Automobile instrument test method and test bench

Abstract

The invention discloses an automobile instrument test method and a test bench, relates to the technical field of automobile instrument test, and aims to improve the reliability of instrument test. The automobile instrument testing method comprises the steps of controlling each target alarm lamp on a to-be-tested instrument to be turned on or turned off according to a pre-configured test case. Wherein the test case comprises the first type and the first position of each target alarm lamp. And acquiring an image of the to-be-measured instrument. And inputting the to-be-tested instrument image into a pre-trained deep learning model to obtain a second type and a second position of each target alarm lamp. And when the second type of each target alarm lamp is the same as the first type and the second position is the same as the first position, determining that the to-be-tested instrument passes the test. And the deep learning model is insensitive to environmental factors, so that the requirement on the test environment is not high, and the test result is more reliable.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile instrument testing, in particular to an automobile instrument testing method and testing bench. BACKGROUND

[0002] With the deepening of the electric, intelligent, networked and shared trends of the automobile industry, the number of various types of warning lights integrated on automobile instruments gradually increases. The traditional automatic testing scheme has the problems of limited applicability and low reliability, which cannot keep up with the instrument software development rhythm of agile development and rapid iteration. SUMMARY

[0003] The embodiments of the present application provide an automobile instrument testing method and testing bench, aiming to improve the reliability of instrument testing.

[0004] The first aspect of the embodiments of the present application provides an automobile instrument testing method, which comprises: controlling each target warning light on a to-be-tested instrument to light up or turn off according to a pre-configured test case. The test case comprises a first type and a first position of each target warning light. An image of the to-be-tested instrument is acquired. The image of the to-be-tested instrument is input into a pre-trained deep learning model to acquire a second type and a second position of each target warning light. When the second type of each target warning light is the same as the first type, and the second position is the same as the first position, it is determined that the to-be-tested instrument passes the test.

[0005] In the embodiments, the type and position of the target warning light in the image of the to-be-tested instrument are identified by the deep learning model, and the type and position of the target warning light in the image of the to-be-tested instrument are compared with the type and position of the target warning light in the test case, so as to realize the automatic testing of the instrument warning light. Since the deep learning model is not sensitive to environmental factors, the requirement for the test environment is not high, so the test result is more reliable.

[0006] In one embodiment, the second type of each target warning light is determined based on the confidence that the target warning light belongs to various types of warning lights, and the second position of each target warning light is determined based on the position of the target element frame corresponding to the type to which the target warning light belongs.

[0007] In another embodiment, the confidence that the target warning light belongs to various types of warning lights is represented by a 1xN matrix, where N is the number of types of warning lights.

[0008] In another embodiment, the second type of the target warning light is the type of warning light corresponding to the maximum confidence value.

[0009] In another embodiment, the position of the target component frame corresponding to the type of the target alarm light is represented by a 2×2×N matrix, where N is the number of alarm light types and the 2×2 matrix represents the position of the component frame corresponding to each type of alarm light.

[0010] A second aspect of this application provides a test bench, which includes an instrument fixing mechanism, a bus signal trigger, a camera, and a control terminal. The instrument fixing mechanism is used to fix the instrument under test (DUT). The bus signal trigger is electrically connected to the DUT and the control terminal, and is used to respond to a trigger command from the control terminal to control the illumination or extinguishing of various target alarm lights on the DUT. The camera is electrically connected to the control terminal, and is used to respond to an image acquisition command from the control terminal to acquire an image of the DUT and transmit the image to the control terminal. The control terminal is used to generate trigger commands according to pre-configured test cases and transmit the trigger commands to the bus signal trigger; wherein, the test cases include a first type and a first position of each target alarm light; transmit an image acquisition command to the camera; input the DUT image from the camera into a pre-trained deep learning model to obtain a second type and a second position of each target alarm light; when the second type of each target alarm light is the same as the first type, and the second position is the same as the first position, the DUT is determined to have passed the test.

