Safety device for a safety zone of a machine tool

Abstract

A method for controlling a machine tool, wherein the machine tool is located in a safety zone, wherein the safety zone comprises an access zone, wherein at least one imaging sensor creates images of the access zone, wherein the imaging sensor is connected to a computing unit and images created by the sensor are processed by the computing unit, wherein the computing unit is designed to identify objects in the images, wherein the computing unit is connected to the machine tool and sends a stop signal to the machine tool if the imaging sensor identifies that a first object has entered the safety zone, wherein the computing unit sends an enable signal to the machine tool if the imaging sensor identifies that a second object has left the safety zone and comparison of the first object with the second object determines that the first object and the second object are identical.

Description

[0001] 2022P00076WO October 27, 2025

[0002] 1

[0003] Applicant:

[0004] TRUMPF Werkzeugmaschinen SE + Co. KG

[0005] Johann-Maus-Str. 2

[0006] 71254 Ditzingen

[0007] Safety device for a safety area of ​​a machine tool

[0008] The invention relates to a safety device for a safety zone of a machine tool. Furthermore, the invention relates to a method for controlling a machine tool within a safety zone.

[0009] State of the art

[0010] From DE10 2009 036 641 A1 a safety device is known in which a shutdown signal is issued to a source of danger when it is detected that a worker is in the danger zone.

[0011] Another safety device is known from DE10 2023 107 857 A1.

[0012] Task

[0013] The invention is based on the objective of improving a safety device of the type mentioned above.

[0014] Solution

[0015] This problem is solved according to the invention by a method according to claim 1 and a safety device according to claim 8. 2022P00076WO October 27, 2025

[0016] 2

[0017] Advantageous embodiments, which can be used individually or in combination with each other, are the subject of the dependent claims.

[0018] A method for controlling a machine tool is disclosed, wherein the machine tool is located in a safety area, the safety area comprising an access area, wherein at least one imaging sensor creates images of the access area, wherein the imaging sensor is connected to a computing unit and the images created by the sensor are processed by the computing unit, wherein the computing unit is configured to detect objects in the images, wherein the computing unit is connected to the machine tool and sends a stop signal to the machine tool when the imaging sensor detects that a first object has entered the safety area, wherein the computing unit sends a release signal to the machine tool when the imaging sensor detects thatthat a second object has left the security area, and a comparison of the first and second objects confirms that they are identical. This comparison ensures that the same objects that entered the security area have now left. Security within the security area is significantly improved compared to simply counting objects.

[0019] In a preferred embodiment, image recognition is used to classify at least one property of both the first and second objects, and these properties are then used to verify whether the first and second objects are identical. It is understood that a large number of properties can also be classified, thereby further increasing the reliability of the recognition. The more properties of the first and second objects are classified as identical, the more certain the determination that the first and second objects are identical becomes. 2022P00076WO October 27, 2025

[0020] 3

[0021] In one embodiment, artificial intelligence, in particular a trained neural network, is used to classify the property. The artificial intelligence can be trained with known training material for object recognition. Preferably, the training material focuses on objects that are frequently found near the safety zone. Since people in the safety zone can be harmed, it is advantageous if the artificial intelligence can distinguish between different people.

[0022] In one embodiment, the property is one of the following:

[0023] - Object type

[0024] - Color

[0025] - Size.

[0026] It is understood that the list of properties is not exhaustive and further properties of the first or second object can be determined using the imaging sensor or other sensors.

[0027] In a preferred embodiment, image recognition, particularly with artificial intelligence, is used to detect that an object is composed of several sub-objects, whereby the comparison of the first object with the second object is applied to the sub-objects. In other words, the sub-objects of the first object are compared with the sub-objects of the second object. By distinguishing the sub-objects, instead of a person entering the security area, a person wearing blue trousers, a green jacket, and carrying a black briefcase can be detected. If the person leaves the security area again but does not have a black briefcase with them, the security area is not released because not all objects have left the security area.

[0028] In one embodiment, the imaging sensor creates a sequence of images with a large number of temporally successive images, and the direction of movement of the first and / or the second object is determined from this sequence. (From 2022P00076WO, October 27, 2025)

[0029] 4

[0030] The direction of movement can be determined to determine whether an object is entering or leaving the safety area, i.e., whether an object is a first object or a second object.

[0031] In a preferred embodiment, a large number of initial objects enter the safety area, whereby the computing unit only sends a release signal to the machine tool when the imaging sensor detects that all objects that entered the safety area have left the safety area again.

[0032] A safety device for a safety area of ​​a machine tool is also disclosed, wherein the machine tool is located in a safety area, the safety area comprising an access area, wherein at least one imaging sensor is configured to create images of the access area, the imaging sensor being connected to a computing unit, and the safety device being provided and configured to carry out a previously described procedure.

[0033] List of characters

[0034] The invention is explained in more detail below with reference to an advantageous embodiment illustrated in the figures. However, the invention is not limited to this embodiment.

[0035] They show:

[0036] Fig. 1: a system comprising a security area with an access area,

[0037] Fig. 2: the system from Fig. 1 at a later time, and

[0038] Fig. 3: An exemplary process. 2022P00076WO October 27, 2025

[0039] 5

[0040] Fig. 1 shows a system comprising a safety zone 1. A machine tool 2 is located within the safety zone 1. The safety zone 1 has an access zone 3. Access to the safety zone 1 is only possible via the access zone 3. In this example, the safety zone 1, with the exception of the access zone 3, is protected by a fence. The access zone 3 is monitored by a camera 4. The camera 4 is connected to a processing unit 6. Using the camera 4, the processing unit 6 can determine whether an object crosses an access boundary 5 of the access zone 3.

