EP3877902A1

EP3877902A1 - Method for controlling the operation of a machine for harvesting root crops

Info

Publication number: EP3877902A1
Application number: EP19808698.5A
Authority: EP
Inventors: Wolfram STROTHMANN; Daniel Bösenberg
Original assignee: Grimme Landmaschinenfabrik GmbH and Co KG
Current assignee: Grimme Landmaschinenfabrik GmbH and Co KG
Priority date: 2018-11-07
Filing date: 2019-11-05
Publication date: 2021-09-15
Also published as: WO2020094654A1; BR112021005034A2; CA3118776A1; EA202191204A1; DE102018127844A1; JP2022506703A; JP7339336B2; CN112997194A; IL282955A; US20210378167A1

Abstract

The invention relates to a method for controlling the operation of a machine (2) for harvesting root crops (4), in which method at least one test image (8) of harvested material comprising root crops (4) moved relative to a machine frame (12) by means of at least one conveying element (10) is recorded by at least one optical image acquisition unit (6), and a conveying speed of the conveying element (10) is adjusted on the basis of a test data set generated with the aid of the test image (8) or formed by same, wherein an evaluation device generates a conveying speed signal independent of a speed of the harvested material on the basis of the test data set in order to adjust the conveying speed. The invention also relates to the aforementioned machine.

Description

Process for regulating the operation of a machine for harvesting root crops

The invention relates to a method for regulating the operation of a machine for harvesting root crops and the machine. According to the method, at least one test image of crops advanced by means of at least one conveyor element relative to a machine frame is recorded by at least one optical image acquisition unit. The crop includes root crops. A conveying speed of the conveying element is set on the basis of a test data record generated on the basis of the test image or formed by the test image.

The test image shows crop that was previously picked up by the machine for harvesting root crops. The conveyor element as part of the machine serves the locomotion of the crop within the machine, with at least part of the crop being in direct contact with the conveyor element. The conveying speed at which the conveying element moves is set with the test data record.

The laid-open specification US 2018/0047177 A1 discloses a method in which a speed of the conveying element is determined on the basis of the test image. The conveying speed of the conveying element is adjusted on the basis of this determined speed. A disadvantage of the known, generic methods is that, depending on the harvesting conditions, associated separation units cause significant damage to the root crops or to a large amount of admixtures among the root crops unloaded from the machine. In addition, it is therefore generally proposed in US 2018/0047177 A1 to change the harvesting rate or one or more configurations of the machine as a function of a server-based evaluation of three-dimensional data of the crop material recorded by the sensors of the machine.

The object of the present invention is to provide a method for optimizing the utilization of the conveying element in favor of an improved protection of the root crops.

According to the invention the object is achieved by a generic method in which an evaluation device based on the test data set generates a conveyor speed signal independent of a speed of the crop to adjust the conveyor speed. On the basis of the evaluation device, who the or the test data sets for the conveying speed signal, on which the speed of the crop shown with the test image and in particular the conveyor element has no influence. The speed of the crop is understood to be the direction-independent speed of the crop transported by means of a conveyor element. Accordingly, under a (conveying) speed of the conveying element, its direction-independent speed amount understood, for example, the speed of rotation of a belt, a screen star or a roller.

The machine is a self-propelled or towed vehicle for harvesting root crops, especially potatoes, beets, carrots or chicory. During the execution of the method according to the invention, the machine is moved in particular in the direction of rows, in particular planting embankments, of the root crops to be harvested and these are taken up as part of the crop in a continuous process from the ground. After picking up the crop, it is at least partially moved relative to the machine frame of the machine by the at least one conveyor element. The conveying element also serves to separate the root crops from additions and is in particular part of a separating device comprising at least one separating element for separating the root crops arranged on the conveying element from additions arranged on the conveying element.

Alternatively, the machine can also be a machine for separating root crops from additions to the crop, e.g. of clods, stones or earth.

In particular, the conveying element is rotating or rotating. The conveying element is preferably designed as a sieve star, sieve belt, hedgehog belt, ring elevator or as a conveyor roller, in particular comprising a roller table. The Förderge speed is in particular a lateral speed of the crop contacting conveyor element section, a circulation or a rotational speed. By changing the conveying speed, the density or filling height of the crop, in particular the root crops, on the conveying element is varied.

For this purpose, the optical image acquisition unit is arranged in a stationary manner on the machine, in particular above the conveyor element. The image acquisition unit is directed at the conveying element and thus during operation at a flow of crops conveyed by the conveying element. The method according to the invention is carried out in particular exclusively during the harvesting with the machine and is preferably repeated cyclically. In particular, transitions between individual conveying elements, e.g. in the form of drop levels, and supply and discharge conveying elements in front of and behind separating elements of separating devices.

The test image is, in particular, a multidimensional, preferably two-dimensional, image, on which at least part of the harvested crop with root crops, quantities and / or the conveying element is depicted. On the basis of the test image recorded by the image acquisition unit, the test data record is either already generated by the image acquisition unit or by the evaluation device. Alternatively, the test data record can be formed by the test image itself. This applies in particular to image acquisition units whose test images already have a format suitable for the subsequent analysis in the evaluation device. The test data record is, in particular, one that is created by processing, for example filtering and / or other images, and is at least temporarily available in the system. gender data set, the information of which, for example, color values, is evaluated in the evaluation device. It can be available as an image file, table, matrix or vector field, for example. The test image or the test data record is transmitted from the image acquisition unit to the evaluation device. The optical image capture unit is designed in particular as a digital photo or video camera for two-dimensional recording of the test image or as a line scan camera. Insofar as reference is subsequently made to the test image in connection with the processing of the image information in the evaluation device, this can be the test data record in this context.

The evaluation device is used to evaluate the test data set. The evaluation device comprises at least one processor and is designed either as a central computing unit or as a decentralized system comprising at least one processor and at least one memory with different positions on components of the machine. It is therefore a local system to carry out any evaluations directly on site and to make the results available directly.

The conveying speed signal for setting the conveying speed is sent by the evaluation device, in particular to a conveying speed control device of the machine or the conveying element itself. The conveying speed signal is preferably a digital data record which is transmitted in a wired or wireless manner. This preferably triggers an increase, a decrease or a maintenance of the present conveying speed. The Conveyor speed signal corresponds in particular to a load characteristic value calculated on the basis of the test data set, which characterizes the workloads of the conveyor element, and is interpreted in particular by the conveyor speed control device. In particular, an electrical signal is emitted by the conveyor speed control device, which in an advantageous embodiment of the invention is comprised of the same computing unit as the evaluation device. By means of the conveying speed control device, in particular by the electrical signal, in particular a hydraulic pressure, a pneumatic pressure, a current, a voltage, a force and / or a moment is adapted to drive the conveying element. In particular, in addition to the conveying speed signal, the conveying speed control device receives further signals, in particular with higher priority, by means of which it adjusts the conveying speed.

The advantage of a conveyor speed signal that is independent of a speed of crop flow objects is that the error entered by determining the crop or conveyor element speed, which makes it difficult or impossible to detect that the conveyor element is too small or too dense, is avoided. Instead, the Fördergeschwin speed signal is dependent on other dynamic, ie related to the movement of the crop or the conveying element, or stationary, ie on a movement of their independent, quantities. In a majority of these, a significant correlation to the utilization of the conveyor element could be determined, so that the utilization can now be reliably controlled using the conveyor speed signal. The necessary for the generation of the necessary signal capacity is also reduced. In particular, with a largely independent of the driving speed and / or of the crop picked up by the machine crop regulation of the occupancy of the conveying element, ie the density of crop on or in the conveying element, can be regulated. This way, the harvesting root crops are conveyed under optimal conditions and are optimally cleaned and damage avoided.

The setting of the conveying speed is carried out in particular automatically and independently of the driving speed of the self-propelled or towed machine, as a result of which operating personnel are less distracted.

The evaluation device preferably compares the test data record with an output data record generated on the basis of an output image or formed by the latter. Particularly preferably, the output data record created on the basis of the output image or formed by it was recorded with the same optical image acquisition unit in time before the test image. The output data record results from processing that is the same as that for processing the test data record. In particular, when comparing the test image and the output data set, brightness values, contrasts or color values are compared. By comparing the test data set with the output data set, an evaluation of the dynamic behavior of the crop is simplified and / or trends in the composition can be identified, as a result of which further information on the operating state of the machine and its development can be determined. On the basis of this information, a more functional conveying speed signal can be provided and it can be a temporary one improved utilization can be achieved. The comparison of brightness values, contrasts or color values can also include a statistical evaluation of these respective values.

