DE102023202823A1 - Route planning system and method for agricultural work vehicles to generate an optimized processing route - Google Patents

Abstract

The invention relates to a method and a route planning system for agricultural work vehicles for generating an optimized processing route on a field by means of a quantum annealing system.

Description

The invention relates to a method and a route planning system for agricultural work vehicles for generating an optimized processing route on a field by means of a quantum annealing system.

Agricultural work vehicles are now highly efficient. Therefore, the operational planning/organization of the work process for high-performance work vehicles such as combine harvesters, but also other agricultural machines such as fertilizer spreaders, seed drills, forage harvesters, mowers, tedders, swathers, etc., is becoming increasingly important. Due to the weather, only a limited number of combine harvesting hours are available during the grain harvest, which are often not used optimally due to poor operational planning.

Precise operational planning is also important for other agricultural machines in order to achieve the theoretical performance of the machines in practical use.

For a specific area, e.g. a field to be worked, it is unclear which route the agricultural work vehicle must take in order to minimize sharp curves and multiple crossings and still cover the entire field.

The state of the art reveals some route planning systems that are particularly optimized with regard to the recorded environment/soil or the harvesting process itself. However, such optimizations often reach their limits, particularly with regard to the number of conditions and in the case of a large or rugged field.

The EP 1795986 A2 discloses a route planning system for agricultural work machines, wherein the agricultural work machine is assigned a defined working width for generating travel routes in a territory and the agricultural work machine comprises at least one crop storage unit for receiving crop quantities and the crop quantity can be loaded onto unloading vehicles, wherein the crop quantity received by a crop storage unit is balanced depending on at least one characteristic parameter and the balancing can be dynamically adapted to changes in the at least one characteristic parameter.

The EP 2267566 A2 discloses a method for planning a route for processing an agricultural area by a mobile machine, comprising the steps of: horizontal, spatially resolved estimation of soil compaction resulting from previous processing of the area by the same or another machine, and taking the determined soil compaction into account when determining the course of a route for the intended processing of the area by the machine.

• Definition einer Referenzzeile mit einer Referenzkontur in einem Arbeitsbereich; Erstellen einer Darstellung der definierten Referenzzeile, wobei die Referenzzeile mindestens eine aus einer allgemein gekrümmten Komponente und einer allgemein linearen Komponente umfasst; und
• Generieren einer oder mehrerer Konturzeilen mit einer Tracking-Kontur, die die Referenzkontur basierend auf einer Fahrzeugbreite und einem Radiusdifferenzparameter, der der gekrümmten Komponente zugeordnet ist, verfolgt oder spiegelt; die Konturzeilen, die durch eine Translationstechnik für die im Allgemeinen lineare Komponente und eine Radiusmodifikationstechnik für die gekrümmte Komponente erzeugt werden; wobei die Generierung die Anwendung eines Suchalgorithmus auf mögliche Kandidatenwerte des Radiusdifferenzparameters umfasst und wobei der Suchraum so begrenzt ist, dass der Radiusdifferenzparameter durch einen möglichen Radiusdifferenzparameter begrenzt wird, der im Allgemeinen gleich oder kleiner als die Fahrzeugbreite ist.

The US 7216033B2 discloses a method for planning the path of a vehicle, comprising:

• Defining a reference line having a reference contour in a work area; creating a representation of the defined reference line, the reference line comprising at least one of a generally curved component and a generally linear component; and
• Generating one or more contour lines having a tracking contour that tracks or mirrors the reference contour based on a vehicle width and a radius difference parameter associated with the curved component; the contour lines generated by a translation technique for the generally linear component and a radius modification technique for the curved component; wherein the generation comprises applying a search algorithm to possible candidate values of the radius difference parameter, and wherein the search space is limited such that the radius difference parameter is limited by a possible radius difference parameter that is generally equal to or less than the vehicle width.

It is therefore an object of the invention to provide a method and a route planning system for agricultural work vehicles for generating an optimized processing route on a field in order to achieve improved utilization of the agricultural work vehicles.

The object is achieved by a route planning system having the features of claim 1 and a method having the features of claim 9.

