested crop from the storage tanks of harvesting units to a mobile or stationary depot; display means for displaying data enabling the harvesting units and/or users of the harvesting units to follow a specified area coverage path plan and/or output means for delivering data enabling the harvesting units and/or users of the harvesting units to follow the area coverage path plan and/or display means for displaying data enabling the transport units and/or users of the transport units to follow a dynamically specified path to a next assigned location, whereby this location can be an unloading point or a loading point.

output means for delivering data enabling the transport units and/or users of the transport units to follow a dynamically specified path to a next assigned location, whereby this location can be an unloading point or a loading point.

Description:
COORDINATION OF HARVESTING AND TRANSPORT UNITS

FOR AREA COVERAGE

TECHNICAL FIELD

[0001] This specification relates to planning and operation of the coordination of harvesting and transport units for coverage of a specified closed working area. In an agricultural setting, such areas correspond to fields growing crops or other vegetation.

BACKGROUND

[0002] In an agricultural setting, the harvesting time of crops is the time of the yearly agricultural work cycle typically requiring activation of both maximal machinery power and maximal human resource power during a short period of time. In practice, weather- related limited time- windows for optimal harvesting conditions and insufficient harvest planning often make the harvesting process a stressful experience for human resources involved. Additional human resources hired, specifically only for the duration of the harverst, often are not familiar with the local area, path network and topology of fields to be harvested. Additional machinery, leased or purchased for the harvesting process, often may not be needed in case of optimized harvest planning and operation of already available machinery. Therefore, there is a need for simple structured methodologies for harvest planning, as well as supporting systems for their implementation such as navigation guidance and scheduling methods that permit improved harvest planning and operation. SUMMARY

[0003] This specification relates to two integrated methods for the planning and operation of the coordination of harvesting and transport units for area coverage. General objective is the time-optimal area coverage. Assumed are a given number of harvesting units.

[0004] According to both of these methods, a layered optimization is conducted, comprising: inputting a given area specifying data including at least area boundary coordinates and boundary coordinates of area parts prohibited from trespassing by harvesting and transport units, for example, tree islands within an agricultural field; determining as the first optimization layer an area coverage path plan for a given number of harvesting units; and determining as a second optimization layer a scheduling plan according to which to efficiently transport crop harvested by said harvesting units and stored in their storage tanks from the working area to a stationary depot located outside of said area by means of transport units, whereby said second optimization layer takes the area coverage path of the first optimization layer into account for its planning.

[0005] According to the first method, harvest operations are planned by grouping of all available harvesting units; assigning groups of transport units to said groups of harvesting units; assigning an area coverage path plan to all harvesting units; each transport unit during each transport tour serving sequentially all harvesting units of its assigned group of harvesting units, whereby a transport tour of a transport unit comprises leaving the depot with an empty storage tank and returning with a storage tank filled up to its capacity limit; thereby sequentially transfering harvested crop from the storage tanks of said group of harvesting units to the storage tank of said transport unit while all harvesting units of said group remain at zero velocity during said transfer process until the storage tanks of all harvesting units of said group are emptied; thereby enabling cyclic scheduling of transport units. Accordingly, said method also determines the minimal number of required transport units to optimally serve said groups of harvesting units. An increase in the number of employed transport units would not decrease the time required for overall area coverage.

[0006] According to the second method, harvesting and transport units are coor- dinated by real-time matching, dismissing offline a priori planning and scheduling. Criteria determining the matching are motivated by the objective of minimizing overall harvesting time, and the fact that any harvesting unit cannot further proceed with harvesting as long as the fill level of its storage tank is saturated at its capacity limit.

[0007] The details of this specification are set forth in the accompanying drawings and the description below. Aspects and advantages will become apparent from the drawings, the detailed description and the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[0008] FIG. 1 displays a flow diagram for the transfer of crop from a working area to a stationary depot for storage. [0009] FIG. 2 illustrates components relevant for the harvesting process.

[0010] FIG. 3 illustrates two general types of area coverage patterns.

[0011] FIG. 4 displays a flow diagram for two general types of area coverage patterns.

[0012] FIG. 5 illustrates an area coverage pattern for harvesting specific rectangular working [0013] FIG. 6 illustrates the coordination scheme for crop transfer from harvesting units to a transport unit during the first headland traversal.

[0014] FIG. 7 illustrates the coordination scheme for the transition of multiple harvesting units from one straight lane to the next.

[0015] FIG. 8 illustrates the coordination scheme for crop transfer from harvesting units to a transport unit during the traversal of specifc straight lanes.

[0016] FIG. 9 displays a flow diagram for the method for area coverage based on cyclic scheduling.

[0017] FIG. 10 illustrates the grouping of multiple harvesting units, and further illustrates corresponding transport unit trajectories for the collection of harvested crop. [0018] FIG. 11 illustrates one cycle of a transport unit: starting with an empty storage tank at a depot; filling said storage tank by collecting harvested crop from a group of assigned harvesting units; returning to a depot for unloading of the collected crop; thereby completing the current cycle; before the next cycle can be started.

[0019] FIG. 12 illustrates the cycle of a transport unit in comparison to the evolution of the fill levels of a group of harvesting units that said transport unit is serving.

[0020] FIG. 13 illustrates the method for the dimensioning of a group of transport units serving a group of harvesting units according to the method of cyclic scheduling.

[0021] FIG. 14 illustrates an in-field navigation path planning method.

[0022] FIG. 15 displays a flow diagram for the method for area coverage based on real-time matching of pairs of harvesting and transport units.