The conveying speed signal for setting the conveying speed, which is generated on the basis of the test data set and the output data set, is preferably independent of the speed of the crop and / or of the conveying element, or at least only additionally. By taking the speed into account in addition, deviating influences relevant to the utilization of special root crops can be determined and an optimized conveying speed signal can be provided for these cases.

In an advantageous embodiment of the invention, the test data set serves as a starting data set for a first execution of the method in a further execution of the method. The test data record of a first embodiment of the method thus resembles the output data record of a further embodiment of the method. Alternatively, both a test image and an output image are recorded each time the method is carried out. In particular, the optical image acquisition unit records images with a frequency between 0.1 and 1000 Hertz, preferably a comparison of the test data record with the output data record with a lower frequency, in particular from 0.1 to 10 Hertz, takes place. With these process features, the conveying speed signal is based on a particularly high-resolution data basis and the process can be carried out efficiently, as it were. In particular, the method according to the invention is characterized in that the evaluation device determines the conveying speed signal on the basis of an evaluation of the optical flow of the crop resulting from the test and the output data set. The optical flow, which results from the test and the output data record, is a data record with movement information of the objects or objects visible in the test image, in particular in the reference system of the imaging optics of the image acquisition unit.

The evaluation device preferably calculates at least one movement characteristic data record, in particular for determining the optical flow. The movement characteristic data set identifies a movement, in particular a direction of movement, of at least one object which is at least partially imaged by the test image, in particular by a part of the test image. In particular, several objects can be imaged on at least part of the test image at the same time, so that the movement characteristic data record at least indirectly indicates the direction of movement thereof. The conveyor speed signal is generated on the basis of the movement characteristic data record.

The movement characteristic data record preferably contains only one piece of information or a numerical value or a plurality of pieces of information or numerical values. The movement characteristic data record is calculated in particular on the basis of both the test data record and the output data record or their comparison, alternatively only on the basis of the test data record. The movement characteristic data record contains an indication by which a movement of the object or objects at least partially depicted is at least partially specified. In particular, the movement characteristic data record has information on the direction. In the case of several objects, which may only be partially shown, the movement data record can have information on a plurality of directions or on an overall direction of movement. The object can be any at least partial representation of an imaged body with a physical extension, in particular in particular at least part of a root crop, a herb, a clod, earth, the conveying element, or combinations thereof.

The movement information of any objects or combinations of objects in the test image and in the output image are determined in the determination of the optical flow by comparing at least partially recoverable areas in both images. These retrievable areas can be, for example, the size of a pixel or can be characterized by a pixel, so that no object recognition in the sense of a detection of objects in the form of root crops, stones or the like is necessary.

By taking into account the movement characteristic data set which characterizes the movement of a depicted object, a detailed conclusion about a movement situation of the harvested crop can be calculated. In particular, a movement situation arises solely from the consideration of the direction of the movement, preferably without considering the speed. In particular, the conveying speed signal and a consequent change in conveying speed can influence the movement situation particularly continuously and in a manner that is particularly error-free and thus achieve an optimal throughput of crop.

The movement characteristic data record preferably contains two numerical values, on the basis of which a vector can be generated. The movement characteristic data record preferably comprises two lines in different directions or alternatively an angle and a line. As a result, at least one vector can be generated, which is preferably displayed to the user with the test image on a visualization unit. The user thereby receives an image of the movement situation and can check the conveying speed change made by the evaluation device, if desired, for its success.

For calculating the at least one movement characteristic data set, a test sub-data set that is generated on the basis of a first partial image area of the test image is compared with a plurality of output sub-data sets that are generated on the basis of further partial image areas of the output image. Alternatively, an output sub-data record that is generated on the basis of a first partial image area of the output image is compared with a plurality of test sub-data records that are generated on the basis of further partial image areas of the test image. For each comparison, a match between the respective test and initial sub-data sets is assessed. In each comparison, in particular exactly one test sub-data record with exactly one gear sub-record compared. A match between a test and an output sub-data record is particularly good when the sub-picture areas described by them have a great visual similarity. To determine the similarity, brightnesses, contrasts and / or color values can be compared.

The correspondence is evaluated in particular only on the basis of the respective test and output data sets, alternatively on the basis of further data of the test and output data sets. In another preferred embodiment, the correspondence is also assessed on the basis of further information that is not part of the test and output data sets and that is recorded in particular by sensors of the machine. In particular, an auxiliary variable, such as a rotational speed of the conveying element, is taken into account for the evaluation of the match. As a result, an expected positional deviation of two partial image areas from the test and the output data record is preferably predetermined and is included in the evaluation of the match.

The correspondence is preferably assessed on the basis of a contrast of the components of the test and output sub-data sets on which the partial image areas are based. In particular, a determined contrast of the first image area is compared with at least partially matching contrasts of the further partial image areas and the agreement of the contrasts is evaluated, in particular on the basis of a brightness or color gradient or a spatial extension of the contrast. Through this form of assessing the match of different Sub-image areas can be assigned to one another in a particularly reliable manner, and at least partially show the same object, and thereby track movement of the crop, regardless of whether complete parts of the crop, such as potatoes, stones or the like, are shown. As a result, the conveying speed signal can be calculated on the basis of a larger amount of information, and the speed of the harvesting machine can thereby be controlled particularly precisely depending on the movement situation.

The movement characteristic data record of an object shown by the first partial image area, in particular a movement direction encompassed by it, is particularly preferably calculated on the basis of position characteristic values of the test data record and the output data record, which are assigned to the two best matching test and output sub-data records. Both the test data record and the output data record thus contain positional values which represent the position under different image areas of the test image or the output image relative to other image areas or image reference markings or absolutely. The direction of movement is calculated in particular on the basis of a calculation of two differing position characteristic values, for which purpose the position characteristic values include position data of at least two different dimensions. The direction of movement thus indicates from where to where an image area or object represented by the test image or by the output image has moved between the recording of the output image and the recording of the test image, and is in particular different by two movement paths Reference directions defined. This allows a make particularly precise statements regarding the movement situation on the conveyor element and in particular determine whether the crop is blocked or unhindered movement.

In particular, the evaluation device divides both the test image and the output image into a plurality of image areas which are preferably of equal size, each image area of the test image or of the output image being assigned an image region of the output image or test image which best matches it. Each image area is based in particular on a test or output sub-data record. This enables a plurality of movement characteristic data sets, in particular directions of movement, to be determined and the movement situation to be determined in a higher resolution.

In an advantageous embodiment of the invention, a degree of correspondence characterizing the degree of correspondence between a test sub-data set and an output sub-data set has an influence on the conveying speed signal. Depending on how great the degree of correspondence between the best matched test and output sub-data sets is, the direction of movement calculated on the basis thereof has a different meaning in the calculation of the conveying speed signal. Thus, a clearly understandable movement of an object has a greater influence on the conveying speed signal than a movement that is supposed to be traced only on the basis of two relatively different test and output sub-data sets could. This increases the meaningfulness of the movement characteristic data sets as a whole and thus the value of the conveyor speed signal.

The evaluation device preferably generates a movement characteristic data set for different objects at least partially represented with the test image or different first image areas, which in particular comprise exactly one pixel of the test and / or the output image. In particular, one movement characteristic data record is determined for a plurality of test and / or output sub-data records, independently of objects shown by the respective images. A movement characteristic data record is particularly preferably generated in each case in particular comprising a movement direction for a plurality of pixels of the test and / or the output image. In particular, a movement characteristic data record is generated for each pixel of the test and / or the output image or, alternatively, preferably at least for each pixel of a selected, contiguous section of the test and / or the output image. This number of characteristic movement data records and the resolution used to determine them enable the movement situation on the conveyor element to be traced particularly precisely and the conveying speed to be set particularly closely based on the movement situation. This further increases the machine's efficiency.

In a first calculation step, the evaluation device preferably calculates a movement characteristic data record for a plurality of image areas each comprising at least a first number of pixels and in a later calculation step and taking into account those calculated in the first calculation step Motion characteristic data sets each a further movement characteristic data record for a higher number of deviating image areas which comprise a smaller number of pixels. In particular, in the first calculation step, the evaluation device calculates a movement characteristic data record for a smaller number of larger image areas and in the later calculation step a larger number of movement characteristic data records for smaller image areas, which, when put together, give the same overall picture as the larger picture areas. Thus, the movement characteristic data sets calculated in the last calculation step, which are assigned in particular to each pixel, are determined by an iterative approximation, and thus the probability of incorrect movement characteristic data sets, which in particular include movement directions that do not correspond to the real movement directions of the objects on the conveying element, minimized.

