Device and method for controlling a material flow at a material flow node

Abstract

A device and a method for controlling a material flow (18) in a storage and order picking system (10) at a sensor-controlled node (12) of the material flow (18) is disclosed, where a decision is to be made regarding a future material flow direction or speed of the material flow (18), wherein at least one input area (28) of the node (12) is detected by one, preferably single, sensor (22), comprising the steps: generating (S10) sensor data (46) by means of the sensor (22) which represent the material flow (18), defined by one or more material flow objects (36) that are approaching the node (12) downstream, in a field of view (26) of the sensor (22), in particular in the input area (28) of the node (12);Check (S16) whether the sensor data (46) for each of the approaching objects (36-1, 36-2) includes at least one associated object feature (40) that allows for the safe identification of the corresponding object (36); and if the sensor data (46) for each of the objects (36-1, 36-2) includes at least one associated feature (40) that allows for safe identification, control (S18) the material flow (18) at the node (12) based on the sensor data (46);or if the sensor data (46) do not include for each of the approaching objects (36-1, 36-2) the at least one associated feature (40) that allows for secure identification (S16), control (S26) of the material flow (18) at the node (12) at least for those of the objects (36-1, 36-2) whose associated feature (40) is not included in the sensor data (46), based on an information fusion (S22) of the sensor data (46) with pre-generated additional data (50) originating from at least one information source (52) that differs from the sensor (22) of the node (12), wherein the additional data (50) include further object features (40) which, together with the features (40) from the sensor data (46), allow for sufficiently secure identification (S24).;

Description

[0001] The present invention relates to a device or a system as well as a method for controlling a material flow in a storage and order picking system at a sensor-controlled node of the material flow, where the material flow divides downstream into at least two output directions.

[0002] Material flow is generally understood as a sequence of storage, transport, and processing processes. To direct and guide material flow objects through a (logistics) storage and picking system, a very large number of (technically simple and inexpensive) sensors, particularly light sensors, light barriers, and scanners, have been used. These sensors are used throughout an overall system, particularly at nodes (intersections, branches, junction points, etc.) of the material flow. At these nodes, several sensors are often used simultaneously. A general goal is to be able to perform tracking and tracing. In (intra)logistics, tracing and tracking refer to the (internal) tracking of material flow objects.

[0003] One goal of tracking is to be able to determine and retrieve the exact location of an object within the overall system at any time. Based on this information, processes can be optimized and unused capacity reduced, for example, through precise planning of material flow, warehouse allocation, goods receipt, or similar processes. Tracking refers to the tracking of an object from a source to a sink. Tracing refers to tracing from the end customer to the point of primary production.

[0004] In addition to simple and inexpensive sensors, such as the light sensors, light barriers and scanners mentioned above, cameras are also increasingly being used for intralogistics purposes.

[0005] A material flow control system using a video camera is disclosed in DE 10 2006 015 689 A1. Status images from the camera are evaluated to determine the actual position of the object to be conveyed in a predetermined area of ​​the system. The video camera is arranged vertically above a conveyor system, for example. The status images are evaluated to determine the actual position of the conveyed goods (material flow objects), which is then compared with a target position to generate corresponding conveyor control signals based on the target-actual comparison.

[0006] In DE 102011 055 455 A1, a video camera is used to determine the positions of objects within a transport container, wherein the transport containers are transported on a conveyor system. DE 10 2011 053 547 A1 discloses the use of video cameras to monitor the transfer process of a conveyed item by a robot positioned between two parallel conveyor systems. In particular, identification marks and dimensions of the conveyed items are determined using the camera data. A similar system is disclosed in DE 10 2010 002 317 A1. Furthermore, in DE 10 2010 001 569 A1, the fill level of a shaft of an automatic order picking machine is monitored with a video camera. Video cameras are therefore already being used for specific purposes in intralogistics. However, it is always implicitly assumed that the identity of the objects to be handled is known with certainty.

[0007] DE 10 2007 035 272 A10 discloses a method for identifying transported goods, particularly luggage. The technological background can be found in the book "Material Flow in Logistics Systems" (6th edition, Berlin, Heidelberg, New York: Springer 2009, pages 329-339).

[0008] DE 10 2015 219 660 A1 concerns a method and system for determining data for an object.

[0009] DE 44 08 650 A1 concerns an identification and control system for means of transport in a processing plant.

[0010] However, there are almost no solutions to the problem of reducing the number of sensors required for tracking and tracing in the overall system, which are used in particular for material flow control within a warehouse and picking system.

[0011] It is therefore an object of the present invention to overcome the above-mentioned disadvantages by means of a corresponding device and a corresponding method.

[0012] This object is achieved by a method for controlling a material flow according to claim 1.