[0011] In this embodiment, the control terminal uses a deep learning model to identify the type and location of target alarm lights in the image of the instrument under test (DUT). The type and location of the target alarm lights predicted by the deep learning model are compared with the type and location of target alarm lights in the test cases. If the type and location of all target alarm lights predicted by the deep learning model match the type and location of all target alarm lights in the test cases, the DUT is determined to have passed the test; otherwise, the DUT is determined to have failed the test. Because the deep learning model is insensitive to environmental factors, the requirements for the testing environment are not high, making the test results more reliable. Furthermore, since no calibration of the test bench or fixture is required during testing, testing efficiency is improved.

[0012] A third aspect of this application provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the method provided in the first aspect.

[0013] A fourth aspect of this application provides a computer program product including computer instructions that implement the method provided in the first aspect when a processor executes the computer instructions.

[0014] It is understood that the beneficial effects of the computer-readable storage medium provided in the third aspect and the computer program product provided in the fourth aspect of the embodiments of this application are substantially the same as the beneficial effects of the method provided in the first aspect, and will not be repeated here. Attached Figure Description

[0015] Figure 1 FIG. 1 is a structural schematic diagram of a test bench according to an embodiment.

[0016] Figure 2 FIG. 2 is a flowchart of a method for testing an automobile instrument according to an embodiment.

[0017] Figure 3 FIG. 3 is a structural schematic diagram of a deep learning model according to an embodiment. DETAILED DESCRIPTION

[0018] It should be noted that the terms "first", "second", "third", "fourth" and the like in the specification, claims and drawings of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. The method disclosed in the embodiments of the present application or the method shown in the flowchart includes one or more steps for implementing the method, and the execution order of the steps can be interchanged with each other without departing from the scope of the claims, and some steps can also be deleted.

[0019] In the embodiments of the present application, the test bench is a device or system used for testing and evaluating products or systems in a controlled environment. The test bench can simulate various working conditions and environments, so as to realize performance testing, reliability testing, durability testing, safety testing, etc. of the object under test. In the automobile industry, the test bench is widely used in the research and development, production and quality detection of instruments, engines, transmissions, braking systems, suspension systems and other components, as well as performance testing and reliability verification of complete vehicles.

[0020] In the field of automobile instrument testing, the traditional test scheme mainly includes an automatic test scheme based on instrument screenshots and an automatic test scheme based on template matching algorithm.

[0021] The automatic test scheme based on instrument screenshots: first, the target alarm light on the instrument under test is triggered by automatically calling the test case script, then the instrument interface screenshot is exported through the internal interface of the instrument software, and finally the instrument interface screenshot is compared with the test case, so as to realize the automatic test of the instrument alarm light. The defect of this test scheme is that it is only suitable for the early stage of instrument software development. In the middle and later stages, mainstream manufacturers usually add some restrictions on safety functions on the instrument, resulting in incomplete instrument interface screenshots, which may lose some alarm information. Moreover, to prevent information leakage, some manufacturers may shield the internal interface of the instrument software, resulting in that this test scheme cannot be used.

[0022] An automated testing solution based on template matching eliminates the need to export screenshots of the instrument interface through the instrument software's internal interface. Instead, it acquires images of the instrument under test (DUT) by capturing images with a camera, and then uses a template matching algorithm to compare the DUT images with test cases, thereby automating the testing of instrument alarm lights. This testing solution is independent of the instrument software's internal interface and can be widely applied to all stages of instrument software development. However, due to the limitations of the template matching algorithm, it has high requirements for the testing environment. Each test requires calibration of the test bench or fixture to ensure consistency. Even slight changes in the testing environment can cause the template matching algorithm to malfunction, thus affecting the accuracy of the test results.

[0023] To improve the reliability of instrument panel testing, this application provides a method and test bench for testing automotive instrument panels. A deep learning model identifies the type and location of target warning lights in the image of the instrument panel under test. The method compares the type and location of the target warning lights in the image with the type and location of target warning lights in the test cases, thereby achieving automated testing of instrument panel warning lights. Since the deep learning model is insensitive to environmental factors, the requirements for the testing environment are low, making the test results more reliable. Furthermore, since no calibration of the test bench or fixtures is required during testing, testing efficiency is improved.