The at least one movement characteristic data record preferably includes at least temporarily a first movement path in a first direction and a second movement path in a second direction and / or a direction and / or a rich direction that deviates from the first, in particular by 90 ° in the image plane total movement distance independent of the direction. In particular, the direction and thus the direction of movement of the movement characteristic data set is calculated on the basis of the first and the second movement distance. The movement distances and / or the total movement distance are specified in particular as relative values which are dependent on the positioning, in particular on the orientation of the image acquisition unit and do not require any separate calibration. In an advantageous embodiment of the invention, the evaluation device calculates a utilization characteristic on the basis of at least one movement characteristic of the at least one movement characteristic data characterizing a direction of movement and in particular on the basis of at least one reference characteristic value assigned to the movement characteristic. In particular, only movement characteristics that characterize a direction of movement are used for the calculation of the movement characteristic data sets and in particular no further data based on the test and / or the output data set. The movement characteristic characterizes in particular only the direction of movement. In particular, the utilization characteristic value is calculated on the basis of its plurality of movement characteristic values, each movement characteristic value being part of a different movement characteristic data record. The movement characteristic value specifies in particular the extent of a movement in a transverse direction deviating from the conveying direction of the conveying element or specifies a direction of movement, for example an angular indication. The movement parameters are either offset as such to the utilization characteristic or initially offset against the assigned reference characteristic. The reference characteristic value specifies in particular an ideal or global direction in which the crop has to move. Deviations of the movement characteristic values from the reference characteristic value or the reference characteristic values are thus preferably used to calculate the utilization characteristic value.

The reference parameters are preferably all the same and indicate the direction in which the crop has to move overall and / or have have different values that assign each image area or test or output sub-record its own direction of movement to be compared.

In the case of a uniform reference characteristic value for several movement characteristic values, if there is no computational correction of a perspective-related distortion of the test image due to a lens of the image acquisition unit, there is in particular a basic deviation between the reference characteristic value and at least a large part of the movement characteristic values, even if no crop, e.g. is represented in the form of beets or potatoes or this has no movement component in a transverse direction deviating from the reference direction.

The evaluation device particularly preferably statistically evaluates a plurality of movement characteristic values, which are comprised by different movement characteristic data sets, in order to calculate the utilization characteristic value. The movement characteristics of the pixels of at least part of the test image are preferably involved. In particular, the evaluation device calculates a standard deviation of the movement characteristic values, which in particular characterize a direction of movement, from the respective reference characteristic values or from the uniform reference characteristic value. For this purpose, the amounts of the deviation from the movement and reference values are used in particular in that an average amount-based deviation is first calculated. The statistical evaluation of the movement characteristic values and thus the determination of the load characteristic value preferably takes place independently of the amount of the speed of the crop represented by the test image. In a preferred embodiment of the invention, the differences between the deviations from the mean deviation are then formed and then squared. The squared differences are summed up and divided by the number of movement parameters and the square root of the result is taken. This form of statistical evaluation correlates the load characteristic value particularly well with the risk of blockages occurring in the area of the conveying element, as a result of which the conveying speed signal can be used particularly reliably to set an actual load factor that comes close to the maximum possible load factor.

As an alternative to calculating the standard deviation, the utilization characteristic value is calculated as the mean value of the movement characteristic values which indicate the direction of movement, or their amounts or their amount-related deviation from the reference characteristic value (s). Again, alternatively, the mean value is calculated for movement characteristic values which indicate a total movement distance or an average quadratic deviation of these movement characteristic values. In these cases too, the load factor has a significant correlation with the actual tendency of the machine to clog.

According to the invention, a statistical evaluation of only the directions of movement of the objects at least partially depicted on the test images is generally provided, in particular using the standard deviation of the direction of movement to one indicated by the main conveying direction (s) of the associated conveying element. given reference direction for determining the conveying speed signal be particularly well suited.

As an alternative or in addition to the calculation of the standard deviation, the utilization characteristic value is calculated as the mean value of the movement characteristic values, which indicate the direction of movement, or their amounts or their amount-related deviation from the reference characteristic value (s). Again, as an alternative, the mean value is calculated for movement characteristic values which indicate a total movement distance or an average quadratic deviation of these movement characteristic values. In these cases too, the load factor has a significant correlation with the actual tendency of the machine to clog. As further statistical characteristics that can describe the flow behavior and can be used individually or together with other values as input variables for speed regulation, percentiles over flow lengths or orientations, statistical, absolute or central moments 1., 2.,. .. kth order, or especially histogram comparisons of the current histograms with pre-configurable normal histograms can be used.

As an alternative or in addition to the above-described embodiments of the method according to the invention, the evaluation device calculates at least one first portion of the test image formed by at least one image area. The at least one image area at least partially depicts a defined component of the crop or the machine. On the basis of the first part, in particular Utilization characteristic or another utilization characteristic is calculated. The load characteristic value is to be equated with the share in many exemplary embodiments.

Before calculating the first share, the component that is statistically represented by the first share is predefined. The test image and / or the test data set are in particular divided into a plurality of preferably equally large image areas. The image areas, which at least partially show the component, together form the first portion. The portion is in particular a portion of these image areas, which at least partially show the constituent parts, of the entire image, the first portion being formed on the basis of a ratio of the number of image areas or on the basis of their common areas.

The first portion is a measure of the extent of image areas that depict the component and thus a measure of the density of the component in the field of view of the image acquisition unit or the portion of the test image to be viewed. The component is, in particular, at least partially a root crop component, as a result of which the first fraction at least approximately indicates a concentration of root crops. An image area is in particular evaluated as a component and is assigned to the first component if at least 50% to 100% of its area shows the component. In particular, the at least one image area can also be assigned only partially to the first portion or preferably partially assigned to different portions. This is particularly advantageous if, within the framework of the preferably model-based classification method, it is not possible to clearly assign the image area to a corresponding component. Before- In this case, probabilities for the allocation to different shares are determined. The image areas are particularly preferably assigned proportionately or partly to different proportions according to the probabilities. As a result, the relationships between the components are mapped even more precisely.

By calculating the first portion, the characteristic value is calculated, which in particular characterizes the composition of the crop. On this basis, the conveying speed can be regulated particularly advantageously, since the cleaning performance of the conveying element or the separating device comprising the conveying element is strongly dependent on the composition of the crop. In the event that the first portion indicates a concentration of admixtures, the conveying speed can preferably be increased with increasing first portion in order to produce a lower occupancy in favor of a larger separating, in particular sieving, effect. The load factor is preferably calculated at least on the basis of the first component or is equal to the first component.

The conveying speed signal is preferably dependent on a speed of the crop or of the conveying element. Using a speed value that represents this speed, it is possible in particular to calculate the utilization characteristic value, which therefore has a different informative value.

The at least one image area, which forms the first portion, is preferably based in particular on a test data generated on the basis of the image area. set, identified as showing the defined component. In particular, the image area is identified on the basis of a test value contained in the test image and / or in the test sub-data record, preferably color information. The color information includes, in particular, black-and-white, gray and / or color channel values of a color space.

The test sub-data set, the test value or the color information is preferably classified by a model-based, statistical classification method. An image area is therefore assigned to the first portion in particular if the result of the classification process is assigned to the defined component of the crop or the machine. The classification method uses in particular a neural network, a random forest, a Bayesian classifier, a support vector machine and / or a decision tree. By using the classification method, the result of the calculation of the first portion, in particular of different portions, is particularly robust and powerful in terms of the composition of the crop.

The test value or the color information is particularly preferably compared with one or more reference values or reference areas and, based on this, an image area is either assigned to the first portion or not. Like the test image, the reference image is preferably to be acquired by the optical image acquisition unit, with a user in particular having to mark different parts of the reference image as different components. This form of differentiation enables particularly reliable identification of a enables the component on the test image. Particularly preferably, at least one of the test values of the test sub-data set, which in particular comprises the color information, is compared with at least one reference value and an image area is particularly assigned to the first component if at least the at least one test value of the test sub-data set lies within an assigned reference value range . This reference value range is limited in particular by a maximum value and by a minimum value, with different test values preferably having to lie in assigned reference value ranges in order to add the image range to the first portion.

In an advantageous embodiment of the invention, the evaluation device automatically or automatically further develops a model on which the classification method is based upon the input of exemplary image areas of the reference image which are to be attributed to the first portion. Alternatively or additionally, the evaluation device automatically calculates or changes the at least one reference value range upon input of exemplary image areas of a reference image that are to be attributed to the first portion. In particular, the reference values, the reference value ranges or the model or model parameters thereof are therefore at least not completely manually predefined by the user. Instead, it is sufficient to enter at least one exemplary image area, which shows the component, for starting up the evaluation device. Based on the image area, the evaluation device automatically determines the at least one reference value, the at least one reference value area or the model or model parameters thereof. Thus, the evaluation device is based on different applications largely independently. The greater the number of image areas entered, the more precisely the reference values, the reference value areas or the model or model parameters thereof can be determined.

