Method and device for intelligently selecting the field of view of cameras on a machine tool

Abstract

The invention relates to a method (34) for monitoring the manufacture of a component (14) using a machine tool (12), wherein a plurality of cameras (20a, b) cover different fields of view (22a, b). An algorithm (26) can create a ranking of the relevance of images from the cameras (20a, b) and display or emphasise only the most relevant image or images on a monitor (28). The algorithm (26) can assign a higher relevance to images that: a) depict a known machine tool part that is mentioned in a status or error message (30); b) have a high optical flow; and / or c) depict an identified machine tool part that is moving. In the event of c), images can be successively assigned higher relevance if an identified machine tool part moves from one field of view (22a, b) belonging to said image to the next. Images assigned lower relevance can be deleted or displayed - and / or saved - at a reduced size. The algorithm (26) can take the form of artificial intelligence. The invention also relates to a device (10) for carrying out such a method (34).

Description

[0001] Method and device for intelligent field of view selection of cameras on a machine tool

[0002] Background of the invention

[0003] The invention relates to a method and a device for producing a component.

[0004] DE 10 2017 121 098 A1 discloses tracking a component to be manufactured through a production line. For this purpose, several cameras are installed in the production line, with the camera monitoring the component changing depending on the camera's field of view. The component's location can be supported by a UWB positioning system.

[0005] It is also known to observe the production of a component in a machine tool by observing it from different fields of view or perspectives using multiple cameras. A user or observer of the production process can be located at a different location than the machine tool and act as a so-called remote controller. However, it is difficult for an inexperienced user to find the relevant camera images from the multitude of available images. Especially in the event of a machine tool failure, the user wants to quickly view the relevant camera images.

[0006] Object of the invention

[0007] It is therefore an object of the invention to provide a method and a device that significantly facilitate the monitoring of a machine tool.

[0008] This object is achieved according to the invention by a method according to claim 1 and a device according to claim 12. The dependent claims represent preferred developments.

[0009] The problem is thus solved by a method for producing a component with a machine tool, wherein the following process steps are carried out:

[0010] A) Images of at least parts of the machine tool are taken by several cameras, the images being fed into an evaluation unit which has an algorithm for evaluating the images;

[0011] B) one or more relevant images are selected by the algorithm and output to a monitor. The relevance of the images is determined by the following criterion(s): a) images showing a machine tool part that is mentioned in a status output (e.g. a sensor value or an axis position) or error output of a machine tool controller are assigned a higher relevance than images showing a machine tool part that is not mentioned in a status or error output; b) images that have a high optical flow are assigned a higher relevance than images that have a lower optical flow; and / or c) images in which a moving machine tool part is identified are assigned a higher relevance than images in which no moving machine tool part is identified.

[0012] The method according to the invention enables a user or observer of the machine tool to concentrate on the "region of interest" of the machine tool. This allows even less experienced users to perform remote maintenance of the machine tool. Errors can be analyzed and located more quickly and easily.

[0013] The method can be configured to display only the most relevant image on the monitor. Alternatively, only a few relevant images can be displayed on the monitor. Alternatively, the most relevant image or a few relevant images can be highlighted on the monitor. For example, less relevant images can be displayed smaller on the monitor.

[0014] The cameras are preferably designed as video cameras; the recordings are then available as video recordings.

[0015] Process steps A) and B) can be repeated, especially multiple times. This allows the user to continuously track the region of interest in the machine tool.

[0016] Preferably, at least one of the cameras is aligned to the area of ​​the machine tool identified by the status output and / or the error output of the machine tool control system. Alignment can be achieved, in particular, by panning and / or zooming the camera. By panning and / or zooming, the camera's field of view can be particularly well aligned to the area of ​​interest. If multiple cameras are aligned to the area, it is possible to view the area from multiple angles. The most relevant image can then be selected from the multiple angles.

[0017] Particularly preferably, when applying criterion c), the algorithm predicts the camera's field of view into which an identified machine tool part will next move and assigns greater relevance to the associated image (of the next field of view) as soon as the identified machine tool part has left the current field of view. This allows the user to easily track a machine tool part. Process step B) can be performed "live," i.e., largely without delay after process step A). ​​Alternatively, process step B) can be performed using saved images, particularly for machine tool error analysis.