DETAILED DESCRIPTION

Notation, objective and layered optimization

[0023] Throughout this specification and due to the agricultural setting, terms "area" and "field" are used interchangedly. The workflow during the harvesting process is visualized in FIG. 1: given an area (an agricultural field) with crop to be harvested according to SI, this crop is harvested by harvesting units S2, before it is transported by transport units S3 to a depot S4 for storage. FIG. 2 visualizes relevant components. Area 1 is a closed working area of specific topological shape that is constrained by its field boundary 2, also referrerd to as field contour. There may be multiple areas 3 within the field, such as tree-islands, pole masts and the like, prohibited from trespassing by any moving ground machinery. At the beginning of the harvesting process, the whole field is covered by the crop to be harvested 5. With progress of the harvest, the harvested area 4, that is crop-free, increases monotonously, until at the end of the harvest, all of the field has been harvested. The harvesting process is conducted by harvesting units (HUs) 6. Harvested crop must be stored at depots 7 located outside of the field and connected to the field by a) at least one field entry point 8, also referred to as a field exit point, at which mobile ground machinery can enter /exit the field, and b) the shortest paths 9 within a path network for each combination of field entry points and depots. The shortest paths and the path network may include both tarred, i.e., well-maintained, roads and rural and gravel paths, thereby possibly permitting different traveling velocities along the paths. We assume that throughout the harvesting process all HUs remain within the field boundary and are maintained and refueled on the field. Thus, mobile machinery is required to transport harvested crop from field entry points to the depots. Therefore, it is distinguished between two cases. The first case 10 involves field and path transport units (FPTUs) 11 that can shuttle between HUs and depots, thereby traveling both within the field area and also along the path network connecting fields and depots. The second case 12 involves field transport units (FTUs) 13 that differ from FPTUs in that they are constrained to operate during har- vest only within field boundaries, thereby tranporting harvested crop from the HUs to tranport trucks (TTs) 14 that are shuttling between the entry points of the field and the depots on the shortest paths along the path network. In both cases, an unloading auger 15 enables the transfer of harvested crop between HUs and FPTUs/FTUs. A similar unloading auger transfers harvested crop between FPTUs/FTUs and TTs if applicable. In the following, for brevity, we refer to TUs (transport units) whenever an in-field operation method is identical to both FPTUs and FTUs, and therefore does not require explicit differentiation between FPTUs and FTUs. In general, harvested crop can be transfered between a HU and a TU "on-the-go", i.e., while both HU and TU are moving at the same speed, or, alternatively, "on-the-spot", i.e., while both are stopped. For the latter case, we assume that a storage tank of limited capacity for the storage of harvested crop is mounted on each HU. For the former case, harvested crop can be transfered directly to the TU while it is harvested. Alternatively, harvested crop stored in the storage tank can additionally be transfered to the TU, while both HU and TU are moving. Which of the two transfer cases is feasible may depend on a variety of influences such as the human operator or automated driving capability, curved path trajectories, traveling speed or field topology. The absence of a storage tank is a special case of above description, with maximal storage tank capacity equal to zero. During harvest, typically multiple fields growing the same crop are harvested sequentially, in a "crop tour". Therefore, the exit point 16 of a given field that connects along the shortest path to the entry point of the field next in the crop tour can be relevant for planning of the area coverage paths of the HUs. The set of HUs shall be denoted by Ή with H = \Ή\ denoting the total number of HUs. For all

^ denote the fill level of the harvesting tank at time t and its maximal capacity, and denote the traveling speed at time t and its maximal limit.

Similarly, the set of TUs is denoted by T, and further analogously characterised by

Similarly, the set of TTs is denoted by Q, and further analogously characterised by

For the remainder, the number of HUs active on the given field is assumed to be fixed. In practice, it may vary from at least one up to more than 10 HUs on very large farms. The number of HUs is typically dependent on a variety of criteria such as financial ressources, and the speed with which it is required to cover a specific number of fields over a specific total area within a given harvesting time window.

[0024] In practice, time-optimal field harvest is of interest due to often weather- related limited time harvesting windows. In the general case, this is achieved when all available HUs keep harvesting continuously and operating at their maximum admissible traveling speed without interruptions throughout the entire harvesting process. This is because the harvesting progress scales linearly with the harvested crop area. Maximum admissible traveling speed refers to a speed at which crop can be still be harvested without damaging or substantially losing the crop to be harvested. The maximum admissibile traveling speed may be time-varying over the harvesting process, and may temporarily also be zero, for example, in the case that harvested crop can be transfered to TUs only at standstill. Ideally, TUs must be operated such that given HUs can operate time-optimally. Thus, TUs must serve HUs. This motivates a layered planning process. The first layer addresses the operation of HUs (HU-layer). The second step addresses the operation of TUs (TU-layer). If applicable, the third step addresses the operation of TTs (TT-layer), whereby TTs must serve the TUs. Since TUs must serve HUs, the first layer can be optimized independently from the second layer. Similarly, the second layer can be optimized independently of the third. Thus, after optimization of the first layer, the second layer is optimized starting from the result of the optimization of the first layer. In the following, the three optimization layers are discussed.

HU-layer: area coverage path planning and formation driving for HUs

[0025] The first optimization layer is discussed assuming multiple HUs. The case of harvesting a field with a single HU is naturally included in that description. For the operation of HUs, the following aspects are relevant for this specification: formation driving, three methods of crop transfer, the role of the unloading auger, the first headland area coverage path plan, and two general types for area coverage path planning.

