Method and device for collision avoidance during workpiece machining by a multi-axis machining center

Abstract

Method (26) for machining a workpiece (14) with a device (10), wherein the device (10) comprises a processing machine (12) which is designed for three-dimensional machining of the workpiece (14) and has a tool head (22) with a tool which is movable about at least a first axis (16) and a second axis (18), wherein the method (26) is carried out by the device (10) and comprises the following method steps: B) Acquiring input data (42) on the contour of the workpiece (14), the contour of the tool head (22), the distance between the tool head (22) and the workpiece (14), the first axis (16) and the second axis (18); C) processing the input data (42) into feature data (46); D) processing the feature data (46) in a machine learning algorithm (28) of the device (10); E) Outputting a prediction from the machine learning algorithm (28) regarding the collision of the tool head (22) with the workpiece (14) and / or another part of the processing machine (12).

Description

Background of the invention

[0001] The invention relates to a method and a device for machining a workpiece.

[0002] During the three-dimensional processing of a workpiece, in particular during laser cutting or laser welding, a part of a processing machine, in particular a tool head of the processing machine, may collide with a workpiece or another part of the processing machine.

[0003] It is known to avoid such collisions by manually checking a machining path. The applicant is also aware of using a brute-force algorithm to avoid collisions during workpiece machining. However, both methods are complex and require very long computing times. Object of the invention

[0004] It is therefore an object of the invention to provide a method and a device which enable collision avoidance in an effective and reliable manner. Description of the invention

[0005] This object is achieved according to the invention by a method according to patent claim 1 and a device according to patent claim 11. The dependent patent claims represent preferred developments.

[0006] The object of the invention is thus achieved by a method for machining a workpiece with a device, wherein the device comprises a processing machine which is designed for 3D machining of the workpiece and has a tool head, wherein the tool head is movable about a first axis and a second axis, wherein the method is carried out by the device and comprises the following method steps: B) Collecting input data on the geometry of the tool head, the workpiece and the distance between the tool head and the workpiece as well as on the first axis and the second axis; C) Processing the input data into feature data; D) processing the feature data in a machine learning algorithm of the device; E) Outputting a prediction from the machine learning algorithm regarding the collision of the tool head with the workpiece and / or another part of the processing machine.

[0007] The method according to the invention enables reliable and inexpensive collision avoidance between the tool head and the workpiece or another part of the processing machine during the processing of the workpiece along a processing path.

[0008] The first axis and the second axis preferably run through the machining point of the tool on the workpiece. The first axis can be a B-axis of the machining center. The second axis can be a C-axis of the machining center.

[0009] The distance between the tool head and the workpiece can be determined from a Z-buffer image.

[0010] The device may include a computer with the machine learning algorithm. The computer may be part of the processing machine.

[0011] The processing machine is preferably designed as a 5- or 6-axis machine. Preferably, three axes are provided for moving the tool to a machining point and two to three axes are provided for adjusting the working position of the tool head.

[0012] The input data is preferably available as a geometric image (black and white, grayscale, RGB) and / or a Z-buffer image (distance image). The feature data can be in the form of a feature vector.

[0013] The output in process step E) can be in the form “collided”, “collides soon” or “does not collide”.

[0014] Before method step B), the machine learning algorithm can be trained in method step A) by inputting verified input data from collisions and / or non-collision events. The verified input data can originate from verified predictions obtained in method step E). Alternatively, the input data can originate from a simulation. The training of the machine learning algorithm represents an independent aspect of the invention.

[0015] The input data in process step B) can be taken at least partially from CAD data or NC data of the processing machine. For example, the CAD data can be converted into simulated sensor data, in particular simulated camera data, in a simulation.

[0016] Alternatively or additionally, the input data may be taken at least partially from sensor data recorded by a sensor.

[0017] The use of sensor data enables collision avoidance with significantly reduced computing power required to create or process the input data. Furthermore, the use of sensor data also makes it possible to react to unforeseen situations, for example, when a holder—particularly unknown to the machining center—is used to clamp the workpiece.

[0018] The sensor is preferably located or formed on the tool head. This means that the sensor data requires little or no further processing.

[0019] In a particularly preferred embodiment of the invention, the sensor is designed in the form of a camera. The camera can be a time-of-flight (TOF) camera or a contour depth (RGB-D) camera.

[0020] In process step E), collision-free input data (values) of the first axis and the second axis can be output by processing the input data from process step B).