• einer Speichereinheit für ein oder mehrere feldspezifische Daten, wobei die feldspezifischen Daten zumindest die Koordinaten für die Feldumrandung als auch die Koordinaten von einer Anzahl von abzufahrenden Punkten aufweisen, wobei die abzufahrenden Punkte das gesamte Feld hinreichend genau abdecken und wobei die abzufahrenden Punkte zumindest einen Startpunkt und einen Endpunkt umfassen,
• und ein oder mehrere arbeitsfahrzeugspezifische Daten, welche zumindest die optimale Geschwindigkeit des landwirtschaftlichen Arbeitsfahrzeugs in Abhängigkeit von einem Verbrauch umfassen,
• wobei das Routenplanungssystem dazu ausgebildet ist, Verbindungskanten anhand erkannter nachbarschaftlicher Beziehungen zwischen allen Punkten als auch dem zumindest einen Endpunkt und dem zumindest einen Startpunkt zu generieren, und wobei das Routenplanungssystem dazu ausgestaltet ist, verschiedene Bearbeitungsfahrwege anhand verschiedener Matrizen auszubilden, welche die jeweiligen abzufahrenden Punkte als Qubits unter Berücksichtigung der Zeit angeben, und wobei das Routenplanungssystem dazu ausgebildet ist, eine mithilfe der Qubits als Kostenfunktion gebildete Energiefunktion, anhand einer Energie für eine Beschleunigung des landwirtschaftlichen Arbeitsfahrzeugs als auch der Abweichung von der vorgegebenen optimalen Geschwindigkeit als auch von diskretisierten Zeitpunkten als Zeit zu generieren und wobei und wobei ein Quantenannealing-System vorgesehen ist, welches mit einer der Energiefunktion entsprechenden Energielandschaft assoziiert ist, und wobei ein Grundzustand des Quantenannealing-Systems eine Minimierung der Energiefunktion unter folgenden Bedingungen kodiert
  • - das landwirtschaftliche Arbeitsfahrzeug fährt alle Punkte zumindest einmal ab,
  • - das landwirtschaftliche Arbeitsfahrzeug fährt den zumindest einen Startpunkt an einem ersten Zeitpunkt und den zumindest einen Endpunkt an einem letzten Zeitpunkt korrekt an,
  • - das landwirtschaftliche Arbeitsfahrzeug befindet sich zu einem diskretisierten Zeitpunkt zwischen dem ersten Zeitpunkt und dem letzten Zeitpunkt an genau einem der abzufahrenden Punkte,
  • - das landwirtschaftliche Arbeitsfahrzeug kreuzt keine Feldumrandung,
• und wobei das Quantenannealing-System zur Annäherung an den Grundzustand ausgebildet ist und wobei das Routenplanungssystem zum Auslesen des Grundzustandes als minimale Energiefunktion und Umsetzen des Grundzustandes in einen optimalen Bearbeitungsfahrweg ausgebildet ist.

The task is solved by a route planning system for agricultural work vehicles to generate an optimized processing route on a field using a quantum annealing system comprising

• a storage unit for one or more field-specific data, wherein the field-specific data contains at least the coordinates for the field border as well as the coordinates of a number of points to be traveled, wherein the points to be traveled cover the entire field sufficiently accurately and wherein the points to be traveled comprise at least one start point and one end point,
• and one or more work vehicle-specific data, which include at least the optimal speed of the agricultural work vehicle depending on consumption,
• wherein the route planning system is designed to generate connecting edges based on recognized neighborhood relationships between all points as well as the at least one end point and the at least one starting point, and wherein the route planning system is designed to form different processing routes based on different matrices which specify the respective points to be traveled as qubits taking time into account, and wherein the route planning system is designed to generate an energy function formed using the qubits as a cost function, based on an energy for an acceleration of the agricultural work vehicle as well as the deviation from the predetermined optimal speed as well as from discretized points in time as time, and wherein and wherein a quantum annealing system is provided which is associated with an energy landscape corresponding to the energy function, and wherein a ground state of the quantum annealing system encodes a minimization of the energy function under the following conditions
  • - the agricultural work vehicle passes all points at least once,
  • - the agricultural work vehicle correctly drives to at least one starting point at a first point in time and to at least one end point at a last point in time,
  • - the agricultural work vehicle is located at exactly one of the points to be travelled at a discretised point in time between the first point in time and the last point in time,
  • - the agricultural work vehicle does not cross the field boundary,
• and wherein the quantum annealing system is designed to approach the ground state and wherein the route planning system is designed to read out the ground state as a minimum energy function and convert the ground state into an optimal processing path.

• - Bereitstellen von ein oder mehreren feldspezifischen Daten, wobei die feldspezifischen Daten zumindest die Koordinaten für die Feldumrandung als auch die Koordinaten von einer Anzahl von abzufahrenden Punkten aufweisen, wobei die abzufahrenden Punkte das gesamte Feld hinreichend genau abdecken und wobei die abzufahrenden Punkte zumindest einen Startpunkt und einen Endpunkt umfassen,
• - Bereitstellen von ein oder mehreren arbeitsfahrzeugspezifischen Daten, welche zumindest die optimale Geschwindigkeit des landwirtschaftlichen Arbeitsfahrzeugs in Abhängigkeit von einem Verbrauch umfassen,
• - Generieren von Verbindungskanten anhand erkannter nachbarschaftlicher Beziehungen zwischen allen Punkten als auch dem zumindest einen Endpunkt und dem zumindest einen Startpunkt,
• - Bereitstellen der verschiedenen Bearbeitungsfahrwege anhand verschiedener Matrizen, welche die jeweiligen abzufahrenden Punkte (N) als Qubits unter Berücksichtigung der Zeit angeben
• - Ausbilden mithilfe der Qubits einer als Kostenfunktion gebildeten Energiefunktion, anhand einer Energie für eine Beschleunigung des landwirtschaftlichen Arbeitsfahrzeugs als auch der Abweichung von der vorgegebenen optimalen Geschwindigkeit als auch von diskretisierten Zeitpunkten als Zeit
• - Assoziieren eines Quantenannealing-Systems mit einer der Energiefunktion entsprechenden Energielandschaft, wobei ein Grundzustand des Quantenannealing-Systems eine Minimierung der Energiefunktion unter folgenden Bedingungen kodiert:
  • - das landwirtschaftliche Arbeitsfahrzeug fährt alle Punkte zumindest einmal ab,
  • - das landwirtschaftliche Arbeitsfahrzeug fährt den zumindest einen Startpunkt an einem ersten Zeitpunkt und den zumindest einen Endpunkt an einem letzten Zeitpunkt korrekt an,
  • - das landwirtschaftliche Arbeitsfahrzeug befindet sich zu einem diskretisierten Zeitpunkt zwischen dem ersten Zeitpunkt und dem letzten Zeitpunkt an genau einem der abzufahrenden Punkte,
  • - das landwirtschaftliche Arbeitsfahrzeug kreuzt keine Feldumrandung,
• - Annähern an den Grundzustand des Quantenannealing-Systems und Auslesen des Grundzustandes als minimale Energiefunktion und Umsetzen des Grundzustandes in einen optimalen Bearbeitungsfahrweg.