[0013] One advantage of the present invention, i.e. the influencing of the material flow at the junction point, is preferably achieved with just a single sensor technology, namely, for example, the (video) camera. The camera replaces a large number of sensors (e.g., light sensors, light barriers, etc.) that were conventionally used. A further advantage is that, for example, stop positions, ejection positions, deflection positions, acceleration positions and the like can be detected and influenced independently of the sensor position due to the (large) field of view of the camera. This is not about controlling the material flow with cameras per se, but about eliminating a major disadvantage of the camera, which is that the camera either cannot see the object-identifying features at all or cannot resolve them sufficiently.This disadvantage is overcome by information fusion, which combines and evaluates the various information sources and combines them to solve the specific problem.

[0014] The invention enables object tracking within the warehouse and picking system with a minimal number of sensors. Low-cost and technologically simple sensors, particularly inexpensive (video) camera systems, can be used, whose limited application area is compensated for by information fusion.

[0015] In a preferred embodiment, the at least one feature is optically recognizable and wherein the at least one feature is preferably a barcode, an individualizing color combination, an individualizing manufacturer's mark, an individualizing contour, and / or an individualizing dimension or size.

[0016] These features can be extracted and identified from the sensor data using conventional image processing. Image processing can rely on well-known image and pattern recognition methods.

[0017] Furthermore, it is preferred if the field of view of the camera is chosen to be so large that visible barcodes cannot be resolved.

[0018] The field of view is then very large, with many of the objects and also many of the nodes, including the objects located there, being monitored simultaneously with a single sensor. The number of sensors required in the overall system is further reduced. The resolution of the cameras becomes less important, thereby reducing the cost. The lack of resolution is compensated for by information fusion. The cameras are preferably used to determine the position and / or orientation of approaching objects.

[0019] The at least one information source includes a material flow control, a warehouse management control, a workstation, and / or a goods receipt.

[0020] These are classic sources of information that are already available in conventional warehouse and picking systems.

[0021] The additional data for each of the approaching objects includes at least one of the following parameters: an order in which the objects approach the node; a relative distance between successive objects; and a time at which a respective object was added to the material flow; a drop-in location where the respective object was added to the material flow; a path that the respective object has traveled through the system so far; a location-dependent speed at which the respective object is moved through the system; a dimension of the respective object; a color combination of the respective object; a manufacturer's identifier of the respective object; and a picking order to which the respective object belongs.

[0022] These parameters are information that in most cases is already available in the warehouse and picking system.

[0023] It is also advantageous if another of the information sources is defined by a light barrier, a light sensor, a (barcode / RFID / NFC) scanner, and / or a scale.

[0024] Such sensors are already present in conventional storage and order picking systems and will continue to be used in the future, at least at critical material flow nodes.

[0025] A further advantage is that controlling the material flow involves operating an actuator at the node.

[0026] In this case, it is possible for the respective object to change its state of motion, in particular its direction and / or its speed, at the node in a desired manner.

[0027] It is further optionally proposed that the checking step comprises: detecting each of the approaching objects from the sensor data; extracting associated features for each of the detected objects; and for each of the detected objects, determining whether the extracted associated features allow the secure identification.

[0028] In particular, the information fusion comprises a data fusion of the sensor data with the additional data by merging and completing a corresponding incomplete sensor data set, which does not include the associated feature required for secure identification, with an additional data set, wherein the additional data set comprises a variable, in particular one of the features, which is present in both the sensor data set and the additional data set.

[0029] The method may further comprise the optional step of checking whether the merged and completed sensor data set includes one of the features that comprises the sufficiently secure identification.

[0030] In particular, sufficiently secure identification can be defined by a variable threshold for the degree of identification. This degree can be, for example, 60%, 70%, 80%, or 90% of a (100%) secure identification, where the identification is unique or even one-to-one.

[0031] Furthermore, the object is achieved by a device according to claim 10.

[0032] It is understood that the features mentioned above and those to be explained below can be used not only in the combination specified in each case, but also in other combinations or on their own, without departing from the scope of the present invention.

[0033] Embodiments of the invention are illustrated in the drawings and explained in more detail in the following description. They show: Fig. 1 a perspective view of a node of a material flow in a partially shown storage and picking system; Fig. 2 a block diagram of a storage and picking system; Fig. 3 a flow diagram of a method for controlling a material flow; and Fig. 4 a top view of two adjacent nodes to illustrate different funding scenarios.

[0034] Fig. Figure 1 shows a perspective view of an exemplary (material flow) node 12 in a storage and order-picking system 10, which is again only partially shown. The node 12 is implemented, for example, by a conveyor system 14 (e.g., roller conveyor 16, belt conveyor, or the like). Fig. 1 shows a roller conveyor 16 on unspecified supports. A material flow 18 is indicated by light arrows on the roller conveyor 16. The junction point 12 is designed as a T-shaped branching point 20 (in plan view).