[0018] To conserve resources, recordings with less relevance can be deleted or transferred or saved at a reduced resolution.

[0019] In a further preferred embodiment of the invention, the algorithm is implemented in the form of a machine learning algorithm. The machine learning algorithm is preferably trained using stored recordings that have been assigned to a specific status or error output. Alternatively or additionally, the machine learning algorithm can be trained by selecting cameras from one or more experienced users.

[0020] The monitoring of the machine tool is further facilitated if, in addition to the recording(s) displayed on the monitor, a status or error output of the machine tool control is provided, particularly on the monitor.

[0021] Furthermore, an interaction option with the machine tool can be provided. Machine data can be displayed in the recordings, and functions can be deactivated / activated directly in the recordings if the situation requires it. Example: A machine tool stops with a "transport control" error. The operator is immediately shown the relevant live recording(s) and can restart the machine tool from the recording if the error is a false alarm. This enables significantly faster operation than having to switch to a control system and search for the function.

[0022] The method according to the invention is particularly suitable for use on a machine tool designed for laser processing, in particular for laser cutting or laser welding, of the component. Alternatively or additionally, the method can be used on an automated bending system or a storage system.

[0023] The object of the invention is further achieved by a device for carrying out a method described herein. The device comprises the machine tool, the cameras—particularly in the form of video cameras—the evaluation unit with the algorithm, and the monitor. Features and advantages described for the method apply accordingly to the device, and vice versa.

[0024] The machine tool preferably has a laser head for laser processing the component, in particular for laser cutting and / or laser welding.

[0025] Alternatively or additionally, the machine tool may have an automated tool changer, particularly for punching or bending machines.

[0026] Further advantages of the invention will become apparent from the description and the drawings. Likewise, the above-mentioned and further-described features can be used individually or in combination in any desired manner. The embodiments shown and described are not intended to be exhaustive, but rather are exemplary in nature for describing the invention.

[0027] Detailed description of the invention and drawing

[0028] Fig. 1 shows a schematic view of a device according to the invention and a method according to the invention.

[0029] Fig. 1 shows a device 10 with a machine tool 12 for producing a component 14. For producing the component 14, the machine tool 12 has, among other things, a laser head 16, here in the form of a cutting head. The production of the component 14 is controlled by a machine tool control system 18. The production of the component 14 is monitored by several cameras 20a, 20b, here in the form of video cameras. Each camera 20a, b monitors its own field of view 22a, 22b. The fields of view are shown separately here, but can also overlap. The images created by the cameras 20a, b, here in the form of videos, are fed into an evaluation unit 24. The evaluation unit 24 can be part of the machine tool control system 18. The evaluation unit 24 has an algorithm 26 that evaluates the images.

[0030] The evaluation unit 24 is configured to use the algorithm 26 to determine the most relevant images, in particular the most relevant image, and transmit them to a monitor 28. The evaluation of the production process is made significantly easier for a user by selecting the most relevant image(s). Particularly in the event of a production error, the user can focus on the relevant images on the monitor 28 and does not have to suppress images of irrelevant parts of the machine tool 12.

[0031] Furthermore, the evaluation unit 24 is configured to align the cameras 22a, 22b to the area of ​​the machine tool 20 designated by a status output or an error output 30 of the machine tool control 18. For this purpose, the cameras 22a, 22b can be pivoted by the evaluation unit 24, or a zoom of the cameras 22a, 22b can be adjusted.

[0032] Algorithm 26 can be implemented as a machine learning algorithm. The machine learning algorithm can be trained, in particular, based on tagged stored recordings and / or user behavior when selecting recordings.

[0033] The algorithm 26 may evaluate the relevance of the images by a) assessing whether the image shows a part of the machine tool 12 for which a status and / or error output 30 is issued by the machine tool controller 18; b) determining the optical flow of the images and assigning high relevance to images with a high optical flow; and / or c) identifying a moving part of the machine tool 12 in the images.

[0034] For example, when applying criterion c), the algorithm 26 can identify the laser head 16 in the field of view 22a and, based on the movement of the laser head 16 in the direction of an arrow 32, recognize that it will next appear in the field of view 22b. Once the laser head 16 has left the field of view 22a, the algorithm 26 can then assign a high relevance to the field of view 22b, allowing the user to continue tracking the movement of the laser head 16 on the monitor 28.