Furthermore, the object is achieved by a method for generating an optimized processing path for agricultural work vehicles on a field by means of a quantum annealing system comprising the steps:

• - Providing one or more field-specific data, wherein the field-specific data comprises at least the coordinates for the field boundary as well as the coordinates of a number of points to be traveled, wherein the points to be traveled cover the entire field sufficiently accurately and wherein the points to be traveled comprise at least one start point and one end point,
• - Providing one or more work vehicle-specific data, which include at least the optimal speed of the agricultural work vehicle depending on consumption,
• - generating connecting edges based on recognized neighborhood relationships between all points as well as at least one end point and at least one start point,
• - Providing the various processing paths using different matrices, which specify the respective points (N) to be traveled as qubits taking time into account
• - Forming an energy function formed as a cost function using the qubits, based on an energy for an acceleration of the agricultural work vehicle as well as the deviation from the specified optimal speed as well as discretized points in time as time
• - Associating a quantum annealing system with an energy landscape corresponding to the energy function, where a ground state of the quantum annealing system encodes a minimization of the energy function under the following conditions:
  • - the agricultural work vehicle passes all points at least once,
  • - the agricultural work vehicle correctly drives to at least one starting point at a first point in time and to at least one end point at a last point in time,
  • - the agricultural work vehicle is located at exactly one of the points to be travelled at a discretised point in time between the first point in time and the last point in time,
  • - the agricultural work vehicle does not cross the field boundary,
• - Approaching the ground state of the quantum annealing system and reading out the ground state as a minimum energy function and converting the ground state into an optimal processing path.

The points to be traveled can be generated, for example, using a classic computer system and, for example, a current field view taken by a drone.

This means that, for example, the nature of the ground can also be taken into account as an additional condition or when determining the optimal speed.

Previous route planning systems for agricultural work vehicles can only generate routes for a few points to be travelled, i.e. smaller fields or no fields with many cuts/craggy fields. In addition, the generation is limited under the various conditions and with regard to the optimization variables, as otherwise conventional computers quickly reach their computing capacity.

Due to computing capacity, it has not been possible to optimize such processing paths using computers. Conventional optimization proves to be very inefficient or unsolvable, particularly for larger fields or fields with many cuts. However, optimization using quantum computers must meet certain requirements/conditions that make optimization possible in the first place.

The invention provides a computer program for optimizing an arrangement of a transmission using quantum computing. The computer program comprises instructions that cause a quantum computing hardware module, a classical hardware module inspired by quantum computing, or a special-purpose hardware module inspired by quantum computing to carry out the steps of the method according to the invention or the route planning system when the computer program runs on one of these hardware modules.

Thus, the route planning system/method may be based on hybrid quantum-classical information processing, which is performed on a hybrid quantum-classical information processing system comprising one or more classical digital computers and the quantum annealing system.

A quantum annealing system can simultaneously explore all potential solutions to a quadratic binary optimization problem (QUBO) using quantum annealing.

According to the invention, the method/route planning system recognizes the neighborhood relationships of the various points to be traveled to, on the basis of which permitted processing routes can be defined by means of binary variables.

Furthermore, the method/route planning system forms a cost function as an energy function, particularly on a digital computer, and associates the quantum annealing system with an energy landscape corresponding to the energy function.

In particular, it is also a technical task to adapt the agricultural problem to be solved in such a way that it can be solved energetically by means of a quantum annealing process.

Furthermore, the constraints are formulated according to the invention in such a way that an optimal processing path is determined, for example for processing a field for an agricultural machine, which is minimized with regard to the time required, the energy, the consumption of the machine on the basis of a negative or positive acceleration.

The constraints can be added to the energy function using a Lagrange multiplier and must be taken into account to reach the ground state.

In adiabatic quantum computing, the problem is translated into the energetic structure of a quantum system. By gradually changing the system, the ground state is reached as a target state, from which the solution can be read and is converted into an optimal processing path by the route planning system/procedure based on the read qubits.

By reaching the basic state, an optimal processing path can be generated taking into account the above-mentioned constraints, even for very large fields with many points to be traveled. The optimal processing path can now be determined by the route planning system/procedure based on the minimum energy function determined.

Furthermore, the route planning system for generating the energy function can have a digital computer with which the quantum annealing system interacts.

By means of such a route planning system/method according to the invention, it is now possible to determine an optimized processing route under the given conditions, in particular in the case of large fields or winding fields, which are characterized by a high number of points to be traveled.

Furthermore, frequent braking/starting as well as time and energy consumption can now be taken into account when generating the optimal machining path.