[0026] It is distinguished between two general types of trajectories for area coverages of the HUs: headland paths and straight lanes. See FIG. 3 for visualization. The first type describes a coverage pattern along headland paths, for brevity also referred to as headlands, and visualized by headland centerline 18, that are generated from erosions (mathematical morphological operations) acting on the field boundary 17. The lateral distances, see 22, between different headlands are constant and here denoted by headland width W. The outmost headland is laterally displaced by W/2 from the field contour, see 21. Width W is determined by the combined operating width of all HUs when moving in parallel. The second general type for area coverages of the HUs involves straight lanes illustrated by 35, or, alternatively, along lanes curvedly aligned to a particular part of the field contour and locally parallel. As illustrated by FIG. 3, after the transition to the final headland at 36, the coverage of that entire headland path, the traversal of straight lanes is illustratively started at location 37.

[0027] According to the method of this specification, throughout the entire harvesting process, HUs drive in formation, whereby priority levels are assigned to the HUs according to their position within the formation. Hence, priority levels are assigned to HUs according to local longitidunal ordering when facing the path direction for area coverage. See FIG. 3 for visualization of three exemplatory HUs, 26, 27 and 28. HUs 26 and 28, which are also labeled by letters A and C, are assigned the highest and lowest priority order, respectively. HUs are ordered within the preferred formation according to a performance metric such as, for example, traveling speed for harvest and storage tank capacity, whereby fastest HUs are always placed most advanced and assigned the highest priority order. Guiding principle is that higher prioritized HUs and TUs must never constrain lower prioritized ones while operating in the aforementioned formation. Then, HUs are displaced longitudinally and laterally such that the edge, in the following referred to as "visual edge", between harvested and not yet harvested area can be exploited visually. Thus, HU 27 can exploit the visual edge generated by 28 to accurately align for harvest of the not yet covered area. Similarly, 28 exploits the visual edge generated by 27. The not yet harvested crop area is indicated by 30. Lateral precise operation is required to maximally exploit the operating width of HUs. Longitudinally, in contrast, a slack distance 32 between any two HUs is permissible and in practice unavoidable. This slack distance shall be kept as small as practically feasible to maintain a tight formation.

[0028] The second reason for the preferred formation driving of HUs is that TUs can operate in parallel on the already harvested area. For visualization in FIG. 3, see TU 34 illustratively operating in parallel to HU 25. In general, there are two exceptions therefore. One is the trajectory of the outmost HU operating along the first headland path. See 23 and 33 for a corresponding HU-TU pairing. The reason for the exception is that, in general, there is no path immediately next to field contour 17 that would permit TU 33 to operate in parallel close to HU 23. The solution for this scenario that still permits harvest transfer from HU to TU is discussed further below. The second exception is discussed further below in the context of straight lane traversal.

[0029] It is distinguished between at least three methods of crop transfer from any of the HUs to any of the TUs. First, crop transfer is possible while both units are at standstill. This method of transfer is always possible, i.e., on curved as well as straight path segments within the field. This method may be preferable due to various reasons, including a challenging hilly field topology that makes it difficult for human drivers or the automated steering system in coordinating a pair of HUs and TUs while being in motion. Second, crop transfer is possible while both units are moving at the same speed, and traveling along approximately straight lanes within the field. Third, crop transfer is possible while both units are moving at the same speed and traveling along arbitrary curvy path segments within the field. This method naturally implies that crop transfer along straight path segments within the field is also possible. In general, for the latter two cases, admissible speeds of any unit pairing may be different when traveling with and without simultaneous crop transfer.

[0030] Crop transfer between a HU and TU is enabled by the unloading auger. For headland traversal, two driving directions are possible. These are clockwise (CW) or counterclockwise (CCW) and are determined by the side on which HUs can operate their unloading auger: if on the left- hand side when facing forward, headlands are traversed in CW-direction; if on the right-hand side when facing forward, headlands are traversed in CCW-direction. If the unloading auger can be operated such that unloading is possible on both sides, both headland traversal directions are permitted. The role of the unloading side will be relevant later on when discussing straight lane traversal according to the method of this specification.

[0031] According to the first general type for area coverages of the HUs, after entering the field at entrance 19, HUs traverse to the first headland 18, before then traversing to the next headland and so forth until the field is covered. Headlands shall be denoted according to traversal order as "first headland", "second headlands" and so forth. [0032] As indicated in FIG. 4, the area coverage along the first headland is common to both general types for area coverage. The area coverage either proceeds along either eroded headland paths according to Step S7, or, alternatively, it proceeds with straight lanes according to Step S9 after possibly traversing some additional eroded headlands beyond the first headland according to Step S8.

[0033] A variation of the first general type for area coverages of the HUs is particularly applicable for field contours hat are approximately rectangular. Motivation for said variation is to avoid crop repression when turning with HUs while simultaneously continuing the harvesting process. See FIG. 5 for visualization. The corresponding global area coverage path is spiral-like, whereby path segments 39 are concatenated forming rectangles, see 40. In order to avoid aforementioned crop repression, a group of HUs, in FIG. 5 exemplified by three HUs led by 38, must drive backwards according to 41 after completion of harvesting any straight lane, before continuing forward motion and harvesting 42 in direction of the next straight lane.