[0035] A preferably single sensor 22 for monitoring the material flow 18 is arranged (e.g. stationary) at the node 12. The sensor 22 is preferably implemented by a (video) camera 24 having a field of view 26 which is in the Fig. 1 is indicated by a dashed line and which has, for example, a conical extension with the camera 24 as the origin. The field of view 26 preferably covers a relatively large area, which, for example, centrally encompasses the node 12. The field of view 26 can be dimensioned such that an entrance area 28 (including a conveyor technology input element), the two exit areas 30 (including the corresponding conveyor technology output elements) as well as a centrally arranged intersection area 32, to which the entrance and exit areas 28 and 30 are adjacent, are reliably recorded by the camera 24. In particular, the movements of material flow objects 36 are recorded that approach the intersection area 32 (in the Fig. 1 from left to right).

[0036] In the intersection area 32, for example, a lifting device 34 in the form of a two-track belt conveyor is provided, which is not described in detail here. The tracks or belts of the belt conveyor are arranged in the spaces between rollers (not shown or described in detail here) of the roller conveyor 16. The lifting device 34 is designed to be raised and lowered relative to the roller conveyor 16. The belts of the lifting device 34 are oriented such that the material flow objects 36 can be discharged perpendicular to the original conveying direction. For this purpose, the previously lower belts between the rollers are raised so that objects 36 located on the belts are lifted by the rollers.

[0037] Instead of the lifting device 34, other actuators can also be provided as an alternative (or in addition). Examples of other actuators include: alignment devices, push-off devices, labeling devices, lifts, or similar devices.

[0038] In the Fig. 1 shows two exemplary objects 36-1 and 36-2 in the area of ​​the node 12. More or fewer objects 36 can also approach the node 12. The objects 36-1 and 36-2 are conveyed at a distance from one another, but they could also be conveyed together without any distance from one another (as a group). The objects 36 are shown here as examples of cuboid-shaped containers, which in the present case are transported via the roller conveyor 16 as conveyed goods without load carriers (e.g., trays, containers, pallets, or the like). It is understood that the objects 36 can be realized by any type of goods (e.g., picking goods, piece goods, containers, etc.) that are handled in the storage and picking system 10. The objects 36 can also be realized by (conveyed) goods including a load carrier (not shown).

[0039] An important function of the sensor 22 is to detect the material flow 18 in the area of ​​the node 12 in order to provide information to a material flow control 38 (see Fig. 2) To provide information about the objects 36-1 and 36-2 in order to direct the objects 36 in a desired direction (e.g., into the exit area 30-1). The material flow 18 comprises in the example of Fig. 1 shows the two objects 36-1 and 36-2 moving toward node 12. The material flow 16 can include, among other things, information about the objects 36, such as their current position and speed on the roller conveyor. Of particular interest, however, is the identity of the objects 36-1 and 36-2.

[0040] In order to correctly control the objects 36-1 and 36-2 at the intersection 12, ie, to correctly influence the material flow (e.g., to move them to a desired destination), the objects 36 should be clearly identified, in particular unambiguously, at the latest in the intersection area 32. This identification is carried out computer-aided in real time, in particular while the objects 36 are (continuously) conveyed into the intersection area 32.

[0041] The identification of objects 36 is generally based on features 40, such as barcodes 42, manufacturer's markings (e.g., logos) 44, or the like, which are described in more detail below and which are optically detected, evaluated, and recognized by the sensor 22 or camera 24 in the entrance area 28 of the node 12. Corresponding sensor data 46 is transmitted in the form of a signal from the sensor 22 to an evaluation device 48. The corresponding signal or data transmission can also be wireless.

[0042] The evaluation device 48 evaluates the sensor data 46 by checking whether one or more features 40 are present for each of the detected objects 36, which allow a secure identification of the respective objects 36. The evaluation device can be configured to analyze the sensor data 46 according to the features 46, as will be explained in more detail below. However, this analysis (extraction) of the features 40 can also be performed by another entity (not shown). The evaluation device 48 is further configured to evaluate the features 40 obtained from the sensor data 46 (if they are present at all) in order to determine whether secure identification is even possible for each of the objects 36-1 and 36-2.

[0043] There are different degrees of "secure identification." The barcode 42 generally allows 100% secure identification of the objects 36, provided the barcode 42 was visible and resolvable by the camera 24. The manufacturer's identifier 44 (e.g., a company-specific logo) does not always result in 100% secure identification. For example, it is possible that multiple objects 36 from the same manufacturer, each different in terms of an object type, are present within the storage and picking system 10, so that a secure, in particular a unique or even one-to-one, identification of the manufacturer's identifier 44 is not sufficient to reliably identify the respective type of object 36 (100%).