In particular, by setting a start point and an end point and minimizing the start point and the end point, transfer points can also be generated in the field to transfer the crops to another machine, for example.

In a further embodiment, a display unit, for example a display, is provided, in particular on the integrated digital computer, which is designed to display the optimized processing route. This can then be transmitted, for example, to the driver in the agricultural work vehicle for processing the field.

In a further development, the route planning system is designed to generate setting parameters for autonomous operation of the agricultural work vehicle based on the optimized processing route. In particular, these can be generated by the digital computer. This requires a description of the vehicle dynamics/actuators of the agricultural work vehicle, which can be stored in the memory unit of the digital computer.

In a further development, a manual input unit is provided for weighting the individual factors of the energy function. This can also be provided in the digital computer, for example.

Furthermore, the work vehicle-specific data can also include the working width of the agricultural work vehicle. This means that, for example, the points generated to be traveled can be adjusted accordingly by the route planning system so that there is as little overlap as possible when traveling the processing route.

The field-specific data may also include additional environmental data from the surroundings/environment of the agricultural work vehicle recorded by a sensor system of the agricultural work vehicle.

This data can be incorporated into the conditions with appropriate modifications, for example the energy function/optimal speed can be adjusted accordingly.

wobei ${\overrightarrow{n}}_i$

i = 1.., N gleich die Anzahl von abzufahrenden Punkten ist, und f eine Anzahl von Punktetupeln, so dass ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

jeweils benachbarte Punkte sind, γ jeweils einen abzufahrenden Bearbeitungsfahrweg darstellt, welcher aus einer Anzahl von Punktetupeln besteht, so dass $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

jeweils benachbarte Punkte und als Verbindungskante

verbunden sind und γ(t_i) der zeitlich diskretisierte Bearbeitungsfahrweg ist und (||γ̇(t_i)|| - v_opt) die Abweichung von der optimalen Fahrgeschwindigkeit ist, und γ̈(t_i) die Kosten für die positive und negative Beschleunigung sind, und wobei $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

sicherstellt, dass jeder Punkt auf dem Feld abgefahren wird, anhand von negativen Kosten.In further development, the energy function H is given by : $$H={\displaystyle \sum _{i=1}^f\left(\Vert \ddot{\gamma }\left(t_i\right)\Vert +\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where ${\overrightarrow{n}}_i$

i = 1.., N is the number of points to be traveled, and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points, γ represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

adjacent points and as a connecting edge

are connected and γ(t _i ) is the time-discretized processing path and (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and where $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that every point on the field is visited, based on negative costs.

The energy function corresponds to a quadratic binary energy function H with binary variables.

wobei ${\overrightarrow{n}}_i,$

i = 1.., N gleich die Anzahl von abzufahrenden Punkten, und f eine Anzahl von Punktetupeln ist, so dass ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

jeweils benachbarte Punkte sind, γ jeweils einen abzufahrenden Bearbeitungsfahrweg darstellt, welcher aus einer Anzahl von Punktetupeln besteht, so dass $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

jeweils benachbarte Punkte und als Verbindungskante

verbunden sind und γ(t_i) der zeitlich diskretisierte Bearbeitungsfahrweg ist, und wobei (||γ̇(t_i)|| - v_opt) die Abweichung von der optimalen Fahrgeschwindigkeit ist, und γ̈(t_i) die Kosten für die positive und negative Beschleunigung sind, und $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

sicherstellt, dass jeder Punkt auf dem Feld abgefahren wird anhand von negativen Kosten und wobei c₀, c₁, c₂ Gewichtungsfaktoren sind.Alternatively, the energy function H is given by : $$H={\displaystyle \sum _{i=1}^f\left(c_0\Vert \ddot{\gamma }\left(t_i\right)\Vert +c_1\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="DE102023202823A1\_0101.png"\; state="\{\{state\}\}">c</figure-callout>}}_2{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where ${\overrightarrow{n}}_i,$

i = 1.., N is the number of points to be traversed, and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points, γ represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

adjacent points and as a connecting edge

and γ(t _i ) is the time-discretized processing path, and where (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that each point on the field is visited using negative costs and where c ₀ , c ₁ , c ₂ are weighting factors.

This allows wear to be minimized as a function of acceleration or energy consumption as a minimal deviation from an optimal travel speed. Furthermore, the optimal processing path can be influenced based on the weighting.

Furthermore, in particular a display unit can be provided for displaying the processing path and for manually entering the weighting factors.

• 1: ein Routenplanungssystem gemäß der Erfindung,
• 2: einen ersten Bearbeitungsfahrweg,
• 3: ein zweiter Bearbeitungsfahrweg.

Further features and advantages of the present invention will become apparent from the following description with reference to the accompanying figures, which schematically show:

• 1 : a route planning system according to the invention,
• 2 : a first processing path,
• 3 : a second processing path.

The route planning system 1 has a storage unit 2 in which several field-specific data are stored. The storage unit 2 can be arranged on an integrated digital computer 5.

des Arbeitsfahrzeugs, wobei die Punkte $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

das gesamte Feld hinreichend genau abdecken.These field-specific data include at least the coordinates of a number ℕ of points to be traveled $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

of the work vehicle, whereby the points $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

cover the entire field with sufficient accuracy.