[0034] As indicated by FIG. 4, a first headland path is always required and common to both general types for area coverage. According to the method of this specification, the first headland traversal involves characteristic maneuvers of HUs for transfer of harvested crop to TUs in case of at least two HUs being employed. As mentioned above, in general, there is no path immediately next to the field contour that permits TUs to operate in parallel close to the highest prioritized HU when traversing the first headland. FIG. 6 illustrates said maneuver for the case of three HUs, 43, 44 and 45. The not yet harvested area and field boundary are indicated by 46 and 47, respectively. Thus, whenever aforementioned HUs reach a specific fill level within their harvesting tank, all three HUs perform a backwards driving maneuver, see exemplatorily 48 and 49, such that a TU can collect harvested crop from HUs C, B and A along a trajectory as visualized by 50. The specific fill level at which said maneuver is to be performed will be specified later when discussing the coordination between HUs and TUs. The order in which harvested crop is collected by a TU from a HU-group is always according to inverse priority. Thus, collection is started at lowest prioritized HU (here C) and is ended at highest prioritized HU (here A). After completion of the crop transfer from HUs to the TU, HUs continue their harvesting process, importantly according to the original formation and priority scheme. As soon as sufficient space is left, the loaded TU can perform a turning maneuver as indicated by 51 and return to the depot for unloading. [0035] Two general types for area coverage patterns of HUs are mentioned above. Under the assumption of arbitrarily shaped field contours, the first case, with coverage patterns along headland paths throughout the complete area coverage, bears three main disadvantages. First, when tracking curved headland paths and simultaneously harvesting crop, part of the crop is always repressed to the side due to the steering motion. Consequently part of the crop is likely lost. Second, for work on the field throughout the year (e.g., for fertilizing and spraying operations), straight lanes with corresponding tractor furrows are typically created. During harvest, these furrows would have to be crossed multiple times. This may be uncomfortable for the HU operator and possible cause increased wear or even damage to the HU. Third, the crop transfer between a HU-TU pairing along curved headland paths may be challenging due to the need for accurate steering and simultaneous caution to not lose precious crop.

[0036] As mentioned, area coverage of the HUs may involve not only movement along headlands but additionally along straight lanes or along lanes curvedly aligned to a particular part of the field contour. For the remainder, the case of lanes curvedly aligned to a particular parth of the field contour shall implicitly be comprised in all of the following discussion for straight lanes. Four questions arise. First, when to start area coverage along straight lanes? Second, what role does the side on which the unloading auger can be operated play for crop transfer between HUs and TUs? Third, how to perform a turn from one straight lane to the next? Fourth, what orientation shall the lanes have with respect to an orthogonal coordinate frame in the two-dimensional plane? All four questions are addressed in the following. According to the method of this specification, if the unloading auger can be operated both to the left- and right-hand side of the HU when facing the direction of travel, then area coverage along straight lanes is not constrained by travel distance along the straight lane. This is because at all time harvested crop can always be transfered to TU on one of the sides of the HU. Thus, straight lane traversal can be initialized as soon as a sufficient number of headlands (at least 1) has been traversed and harvested such that there is enough space on the already harvested area for turning of the HUs when traversing from one straight lane to the next. In contrast, if the unloading auger can be operated on only one side, in practice typically on the left-hand side, then, the maximum length of straight lanes must be such that any of the employed HUs can harvest along that lane without reaching the storage tank capacity limit. This is because throughout its traversal, harvested crop cannot be transfered to any TU. See 61 in FIG. 7 for illustration. Only if unloading auger 62 can be operated to the right-hand side of the HU, harvested crop could be transfered to TU 61. In general, in real-time operation of HUs, it cannot be guaranteed that HUs can always be emptied precisely before traversing a straight lane. Therefore, a generic logic is needed to organize crop transfer also along any straight lane, even if traversing long straight lane segments, that do not admit storage of harvested crop within storage tanks of limited capacity of HUs that only permit operation of the unloading auger on one side. For visualization of the method, see FIG. 8. The not yet harvested and the already harvested areas are indicated by 63 and 64, respectively. It is assumed that the unloading auger of all three HUs, A, B and C, can only be operated to the left-hand side of each HU when facing forward according to direction 65. Thus, according to the method of this specification, once a specified fill level is reached by the HUs, they drive backwards as indicated in FIG. 8. Specifically, HU C, i.e., the least-prioritized HU must perform a steering maneuver as indicated by 67 to make sufficient space laterally with respect to not yet harvested crop boundary 66 such that the designated TU can align and start the unloading process. An important remark is made that motivated the devised method, and specifically also the proposed method to perform the said backwards driving and steering maneuvers of HUs precisely as indicatedin FIG. 8 and, moreover, to start collecting harvested crop from the least prioritized HU. The trajectory taken by the TU for the collection of harvested crops at all three HUs in the order of C, B and A, before returning to the depot, is indicated by 68. A preferred concept according to the method of this specification is to always maintain proposed formation in order to enable all HUs to drive along the visual edge between harvested and not yet harvested crop area. Suppose only HU A drove backwards as indicated and B would continue harvesting and moving forward. Then, the visual edge would quickly be lost once B passes the height along the straight lane at whih A stopped harvesting. This motivated to maintain all HUs at rest, throughout the entire process of crop transfer to the TU, before then continuing harvesting and according to the original formation and according to direction 69. [0037] According to the method of this specification, a turn of a group of multiple HUs from one straight lane to the next is realized as described below. Fundamental notions are a) to always harvest along the visual edge differentiating between already harvested and not yet harvested crop, and b) to maintain the aforementioned formation and ordering of HUs. The first notion motivates to traverse straight lanes in sinuous form as visualized by 35 in FIG. 3. For the second notion, the methodology of the turn is summarized in Algorithm 1.

For visualization of Algorithm 1, see FIG. 7 where 58 indicates the not yet harvested crop area. HUs are ordered according to priority as 52, 53 and 54. The highest prioritized HU stops at location 55 to let lower prioritized HUs pass. Similarly, the HU labeled with B stops at 56 to let HU labeled with C pass by at position 57. HU A starts its new straight lane traversal as soon as it can, thereby harvesting along the visual edge 59, generated previously by HU C. The remaining HUs follow suit according to their priority order. Besides the two notions mentioned, an additional benefit of the devised method is that by its employment all HUs have a consistent and large spacing equal to the combined harvesting width that can be used for traversing from one straight lane to the next.