The method is particularly robust when the entered image areas show the component under different brightness and / or soil conditions. The method can therefore be used reliably even under different application conditions. The evaluation device particularly preferably adjusts the at least one reference value or the reference value ranges during the repeated execution of the method, if necessary with exemplary identification of relevant components by the operator, from which training data for the algorithm can be mapped.

The evaluation device automatically extends the scope of the reference data, in particular on the basis of further sensors such as brightness sensors for measuring the ambient brightness, which the evaluation device assigns to test data records recorded essentially at the same time. Alternatively or additionally, the user of the method, ie in particular the driver or operator of the machine or a machine coupled to it, has the option of manually marking the at least one component on visualized test images in order to expand the scope of the reference data of the evaluation device. For example, based on the information provided by the user or on the basis of data stored in the evaluation device, the user can differentiate between, for example, potatoes, cabbage, stones and clods and calculate the respective proportions. The method according to the invention is preferably carried out automatically after it has started, with the exception of entering any training data that may be present in the form of marking components. The driver or operator of the machine is easier to operate.

The image areas forming the first portion are preferably identified on the basis of image sub-data sets generated on the basis of respectively adjacent image areas or formed thereby. In particular, color information, preferably also comprising black-and-white and / or gray values, included in the test sub-data sets is used for this purpose. The evaluations of the image areas therefore do not take place solely on the basis of the data assigned to them, but will also use additional data assigned to the surrounding image areas. In this way, brightness and / or color gradients can be determined and the identification takes place on the basis of a broader data basis.

The different image areas are preferably weighted differently when calculating the first portion. The contribution of the image areas forming the first part is therefore different. This makes it possible not to calculate the first portion purely on the basis of the perspective representations of the test image, but rather to give particular weight to image regions which show a component of the crop that is further away from the image acquisition unit than image regions which show a component closer to the image acquisition unit . This allows form a perspective-adjusted first part and thus achieve a particularly realistic picture of the crop composition on the conveyor element.

The entire test image or a coherent part of the test image is preferably subdivided into, for example, previously described partial image areas. The partial image areas in particular each comprise the same number of pixels of the test image, preferably exactly one pixel. The test image part is a part or section of the test image which comprises a plurality of partial image areas. To calculate the first portion, only the image areas showing the portion that belong to the part of the test image are taken into account. For this purpose, the part of the test image is particularly defined so that it depicts sensitive and monitored zones within the machine. The image area forming the first portion thus comprises, in particular, a plurality of partial image areas of a test image part.

The test image or part of the test image is in particular rastered into a plurality of partial image areas, each of which is preferably rectangular. When the sub-image areas are formed by exactly one pixel, a particularly large database for evaluating the condition of the crop with regard to its individual components is created, and thus a particularly sensitive control of the operating parameter is made possible. At the same time, the data supplied by conventional 2D digital cameras with a maximum of a few million pixels as a rule can be processed promptly for an evaluation device equipped with one or more current processors. The test image preferably comprises a plurality of test image parts, for each of which the evaluation device calculates a first portion, in particular a plurality of portions of image areas. The test image parts in particular show different sections of the same conveying element or different conveying elements. In particular, the test image parts show sections of a conveying element, one of which is arranged in the conveying direction in front of a separating device or a separating element thereof and another behind the separating device or a separating element thereof. Alternatively, the test image parts show different conveying elements, which represent alternative conveying paths for different components of the crop (for example, a conveying element for cleaned root crops, a conveying element for sorted additives). By calculating the first proportion for these different test pattern parts, the cleaning or separating performance of the associated separating device can be evaluated particularly comprehensively. In particular, the first portion of a crop flow to the separating device can be compared with the first portion of a root crop mixture outflow from the separating element or the separating device and thus the effectiveness of the separating device can be determined. Depending on the effectiveness, in particular the conveying speed is set so that, for example, congestions occurring at short notice can be remedied by a slower request or faster removal. By defining different parts of the test image in front of a separating or deflecting device, the accumulation of crop can be assessed particularly well. For example, for a part of the test image located directly in front of the deflection device, an occupancy with crop can be set in relation to an occupancy in the run-up to this area, for example to increase the conveying speed if the occupancy in front of the deflection device is too low. Likewise, the test image parts depicted or present in the respective test data records can show part of a conveying element in front of a separating or deflecting element of the separating device and part of the conveying element after the separating or deflecting element. If the image analysis shows that too large proportions of e.g.

If root crops appear in an undesired area behind a deflecting element, which may indicate insufficient separation due to congestion, the conveying speed can be adjusted accordingly.

In a further embodiment of the invention, the test image parts preferably show different conveying elements after a separating device, in particular a conveying element for discharging a root crop mixture and a conveying element for discharging additions after the same separating device. In this case, a first proportion of a component, for example root crops, is preferably determined for both test pattern parts. Alternatively, different proportions are calculated for the different parts of the test pattern. This makes it possible, for example, to compare a portion of admixtures in the root crop mixture outflow with a portion of root crops in a stream of sorted admixtures and, based on this, adjust the speed of a conveying element assigned to the separating device.

The image areas forming the first portion preferably show root crops or parts thereof and a second portion forming image areas or portions thereof. The evaluation device thus calculates at least two different parts. The evaluation device particularly preferably calculates at least four shares le includes a share for machine components, a share for root crops, a share for herb components, a share for soil or clods, a share for stones and / or a share for damaged areas. Depending on the application, in particular only a part of the stated proportions is calculated and or several of the stated proportions are combined. The total of the shares is in particular <1.

A plurality of parts in the calculation of the evaluation device can provide a more precise picture of the composition of the crop or the occupancy of the conveying element. As an alternative to identifying image areas on the basis of limit values, all image areas of the test image or of a test image part are necessarily assigned to a portion. In this case, a degree of agreement between test sub-data sets calculated on the basis of the image areas and reference sub-data records is preferably evaluated and each image area is assigned to the part for which the agreement is greatest.

In an advantageous embodiment of the invention, the load factor is based on a deviation of the first portion from a threshold value calculated by the evaluation device. In particular, the threshold value characterizes an optimal utilization of the conveying element, a deviation from a defined amount triggering an increase or decrease in the conveying speed. The load factor is based in particular on a plurality of components and preferably further data, in particular sensor data. In an advantageous embodiment of the invention, the conveying speed signal is calculated on the basis of a plurality of utilization characteristic values, in particular calculated chronologically one after the other, or at least one previously calculated utilization characteristic value is included in the calculation of the utilization characteristic value. In particular, a moving average of the load characteristic is calculated and is the basis of the conveying speed signal or the load characteristic curve is smoothed, in particular using a low-pass filter. As a result of these measures, the method according to the invention is particularly susceptible to faults and can therefore be used particularly robustly.

In an advantageous embodiment of the invention, at least one sensor transmits sensor data to the evaluation device, which are incorporated into the calculation of the conveying speed signal. The sensor is in particular a sensor, preferably a touch sensor or an ultrasonic sensor, for measuring a crop layer thickness on the conveying element and / or a rotational speed sensor, in particular for measuring a rotational speed of a conveying element drive. It is preferably a sensor for measuring the drive power, for example in the form of a pressure sensor for measuring a hydraulic oil pressure. In particular, a slip of the conveying element is determined on the basis of the speed sensor and is transmitted to the evaluation device in the form of the sensor data. Using a moisture sensor, information can also flow into the calculation of the conveying speed signal. On the basis of this additional information available in the sensor data, which goes beyond that provided on the basis of the test image, the evaluation device has a much more precise picture of the utilization situation in the area of the conveyor element, which in turn can better influence the conveyor speed.

The evaluation device preferably triggers either acceleration or deceleration of the conveying speed of at least individual conveying elements of the harvesting machine by means of different conveying speed signals. In particular, the evaluation device or the conveying speed control device comprises a three-point controller, which alternatively triggers either the acceleration, the deceleration or a maintenance of the current conveying speed. Acceleration is triggered in particular when the load factor exceeds a predefined first threshold value, a deceleration is triggered accordingly when the load factor falls below a predefined second threshold value. In the calculation of the actual increase or decrease in the conveying speed on the basis of the conveying speed signal, the current absolute value can advantageously be taken into account again.