Therefore $$\mathbb{N}=\left\{{\overrightarrow{n}}_1,\dots ,{\overrightarrow{n}}_N\right\}$$

Furthermore, the storage unit 2 has the coordinates for the field boundary FU.

auf: $$\mathbb{S}\subset \left\{\mathrm{1,}\dots ,N\right\}$$

und zumindest einen oder eine Menge von Endpunkten IF, welche durch: $$\mathbb{F}\subset \left\{\mathrm{1,}\dots ,N\right\}$$

gegeben ist.In addition, the storage unit 2 has at least one or a set of starting points

on: $$\mathbb{S}\subset \left\{\mathrm{1,}\dots ,N\right\}$$

and at least one or a set of endpoints IF, which are defined by: $$\mathbb{F}\subset \left\{\mathrm{1,}\dots ,N\right\}$$

is given.

einem Endpunkt

als auch den abzufahrenden Punkten $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

als auch die Feldumrandung FU. 2 as well as 3 shows a field with a starting point

an endpoint

as well as the points to be traveled $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

as well as the field border FU.

werden Verbindungskanten

anhand erkannter nachbarschaftlicher Beziehungen zwischen allen Punkten $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

als auch dem zumindest einen Endpunkt

und dem zumindest einen Startpunkt

generiert. Dies bedeutet, dass zunächst die jeweiligen Nachbarspunkte eines Punktes erkannt werden, und anhand dessen Verbindungskanten

generiert werden.Based on the points to be traveled $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

connecting edges

based on recognized neighborly relationships between all points $\left\{{\overrightarrow{n}}_i,i=\mathrm{1,}\dots ,N\right\}$

as well as at least one endpoint

and at least one starting point

This means that first the respective neighboring points of a point are recognized and based on this connecting edges

generated.

stellen somit die fahrbaren Verbindungen zwischen den nächsten Nachbarspunkten $\left\{{\overrightarrow{n}}_i,{\overrightarrow{n}}_j,i,j=\mathrm{1,}\dots ,N\right\}$

von ℕ dar, wobei die Verbindungskanten

gegeben sind durch: $$\mathbb{E}=\left\{\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_j\right),\dots ,\right\}$$

The connecting edges

thus provide the mobile connections between the nearest neighboring points $\left\{{\overrightarrow{n}}_i,{\overrightarrow{n}}_j,i,j=\mathrm{1,}\dots ,N\right\}$

of ℕ, where the connecting edges

are given by: $$\mathbb{E}=\left\{\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_j\right),\dots ,\right\}$$

A connection may or may not be allowed.

kann als Matrix dargestellt werden: $$A_{{\overrightarrow{n}}_i,{\overrightarrow{n}}_j}=\{\begin{array}{c}1wenn\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_j\right)\in E\\ 0sonst\end{array}$$

The set of allowed connecting edges

can be represented as a matrix: $$A_{{\overrightarrow{n}}_i,{\overrightarrow{n}}_j}=\{\begin{array}{c}1wenn\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_j\right)\in E\\ 0sOnst\end{array}$$

erlaubt, wenn es sich um benachbarte, befahrbare Punkte handelt. Eine Verbindungskante ist beispielsweise nicht erlaubt, wenn es sich um zwei Punkte zwischen denen die Feldumrandung FU liegt handelt, d.h. eine Verbindungskante

darf die Feldumrandung FU nicht kreuzen.A connecting edge

allowed if they are adjacent, drivable points. For example, a connecting edge is not allowed if it is two points between which the field boundary FU lies, ie a connecting edge

must not cross the field boundary FU.

welche die Diskretisierung der Zeit darstellt.Furthermore, there is a discrete set of time steps T: $$T_f=\left\{t_1,\dots ,t_f\right\}$$

which represents the discretization of time.

definiert werden, wobei $\left({\overrightarrow{n}}_{i,l},{\overrightarrow{n}}_{i,l+1}\right)\subset \mathbb{E}$

jeweils benachbarte Punkte und als Verbindungskante

verbunden sind: $$\gamma =\left({\overrightarrow{n}}_{i1},{\overrightarrow{n}}_{i2},\dots {\overrightarrow{n}}_{il}\right)$$

welcher durch eine zeitliche Diskretisierung als Bearbeitungsfahrweg γ(t_i) unterteilt werden kann.Then a processing path y can be defined by point tuples ${\overrightarrow{n}}_{il}\in N$

be defined, where $\left({\overrightarrow{n}}_{i,l},{\overrightarrow{n}}_{i,l+1}\right)\subset \mathbb{E}$

adjacent points and as a connecting edge

are connected: $$\gamma =\left({\overrightarrow{n}}_{i1},{\overrightarrow{n}}_{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="DE102023202823A1\_0101.png"\; state="\{\{state\}\}">i</figure-callout>}2},\dots {\overrightarrow{n}}_{il}\right)$$

which can be divided into the processing path γ(t _i ) by a temporal discretization.

The machining path γ(t _i ) can be represented as binary variables using a matrix M: $$M:=\left(\begin{array}{ccc}{x}_{t\left(1\right),n\left(1\right)}& \cdots & x_{t\left(1\right)n\left(N\right)}\\ ⋮& \ddots & ⋮\\ x_{t\left(f\right),n\left(1\right)}& \cdots & x_{t\left(1\right),n\left(1\right)}\end{array}\right)$$

an.Here x is the location of the vehicle at time t _i , i = 1,.. f at a point to be traveled $N=\left\{{\overrightarrow{n}}_1,\dots ,{\overrightarrow{n}}_N\right\}$

to.