[0044] In this (insufficient) case, a “sufficiently secure” identification is carried out on the basis of information fusion, as will be explained in more detail below.

[0045] During information fusion, the evaluation device 50 uses additional data 50 in the evaluation and assessment in addition to the sensor data 46, i.e., in particular, in addition to the features 40 extracted therefrom, which originate from at least one information source 52 that differs from the sensor 22 of the node 12. The additional data 50 were generated in advance, in particular upstream relative to the sensor 22, and comprise - e.g., in addition to the manufacturer's identification 44 - further object features 40, which, together with the feature(s) 40 from the sensor data 46, allow the sufficiently secure identification for each of the objects 36-1 and 36-2. The sufficiently secure identification does not necessarily have to be a 100% secure identification. Threshold values ​​for the degree of identification can be defined (such as, for example,60%, 70%, 90% or the like) in order to allow the evaluation device 48 to make a decision as to which of the output areas 30-1 and 30-2 the corresponding object 36 is to be directed.

[0046] In general, it is possible that the evaluation device 48 can, for example, make a reliable identification for the object 36-1, but not for the other object 36-2. In this case, the object 36-1 can be controlled with or without information fusion in terms of material flow. The object 36-2, however, is always controlled by means of information fusion (based on the sensor data 46 and the additional data 50). The evaluation device 48 makes such a decision for each of the objects 36-1 and 36-2 approaching the node 12, as described with reference to Fig. 3 will be explained in more detail.

[0047] The term "information fusion" generally refers to methods by which data from different sensors, i.e., from one or more information sources 52, are linked with the aim of obtaining new and more precise knowledge about measured or detected values ​​(here, the images contained in the sensor data 46) and events. Information fusion can include sensor fusion, data fusion, and distributed detection or measurement. Basic principles of information fusion have long been known and will be described in more detail below.

[0048] Information fusion is used to control the material flow of objects 36 at node 12 (e.g., redirect, stop, accelerate, align, etc.) whose associated characteristics 40, which are object-specific or (uniquely) individualizing, can either not be detected at all or not with sufficient accuracy by sensor 22. For example, object 36-1 could be positioned on the roller conveyor 16 with exactly the side on which the barcode 42 is applied, allowing secure identification. In this case, however, camera 24 would not be able to detect barcode 42. In another case, the resolution of camera 24 might not be sufficient to make the (nevertheless visible) barcode 42 usable. The resolution of sensor 22 depends, among other things, on the size of field of view 26. Preferably, the size of field of view 26 is chosen to be very large in order to use as few sensors 22 as possible in the overall system 10.The field of view 26 is preferably at least large enough to encompass the input area 28 and the two output areas 30-1 and 30-2. However, it is also possible to make the field of view 26 even larger, so that, for example, several nodes 12 are detected simultaneously by the same sensor 22.

[0049] The one or more other information sources 52 are preferably controlled by the material flow control 38 and / or a warehouse management control 54 (cf. Fig. 2) is implemented. In the example of Fig. 1, the material flow control 38 (not shown) has, for example, information that the smaller, second object 36-2 is at a drop-off point not shown or designated in detail here (e.g. in the goods receipt 56 or at a work station 58, cf. Fig. 2) was moved onto the roller conveyor 16 after the first object 36-1. Thus, (supplementary) information on a geometric relative distance and a sequence of the objects 36-1 and 36-2 may be available. In the goods receiving area 56 or at the workstation 58, both objects 36-1 and 36-2 were uniquely – and thus reliably – identified, for example, by scanning their respective barcode 42 (barcode 42 of the second object 36-2 is in Fig. 1 not visible), for example with a hand-held scanner not shown or identified in detail here. The scanner may have been operated by a person who placed the objects 36-1 and 36-2 on the roller conveyor 16. Alternatively, another camera 24 could have been used at the goods receipt location 56 or the workstation 58 to clearly and reliably identify the objects 36-1 and 36-2 there (e.g. because both barcodes 42 were visible and resolvable there). The time interval that elapsed between the placement of the first object 36-1 and the placement of the second object 36-2 on the roller conveyor 16 represents a further exemplary piece of information or a feature 40 for the special object pair 36-1 / 36-2, which can be used by the evaluation device 48 when identifying the respective objects 36-1 and 36-2.

[0050] Furthermore, in a material flow computer 60 (cf. Fig. 2) and / or in subordinate control units, such as a programmable logic controller (PLC) 62 of a conveyor module not shown or designated in detail here, the order in which the objects 36-1 and 36-2 are transported downstream to the node 12 can be stored.