A conveying speed gradient triggered by the conveying speed signal and / or the difference in conveying speeds before and after acceleration or deceleration depending on the utilization is particularly preferred. In particular, the magnitude of the speed gradient during the Deceleration greater than during acceleration in order to avoid blockages as reliably as possible and at the same time to avoid vibrations of the crop on the conveyor element. In addition, a speed increment is preferably greater, the smaller the load factor is and / or smaller, the larger the load factor is. Alternatively, exactly one conveying speed is assigned to each load characteristic, which is continuously adjusted to the load characteristic.

Preferably, after triggering a change in conveying speed for a defined period and / or a defined conveying distance, no further conveying speed change is triggered. As a result, after a change in the conveying speed signal triggered by the conveying speed signal, in particular an acceleration or a deceleration, there is no further change until an effect of the triggered conveying speed change can be assessed. At the same time, work is done more gently. To determine the period between the change in conveying speed and the time at which the crop located in front of the conveying element during the conveying speed change reaches the test image, the evaluation device receives in particular a signal from a speed or speed sensor which is used to monitor the rotational speed of the conveying element. The signal or the circulation speed can be used to calculate how long the period is theoretical. This configuration of the method avoids over-regulation of the conveying speed and the necessary inertia Changes in the movement situation of the crop and thus the load factor are taken into account.

The conveying speed signal is preferably transmitted by wire, in particular by means of CAN bus or Ethernet, or wirelessly to the conveying speed control device, the setting of the conveying element preferably having to be released beforehand by an operator via an input at an interface. This form of data transmission makes it particularly easy to integrate the conveyor speed signal into existing data infrastructures and thus easily change the conveyor speed based on the conveyor speed signal. The reliability of the method is increased in particular by the fact that an operator is shown the resulting or to be made setting of the conveying element instead of an automatic setting in particular in the driver's cabin and at an interface (e.g. in the form of a Fluman interface device) via a corresponding Has to release input.

The object is also achieved by a machine for harvesting root crops. The machine comprises a machine frame, at least one conveying element, at least one optical image acquisition unit and an evaluation device and is designed to carry out the method described above. The image capture unit is in particular a 2D or 3D camera, preferably a photo or video camera for recording color or black and white images. The image capture unit is preferably assigned at least one light source which illuminates the objects represented by the test image during operation. This allows the movement characteristics rates, in particular on the basis of the contrasts to be determined, are simplified and calculated more reliably.

The evaluation device preferably comprises a graphic processor unit, in particular a GPU (Graphical Processing Unit) or GPGPU (General Pur pose Graphical Processing Unit) and / or an FPGA (Field Programmable Gate Array) -based processor unit. This type of evaluation device allows the test data set to be evaluated in a particularly resource-saving manner.

In an advantageous embodiment of the invention, the machine has at least one sensor coupled to the evaluation device, in particular a tactile or ultrasonic sensor for measuring a crop layer thickness on the conveying element, a sensor for measuring a drive power, for example a pressure sensor for measuring a hydraulic oil pressure, and a moisture sensor and / or a speed sensor arranged on a conveyor element. With this sensor, the conveying speed signal can be calculated in addition to the movement characteristic data sets also on the basis of measured, physical quantities, whereby the meaningfulness of the quantities calculated with the evaluation device is significantly increased and their susceptibility to errors is reduced.

The machine preferably has a plurality of image recording units which record at least one test image from the same conveying element during operation. Alternatively, the machine preferably has several, in operation, at least one test image of image capturing units accommodating different conveying elements. Through the A plurality of image acquisition units can be used to track the composition of the crop, in particular a course of the first portion along a conveyor line of the machine. In particular, the conveying speeds of different conveying elements can thus be set using different, first shares.

The conveying element is preferably designed as a sieve belt or as a hedgehog belt. The crop lies on it at least temporarily during operation. In operation, the conveying element runs in particular under at least one deflecting roller which extends transversely across the conveying element and crops deflecting therefrom. In order to the steering roller rotates in particular about an axis of rotation, which is set in a plan view of the conveying element by <90 ° to the conveying direction of the conveying element. Together, the conveying element and the deflecting roller form a separating device in this case, which is to be monitored using the method according to the invention. Alternatively, the conveying element is designed as a screen star or conveying roller, the conveying speed being a rotational speed thereof. The sieve star promotes the harvested crop in particular by lying on the sieve star with the sieve star rotating at least 135 °, in particular at least 180 °, the axis of rotation of which extends essentially vertically. When the conveying element is designed as a conveying roller, it is in particular encompassed by a roller table, with an axis of rotation of the conveying roller in particular being arranged essentially horizontally. The image acquisition unit is preferably arranged such that the test image shows at least two alternative conveying paths for different crop components. As a result, two conveying elements can be monitored with the aid of an image capturing unit, one test image part of the test image each depicting a section of the different conveying elements or crops thereon. In particular, one of the conveying elements is designed to convey sorted additions and another of the conveying elements is designed to convey cleaned root crops. This enables a particularly comprehensive picture of the cleaning performance and thus the utilization of the conveying element and / or the separating device comprising the conveying element to be captured.

Further details and advantages of the invention are described in the following, schematically illustrated embodiments; show it:

1 shows a program flow chart of a method according to the invention,

2 shows a detailed view for the determination of crop components on a monitored conveyor section,

3 shows a program flow chart for the calculation of the conveying speed signal,

4 shows a program flow chart for evaluating the conveying speed signal, 5 is a view of a test image and its partial evaluation,

6 and the further possible, partly from evaluation,

7 shows an object according to the invention,

Fig. 8 u. 9 the object according to FIG. 7 in different side views,

Fig. 10 likes a partial view of the object. 7 element with a Förderele,

11 is a detailed view of a portion shown in FIG. 10 area of the device according to FIG. 7,

12 the object according to FIG. 11 from a different perspective,

13 is an illustration of the test image of the image acquisition unit according to FIG. 11;

Fig. 14 like a separator of the machine. 7 with an image acquisition unit, 15 is a schematic test image taken from the perspective of the image acquisition unit shown in FIG. 14;

Fig. 16 like another separator of the machine. 7 with an image acquisition unit,

17 is a schematically illustrated test image taken from the perspective of the image acquisition unit shown in FIG. 16,

18 likes a further detailed view of a machine. 7 with a wide ren image acquisition unit,

19 is a schematic illustration of a test image viewed from the perspective of the image acquisition unit according to FIG. 18,

20 shows a detailed view of a further device according to the invention.

Parts with the same or similar effect are provided with identical reference numerals, where appropriate. Individual technical features of the exemplary embodiments described below can also lead to further developments according to the invention with the features of the above-described exemplary embodiments, but always at least in combination with the features of one of the independent claims. The items listed in the list of figures are sometimes only partially shown in individual figures. The method according to the invention serves to regulate the operation of a machine 2 for harvesting root crops 4 (cf. FIGS. 6 to 8). In the method, at least one test image 8 is recorded by at least one optical image acquisition unit 6, which shows at least one conveying element, initially generally numbered 10, relative to a machine frame 12 of the machine 2, showing crops comprising root crops 4.

The test image 8 is transmitted to an evaluation device, which generates a separator setting signal for setting at least one operating parameter of a separating device of the machine 2 on the basis of a test data record generated on the basis of the test image 8 or formed by it. The images shown as test images or output images only show schematically the parts relevant for the invention without any borders or boundaries. Digital images taken by a camera, in particular, may have additional information not shown in the figures. For example, these can already be masked or filtered on the camera side or when a test data record is being created or edited.

In one exemplary embodiment according to the invention, a crop flow 1.1 of a separating device is captured by means of a first image capturing unit 6 (block 1.2, FIG. 1). In addition, the crop flow from two further optical image acquisition units is additionally monitored (blocks 1.3 and 1.4), for example after the exit of the separating device and in the area of a discharge belt for admixtures 5, which by means of of the separator are disconnected. By means of the method according to the invention, utilization characteristic values LS_1 to LS_3 are determined for the respective measuring points or areas captured by the image acquisition units 6 (blocks 1.5, 1.6,

1.7). These are offset against one another in block 1.8, which leads to a speed signal for at least one conveying element 10 of the separating device. This sets the conveying speed of the conveying element (block 1.9), which optimizes the crop flow 1.1 in the separating device.