Furthermore, a velocity vector ̇γ̇(t _i ) of the discretized machining path γ(t _i ) can be defined as: $$\dot{\gamma }\left(t_i\right):=\{\begin{array}{c}\frac{\gamma \left(t_i\right)-\gamma \left(t_{i-1}\right)}{t_i-t_{i-1}},wenni>1\\ {\left(\mathrm{0,0,0}\right)}^T,sOnst\end{array}$$

Furthermore, an acceleration vector γ̈(t _i ) of the discretized machining path γ(t _i ) can be defined as: $$\ddot{\gamma }\left(t_i\right):=\{\begin{array}{c}\frac{\dot{\gamma }\left(t_i\right)-\dot{\gamma }\left(t_{i-1}\right)}{t_i-t_{i-1}},wenni>2\\ {\left(\mathrm{0,0,0}\right)}^T,sOnst\end{array}$$

wobei

• (||γ̇(t_i)|| - v_opt) die Abweichung von der optimalen Fahrgeschwindigkeit v_opt darstellt,
• γ̈(t_i) die Energiekosten für die positive und negative Beschleunigung sind, beispielsweise die Energiekosten für scharfe Kurven, d.h. die Beschleunigung oder die Krümmung des Weges,
• $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$
  
  sicherstellt, dass jeder Punkt auf dem Feld abgefahren wird anhand von negativen Energiekosten.
• und wobei c₀, c₁, c₂ Gewichtungsfaktoren sind.

The energy function H can then be generated as follows: $$H={\displaystyle \sum _{i=1}^f\left(c_0\Vert \ddot{\gamma }\left(t_i\right)\Vert +c_1\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="DE102023202823A1\_0101.png"\; state="\{\{state\}\}">c</figure-callout>}}_2{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where

• (||γ̇(t _i )|| - v _opt ) represents the deviation from the optimal driving speed v _opt ,
• γ̈(t _i ) are the energy costs for the positive and negative acceleration, for example the energy costs for sharp curves, i.e. the acceleration or the curvature of the path,
• $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$
  
  ensures that every point on the field is visited using negative energy costs.
• and where c ₀ , c ₁ , c ₂ are weighting factors.

The route planning system 1, in particular the digital computer 5, can have a display unit 3 for manually entering the weighting factors c ₀ , c ₁ , c ₂ .

According to the invention, the energy function is now formulated as a quadratic binary energy function H with binary variables.

The energy function H is thus formulated as a quadratic binary optimization problem (QUBO) with the corresponding constraints.

befindet.The constraint is that the agricultural work vehicle is at a discretized time t _i , i = 1,.. f between the first time and the last time at exactly one of the points to be travelled $\mathbb{N}=\left\{{\overrightarrow{n}}_1,\dots ,{\overrightarrow{n}}_N\right\}$

is located.

This can be expressed by: $${\displaystyle \sum _{i=1}^N{x}_{t\left(i\right),n\left(j\right)}=1\forall iϵ}\left\{\mathrm{1,}\dots f\right\}$$

The constraint can be added to the energy function H using a Lagrange operator λ; and must be taken into account when reaching the ground state: $$H+\lambda {\displaystyle \sum {}_{i=1}^N{x}_{t\left(i\right),n\left(j\right)}-\lambda \ast 1}$$

The condition also applies that the agricultural work vehicle visits all points at least once.

This can be expressed by: $${\displaystyle \sum _{i=1}^f{x}_{t\left(i\right),n\left(j\right)}\ge 1\forall jϵ}\left\{\mathrm{1,}\dots N\right\}$$

The constraint can be added to the energy function H using a Lagrange operator and must therefore be taken into account when reaching the ground state.

gesetzt werden durch: $${\displaystyle \sum _{j\in \mathbb{S}}{x}_{t\left(1\right),n\left(j\right)}=1}$$

Furthermore, the starting point

be set by: $${\displaystyle \sum _{j\in \mathbb{S}}{x}_{t\left(1\right),n\left(j\right)}=1}$$

The constraint can be added to the energy function H using a Lagrange operator β; and must also be taken into account when reaching the ground state: $$H+\beta {\displaystyle \sum {}_{j\in \mathbb{S}}{x}_{t\left(1\right),n\left(j\right)}}-\beta \ast 1$$

gesetzt werden durch: $${\displaystyle \sum _{j\in \mathbb{F}}{x}_{t\left(f\right),n\left(j\right)}}=1$$

Likewise, the endpoint

be set by: $${\displaystyle \sum _{j\in \mathbb{F}}{x}_{t\left(f\right),n\left(j\right)}}=1$$

The constraint can also be expressed with a Lagrange operator θ to the energy function H must be added; and must be taken into account when reaching the ground state. $$H+\theta {\displaystyle \sum {}_{j\in \mathbb{F}}{x}_{t\left(f\right),n\left(j\right)}-\theta \ast 1}$$

an einem ersten Zeitpunkt t(1) und den zumindest einen Endpunkt

an einem letzten Zeitpunkt t(f) korrekt an.Under these conditions, the agricultural work vehicle drives to at least one starting point

at a first time t(1) and the at least one end point

at a final time t(f) correctly.