[0024] The test bench of the present application embodiment will be briefly described below.

[0025] like Figure 1 As shown, the test bench 100 includes an instrument fixing mechanism 110, a bus signal trigger 120, a camera 130, and a control terminal 140.

[0026] The instrument fixing mechanism 110 is used to fix the instrument under test. The instrument fixing mechanism 110 may include a clamp.

[0027] The bus signal trigger 120 is electrically connected to the instrument under test (DUT) and the control terminal 140. The bus signal trigger 120 responds to trigger commands from the control terminal 140, controlling the illumination or extinguishing of various target alarm lights on the DUT. The trigger command instructs the bus signal trigger 120 to activate each target alarm light. The trigger command may include a first type and a first position for each target alarm light.

[0028] It is understandable that the number of alarm lights on the target instrument is less than or equal to the total number of alarm lights on the instrument under test.

[0029] Camera 130 is electrically connected to control terminal 140. Camera 130 is used to respond to image acquisition commands from control terminal 140, acquire images of the instrument under test, and transmit the images of the instrument under test to control terminal 140. The image acquisition command instructs camera 130 to capture images of the instrument under test.

[0030] It is understandable that the camera 130's shooting range can cover the entire instrument under test to ensure the integrity of the image information.

[0031] The control terminal 140 is used to generate trigger commands according to pre-configured test cases, transmit trigger commands to the bus signal trigger 120, transmit image acquisition commands to the camera 130, and execute the automotive instrument testing method of this application embodiment. The test cases include a first type and a first position of each target warning light.

[0032] The control terminal 140 may include a processor 141 and a memory 142. When the processor 141 executes computer instructions stored in the memory 142, it implements the vehicle instrument testing method of the present application embodiment.

[0033] The automotive instrument testing method according to the embodiments of this application will be described in detail below.

[0034] like Figure 2 As shown, the automotive instrument testing method includes the following steps: S101. Control the lighting or extinguishing of each target alarm light on the instrument under test according to the pre-configured test cases, wherein the test cases include the first type and first position of each target alarm light.

[0035] In this embodiment, the control terminal 140 can execute a pre-configured test case script, generate trigger instructions by calling a bus simulation tool, and send the trigger instructions to the bus signal trigger 120 to trigger the various target alarm lights on the instrument under test.

[0036] As is understandable, bus simulation tools are primarily used to simulate and analyze communication behavior in in-vehicle networks, supporting various bus types such as CAN, CAN FD, LIN, FlexRay, and automotive Ethernet. Bus simulation tools include, but are not limited to, CanEasy, Vehicle Bus Analyzer (VBA), QEMU CAN device emulator, Hongke CanEasy, and TOSUN TC1038Pro.

[0037] S102. Obtain the image of the instrument under test.

[0038] In this embodiment, after the control terminal 140 finishes executing the test case script, the control terminal 140 can send an image acquisition command to the camera 130 to control the camera 130 to acquire an image of the instrument under test. After acquiring the image of the instrument under test, the camera 130 transmits the image of the instrument under test to the control terminal 140.

[0039] S103, input the to-be-tested instrument image into the pre-trained deep learning model to obtain a second type and a second position of each target alarm lamp.

[0040] In this embodiment, the deep learning model is trained based on multiple instrument sampling images, and the instrument sampling images include images in which various types of alarm lamps are in a lighting state and an extinguishing state respectively.

[0041] As shown in Figure 3 , the deep learning model includes a classification module and a labeling module, wherein the classification module is configured to predict the type of the target alarm lamp, and the labeling module is configured to label the position of the element box according to the type of the target alarm lamp.