[0035] The device 10 shown in Fig. 1 or a method 34 shown in Fig. 1 enable the concentrated tracking of the most relevant images. Irrelevant images can be deleted or transmitted at reduced resolution to save storage space, computing capacity, and / or bandwidth.

[0036] Looking at the drawing, the invention relates in summary to a method 34 for monitoring the production of a component 14 using a machine tool 12, wherein a plurality of cameras 20a, b cover different fields of view 22a, b. An algorithm 26 can create a relevance ranking of images from the cameras 20a, b and display or highlight only the most relevant one or more on a monitor 28. The algorithm 26 can assign a higher relevance to images that a) show a known machine tool part mentioned in a status or error output; b) have a high optical flow; and / or c) show an identified machine tool part that is moving. In case c), images can be assigned a higher relevance one after the other if an identified machine tool part moves from one field of view 22a, b belonging to this image to the next.Images deemed less relevant can be deleted or displayed—and / or saved—at a reduced size. The algorithm 26 can be implemented in the form of artificial intelligence. The invention further relates to a device 10 for implementing such a method 34. List of reference symbols.

[0037] 10 Device

[0038] 12 Machine tool 14 Component

[0039] 16 Laser head

[0040] 18 Machine tool control

[0041] 20a, b Camera

[0042] 22a, b Field of view 24 Evaluation unit

[0043] 26 Algorithm

[0044] 28 monitors

[0045] 30 Status and / or error output

[0046] 32 Direction of movement of the laser head 16 34 Procedure

Claims

Patent claims 1. Method (34) for producing a component (14) with a machine tool (12), comprising the method steps: A) Simultaneously creating a plurality of images of at least parts of the machine tool (12) by a plurality of cameras (20a, b) with different fields of view (22a, b) and feeding the images into an evaluation unit (24) with an algorithm (26); B) Computer-based selection of one or more relevant recordings by the algorithm (26) and output of these recordings to a monitor (28), wherein the relevance of the recordings created is determined based on one or more of the following criteria: a) The respective recording shows an area of ​​the machine tool (12) which is identified by a status output and / or an error output (30) of a machine tool control (18); b) The respective recording shows a comparatively high optical flow; c) The respective recording shows a moving part of the machine tool (12) identified in the recording.

2. Method according to claim 1, in which the method steps A) and B), in particular several times, are repeated.

3. Method according to claim 1 or 2, wherein at least one of the cameras (20a, b) is aligned, in particular by panning and / or zooming, with the area of ​​the machine tool (20) designated by the status output and / or the error output (30) of the machine tool control (18).

4. Method according to claim 2 or 3, wherein the algorithm (26) in criterion c) predicts into which next field of view (22a, b) the identified part of the machine tool (12) will be moved, wherein the recording of this next field of view (22a, b) is evaluated with a high relevance as soon as the identified part has left the current field of view (22a, b).

5. Method according to one of the preceding claims, in which method step B) is carried out with stored recordings.

6. Method according to one of the preceding claims, in which a recording with low relevance is deleted or transmitted and / or stored with reduced resolution.

7. Method according to one of the preceding claims, wherein the algorithm (26) is designed in the form of a machine learning algorithm.

8. Method according to one of the preceding claims, in which a status output and / or an error output (30) of the machine tool control (18) is output for the image(s) displayed on the monitor (28).

9. Method according to one of the preceding claims, in which an interaction option for controlling the machine tool (12) is output on the image(s) displayed on the monitor (28).

10. Method according to one of the preceding claims, wherein the production of the component (14) comprises laser processing by the machine tool (12).

11. The method according to claim 10, wherein the production of the component (14) comprises laser cutting the component (14).

12. Device (10) for carrying out a method (34) according to one of the preceding claims, wherein the device (10) comprises the machine tool (12), the cameras (20a, b), the evaluation unit (24) with the algorithm (26) and the monitor (28), wherein the algorithm (26) is designed to determine the relevance of the images created with the cameras (20a, b) based on one or more of the criteria a) to c) and to output one or more relevant images to the monitor (28). Device according to claim 12 in conjunction with claim 10, wherein the machine tool (12) has a laser head (16). Device according to claim 12 or 13, wherein the machine tool (12) has an automated tool changer.

Images