[0051] The warehouse management control 54 (cf. Fig. 2) is typically implemented by a warehouse management computer (LVR) 64. The warehouse management controller 54 manages object-specific data or object data 66, which can be used as the additional data 50. The object data 66 includes the (object-specific or associated) characteristics 40. The characteristics 40 represent one or more of the following variables or parameters: a weight, a shape or external contour, a dimension, a packaging (e.g., cardboard, plastic film, etc.), a color, a pattern, and / or the manufacturer's mark 44. It is understood that other characteristic types, variables or parameters can be considered and the above list is not exhaustive.

[0052] Further information can be obtained, for example, from a storage location not specified here of one of the objects 36 in a warehouse 68 (cf. Fig. 2). The warehouse 68 represents a material flow source / sink 70 for the objects 36 within the storage and picking system 10 of the Fig. 2. Information is stored in a warehouse management software about which of the objects 36 are stored in which storage location within the warehouse 68. If the warehouse management controller 54, for example, retrieves a container from the warehouse 68 that contains several objects 36 that correspond to the type of the first object 36-1, it may be sufficient, for example, to track this container by means of the material flow controller 38 (tracking / tracing) while it is transported through the storage and picking system 10 by means of a transport system 72 in order to later allow reliable identification. The roller conveyor 16 can be part of the transport system 72. The transport system 72 can further comprise storage and retrieval machines (SRM) 74, driverless transport systems (FTS) 76 and the like. The material flow 18 can be monitored by means of further sensors 22 upstream and / or downstream to the node 12 of the Fig. 1. These sensors 22 can be of the same type or of a different type. Another type of sensor could, for example, be implemented by a scale for determining the weight of the respective object 36.

[0053] An order management 78 (cf. Fig. 2) represents a further potential source of information 52. Information about the material flow 18 at the node 12 can also be obtained or extracted from the corresponding data of the order management 78, as described below by way of example. A specific order comprises, for example, five objects 36, which are reflected in one or more corresponding order lines. Four objects 36 of these five objects 36 have already been recognized at the node 12. Under the further prerequisite that the material flow 18 is order-oriented, i.e. that the objects 36 for this specific order are always transported as a group, the above-mentioned information fusion can be used to determine which object 36 the unidentifiable object is. It must be the fifth object 36 of this order.

[0054] The aim of the present invention, i.e. influencing the material flow 18 at the junction 12, is preferably achieved with only a single sensor technology, namely, for example, with the (video) camera 24. The camera 24 replaces a large number of sensors (e.g., light sensors, light barriers, etc.) that were conventionally used. A further advantage is that, for example, stop positions, discharge positions, deflection positions, and the like can be detected and influenced independently of the sensor position due to the (large) field of view 26 of the camera 24. This is not about controlling the material flow 18 with cameras 24 per se, but about eliminating a major disadvantage of the camera 24, which can be seen in the fact that the camera 24 either cannot see the object-identifying features 40 at all or cannot resolve them sufficiently.This disadvantage is remedied by information fusion, by combining and evaluating the various information sources 52 and combining them to solve the specific problem.

[0055] As a result, a control decision (e.g., path decision, stop decision, etc.) can be made for an object 36 that could not be detected or identified conventionally (only by the camera 24). The technological requirements for the sensors (e.g., image resolution, image processing, and the like) are reduced. This generally allows the use of cameras 24 that are significantly more cost-effective and can already be implemented with current technology.

[0056] It is understood that the technology described here is not only used at branching points 20 (cf. Fig. 1), but also at workstations 58 (e.g., picking workstations) or in goods receipt 56 or goods issue. The technology can even be used in buffer sections, where the objects are stopped (in phases) at predetermined points on the conveyor system 14, for example, to close gaps between adjacent objects 36 or to generate gaps between adjacent objects 36. In this case, the junction point 12 is therefore not a branching point 20, but merely a decision point (stop or convey?).

[0057] A further aspect of the described technology can be seen in a visualization. Both the entire storage and picking system 10 and just parts of it can be visualized. Disturbances (such as a jam on the roller conveyor 16) can thus be visualized in a simple manner. In this case, it is even possible to clearly identify the jammed objects 36. The material flow 18 generally includes information about transport processes and procedures as well as storage and unwanted stays of all objects 36 in the system 10. By means of a material flow analysis, weak points in the material flow 18 can be identified and their causes investigated. The graphic representation (visualization) of the material flow 18 serves to highlight and clarify conditions and processes in order to identify specific data and weak points and to make them easier to assess.The graphical representation offers the advantage of better clarity and ensures that the essentials can be quickly grasped.

[0058] In a further alternative, the sensor 22 of the Fig. 1 may not be installed in a fixed location. The sensor 22 may, for example, be mounted on a movably mounted arm to move the sensor 22 to an area of ​​the system 10 that is of particular interest at a given time. The sensor 22 could also be mounted on a drone, for example. In this case, it would be possible to quickly and flexibly solve material flow problems that previously required the deployment of people on site.