The determination of the conveying speed signal is shown in a greater degree of detail in FIG. 2. Accordingly, a test image 8 comprising crop is first recorded on a conveyor element 10 (block 3, corresponds to block 1.1) by the image acquisition unit. After the recording of the test image 8, a relevant image section or part of the test image 8 is extracted according to a first method sequence according to the invention by means of a corresponding filtering or masking. For this purpose, based on the position of the image acquisition unit, a mask or a region of interest (ROI) is predefined, on the basis of which sections of the test image 8 to be taken into account and not taken into account are distinguished (block 13.1). On the basis of the relevant image section of the test image 8 and the associated test data set, a movement characteristic value is calculated for a plurality of image areas, in particular for each pixel of the image section (block 13.2). The movement characteristic includes in particular a direction of movement. The majority of movement parameters are then statistically evaluated (block 13.3). For this purpose, the movement parameters are each compared with assigned reference values, which are preferably machine-specific and in particular Update-capable database are provided (block 13.4), and a difference between them is calculated or the movement parameters are compared with a uniform reference parameter and a deviation therefrom is calculated. The movement characteristic values or the calculated deviations are statistically evaluated by the evaluation device, in particular a standard deviation of the movement characteristic values from the reference characteristic values is calculated.

A low-pass filter then goes over the continuously evaluated statistics to smooth the determined values (block 13.5). For this purpose, a predefined and, in particular, predeterminable filter time constant is used (block 13.6), which specifies the extent of smoothing.

On the basis of the filtered or smoothed statistics of the above-described deviations, a load factor, generally designated LS, is used in the

Check image of the conveyor section shown (block 13.7). This represents the movement situation of the crop or the crop flow in the area of the separating device, in particular on the conveyor element or in the transition area of two conveyor elements.

With reference to the second, supplementary or alternative path, the relevant parts of the test image are first extracted (block 2.1). For this purpose, a mask or region of interest (ROI) can be predefined based on the position of the image acquisition unit 6 (block 2.2), on the basis of which the distances between the test image 8 to be taken into account and not to be taken into account can be distinguished. On On the basis of the relevant image section of the test image 8 and the test data record now available for processing, the proportions of the image regions showing individual crop components are now calculated (block 2.3). In particular, the color information can be evaluated here. These values can be taken from a reference table or can also be specified by operating personnel (block 2.4).

On the basis of a threshold value definition (block 2.5), the deviations of the calculated proportions from the threshold value are calculated (block 2.6). The threshold value is, for example, an ideal value for the proportion considered (e.g. root crop, quantity 1, quantity 2). This is followed by low-pass filtering in order to smooth the determined deviations (block 2.7). This will be a pleasure. Block 2.8 defined filter time constant used. Subsequently, a further load factor or the load factor LS is calculated based on the smoothed values of the deviations for the individual positions along the conveyor line and the respective proportions (block 2.9).

Subsequently, the Förderge speed signal will be generated based on the or the load characteristic LS, for example by means of a three-point controller described below (block 2.10).

Fig. 3 shows a program flow chart for the calculation of the utilization characteristic LS to the conveying speed signal. In this embodiment, the load factor LS has a value of -1, 0 or 1 and has been described as above generated. After the start of the method, the device listens to or waits for a new utilization characteristic value LS (block 14.1). It goes without saying that, for the purpose of programming, the respective utilization characteristic values, which in the present case are also simply referred to as “LS”, must be differentiated and are therefore referred to in FIG. 3 as LS_x. After the load factor has been transferred, the process continues depending on its size. A load characteristic value LS_x of 0 represents a desired load of the separator, a load characteristic of -1 an underload, ie insufficient load of the separator and a load characteristic of 1 an overload, ie an excessive load with a risk of constipation. In the event that the utilization characteristic value is 0, this is entered in the memory 14.2 of the last utilization characteristic values (block 14.3) without sending out a conveying speed signal for changing the conveying speed. In the event that the utilization characteristic value is 1, a previous utilization characteristic value stored in the memory 14.2 is queried (block 14.4) and then it is determined whether an overload has already been determined after the last stored utilization characteristic value of 0 (block

14.5). If this is not the case, the evaluation device sends out a conveying speed signal to reduce the speed (delay signal, block.

14.6). If this is the case, the new utilization characteristic value is entered in the memory 14.2 and no (further, the conveying speed reducing) conveying speed signal is transmitted. The speed control according to the invention results from the conveyor speed signal according to block 14.6 (block

14.7), ie the adaptation of the conveying speed to the utilization of the individual monitored areas of the conveying line or the separating device. If the capacity utilization value has a value of -1, a utilization characteristic value entered in the memory 14.2 is again queried (block 14.8) and a decision is made in accordance with the above-described distinction as to whether a conveying speed signal for accelerating the conveying speed is being transmitted or has already been transmitted. Optionally, the program flow can be optimized by accelerating only after a certain sequence in number of load characteristic values indicating an underload or underload. For example, it is monitored for the respective areas of the conveyor line whether there is an underload (block 14.9) and only then is an acceleration pulse sent (block 14.10).

Fig. 4 shows a program flow chart for evaluating the conveyor speed signal. In the process shown, a conveying speed increment or decrement for changing the conveying speed is calculated on the basis of the conveying speed signal 17.1 (block 17.2). Based on an existing, in particular specifiable and variable rule base (block 17.3), values such as the extent of the load factor can be included in the calculation. For calculating the increment or decrement, it can also be taken into account whether the machine is moving further away from the in a fine control range of the speed, e.g. close to the load limit (e.g. difference less than 10%) or in a rough control range Utilization limit is found. The utilization limit can preferably be in the evaluation device as the value is defined from which an excessively large deviation signaling a build-up of material occurs.

The conveyor speed increment or decrement is converted by the evaluation device into a speed specification for a separator drive (block 17.5). The resulting conveyor speed signal is sent to the separator device drive (block 17.6). This results in a conveying speed of the separating device (17.4).

Figures 5 and 6 show an example of the evaluation of individual test images. An output image 9 and a test image 8 are shown schematically in FIG. 5, each with root crops 4 on a conveyor line comprising two conveyor elements 10A and 10B. In the following, for the purpose of simplification, funding elements are generally also numbered “10”. A conveyor element 10 are then one or more conveyor elements from the set of conveyor elements (10A, 10B, 10C, 10D, ...).

In a preferred embodiment of the method according to the invention, the evaluation device compares the output image 9 with the test image 8 in that the directions of movement of objects shown on the images are determined. An object does not necessarily identify a coherent body, but merely represents in the test image 8 an area which can be identified with regard to its movement and which in particular has the size of an area reproduced via a pixel of the test image 8. In particular, the evaluation device thus calculates a direction of movement for each pixel of the Test image 8, evaluates their deviation from a reference direction known for each area - in particular each pixel - and evaluates these deviations statistically. 5 shows an example of a calculated direction of movement in the form of a superimposed vector per root crop 4, regardless of the observation of the movement at the pixel level. Each arrow represents a characteristic movement value

20th

To calculate the load characteristic value LS, the movement characteristic values 20 are statistically evaluated. In this case, the movement parameters 20 only comprise a movement direction, not a movement distance possibly indicated by the arrow length. FIG. 5 also shows a histogram with one column per movement characteristic 20. Each column indicates a deviation in the amount of the corresponding movement characteristic 20 from a uniform reference characteristic 22.

To calculate a utilization characteristic value LS, indicated by line 14, a standard deviation of these movement characteristic value deviations from the reference characteristic value 22 is formed in particular. For this purpose, the deviations can in particular be squared and then summed up. This sum is then divided by the number of movement parameters 20 and the square root of the resulting quotient is formed. The resulting value is the load factor LS, which is given in the histogram shown. Advantageously, in order to calculate the characteristic movement values 20, first image areas 16 of the test image 8 are compared with further image areas 18 of the output image 9, each image area 16, 18 comprising the same number of pixels and in particular being rectangular. For a simplified illustration, only a few exemplary image areas 16, 18 are shown in FIG. 9. This results in a motion characteristic value 20 for each image area 16, in particular for each pixel of the test image 8.

Depending on the conveyor section, the evaluation device can determine which load leads to a reduction or increase in the conveyor speed. For example, a speed increase can be carried out with a standard deviation of less than 10 °, the speed can be maintained with a standard deviation of 10 ° to 20 ° and the conveying speed can be reduced with a larger standard deviation. Correspondingly, it can be determined for the present case as a drop between two conveyor elements 10A and 10B designed as sieve belts based on the evaluation of the recognized directions and their standard deviation whether a crop jam occurs on the conveyor element 10B below. Should a corresponding condition be determined, for example, due to the exceeding of a threshold value R indicating a jam, the evaluation device outputs a conveying speed signal for accelerating the conveying element 10B, alternatively or in addition to a reduction in the conveying speed of the conveying element 10A to be conveyed. Fig. 6 shows an example in the upper part of the figure, a test image 8, which also represents the transition from the conveying element 10A to the conveying element 10B. On this conveyor section there are root crops 4 and 5 seeds, which can include stones and herbs. According to the classifiers defined in the training of the algorithm or specified via a database, for example a table with color values in HSV format, individual partial image areas 16 are checked for the presence of the same components. Thus, due to the assignment of the respective image areas to the individual portions, exemplarily shown at the bottom left in FIG. 5, there is a portion distribution of individual portions of root crops 4 and admixtures 5 in test image 8. A1 thus shows the portion of root crops 4 in test image 8 and the corresponding test data set, A2 the percentage of cabbage and A3 the percentage of stones. This assignment is preferably carried out on the basis of the color information, preferably also comprising black, white and / or gray values of the individual pixels, ie an image area 19 which is assigned to a portion corresponds in particular to an area of a pixel.