[0038] The orientation of straight lanes for harvesting is selected according to the same orientation of lanes used throughout the year, e.g., for fertilizing and spraying.

[0039] To summarize, it is distinguished between two main types for area coverage: area coverage along (eroded) headland paths only, and the inclusion of straight lanes for area coverage. According to the method of this specification, for both types, the operation and path planning of HUs is defined as outlined above.

TU-layer: coordination through cyclic scheduling

[0040] The second optimization layer is discussed. Step S10 of FIG. 9 is determined according to the first and aforementioned optimization layer. Steps Sll, S12 and 13 are discussed in the following, whereby Sll and S12 are offline planning steps. According to Step Sll, all available HUs are first grouped. For this grouping a large set of potentially available TUs T is assumed, that may stem from multiple sources including owned TUs and TUs from leasing partners such as contractors. This initial and conservative number will later, in Step S12, be reduced to the optimal number. Let G disjoint subsets be defined by

Throughout, the partitioning of is selected such that the priority order is maintained. For example, and imply that Hi always holds the first three

highest prioritized HUs, and always holds the fourth to sixth highest prioritized

denote the subset of TUs assigned to serve H g , whereby all

h G H g are ordered according to priority within the driving formation. Let the cardinality of a set be denoted by The modulo

operation mod(x, n) returns the remainder after division of number x by number n. The ceiling operator rounds its argument to the nearest larger integer. Thus, for Sll the following steps are conducted:

Algorithm 2: Grouping of HUs

The outcome and purpose of Algorithm 2 is to ensure that any TU can always leave its assigned group of harvesting units, with a storage tank filled to its capacity limit As outlined in Algorithm 2, the maximal number of HUs per group is 3, which is a parameter choice motivated by practical considerations, i.e., by modern storage tank dimensions of HUs and TUs. Another parameter choice (e.g., 4) is feasible and solely requires adaptation of Step 2. An important prerequisite for group travel is the ability of all HUs of the same group to travel at the same speed. In contrast, varying intake rates (e.g., because of varying operating widths) are not necessarily constrainig and therefore in general admissible.

[0041] FIG. 10 illustrates the desired implications of a grouping of HUs and an assignment of TUs. Each TU that is assigned to a specific group of HUs (in general, this group may also consist of only one HU) is meant to collect harvested crop from all of the HUs of that specific group. As illustrated in FIG. 10, the first group composed of HUs 70, 71 and 72 and facing direction 73 consequently results in a trajectory of a TU assigned to that group as indicated by 74, starting and ending the collection process at lowest (here 72) and highest (here 70) prioritized HU of that group. Similar trajectories, 75 and 76, result for the other two groups of HUs displayed.

[0042] FIG. 11 visualizes the steps taken by any TU when shuttling between a) a HU or a group of HUs for loading, and, b) either a depot, a TT or a group of TTs for unloading. The time-axis is indicated by 77. For brevity, in the following we refer to the unloading station as a "depot". TTs and a group of TTs are regarded as a mobile depot. Let 78 denote i.e., the time when begins to travel from a depot

and with empty tank and for the /cth cycle, whereby A cycle here denotes one tour of a TU from the depot to its HU-group for loading, and back to the depot

for unloading, such that a new cycle can be started. Correspondingly, the kth cycle time is given by indicated by 84 in FIG. 11. Let 79 denote

i.e., the time of arrival at the first HU, the least-prioritized HU of

. Accordingly, 81 denotes for the last the highest-prioritized

HU according to the priority ordering. Let 80 denote denote the corresponding

time of leaving after the transfer of harvested crop. Thus, 82 denotes

for the last according to the priority ordering. Let 83 denote i.e., the

time of arrival at the depot for unloading. The storage capacity of any HU and TU is denoted by and for all Let the fill

level of HU j at any time t be denoted by

[0043] Step S12 of FIG. 9 for the dimensioning of the TU fleet and the scheduling thereof, may be implemented as follows, whereby k = 0 indicates the beginning of harvest of a given field:

Algorithm 3: Dimensioning of the TU-fleet and cyclic scheduling of the TUs

1. Input:

2. For all g

Initialize:

3. While the given field is not completely harvested:

4.

The corresponding schedule according to models

The outcome and purpose of this algorithm is a) to select the minimal number of required TUs that guarantees continuous unconstrained operation of all available HUs, and b) to obtain the corresponding planned schedule for all HUs and all TUs involved. Continuous unconstrained operation here refers to a time schedule that avoids that HUs ever have to stop harvesting because of a full storage tank that would require the corresponding HUs to wait for a TU to come for emptying. By increasing the number of TUs beyond the minimal number of required TUs in combination with appropriate scheduling, no additional benefit is gained since unconstrained operation is already satisfied. Any additional owned TUs that are not required for continuous unconstrained operation could, in principle, be rented out to generate additional revenue. Principal characteristic (and likewise the major difference with respect to current real- world working practice) is that all HUs of Ή 9 remain stopped until they are emptied, i.e., their harvested crop is transfered to an assigned TU. This methodology enables a predictable and cyclic system behavior.