Furthermore, the condition applies that the processing path γ(t _i ) must not cross a field boundary FU at any time.

This can be expressed by: $${\displaystyle \sum _{j,j\text{'}}^N{x}_{t\left(i\right),n\left(j\right)}{x}_{t\left(i+1\right),n\left(j\text{'}\right)}{A}_{j,j\text{'}}}=1\forall i=\left\{\mathrm{1,}\dots ,f\right\}$$

The constraint can also be added to the energy function H using a Lagrange operator; and must be taken into account when reaching the ground state.

The energy function H is then minimized using the quantum annealing system 4.

At the beginning of the calculation, the qubits in the quantum annealing system 4 are associated with an energy landscape corresponding to the energy function H. This can be done using the digital computer 5. The energy function H is then uniformly changed towards the ground state. The unknown ground state of the energy function H encodes the minimization by measuring the completed evolution of the quantum bits.

The quantum annealing system 4 explores all potential solutions simultaneously in the quadratic binary optimization problem (QUBO) by quantum annealing to find the minimum energy state.

Digital annealing maps the effects of quantum computing, such as tunneling effects or superposition, in special hardware. This results in significant speed advantages over classic IT architectures; in addition, the available computing capacity does not generally allow the above problem to be solved.

Furthermore, the route planning system 1 can read out the basic state generated in this way as a minimum energy function and convert the read out basic state into an optimal processing path.

The optimal processing route can then be displayed to the driver in the agricultural work vehicle.

If it is an autonomously operated agricultural work vehicle, the route planning system 1/ the procedure can generate setting parameters and transmit them to the autonomously operated agricultural work vehicle.

2 and 3 show two different machining paths. The machining path in 2 has many curves and many points that are crossed twice, which is inefficient and costly.

The processing path in 3 on the other hand, has fewer curves and fewer points that are approached twice.

The route planning system 1/the method finds a processing route which is optimized even for large fields with a high number of points to be traveled.

List of reference symbols

1

route planning system

2

storage unit

3

display unit

4

quantum annealing system

5

digital computer

ti, i = 1,.. f

discretized points in time

FU

field border

γ(ti)

Discretized machining path

γ̈(ti)

velocity vector

γ̈(ti)

acceleration vector

starting point

endpoint

ti, i = 1,..f

discretized points in time

connecting edge

c0, c1, c2

weighting factors

H

energy function

points to be traveled

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• EP 1795986 A2 [0006]
• EP 2267566 A2 [0007]
• US 7216033 B2 [0008]

Claims (12)

Route planning system (1) for agricultural work vehicles for generating an optimized processing route on a field by means of a quantum annealing system (4) comprising a storage unit (2) for one or more field-specific data, wherein the field-specific data have at least the coordinates for the field border (FU) as well as the coordinates of a number of points (ℕ) to be traveled, wherein the points (ℕ) to be traveled cover the entire field sufficiently accurately and wherein the points (ℕ) to be traveled have at least one starting point $\left(\mathbb{S}\right)$

and an endpoint $\left(\mathbb{F}\right)$

and one or more work vehicle-specific data, which include at least the optimal speed of the agricultural work vehicle depending on consumption, wherein the route planning system (1) is designed to connect edges $\left(\mathbb{E}\right)$

based on recognized neighborhood relationships between all points (ℕ) as well as at least one endpoint $\left(\mathbb{F}\right)$

and at least one starting point $\left(\mathbb{S}\right)$

to generate, and wherein the route planning system (1) is designed to form different processing routes based on different matrices which specify the respective points (ℕ) to be traveled to as qubits taking time into account, and wherein the route planning system (1) is designed to generate an energy function (H) formed using the qubits as a cost function, based on an energy for an acceleration of the agricultural work vehicle as well as the deviation from the predetermined optimal speed as well as from discretized points in time (t _i ) as time, and wherein a quantum annealing system (4) is provided which is associated with an energy landscape corresponding to the energy function (H), and wherein a basic state of the quantum annealing system (4) encodes a minimization of the energy function (H) under the following conditions - the agricultural work vehicle travels to all points (ℕ) at least once, - the agricultural work vehicle travels to the at least one starting point $\left(\mathbb{S}\right)$

at a first point in time and at least one end point $\left(\mathbb{F}\right)$

correctly at a last point in time, - the agricultural work vehicle is at exactly one of the points to be traveled (ℕ) at a discretized point in time between the first point in time and the last point in time, - the agricultural work vehicle does not cross any field boundary (FU), wherein the quantum annealing system (4) is designed to approximate the ground state and wherein the route planning system (1) is designed to read out the ground state as a minimum energy function (H) and to convert the ground state into an optimal processing route. Route planning system (1) to Claim 1 , characterized in that the route planning system (1) comprises a digital computer (5) with a display unit (3) and a coupling to the quantum annealing system (4), wherein the display unit (3) is designed to display the optimized processing route. Route planning system (1) to Claim 1 or 2 , characterized in that the route planning system (1) is designed to generate setting parameters for an autonomous operation of the agricultural work vehicle on the basis of the optimized processing route. Route planning system (1) according to one of the preceding claims, characterized in that a manual input unit is provided for weighting the individual factors of the energy function (H). Route planning system (1) according to one of the preceding claims, characterized in that the agricultural work vehicle has a defined working width and the work vehicle-specific data additionally include the working width of the agricultural work vehicle. Route planning system (1) according to one of the preceding claims, characterized in that the field-specific data additionally comprise environmental data of the surroundings of the agricultural work vehicle recorded by a sensor system of the agricultural work vehicle. Route planning system (1) according to one of the preceding claims, characterized in that the energy function (H) is given by: $$H={\displaystyle \sum _{i=1}^f\left(\Vert \ddot{\gamma }\left(t_i\right)\Vert +\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where ${\overrightarrow{n}}_i,$