The load factor, generally designated LS, is based on an example, although preferably based on a deviation of the first portion A1 from a threshold value, again generally designated R, which indicates an optimal share distribution of root crops on the observed point on the conveyor line. For example, the load characteristic value LS is set to 1 in the event of a deviation of> 50% from the cleaning threshold value and to 0 in the event of a deviation of <50% from the cleaning threshold value. These values are then stored accordingly or processed in the further program flow according to FIGS. 1 to 4. An arrangement of the optical image capturing units 6 is disclosed in FIG. 8. The machine 2 according to the invention is designed as a pulled potato harvester, with a large number of conveying elements 10 and their associated separating devices being held by a machine frame 12 which is only partially numbered. Along the conveying route there are a plurality of image capturing units 6 which include the crops transported on the conveying elements 10 comprising root crops 4. The optical image acquisition units 6 form individual measuring points for monitoring the respective separation devices.

The positions for image acquisition units 6 indicated in FIG. 7 are an area immediately after a lifting device 29 (measuring point MS1), a transition from a first conveying element 10A in the form of a wire belt to a second conveying element 10B in the form of a wire belt, which is additionally enclosed by a coarse herb belt (measuring point MS2), around the transition from this second sieve belt 10B to a further conveying element 10C, comprising a further separating device (measuring point MS3). In addition, on the output side of this separating device, a conveying element 10E leading to the picking table is monitored with a further image acquisition unit 6 (measuring point MS4), at the same time a further conveying element 10F provided for residues of admixtures 5, in particular stones, is detected. Finally, another optical image acquisition unit 6 is present at the picking table 45 (measuring point MS5).

An evaluation device can be positioned at any, but preferably in the vicinity of the reading table, centrally accessible location. From The evaluation device can, for example, be given information relating to the setting of the separating devices to an operator on a towing vehicle via a cable 12.1 which can be seen in FIG. 7.

The machine 2 shown in FIGS. 8 and 9 in a side view clarifies the positions of the optical image acquisition units 6. In particular, the image acquisition unit 6 located at the reading table 45 in the United States can be arranged directly on a drop stage leading to a bunker 33.

10 and 11 show the arrangement of an optical image acquisition unit 6 arranged on the frame side above a first drop stage between a conveying element 10A and a conveying element 10B, the field of vision of which is directed downward (measuring point 2). A light source 7 provides illumination of the field of view in order to capture a sufficiently illuminated test image 8. The conveying element 10A is a screen belt which, coming from a digging device 29, already screens part of the admixtures 5, in particular earth, and via a drop stage to a wide range of res conveyed as a belt conveyor element 10B. This conveying element 10B additionally has a coarse herb belt which is provided for separating the herb present on the potatoes or in the crop. Correspondingly, stripping devices 32 are arranged across the width of the conveying element 10B.

A height H of the stripping device 32 above the conveying plane of the conveying element 10B is also adjustable. The conveying speeds of the conveying elements 10A and 10B can be set using the method according to the invention. In For the sake of clarity, FIG. 11 shows only a coarse herb belt 43 and not the actual conveying element 10B which is designed in the form of a sieve belt (cf. FIG. 13).

A test image 8 resulting from the field of view of the optical image acquisition unit 6 shown in dashed lines in FIG. 12 is shown in detail in FIG. 13 (without crop). On the basis of a test data record created from this test image 8, the evaluations described above are carried out on the basis of the detected movement directions of the crop and / or on the basis of the respective proportions of the crop components and, if necessary, the conveying speeds of the conveying elements are adapted.

Starting from the conveying element 10B, the crop still present is transferred to a further conveying element 10C with a conveying direction 1C. This is a separator in the form of several superimposed, rotating Umlenkwal zen 24 assigned. The crop is transported in the direction of the conveying element 10D via a pulse exerted by the latter (FIG. 14).

A distance H between the conveying element 10C and the lower deflection roller 24 can be set by the operating personnel for the purpose of varying a separation performance. Using the method according to the invention, the speeds of at least the conveying and conveying conveyor elements 10C and 10D designed as sieve belts can be varied. In addition, according to an advantageous further development a variation of the separation performance or deflection can be achieved via the adjustability of the circulation speeds of the deflection rollers 24.

The image acquisition unit 6 shown in FIG. 14 generates the test image shown in FIG. 15, in which a test image part 8A is defined by filtering or masking. In addition, a test image part 8B, which is viewed behind the deflection rollers 24 from a conveying direction 1C, is defined by filtering. For the conveying speed setting, the area of the conveying element 10C located in front of and behind the separating element formed by the deflecting rollers 24 is thus monitored. It is possible to create respective test data sets for both image areas 8A and 8B, or the respective evaluations for the two image areas 8A and 8B can result from the corresponding partial areas of a test data set.

If an associated target value for the test image part 8A results in a build-up of material in front of the deflection rollers 24, the conveying speed of the conveying element 10C is increased.

Alternatively, the evaluation can only be based on the areas 8A and 8C located in front of the deflecting rollers 24 and bordered by broken lines. For these two areas, for example, permissible component densities can be defined via the respective threshold values R. If, for example, a portion associated with excessive build-up immediately in front of the deflecting rollers, for example of root crops 4, is exceeded, the conveyor belt 10C can be run faster, alternatively or additionally a conveyor belt can be run faster.

A height H of the lower ends of fingers 26 of a separator designed as a finger band 26.1 can also be set by the operating personnel as one of several operating parameters. The height H describes the distance of the fingers 26 from the upper edge of the conveyor element designed as a hedgehog belt. In addition, an angle of attack of the finger band 26.1 can be designed to be adjustable to a perpendicular to the conveying plane of the conveying element. The same applies to the running speed of the finder tape 26.1.

Another optical image acquisition unit 6 arranged in the area of the conveyor belts 10C and 10D is shown in FIG. 16. This image acquisition unit 6 can be used in addition to the image acquisition unit according to FIG. 14 in order to monitor the crop transport in the inspection image area 8D. In particular, it also serves to monitor the effect of the separating and deflecting device formed by the deflecting rollers 24. In particular, depending on the results of the evaluation of the test images 8 of the optical monitoring unit 6 according to FIG. 14, the conveying element 10D can be adjusted in its conveying speed. A monitoring unit 6 according to FIG. 16 is also assigned a light source 7 for better illumination of the monitored area 8D.

Another optical image acquisition unit 6 is provided with an associated light source 7 above a reading table with a view of a conveying element 10E and a conveying element. derelement 10F arranged (Fig. 18). Here, as described above, a perspective cleanup is carried out on the basis of the “fish-eye” representation of the image acquisition unit 6. By means of masking, the test image parts 8A and 8B shown in the test image 8 according to FIG. 19 are selected which, on the one hand, serve as a conveying path, the conveying element 10E with conveying direction 1E for the removal of root crops, and, on the other hand, as another conveying path, the conveying element 10F with conveying direction 1 F for conveying away additions 5 monitor in the form of stones. The evaluation described above is used to check whether the proportions of root crops 4 on the conveying element 10F are too large. If this is the case, the upstream conveying element 10D is set more slowly by means of the evaluation device by means of the method according to the invention. In addition, in a further development of the invention, depending on corresponding control specifications, the finger band 26.1 shown above the conveyor element 10D designed as a hedgehog belt can be adapted with respect to its separating effect with the fingers 26 shown by way of example and in broken lines behind the cover 40 located in front of it. For example, the distance of the fingers 26 from the conveying element 10D is reduced in order to convey more crop material in the form of root crops 4 onto the conveying element 10E via an associated chute 41.