[0044] The two modeling functions,

G, are discussed in more detail. The fill level of the harvesting tank at time t is

The simplest and most practical model for predictions is assuming a

constant and traveling speed-dependent harvesting rate. The second modeling function returns the time at which a g can return at the earliest

possible time back to In detail, the difference involves all of

the following timings: 1) the time for transfering the harvested crop from all HUs of group This time also includes time for transitioning between the HUs

of that group; 2) the time for traveling between the last loaded HU of group Ή 9 and the depot. In case of a stationary depot, this time is partitioned into a) the traveling time along the path network outside the field between the best field entrance and the stationary depot, whereby the best field entrance is characterized by resulting in the overall shortest path from the position of crop transfer to the stationary depot, and b) the traveling time inside the field between the current position and the field

entrance. In case of a mobile depot (TT) , this time only involves the latter part; 3) the time for transfer of the harvested c rop from the TU to the depot; 4) the time for traveling from the depot back to group whereby, in the meantime, this group has continued to harvest along the area coverage path plan and according to the scheduling with the other TUs assigned to

[0045] The simplest and most practical model for predictions assumes a) constant unloading rates for each HU (nominally provided by its manufacturer) , b) transitioning times for any TU moving between the HUs of the same group (these time are small since all HUs are at rest and in close formation), c) constant traveling speeds along the path network and for in-field travel, and d) constant unloading rates for each TU (required for unloading at the depot). All parameters are preferably identified and learnt from real-world field experiments. Specifically, measurements taken from the area coverage of the previously harvested fields within the sequence of all fields of the same crop may be used for the parameter identification process. Accurate predictions of in- field traveling times are enabled by the planned and therefore known area coverage path for HUs according to the first optimization layer.

[0046] FIG. 12 illustrates characteristic time evolutions of fill-levels

of an exemplatory group of three HUs. Specifically, it is illustrated that

all HUs of that group remain stopped until all HUs are emptied, i.e. , their harvested crop is transfered to an assigned TU, in order to ensure a cyclic system behavior suitable for scheduling. The x-axis 85 of FIG. 12 denotes time t. In reference to FIG. 11, 89 denotes the time at which a specific TU has to start at the depot

to reach the first HU of at time 92. Time is indicated by 91. The second HU

is reached at time 93. At time 94, all HUs are emptied and group can continue

harvesting along the area coverage path plan in close formation. The time-evolution of the fill levels of the three exemplatory harvesters is indicated by y-axes 86 for

87 for and 88 for respectively. Intake rates of harvested crops, 95,

97 and 99 may be different. Nevertheless peak fill levels 96, 98 and 100 are such that at time given by 92, the combined fill level is

Note that between times 92 and 94, no further crop is harvested (no further increase in peak fill levels). At time 94, all HUs are emptied. Crop harvest can be continued and harvesting tanks are being filled again. Intake rates 101, 103 and 105 may vary from previous 95, 97 and 99 due to possibly spatially different crop yields within the field. Similarly, peak levels 102, 104 and 106 may also vary from previous 96, 98 and 100, since they are dependent on the next assigned TU to collect harvested crop.

The loading capacity of that TU may vary from that TU assigned previously for the collection between times 92 and 94. [0047] FIG. 13 further illustrates Step 2 of Algorithm 3. The x- and y-axes, 107 and 108, illustrate time t and the combined fill level of an exemplatory

HU group Thus, 109 indicates The combined fill level

rises cyclically according to 110, until it reaches its peak level 111 at time 115, which is determined by the capacity of the currently serving before it falls

according to 112 because of the emptying and transfer process of harvested crop from the HUs to the currently serving TU. Note that according to the method x may vary dependent on the currently serving TU, thereby admitting varying peak levels, for example, with 113 different from 111. Here, 110 and 112 are modeled exemplatorily as linear. In practice, 110 is monotonously increasing, but only approximately linear becase of spatial crop yield variations. In contrast, in practice 112 is monotonously decreasing according to the profile of the superposition of the unloading functions of all HUs of any group during the time period of the entire unloading process, i.e.,

with reference to FIG. 12 between time instances 92 and 94. The complete emptying of all HUs, until a combined fill level of zero is achieved for all of the HU-group, allows cyclic scheduling. In FIG. 13, four TUs are required to cyclically serve the exemplatory HU-group. The starting times at which each of the four TUs leaves the depot are denoted by 114, 121, 122 and 123, respectively. Four TUs are required. After the fourth TU, the first TU again leaves the depot at time 116, and can return at time 117 to serve the HU-group; eventhough the next needed service time is only at time 118. The corresponding time margin 119, i.e., the difference between time instances 118 and 117, is characteristic and beneficial for the method for robustness reasons. It allows, for example, to compensate for modeling errors and unscheduled delays in the TU-operation. For the case that the corresponding TU j e T 9 indeed arrives at time 117, the TU is, according to the method, enforced to wait until time 118, i.e., until The next time at which the corresponding TU leaves the depot is 120.

[0048] For real-time operation following the offline planning, Step S13 of FIG. 3 may be implemented as follows:

If all harvesters of each group are of identical dimensions (in particular, identical intake rate), target harvesting capacities are set

indicated and according to the method, every HU-group, besides the offline preplanned area coverage path plan, requires information about the loading capacity of the next TU assigned to that HU-group. [0049] For the real-time operation of all TUs according to S13, all TUs are equipped with at least the following: 1) a path planner that can navigate each TU from its current location towards either the predicted location of its assigned group H g , or to a depot. The navigation part can be decomposed into two parts, i.e., out- and in-field navigation. The former comprises routing along the path network connecting the given field and the stationary depot. For this, shortest path algorithms can be employed (see D. Bertsekas, Dynamic programming and optimal control. Vol. 1. No. 2. Belmont, MA: Athena Scientific, 1995.). Note that the path network may have to be generated and tailored manually since rural and gravel roads as well as field entrances may need to be defined manually. In contrast, for in- field navigation, an online path planner is required for path planning within an area with time-varying areas prohibited from trespassing (the not yet harvested crop). Such path planner is presented below. The same path planner can be used in the distributed and alternative scheme for the coordination of HUs and TUs discussed below; 2) a plan for the sequence in which to load harvested crop from all According to the method of this

specification, the loading sequence is the inverse priority order of each group of HUs. Thus, the lowest prioritized HU within is served first and the highest prioritized

HU last; 3) the predicted time t at which each has to arrive at such that This time may be used for velocity control and reassurance about

scheduling times. In situations where TUs are operating ahead of schedule, it may also be used to operate the TU in an environmental friendly manner and for reduced fuel consumption.