[0059] Information fusion will be examined in more detail below. To implement information fusion, artificial intelligence or a neural network can be used in the evaluation device 48.

[0060] The data fusion mentioned above refers in this case to the merging and completion of incomplete data sets. Data fusion is a component of information integration. A "donor data set" (additional data 50) provides supplementary data for a "recipient data set" (sensor data 46). For example, the donor data set consists of the variables (G, S), and the recipient data set consists of the variables (G, E). The common variable G is therefore present in both data sets, while the donor variable S and the recipient variable E are each present in only one of the data sets. Based on the donor data set, a model is created to calculate the value S from the variable G. This model is applied to the recipient data set, creating a new fused data set (G, E, S). Statistical methods can be used in this context. Data fusion allows several partially incomplete data sets to be combined.The preferred variables are attributes 40. Before merging data from two information sources 52 is possible, the data may need to be aligned to a common schema. Missing attributes or variables can be filled with zero (for "no value"). However, a common identification of the attribute as an identifier is usually required. A merge operator can be used, for example, to merge incomplete data sets that cross over.

[0061] Furthermore, it may happen that more than two data sets from different sensors (including sensor 22) are fused together. This is then referred to as sensor data fusion. In sensor data fusion, output data from multiple sensors or information sources 52 are generally linked. The goal is to obtain information of better quality. Better quality means that the objects 36 at the node 12 can be reliably identifiable and, in particular, better classified. In this way, the reliability of the material flow control 38 can generally be increased by using several of the information sources 52. Using different sensor types also increases the overall detection probability. This means that the objects 36 can be detected even if individual sensors that differ from sensor 22 are limited in their perception or are "blind" due to environmental conditions.To increase identifiability, the measurement errors of the sensors should not correlate with each other, but should satisfy certain probability density functions (e.g., normal distribution) or be otherwise identifiable and quantifiable. Since the field of view (e.g., field of view 26) of the sensors is usually limited, the use of multiple sensors increases the field of view of the overall system accordingly. Ambiguous identification is easier to resolve. Additional sensors often provide additional information, thus expanding the knowledge of the overall system. Sensor data fusion can also lead to cost reduction, for example, because several cheaper sensors can replace a particularly expensive sensor.

[0062] With reference to Fig. 3, a flow diagram of a method for controlling the material flow 18 in the storage and order picking system 10 at the sensor-controlled node 12 of the material flow 18 is described below, where a decision is to be made about a future material flow direction and / or about a speed of the material flow 18.

[0063] In a step S10, the additional data 50 is generated (in advance). One or more (data) sets of additional data 50 are generated by one or more of the information sources 52 that differ from the sensor 22 of the node 12. The additional data 52 are generated in advance in order to be able to later perform the information fusion with the sensor data 46, as will be explained in more detail below. Step S10 is optional. It is understood that the additional data 50 does not have to be generated explicitly, but can also already be present as data sets in the system 10.

[0064] In a step S12, the sensor data 46 are generated at the node 12. The sensor data 46 are generated by the sensor 22 by detecting the objects 36 in the field of view 26 of the sensor 22, which is implemented in particular by the (single) camera 24.

[0065] In a step S14, the features 40 of the detected objects 36 are obtained from the sensor data 46 by the evaluation device 48, i.e., extracted. During the extraction, image recognition is preferably performed. Image recognition is a sub-area of ​​pattern recognition and image processing. During pattern recognition, regularities, repetitions, similarities, and / or laws are recognized in the set of sensor data 46. In particular, the above-mentioned parameters (e.g., barcode 42, manufacturer's mark 44, dimensions, color patterns, packaging types, and the like) of the objects 36 are recognized that approach the node 12 downstream and enter the field of view 26 of the sensor 22, and in particular, the entrance area 28 of the node 12. A central point in this context is the recognition of the patterns, i.e.,in particular the features 40 which are common to all objects 36 of a category and which distinguish these objects 36 from the content of other categories.

[0066] During image processing, the sensor data 46 is processed, which can be (individual) images from video data of the camera 24. The result of the image processing can be a set of features 40 of the input image. In most cases, the images are viewed as a two-dimensional signal, so that conventional methods from signal processing can be applied. The objects 36 can, for example, be measured. A position of the objects 36 relative to the roller conveyor 16 can be determined. A relative distance between neighboring objects 36 can be measured. The barcode 42 can be "measured" and thus recognized. By means of image processing, coded information (barcode 42) can therefore be automatically read out from the images in order to identify the objects 36.

[0067] In step S16, a check is performed—for example, via a feature comparison—to determine whether the extracted feature(s) 40 are present for the reliable identification of the respective detected objects 36. If these features 40 are present, the material flow 18 is controlled, preferably exclusively, based on the sensor data 46, as shown in step S18. Otherwise, the additional data 50, which in turn includes supplementary features 40, is retrieved in step S20.