[0021] Alternatively, if a collision is predicted in process step E), the following process steps can be carried out after process step E): F) Variation of the input data of the first axis and / or the second axis; G) Carrying out process steps B) to E); H) Carrying out method steps F) and G) until no collision is predicted in method step D) or a predetermined number of runs of method steps F) and G) is reached, wherein if no collision occurs, the input data of the first axis and second axis used in method step B) are output.

[0022] In a particularly preferred embodiment of the invention, the machine learning algorithm is implemented in the form of a neural network. The neural network may have an input layer, several hidden layers, and an output layer. The feature data may be fed to the input layer of the neural network.

[0023] The method according to the invention is particularly advantageous when the processing machine is designed in the form of a cutting machine and / or a laser processing machine. The laser beam of a laser tool of the laser processing machine preferably has the processing point of the tool (here, the laser beam) at its focal point.

[0024] The object of the invention is further achieved by the device described here for carrying out the method described here.

[0025] 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 combinations according to the invention. The embodiments shown and described are not intended to be exhaustive, but rather are exemplary in nature for describing the invention. Detailed description of the invention and drawing Fig. 1 shows a processing machine with a tool head for machining a workpiece. Fig. 2 shows the processing machine from Fig. 1, where the tool head collides with the workpiece when machining a specific point on the workpiece. Fig. 3 shows the processing machine from Fig. 2, where the tool head is positioned when machining the same location as in Fig. 2 does not collide with the workpiece. Fig. 4 shows a method for detecting and avoiding a collision by a machine learning algorithm and for training the machine learning algorithm. Fig. Figure 5 shows an alternative method for detecting and avoiding a collision using a sensor. Fig. Figure 6a shows a close-up of a collision of a tool head with a workpiece. Fig. 6b shows an edge image of the situation according to Fig. 6a. Fig. 6c shows a filtered Z-buffer image of the situation according to Fig. 6b.

[0026] Fig. Figure 1 shows a device 10 with a processing machine 12 for processing a workpiece 14. The processing machine 12 is multi-axis. The processing machine 12 has a first axis 16 and a second axis 18. Three additional axes are provided for moving to a processing point 20.

[0027] Fig. 2 also shows the device 10. From Fig. 2 shows that the processing machine 12 has a tool head 22 that collides with the workpiece 14 in a collision area 24. The aim of the present invention is to avoid such collisions in a particularly efficient manner.

[0028] Fig. 3 shows the device 10 with the processing machine 12 and the workpiece 14. From a synopsis of the Fig. 2 and Fig. 3 it is evident that in Fig. 3 the same machining of the workpiece 14 is carried out as in Fig. 2 shown collision by pivoting the axes 16 and 18 (see Fig. 1) is however avoided.

[0029] Fig. 4 shows a method 26 for detecting and avoiding a collision with a machine learning algorithm 28. In this case, CAD data of the workpiece 14 (see Fig. 3) are converted into simulated sensor data. Depending on the simulation of the workpiece processing, a collision 32 or a resolved collision 34 can occur. For these scenarios, a Z-buffer image (distance image) 36a, 36b, a geometry image 38a, 38b and the associated values ​​40a, 40b of the axes 16, 18 (see Fig. 1). This input data 42 (see also Fig. 6a-c) are processed in an extraction unit 44 into feature data 46, here in the form of a feature vector 48.

[0030] The feature vector 48 is input into the machine learning algorithm 28, here in the form of a neural network. In particular, the feature vector 48 is passed to an input layer 50 and processed by hidden layers 52. An output layer 54 outputs collision-free values ​​40c of the axes 16, 18 (see Fig. 1).

[0031] To train the machine learning algorithm 28, these values ​​40c can be verified and fed back as feedback 55.

[0032] Fig. 5 shows a method 26 in which the input data 42 is generated at least partially by a sensor 56, here in the form of a camera. The sensor 56 can be arranged on the tool head 22 of the processing machine 12. Particularly preferably, the sensor 56 is arranged in the same position in which a virtual sensor was located when data for training the machine learning algorithm 28 was generated by a simulation.

[0033] The device 10 comprises a control 58, which firstly determines the values ​​40a of the first axis 16 or second axis 18 (see Fig. 1). From the input data 42, the feature data 46 or the feature vector 48 are generated in the extraction unit 44 and transferred to the machine learning algorithm 28, which outputs collision-free values ​​40c. These collision-free values ​​40c can be transferred to the controller 58 to correct the machining path of the workpiece 14 if necessary.