i = 1.., N is the number of points to be traversed, and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points (ℕ), y represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

neighboring ones points (ℕ) and as connecting edge $\left(\mathbb{E}\right)$

and γ(t _i ) is the time-discretized processing path and where: (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that every point on the field is visited using negative costs. Route planning system (1) according to one of the preceding Claims 1 until 6 , characterized in that the energy function (H) is given by: $$H={\displaystyle \sum _{i=1}^f\left(c_0\Vert \ddot{\gamma }\left(t_i\right)\Vert +c_1\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-c_2{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where ${\overrightarrow{n}}_i,$

i = 1.., N is the number of points to be traversed (ℕ), and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points (ℕ), y represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

adjacent points (ℕ) and as a connecting edge

are connected, and γ(t _i ) is the time-discretized processing path, and where (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that each point on the field is visited using negative costs, and where c ₀ , c ₁ , c ₂ are weighting factors. Method for generating an optimized processing route for agricultural work vehicles on a field by means of a quantum annealing system (4), comprising the steps of: - providing one or more field-specific data, wherein the field-specific data have at least the coordinates for the field boundary (FU) as well as the coordinates of a number of points (ℕ) to be traveled, - wherein the points (ℕ) to be traveled cover the entire field with sufficient accuracy and wherein the points (ℕ) to be traveled have at least one starting point $\left(\mathbb{S}\right)$

and an endpoint $\left(\mathbb{F}\right)$

- providing one or more work vehicle-specific data, which include at least the optimal speed of the agricultural work vehicle depending on consumption, - generating connecting edges $\left(\mathbb{E}\right)$

based on recognized neighborhood relationships between all points (ℕ) as well as at least one endpoint $\left(\mathbb{F}\right)$

and at least one starting point $\left(\mathbb{S}\right),$

- Providing the various processing routes using different matrices which specify the respective points (ℕ) to be travelled to as qubits taking time into account - Using the qubits to form an energy function (H) formed as a cost function, based on an energy for an acceleration of the agricultural work vehicle as well as the deviation from the specified optimal speed as well as discretised points in time (t _i ) as time - Associating a quantum annealing system (4) with an energy landscape corresponding to the energy function (H), wherein a ground state of the quantum annealing system (4) encodes a minimisation of the energy function (H) under the following conditions: - the agricultural work vehicle travels to all points (ℕ) at least once, - the agricultural work vehicle travels to at least one starting point $\left(\mathbb{S}\right)$

at a first point in time and at least one end point at a last point in time, - the agricultural work vehicle is at exactly one of the points to be traveled (ℕ) at a discretized point in time between the first point in time and the last point in time, - the agricultural work vehicle does not cross any field boundary (FU), - approaching the ground state of the quantum annealing system (4) and reading out the ground state as a minimum energy function (H) and converting the ground state into an optimal processing path. procedure according to claim 9 , characterized in that a weighting of the individual conditions is achieved. procedure according to claim 9 or 10 , characterized in that the energy function (H) is given by: $$H={\displaystyle \sum _{i=1}^f\left(\Vert \ddot{\gamma }\left(t_i\right)\Vert +\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

Where ${\overrightarrow{n}}_i,$

i = 1.., N is the number of points to be traversed, and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points (ℕ), γ represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

adjacent points (ℕ) and as a connecting edge

tied together and γ(t _i ) is the time-discretized processing path and where (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that every point on the field is visited using negative costs. Method according to one of the preceding Claims 9 or 10 , characterized in that the energy function (H) is given by: $$H={\displaystyle \sum _{i=1}^f\left(c_0\Vert \ddot{\gamma }\left(t_i\right)\Vert +c_1\left(\Vert \dot{\gamma }\left(t_i\right)\Vert -v_{Opt}\right)-c_2{\displaystyle \sum _{j=1}^N{x}_{i,j}}\right)}$$

where ${\overrightarrow{n}}_i,$

i = 1.., N is the number of points to be traversed, and f is a number of point tuples such that ${\overrightarrow{n}}_i,{\overrightarrow{n}}_j$

are adjacent points, y represents a processing path to be traveled, which consists of a number of point tuples, so that $\left({\overrightarrow{n}}_i,{\overrightarrow{n}}_{i+1}\right)\subset E$

adjacent points and as a connecting edge

are connected, and γ(t _i ) is the time-discretized processing path, and where (||γ̇(t _i )|| - v _opt ) is the deviation from the optimal travel speed, and γ̈(t _i ) are the costs for the positive and negative acceleration, and $-{\displaystyle \sum {}_{j=1}^N}{x}_{i,j}$

ensures that each point on the field is visited using negative costs, and where c ₀ , c ₁ , c ₂ are weighting factors.

Images