20 illustrates the arrangement of optical image acquisition units 6 on pointing measuring points MS1 to MS5 in the case of a schematically illustrated conveying path of a machine 2 designed as a beet harvester. The image acquisition units 6 are in the wake of a lifting device 29 above a roller table 10M and at the end of a roller conveyor Conveyor element 10N arranged (measuring represent MS1 and MS2). A further optical image acquisition unit 6 monitors in particular a conveyor element 10P (measuring point MS3) designed as a screen star. The subsequent conveyor element 10Q, which is designed as a star, is also monitored in the same way as a conveyor element 10R, which is designed as a ring elevator (measuring points MS4 and MS5). If, for example, a congestion is detected on one of the conveying elements 10M, 10N, 10P, 10Q, 10S, this can be made to run faster at the instigation of the evaluation device. According to one of several possible control scenarios, this can initially only take place for a certain time until a next check or until a monitoring system detects the removal of the critical state.

Claims

Expectations

1 . Method for controlling the operation of a machine (2) for harvesting root crops (4), in which at least one test image (8) of at least one test image (8) is moved relative to a machine frame (12) by means of at least one conveying element (10) Harvested crop includes root crops (4) and a conveying speed of the conveying element (10) is set on the basis of a test data record generated or formed by the test image (8), characterized in that an evaluation device based on the test data record is one of a speed of Harvested independent conveying speed signal for setting the conveying speed is generated.

2. The method according to claim 1, characterized in that the Förderge speed signal continues to be determined independently of the speed of the conveying element (10).

3. The method according to claim 1 or 2, characterized in that the evaluation device compares the test data record with an output data record generated on the basis of an output image (9) or formed by this.

4. The method according to claim 3, characterized in that the test data set of a first embodiment of the method serves as an output data set for a further execution of the method.

5. The method according to one of the preceding claims 3 or 4, characterized in that the evaluation device determines the conveying speed signal on the basis of an evaluation of the optical flow of the crop resulting from the test and the output data set.

6. The method according to any one of claims 3 to 5, characterized in that the evaluation device calculates at least one movement characteristic data set, which characterizes a movement, in particular a direction of movement, of at least one object represented at least partially by the test image (8), based on the movement characteristic data set the conveying speed signal is generated.

7. The method according to claim 6, characterized in that the evaluation device each have a movement characteristic data set for different with the

Test image (8) objects or different, first partial image areas (16) generated, which in particular comprise exactly one pixel of the test (8) and / or the output image (9).

8. The method according to claim 7, characterized in that the evaluation device in a first calculation step for a plurality of partial image areas (16, 18) comprising at least a first number of pixels, one movement each. calculated characteristic data record and in a later calculation step and taking into account the motion characteristic data records calculated in the first calculation step, a further motion characteristic data record is calculated for a higher number of different partial image areas (16, 18) which comprise a smaller number of pixels.

9. The method according to any one of claims 6 to 8, characterized in that the evaluation device a utilization characteristic (LS), in particular exclusively, on the basis of at least one movement characteristic Be characteristic movement (20) of the at least one movement characteristic data set and in particular based at least a reference characteristic value (22) assigned to the movement characteristic value (20) is calculated.

10. The method according to claim 9, characterized in that the evaluation device statistically evaluates a plurality of movement characteristic values (20), which are encompassed by different movement characteristic data sets, for calculating the utilization characteristic value (LS), in particular a standard deviation of the movement characteristic values (20) from the associated one Reference characteristic value (22) or the assigned reference characteristic values (22) are calculated, and in particular the utilization characteristic value (LS) is independent of the amount of the speed of the crop represented by the test image (8).

1 1. Method according to one of claims 9 or 10, characterized in that the load factor (LS) by means of a by the evaluation device calculated deviation of the first portion (A1) from a threshold (R) is determined.

12. The method according to any one of the preceding claims 9 to 1 1, characterized in that the conveying speed signal is calculated on the basis of a plurality of utilization characteristics (LS), in particular calculated chronologically one after the other, or at least one previously calculated utilization characteristic (LS) in the calculation of the utilization characteristic ( LS) flows in.

13. The method according to any one of the preceding claims, characterized in that the evaluation device calculates at least one first portion (A1) of the test image (8) formed by at least one image area (19), the at least one image area (19) at least partially depicts a defined component of the crop or the machine (2), one or a further utilization characteristic (LS) being calculated in particular on the basis of the first portion (A1).

14. The method according to claim 13, characterized in that the at least one image area (19) forming the first portion (A1) on the basis of a test sub-data record generated on the basis of the image area (19), in particular at least one color information comprised thereof, as the defined component of the Harvested crop or the machine (2) is identified.

15. The method according to claim 13 or 14, characterized in that the test data record, in particular at least one test value included therein, preferred the color information is classified by a model-based statistical classification method in particular and an image area (19) is assigned to the first portion (A1) in particular if the result of the classification method is assigned to the defined component of the crop or the machine (2) .

16. The method according to any one of claims 13 to 15, characterized in that the at least one test value of the test sub-data set, in particular the color information, is compared with at least one reference value and an image area (19) is then assigned to the first portion (A1) in particular, if at least the at least one test value of the test sub-data set lies within an assigned reference value range.

17. The method according to claim 15 or 16, characterized in that the evaluation device upon input of exemplary image areas (19) of a reference image to be attributed to the first portion (A1) automatically further develops a model on which the classification method is based and / or automatically calculates the at least one reference value range or changed.

18. The method according to any one of claims 13 to 17, characterized in that different image areas (19) are weighted differently when calculating the first portion (A1).

19. The method according to any one of claims 13 to 18, characterized in that the entire test image (8) or a coherent part of the test image (8A) is divided into partial image areas (16) which in particular each have the same number of pixels of the test image (8), preferably exactly one pixel.

20. The method according to any one of claims 13 to 19, characterized in that the test image (8) comprises a plurality of test image parts (8A, 8B), for each of which the evaluation device has a first portion (A1), in particular several portions of image areas (19 ) calculated, preferably the test image parts (8A, 8B) depicting crop under different areas of a conveying element (10) located in front of a separating or deflecting device.

21st Method according to one of claims 13 to 20, characterized in that the image areas (19) forming the first portion (A1) show root crops (4) or parts thereof and a second portion forming image areas show admixtures (5) or parts thereof.

22. The method according to any one of the preceding claims, characterized in that at least one sensor, in particular sensor for measuring a drive power, an ultrasonic or tactile sensor, for measuring a crop layer thickness on the conveying element, a moisture sensor and / or a speed sensor , Sensor data transmitted to the evaluation device, which flow into the calculation of the conveying speed signal.

23. The method according to any one of the preceding claims, characterized in that the evaluation device triggers either an increase or a decrease in the conveying speed of at least individual conveying elements by means of different conveying speed signals.

24. The method according to claim 23, characterized in that a conveying speed gradient triggered by the conveying speed signal and / or the difference of the conveying speeds before and after acceleration or deceleration is dependent on the load characteristic value (LS).

25. The method according to claim 23 or 24, characterized in that after triggering a change in conveying speed for a defined period of time or a defined conveying path, no further change in conveying speed is triggered.

26. The method according to any one of the preceding claims, characterized in that the conveying speed signal is wired, in particular by means of CAN bus or Ethernet, or wirelessly to a conveying speed control device, preferably the setting of the conveying element (10) in advance by an operator an input must be released at an interface.

27. The method according to any one of the preceding claims, characterized in that the evaluation device evaluates the test data records locally on the machine or a directly connected towing vehicle.

28. Machine for harvesting root crops (4) which has at least one machine frame (12), at least one conveying element (10), at least one optical image acquisition unit (6) and an evaluation device and designed to carry out the method according to one of the preceding claims is.

29. Machine according to claim 28, characterized in that the evaluation device comprises a graphic processor unit, in particular a GPGPU, and / or an FPGA-based processor unit.

30. Machine according to claim 28 or 29, characterized by at least one sensor coupled to the evaluation device, in particular an ultrasound or touch sensor for measuring a crop layer thickness on the conveying element (10), a sensor for measuring a drive power, and / or one Conveying element (10) arranged speed sensor.

31. Machine according to one of claims 28 to 30, characterized by a plurality of image capturing units (6) receiving at least one test image (8) of the same conveying element (10) or different conveying elements (10) during operation.

32. Machine according to one of claims 28 to 31, characterized in that the conveying element (10) as a sieve belt (10A, 10B, 10E) or hedgehog belt (10C,

10D), which runs in operation, in particular, under at least one deflecting roller (24) which extends across the conveying element (10) and deflects the crop therefrom, or sieve star (1 OP, 10Q, 10S) or conveying roller, in particular comprising a roller table is.

33. Machine according to one of claims 28 to 32, characterized in that the image capture unit (6) is arranged such that the test image (8) at least two alternative conveying paths for different crop components, in particular a conveying path at least for root crops (4) and a conveying path for admixtures (5).