[0034] Fig. 6a shows a close-up of a geometry image 38c with a collision area 24.

[0035] Fig. 6b shows the geometry image 38c from Fig. 6a as edge image 60.

[0036] Fig. 6c shows the edge image 60 according to Fig. 6b as filtered Z-buffer image 36c.

[0037] The representations 38c, 60 and / or 36c can be used to determine a collision, with the Z-buffer image 36 being particularly well suited for processing by the machine learning algorithm 28 (see Fig. 4 and Fig. 5) is suitable.

[0038] Taking a synopsis of all the figures in the drawing, the invention relates in summary to a method 26 and a device 10 for avoiding collisions between a part of a processing machine 12 and a workpiece 14 being machined by the processing machine 12. For this purpose, distances between at least a part of the processing machine 12 and the workpiece 14 at the values ​​40a, b of at least a first axis 16 and a second axis 18 of the processing machine 12 during workpiece processing are simulated and / or measured by a sensor 56. These input data 42 are preferably processed into a feature vector 48 and fed to a trained machine learning algorithm 28. The machine learning algorithm 28 determines collision-free values ​​40c of the axes 16, 18 as a function of the feature vector 48.After their verification, these values ​​40c can be fed back to the machine learning algorithm 28 as feedback 54, either directly or indirectly, for training the machine learning algorithm 28. List of reference symbols 10 Device 12 processing machines 14 Workpiece 16 first axis 18 second axle 20 processing points 22 Tool head 24 Collision zone 26 procedures 28 Machine Learning Algorithm 30 Simulator 32 Collision 34 resolved collisions 36a-c Z-buffer image 38a-c Geometry image 40a-c Values ​​of the first axis 16 and second axis 18 42 input data 44 Extraction unit 46 feature data 48 feature vector 50 input layers 52 hidden layers 54 output layers 55 Feedback 56 Sensor 58 Control 60 edge image

Claims

[1] Method (26) for machining a workpiece (14) with a device (10), wherein the device (10) comprises a processing machine (12) which is designed for three-dimensional machining of the workpiece (14) and has a tool head (22) with a tool which is movable about at least a first axis (16) and a second axis (18), wherein the method (26) is carried out by the device (10) and comprises the following method steps: B) Acquiring input data (42) on the contour of the workpiece (14), the contour of the tool head (22), the distance between the tool head (22) and the workpiece (14), the first axis (16) and the second axis (18); C) processing the input data (42) into feature data (46); D) processing the feature data (46) in a machine learning algorithm (28) of the device (10); E) Outputting a prediction from the machine learning algorithm (28) regarding the collision of the tool head (22) with the workpiece (14) and / or another part of the processing machine (12). [2] Method (26) according to claim 1, comprising the following method step carried out before method step B): A) Training the machine learning algorithm (28) by inputting verified input data (42) of collisions and / or non-collision of the tool head (22). [3] Method (26) according to one of the preceding claims, in which the input data (42) in method step B) are at least partially taken from CAD data. [4] Method (26) according to one of the preceding claims, in which the input data (42) in method step B) are at least partially taken from sensor data of a sensor (56). [5] The method (26) of claim 4, wherein the sensor (56) is installed on the tool head (22). [6] Method (26) according to claim 4 or 5, wherein the sensor (56) is in the form of a camera. [7] Method (26) according to one of claims 1 to 6, wherein, in the case of a collision predicted in method step E), collision-free input data (42) of the first axis (16) and second axis (18) are further output in method step E). [8] Method (26) according to one of claims 1 to 6, wherein, if a collision is predicted in method step E), the following method steps are carried out after method step E): F) varying the values ​​(40a-c) of the first axis (16) and / or the second axis (18) in the input data (42); G) Carrying out process steps B) to E); H) Executing the method steps F) and G) until no collision is predicted in the method step E) or a predefined number of runs of the method steps F) and G) is reached, wherein if there is no collision, the values ​​(40a-c) of the input data (42) of the first axis (16) and second axis (18) used in the method step B) are output. [9] Method (26) according to one of the preceding claims, in which the machine learning algorithm (28) is designed in the form of a neural network. [10] Method (26) according to one of the preceding claims, wherein the processing machine (12) is designed in the form of a cutting machine and / or in the form of a laser processing machine. [11] Device (10) for carrying out a method (26) according to one of the preceding claims.

Images