[0050] The aforementioned path planner for in-field navigation between a start and end location, s and e, within an area with time-varying areas prohibited from trespassing (the not yet harvested crop) may be implemented as follows:

Two remarks can be made. First, the path planned according to Algorithm 5 is piecewise affine (PWA). For the in- field navigation provided to a human driver, a visually displayed PWA path is sufficient to serve as a navigation reference. For the navigation of automated vehicles, in contrast, it must be smoothed sequentially taking vehicle dimensions and vehicle agility capabilities into account. Second, the obstacle area, i.e., the not yet harvested crop area is time- varying and monotonously shrinking. This can be taken into account. Assuming knowledge of predicted trajectories of all HUs (given by the global area coverage path plan), the obstacle area evolution can be predicted over time. Assuming further a velocity profile (e.g., a constant traveling speed) for each TU, the time for each TU upon reaching a specific obstacle corner can be predicted. Then, in principle, a more detailed path plan can be determined, comparing times upon reaching obstacle corners or boundaries, and the time these are eroded because of harvesting progress. Nevertheless, in practice and with respect to in-field planning for fully automated vehicles, the most conservative and safest method is to use the static obstacle size recorded at the time of executing the planning. It is safest since unexpected events may halt the predicted obstacle area evolution, and, therefore, in the worst case, cause TUs to drive into unharvested crop area because of the predicted obstacle area evolution, thereby destroying precious crop.

[0051] FIG. 14 visualizes Algorithm 5. Since an in-field path planner such as Al- gorithm 5 is also central to the second method of coordination HUs and TUs to be discussed below, FIG. 14 illustrates a very general scenario. The not yet harvested area is partitioned according to 126 and 127. The field entrance and contour are 124 and 125, respectively. HUs such as 128 move in the directions as indicated by arrows. Assuming TU 129 is assigned to 130, the two in-field paths planned according to Algorithm 5 are indicated by 132 and 133, whereby the straight line connecting 129 and 130 is indicated by 131.

[0052] Throughout each shuttling cycle, different are in

practice not affected by each other until they reach the depot for unloading. The principle applies for both stationary and mobile depots. TUs of different groups may arrive shortly after one and another at the same depot. Thereby, waiting times may result. For example, there may be only a single conveyer belt or similar device that does not permit simultaneous transfer of harvested crop from the TUs to the depot. According to the method of this specification, these waiting times are accounted for by the aforementioned time buffer,

[0053] Motivation and benefits of the first devised method according to this speci- fication for the coordination of HUs and TUs are summarized as follows:

Loading capacities of HUs are typically smaller than the ones of TUs. Consequently in order for TUs to fill their storage tanks up to their capacity limit, each one of them must do one of the following:

- drive to at least a second additional HU for additional transfer;

- remain with the HU from which harvested crop was transfered last, and wait until that HU has again harvested a sufficient amount for the TU to fill its storage tank up to its capacity limit.

The total time for transfering harvested crop from the storage tank of a HU via its unloading auger to a TU is much faster than the time for the HU to harvest the same amount of crop. This motivates TUs to drive to additional other HUs for filling up their free capacities, rather than remaining with one HU and waiting until the HU has again harvested a sufficient amount.

The closer a group of HUs can operate in formation, the smaller the distances that TUs need to cover between the transfers of harvested crop from the corresponding HUs, thereby minimizing overall time for a TU to fill its storage tank up to its capacity limit by serving all out of the group of HUs driving in close formation.

In order to ensure a cyclic and predictable scheduling of TUs between a group of HUs and a depot, each HU of said group must be emptied to zero storage tank level during the unloading step when transfering harvested crop to any TU. Without such cyclic reinitialization, harvesting of HUs in tight formation will eventually not be feasible anymore because of asymmetrically increasing storage tank fill levels. To ensure tight formation driving and simultaneously cyclic reinitialization, all HUs of that group of HUs must remain stopped throughout the complete process of transfering harvested crop from each HU of said group to its assigned TU.

On the first headland tour, and possibly along straight lane traversals, HUs may have to drive backwards for a short period of time in order to position themselves such that a TU can be aligned for unloading. In such situations, the transfer of harvested crop from the corresponding HU to a TU must be conducted while both HU and TU are at rest. This is favorable in view of the invention. It naturally

is in line with the need of the method according to the invention to stop for crop transfer when serving a group of HUs by one TU in order to maintain cyclic and predictable operation. Second method: coordination of HUs and TUs by real-time matching

[0054] According to a second method of this specification, HUs and TUs are coordinated by real-time matching without any a priori planning and scheduling. The method is visualized in FIG. 15. According to Step S14, HUs follow an area coverage path plan for harvest. Once a TU enters the field of interest it is assigned in real-time to a suitable HU according to Step S15.

[0055] The fundametal property of the first HU-TU coordination scheme discussed before is that all HUs of group must remain at stop during the entire transfer

process of harvested crop from all HUs of that group to a corresponding TU. This property is absent for the second method. Specifically, the transfer of harvested crop between a HU and TU can be realized while both units are in motion. This is similar to current working practice. In contrast, to current working practice, TUs are explicitly assigned to HUs in real-time according to a matching criterion.