[0068] In a step S22, the above-mentioned information fusion of the sensor data 46 with the additional data 50 then takes place. Subsequently, in a step S24, a check is carried out to determine whether features 40 are now present for a sufficiently reliable identification. If a sufficiently reliable identification cannot yet be performed, further additional data 50 is retrieved by returning to step S20, and further additional data 50 from further information sources 52 is taken into account. Otherwise, the material flow 18 is controlled based on a combination of the sensor data 46 and the additional data 50.

[0069] This concludes procedure 100.

[0070] Fig. Figure 4 shows a schematic plan view of a section of a conveyor system 14, which is formed, for example, by a main line 80 and two secondary lines 82-1 and 82-2. The secondary lines 82 extend perpendicular to the main line 80 and can be used as feed conveyors and / or discharge conveyors. The areas where the main line 80 is connected to the secondary lines 82 represent the nodes 12-1 and 12-2. Each of the nodes 12-1 and 12-2 is monitored by a camera 24 (not shown in detail). Fig. 4, the corresponding fields of view 26-1 and 26-2 are indicated by (exemplary) circular dashed lines. The fields of view 26-1 and 26-2 are selected to be large enough to capture the above-mentioned areas 28-32 of the intersections 12-1 and 12-2, which are also not shown in detail here. In the intersection area 32 of the left intersection 12-1, for example, an excavation device 34-1 of the type described above is provided. The area of ​​the right intersection 12-2 comprises, by way of example, a scale 84, indicated by a dashed rectangle, as a further sensor 22 next to the camera 24. Furthermore, a further excavation device 34-2 is provided.

[0071] Dark and light arrows indicate Fig.4 shows two different scenarios as examples. In a first scenario, the dark arrows are considered. The objects 36-1 to 36-3 are transported from left to right, with the node 12-1 being shown at a first point in time and the node 12-2 being shown at a later second point in time. In a second scenario, the light arrows are considered, with the objects 36 being transported from right to left and the node 12-2 showing a snapshot that is earlier in time than the snapshot at the node 12-1. In both scenarios, several information or sensor sources are used in order to identify the conveyed goods. These are synchronized with the image processing information of the cameras 24, among other things, in order to control the motors and, if applicable, other actuators at the respective nodes 12-1 and 12-2. This means that the camera(s) and the resultingThe temporal synchronization obtained between objects 36 and locations also results in a synchronization of the sensor data. List of reference symbols 10 Storage and picking system 12 Junction 14 Conveyor technology 16 roller conveyors 18 Material flow 20 branching point 22 Sensor 24 Camera 26 Field of view 28 Entrance area 30 Exit area 32 intersection area 34 Excavation equipment 36 (Material flow) object 38 Material flow control 40 feature 42 barcode 44 Manufacturer markings 46 sensor data 48 Evaluation device 50 additional data 52 Source of information 54 Warehouse management control 56 Goods receipt 58 workplaces 60 material flow calculators 62 PLCs 64 LVR 66 Object data / objective-specific data 68 camps 70 Source / Sink 72 Transport system 74 RGB 76 FIS 78 Order management 80 main line 82 branch line 84 scales