[0041] A first object 7, here a ball, moving through access area 3, crosses access boundary 5. Camera 4 generates an image 8 at regular intervals. These images 8 are stored as a video stream in memory 9. The processing unit 6 evaluates the images 8 generated by camera 4 and stored in memory 9. The processing unit 6 uses image processing algorithms for this purpose. Using these algorithms, for example, the direction of movement 10 of the first object 7 can be determined. The algorithms can also determine properties of the first object 7, such as its type, color, and / or size. The algorithms can incorporate artificial intelligence, for example, in the form of a neural network. In this example, the processing unit 6 recognizes that the first object 7 has entered safety area 1 and sends a stop signal to machine tool 2.The properties of the first object 7 are saved.

[0042] Figure 2 shows the system from Figure 1 at a later stage. The processing unit 6 detects that a second object 11, moving through the access area 3, has crossed the access boundary 5. The camera 4 continues to generate an image 8 at regular intervals. The processing unit 6 evaluates the images 8 generated by the camera 4. The processing unit recognizes that the second object 11 is leaving the security area 1. The processing unit 6 determines the properties of the second object 11 and recognizes that the second object 11 is composed of several sub-objects, in this case, a person and a briefcase. The processing unit 6 compares the properties of the second object 11 with the stored properties of the first object 7. Since the second object 11 and the first object 7 are not identical, no release signal is sent to the machine tool. 2022P00076WO October 27, 2025

[0043] 6

[0044] Had the computing unit 6 determined through comparison that the second object 11 was identical to the first object 7, the computing unit 6 would have sent a release signal to the machine tool 2.

[0045] Figure 3 shows an exemplary procedure. In the first step 110, the processing unit 6 detects that an object 7, 11 crosses the access boundary 5. In a second step 120, the processing unit 6 determines the direction of movement of the object 7, 11. If the object 7, 11 enters the safety zone 1, it is recognized as the first object 7, and in a third step 130, a stop signal is sent to the machine tool 2, and the properties of the first object are stored. If the object 7, 11 leaves the safety zone 1, it is recognized as the second object 11, and in a fourth step 140, the properties of the second object 11 are compared with the stored properties of the first object 7.If the properties of the second object 11 match those of the first object 7, the second object 11 is assumed to be identical to the first object 7, and in a fifth step 150 a release signal is sent to the machine tool 2.

[0046] The features disclosed in the foregoing description, claims and drawings may be important for the realization of the invention in its various embodiments, both individually and in combination.

[0047] 2022P00076WO October 27, 2025

[0048] 7

[0049] Reference symbol list

[0050] 1 Security area

[0051] 2 machine tools

[0052] 3 Access area

[0053] 4 cameras

[0054] 5 Access boundary

[0055] 6 Calculation unit

[0056] 7 First object

[0057] 8th image

[0058] 9 storage

[0059] 10 Direction of movement

[0060] 11 Second object

Claims

2022P00076WO October 27, 2025 8 Patent claims 1. Method for controlling a machine tool (2), wherein the machine tool (2) is located in a safety area (1), the safety area (1) comprising an access area (3), wherein at least one imaging sensor (4) creates images (8) of the access area (3), wherein the imaging sensor (4) is connected to a computing unit (6) and the images (8) created by the sensor (4) are processed by the computing unit (6), wherein the computing unit (6) is configured to detect objects (7, 11) in the images (8), wherein the computing unit (6) is connected to the machine tool (2) and sends a stop signal to the machine tool (2) when the imaging sensor (4) detects that a first object (7) has entered the safety area (1), wherein the computing unit (6) sends a release signal to the machine tool (2) when the imaging sensor (4) detects thatthat a second object (11) has left the security area (1) and a comparison of the first object (7) with the second object (11) reveals that the first object (7) and the second object (7) are identical.

2. Method according to claim 1, wherein at least one property of the first object (7) and the second object (11) is classified by means of image recognition and it is checked on the basis of the properties whether the first object (7) and the second object (11) are identical.

3. Method according to claim 2, wherein an artificial intelligence, in particular a trained neural network, is used to classify the property.

4. Method according to claim 1 or 2, wherein the property is one of the following: 2022P00076WO October 27, 2025 9 - Object type - Color - Size.

5. Method according to one of the preceding claims, wherein image recognition, in particular with artificial intelligence, is used to recognize that an object (7, 11) is composed of several sub-objects, wherein the comparison of the first object (7) with the second object (11) is applied to the sub-objects, in particular the sub-objects of the first object (7) are compared with the sub-objects of the second object (11).

6. Method according to one of the preceding claims, wherein a sequence of images with a plurality of temporally successive images (8) is created by means of the imaging sensor (4) and a direction of movement of the first and / or the second object (7, 11) is determined from the sequence of images.

7. Method according to one of the preceding claims, wherein a plurality of first objects (7) have entered the safety area (1 ), wherein the computing unit (6) only sends a release signal to the machine tool (2) when it is detected by means of the imaging sensor (4) that all objects (7, 11 ) that have entered the safety area (1 ) have left the safety area (1 ).

8. Safety device for a safety area (1) of a machine tool (2), wherein the machine tool (2) is located in the safety area (1), wherein the safety area (1) comprises an access area (3), wherein at least one imaging sensor (4) is configured to produce images (8) of the access area (3), wherein the imaging sensor (4) is connected to a computing unit (6), 2022P00076WO October 27, 2025 10 wherein the safety device is provided and configured to carry out a method according to one of the preceding claims.

Images