[0056] Given a measured and predicted harvesting intake rate and a measured fill level F j (t) , the estimated next time at which ¾ would reach its capacity limit

shall be denoted by The estimated time at which the projected position

of HU j e ¾ can be reached given the current position of T is denoted by whereby shall denote the corresponding connecting path in form of a seq uence

of coordinates. This path can be computed according to the aforementioned path planner for in- field navigation. While the field of interest is not yet entirely harvested, the following Algorithm can be conducted at every saming time t:

Once a TU is assigned to a HU, the maximum amount of harvested crop that is available at said HU at the moment of transfer, and that still fits the loading capacity of the corresponding TU, is transfered. With regard to Step 3 of Algorithm 6, various criteria are possible. Two examples are given. First, is assigned to HU

indexed by i.e., always to the HU that is predicted to

next reach its storage tank capacity limit if not emptied beforehand. This criterion disregards spatial proximity between HUs and TUs. Therefore, a second criterion is

In practice, a trade-off between

these two examples may preferably be selected: the first criterion may apply when a enters the field with and the second criterion may apply for TUs

that still have some, but limited, free residual loading capacities.

In-field transport by unmanned aerial vehicles

[0057] According to an embodiment, TUs may not necessarily be ground vehicles. Instead, unmanned aerial vehicles (UAVs), that shuttle between HUs and a depot, may be employed. Their employment bears the advantage that no aforementioned backwards driving of any HU is required to make space for a TU to align on one side of the HU such that the unloading auger can be operated for the transfer of harvested crop. In case UAVs are employed, the transfer of harvested crop from a HU to an UAV may or may not be conducted by means of an augmenting auger. It may also be conducted in form of the UAV collecting at least one container filled and carried by the HU and storing the harvested crop. In case UAVs are employed in combination with TTs, the required operation range of the UAVs is constrained to the interior of the working area, thereby also limiting the risk of operation UAVs to that area. TT-layer (if applicable): transport trucks operation

[0058] Similarly to TUs, the purpose of TTs is to ultimately serve HUs, however, indirectly with TUs serving as intermediaries. TTs may be assigned to travel between the area entrance closest to the HUs that are served, and the closest stationary depot. TTs may simultaneously serve all HUs Ή, or, alternatively, TTs may also be grouped to serve groups of TUs. These groups may not necessarily coincide with HU- and corresponding

[0059] If the loading capacity of each TT equals precisely the load that is transported from the TUs that simultaneously serve all HU-groups then a scheduling scheme analogous to the case of TUs serving HUs can be devised for determining the number of required TTs and for their scheduling. Then, by construction, each TT leaves the field entrance with a fully loaded tank.

[0060] If the loading capacity of each TT is larger than the load that is transported from the TUs that simultaneously serve all HU-groups two options are

possible. Either, the aforementioned scheme is still employed, whereby, however, each TT leaves the area entrance with a not fully loaded tank. The missing exploitation of the complete available loading capacity is hence the disadadvantage of this option. The fact that a cyclic schedule can be determined is its benefit. Alternatively, each TT is assigned to only leave the area entrance once it is fully loaded. In that case, a precise scheduling is not possible because of the absence of cyclic loading methodology in general. The second option is prefered in practice for consideration of operating costs of TTs.

[0061] If the loading capacity of each TT is smaller than the load that is transported from the TUs that simultaneously serve all HU-groups then multiple TTs

are required, for which the same considerations above apply. [0062] In practice, for the complete time of field harvest, the combined loading capacity of TTs waiting at the field entrance must exceed the load that possibly can arrive simultaneosly at the TTs. If this condition is satisfied for the complete time of field harvest, it is guaranteed that the TU-operation of any TU is never constrained by the TT-layer. Implementation with and without mobile computing devices

[0063] The devised methods can be implemented by means of "mobile computing devices" or "handsets", i.e., portable computing devices that can correspond to smartphones, tablet devices, laptop computers and the like, that can provide processing resources, access a database, provide a user interface to a human operator, and provide network connectivity to connect to other mobile computing devices, and to access remote servers hosted on the Internet to store, manage, and process data (cloud computing) .

[0064] According to the specification, the first devised method can be implemented without the need of any mobile computing devices. This is because the operation of all HUs and TUs is deterministically planned according to the offline planned area coverage path plan for HUs and the offline planned schedule for TUs. The method is devised such that clear-cut rules result that apply independent of delays in operation. In a practical implementation, without any mobile computing devices, the method comprises 1) each HU-group receiving a list of TUs and their corresponding loading capacities according to the planned schedule, and 2) during area coverage: each HU- group stopping harvesting once the combined sum of the harvested crop of said HU- group equals the loading capacity of the TU scheduled to next serve said HU-group; awaiting unloading of all HUs of said HU-group, before continuing harvesting; and repeating Step 2) until the completion of area coverage. [0065] Nevertheless, an implementation of the devised method by means of mobile computing devices may be preferable and increase efficiency in its realization. There

[0066] For the implementation of the devised methods by means of mobile computing devices, the following computations are required in real-time at every system sampling

each TU; an out-field navigation path plan for each TU in case these are shuttling until the stationary depot, or alternatively, such out-field path plan for each TT; All of these computations are typically conducted on a remote server. They are based on, firstly, the information that HUs and TUs are sending in real-time, and, secondly, the offline planned area coverage path plan of HUs, the grouping of HUs and TUs, and the scheduling sequences of TUs for each group.

[0067] It is to be understood that the following claims are intended to cover all of the generic and specific features of the invention herein described and all statements of the scope of the invention which, as a matter of language, might be said to fall therebetween. It is intended that all matter contained in the above description and shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.