Tool state learning device, tool state estimation device, control device, tool state learning method, and tool state estimation method
A tool state learning device has a storage unit for storing an arbitrary image captured by an imaging device imaging a machined surface of an arbitrary workpiece cut using an arbitrary tool, and a teacher data acquisition unit for acquiring, as input data, an arbitrary image stored in the storage unit, and acquiring, as a label, the state of the tool annotated according to prescribed levels indicating the degree of wear of the tool on the basis of the arbitrary image. The tool state learning device also has a tool state learning unit for using the acquired label and input data to perform supervised learning, and generating a learned model in which a machined surface image of the machined surface of a workpiece imaged by the imaging device is inputted and the state of the tool that cut the machined surface of the workpiece is outputted.
1 . A tool state learning device comprising:
a memory configured to store a program and an arbitrary image of a machined surface of an arbitrary workpiece captured by an imaging device, the machined surface having been cut with an arbitrary tool; and
a processor configured to execute the program stored on the memory to cause the tool state learning device to:
acquire, as input data, the arbitrary image stored in the memory and acquire, as a label, a state of the tool, the state being annotated in accordance with predetermined grades each indicating a degree of tool wear based on the arbitrary image;
perform matching processing by using at least CAD data and the arbitrary image, specify the image region of the machined surface, mask, in the arbitrary image, an image region other than the specified image region of the machined surface, and eliminate the effect of the reflection from the image region other than the specified image region of the machined surface;
acquire the arbitrary image having the image region masked, as input data; and
perform supervised learning using the input data and the label that have been acquired, and generate a trained model configured to input a machined surface image of a machined surface of a workpiece captured by the imaging device, and to output a state of a tool with which the machined surface of the workpiece has been cut.
2 . The tool state learning device according to claim 1 ,
wherein the processor displays the arbitrary image to allow a state of the tool to be annotated in accordance with the predetermined grades on a display.
3 . The tool state learning device according to claim 1 ,
wherein the processor updates the trained model by performing the supervised learning using, as input data, a machined surface image newly captured by the imaging device and using, as a label, a state of the tool outputted from the trained model with respect to the newly captured machined surface image.
4 . The tool state learning device according to claim 1 ,
wherein the processor acquires, as input data, tool information including at least one of a material, a shape, or a pre-machining state of the tool, together with the arbitrary image, and
wherein the processor generates a trained model configured to input the machined surface image captured by the imaging device and tool information regarding the tool with which the machined surface shown in the machined surface image has been cut, and to output a state of the tool.
5 . A tool state estimation device comprising:
a memory configured to store a program and a trained model generated by the tool state learning device according to claim 1 , the trained model being configured to input a machined surface image of a machined surface of a workpiece captured by an imaging device and to output a state of a tool with which the machined surface of the workpiece has been cut; and
a processor configured to execute the program stored on the memory to cause the tool state estimation device to:
input a machined surface image of a machined surface captured by the imaging device;
input the inputted machined surface image to the trained model and thereby estimate degree of tool wear as a state of a tool with which the machined surface shown in the image has been cut; and
determine whether or not time for replacement of the tool has been reached, based on the estimated degree of the tool wear,
wherein the processor determines that the time for replacement of the tool has been reached, when the estimated degree of tool wear is higher than a predetermined threshold value.
6 . The tool state estimation device according to claim 5 ,
wherein the trained model is provided in a server that is accessible from the tool state estimation device via a network.
7 . The tool state estimation device according to claim 5 , further comprising:
a tool state learning device comprising:
a memory configured to store a program and an arbitrary image of a machined surface of an arbitrary workpiece captured by an imaging device, the machined surface having been cut with an arbitrary tool; and
a processor configured to execute the program stored on the memory to cause the tool state learning device to:
acquire, as input data, the arbitrary image stored in the memory and acquire, as a label, a state of the tool, the state being annotated in accordance with predetermined grades each indicating a degree of tool wear based on the arbitrary image;
perform matching processing by using at least CAD data and the arbitrary image, specify the image region of the machined surface, mask, in the arbitrary image, an image region other than the specified image region of the machined surface, and eliminate the effect of the reflection from the image region other than the specified image region of the machined surface;
acquire the arbitrary image having the image region masked, as input data; and
perform supervised learning using the input data and the label that have been acquired, and generate a trained model configured to input a machined surface image of a machined surface of a workpiece captured by the imaging device, and to output a state of a tool with which the machined surface of the workpiece has been cut.
8 . A control device comprising:
the tool state estimation device according to claim 5 .
9 . A tool state estimation method comprising:
an input step including inputting a machined surface image of a machined surface captured by an imaging device;
an estimation step including inputting the machined surface image that has been inputted in the input step to a trained model generated by the tool state learning device according to claim 1 , the trained model being configured to input a machined surface image of a machined surface of a workpiece captured by the imaging device and to output a state of a tool with which the machined surface of the workpiece has been cut, and estimating degree of tool wear as a state of the tool with which the machined surface shown in the machine surface image has been cut; and
a determination step including determining whether or not time for replacement of the tool has been reached, based on the estimated degree of the tool wear,
wherein the determination step determines that the time for replacement of the tool has been reached, when the estimated degree of tool wear is higher than a predetermined threshold value.
10 . A non-transitory computer readable medium configured to store a program for causing a processor to control a tool state learning device to execute a tool state learning method, the tool state learning method comprising:
acquiring, as input data, an arbitrary image of a machined surface of an arbitrary workpiece captured by an imaging device, the machined surface having been cut with an arbitrary tool, and acquiring, as a label, a state of the tool, the state being annotated in accordance with predetermined grades each indicating a degree of tool wear based on the arbitrary image;
performing matching processing by using at least CAD data and the arbitrary image, specifying the image region of the machined surface, masking, in the arbitrary image, an image region other than the specified image region of the machined surface, and eliminating the effect of the reflection from the image region other than the specified image region of the machined surface;
acquiring the arbitrary image having the image region masked, as input data; and
performing supervised learning using the input data and the label that have been acquired, and generating a trained model configured to input a machined surface image of a machined surface of a workpiece captured by the imaging device, and to output a state of a tool with which the machined surface of the workpiece has been cut.