Claims

[1] Method for controlling a material flow (18) in a storage and order picking system (10) at a sensor-controlled node (12) of the material flow (18), where a decision is to be made about a future material flow direction and / or about a speed of the material flow (18), wherein at least one input area (28) of the node (12) is detected by a detection field (26) of a single sensor (22) and wherein the node (12) is implemented by a conveying technology (14), comprising the steps: by means of the sensor (22) generating (S10) sensor data (46) that represent the material flow (18) defined by several material flow objects (36) approaching the node (12) downstream in the detection field (26) of the sensor (22), in particular in the input area (28) of the node (12); Check (S16) whether the sensor data (46) for each of the approaching objects (36-1, 36-2) includes at least one associated object feature (40) that allows for the reliable identification of the corresponding object (36); and if the sensor data (46) for each of the objects (36-1, 36-2) includes at least one associated feature (40) that allows secure identification, control (S18) of the material flow (18) at the node (12) based on the sensor data (46); or if the sensor data (46) do not include for each of the approaching objects (36-1, 36-2) the at least one associated feature (40) that allows secure identification (S16), control (S26) of the material flow (18) at the node (12) at least for those of the objects (36-1, 36-2) whose associated feature (40) is not included in the sensor data (46), based on an information fusion (S22) of the sensor data (46) with pre-generated additional data (50) originating from at least one information source (52) that differs from the sensor (22) of the node (12), wherein the at least one information source (52) includes a material flow control (38), a warehouse management control (54), a work center (58), and / or a goods receipt (56), wherein the additional data (50) includes further object features (40),which together with the features (40) from the sensor data (46) allow sufficiently reliable identification (S24) and wherein the additional data (50) for each of the approaching objects (36) include at least one of the following parameters: , a sequence in which the objects (36) approach the node (12); a relative distance between successive objects (36); a time at which a respective object (36) was added to the material flow (18); a delivery point where the respective object (36) was added to the material flow (18); a path that the respective object (36) has traveled so far through the system (10); a location-dependent speed at which the respective object (36) is moved through the system (10); a dimension of the respective object (36); a color combination of the respective object (36); a manufacturer's identification mark (44) of the respective object (36); and / or a picking order to which the respective object (36) belongs; and wherein the only sensor (22) is a camera (24). [2] Method according to claim 1, wherein the at least one feature (40) is optically identifiable and is preferably a barcode (42), an individualizing color combination, an individualizing manufacturer's mark (44), an individualizing contour, and / or an individualizing dimension or size. [3] Method according to claim 1 or 2, wherein the field of view (26) of the camera (24) is selected to be so large that visible barcodes (42) cannot be resolved. [4] Method according to one of claims 1 to 3, wherein a further information source (52) is a light barrier, a light switch, a (barcode / RFID / NFC) scanner, and / or a scale. [5] Method according to any one of claims 1 to 4, wherein controlling the material flow (18) comprises sensor data synchronized actuation of an actuator provided at the node (12), (12) so that the respective object (36) changes its state of motion, in particular its direction and / or its speed. [6] Method according to any one of claims 1 to 5, wherein the testing (S16) comprises: Detect each of the approaching objects (36) from the sensor data (46); Extracting associated features (40) for each of the detected objects (36); and For each of the identified objects (36) Determine whether the extracted associated features (40) allow for secure identification. [7] Method according to any one of claims 1 to 6, wherein the information fusion (S22) comprises a data fusion of the sensor data (46) with the additional data (50) by merging and completing a corresponding incomplete sensor data set, which does not include the associated feature (40) required for secure identification, with an additional data set, wherein the additional data set comprises a variable, in particular one of the features (40), which is present in both the sensor data set and the additional data set. [8] The method of claim 7, further comprising the step of: Check whether the merged and completed sensor data set includes one of the features (40) that provides sufficiently reliable identification. [9] Method according to any one of claims 1 to 8, wherein the step of controlling (S18, S26) the material flow (18) at the node (12) comprises synchronous control of at least one actuator (34) provided within the area of ​​the node (12). [10] Device for controlling a material flow (18) in a storage and order picking system (10) at a sensor-controlled node (12) of the material flow (18), where a decision is to be made about a future material flow direction or about a speed of the material flow (18), wherein the node (12) is implemented by a conveyor technology (14) and wherein the device comprises: a single sensor (22) which is a camera (24) and which has a field of view (26), wherein the sensor (22) is positioned such that the field of view (26) captures at least one entrance area (28) of the node (12) where several material flow objects (36) simultaneously approach the node (12) downstream, wherein the sensor (22) generates corresponding sensor data (46); at least one information source (52) for information fusion of the sensor data (46) with pre-generated additional data (50) originating from the at least one information source (52), wherein the at least one information source (52) is distinct from the sensor (22) at the node (12), and wherein the at least one information source (52) comprises a material flow control (38), a warehouse management control (54), a work center (58), and / or a goods receipt (56); and an evaluation facility (48) that is set up: to check whether the sensor data (46) for each of the approaching objects (36-1, 36-2) includes at least one associated object feature (40) that allows for the reliable identification of the corresponding object (36); and if the sensor data (46) for each of the objects (36-1, 36-2) includes at least one feature (40), to output a signal to control the material flow (18) at the node (12) based on the sensor data (46); or If the sensor data (46) do not include the at least one feature (40) for each of the approaching objects (36-1, 36-2), a signal for controlling the material flow (18) at the node (12) is to be output at least for those of the objects (36-1, 36-2) whose associated feature (40) is not included in the sensor data (46), based on an information fusion of the sensor data (46) with additional data (50) originating from the at least one information source (52), wherein the additional data (50) include further object features (40) which, together with the features (40) from the sensor data (46), allow for sufficiently reliable identification; and to perform information fusion; wherein the additional data (50) for each of the approaching objects (36) include at least one of the following parameters: a sequence in which the objects (36) approach the node (12); a relative distance between successive objects (36); a time at which a respective object (36) was added to the material flow (18); a delivery point where the respective object (36) was added to the material flow (18); a path that the respective object (36) has traveled so far through the system (10); a location-dependent speed at which the respective object (36) is moved through the system (10); a dimension of the respective object (36); a color combination of the respective object (36); a manufacturer's identification mark (44) of the respective object (36); and / or a picking order to which the respective object (36) belongs.

Images