[0042] The classification module can determine the type of the target alarm lamp based on the confidence that the target alarm lamp belongs to various types of alarm lamps. The confidence is a kind of probability value, which represents the degree of trust of the deep learning model for the classification result. The confidence that the target alarm lamp belongs to various types of alarm lamps can be represented by a 1xN matrix, where N is the number of types of alarm lamps. The classification module can determine the type of the target alarm lamp according to the maximum value of the confidence in the 1xN matrix. For example, the instrument includes five types of alarm lamps, and the 1x5 matrix output by the classification module is {0.7, 0.8, 0.9, 0.7, 0.8}, i.e., the confidence corresponding to the first type of alarm lamp to the fifth type of alarm lamp is 0.7, 0.8, 0.9, 0.7, 0.8, and since the confidence corresponding to the third type of alarm lamp is the maximum value, the classification module determines that the type of the target alarm lamp is the third type.

[0043] The labeling module can determine the position of the target alarm lamp based on the position of the target element box corresponding to the type of the target alarm lamp. The element box is a rectangular box drawn based on the boundary of the alarm lamp region in the instrument image, and the position of the element box includes the vertex coordinates of the element box. The position of the target element box corresponding to the type of the target alarm lamp can be represented by a 2x2xN matrix, where N is the number of types of alarm lamps, 2x2 matrix is the position of the element box corresponding to various types of alarm lamps, and 2x2 matrix includes two diagonal vertex coordinates of the element box. For example, the instrument includes five types of alarm lamps, and the 2x2x5 matrix output by the labeling module is shown in Matrix (1), i.e., the position of the element box corresponding to the first type of alarm lamp to the fifth type of alarm lamp is shown in Matrix (2) to Matrix (6), and when the type of the target alarm lamp is the third type, the labeling module determines the position of the target alarm lamp as shown in Matrix (4).

[0044] wherein (X 11 ,Y 11 ) and (X 12 ,Y 12) are respectively the left upper corner vertex coordinates and the right lower corner vertex coordinates of the element frame corresponding to the first type of alarm lamp, 21 21 ) are respectively the left upper corner vertex coordinates and the right lower corner vertex coordinates of the element frame corresponding to the second type of alarm lamp, 22 22 ) are respectively the left upper corner vertex coordinates and the right lower corner vertex coordinates of the element frame corresponding to the third type of alarm lamp, 31 31 ) are respectively the left upper corner vertex coordinates and the right lower corner vertex coordinates of the element frame corresponding to the fourth type of alarm lamp, 32 32 ) are respectively the left upper corner vertex coordinates and the right lower corner vertex coordinates of the element frame corresponding to the fifth type of alarm lamp. 41 41 42 42 51 51 52 52

[0045] It can be understood that the deep learning model includes Convolutional Neural Networks (CNNs), which can include, but are not limited to, LeNet, AlexNet, VGGNet, ResNet, Inception (GoogleNet), YOLO.

[0046] S104, determining whether the second type of each target alarm lamp is the same as the first type.

[0047] If yes, step S105 is executed; if no, step S107 is executed.

[0048] In this embodiment, when the second type of the target alarm lamp is the same as the first type, it indicates that the type of the target alarm lamp predicted by the deep learning model is consistent with the test case.

[0049] S105, determining whether the second position of each target alarm lamp is the same as the first position.

[0050] If yes, step S106 is executed; if no, step S107 is executed.

[0051] In this embodiment, when the second position of the target alarm lamp is the same as the first position, it indicates that the position of the target alarm lamp predicted by the deep learning model is consistent with the test case.

[0052] S106, determining that the instrument under test passes the test. ​​​​​​​​​​​​

[0053] S107, determining that the to-be-tested instrument fails the test.

[0054] In this embodiment, the control terminal 140 identifies the types and positions of the target warning lights in the to-be-tested instrument image through the deep learning model, compares the types and positions of the target warning lights predicted by the deep learning model with the types and positions of the target warning lights in the test case, and determines that the to-be-tested instrument passes the test when the types and positions of all the target warning lights predicted by the deep learning model are consistent with the types and positions of all the target warning lights in the test case respectively, or determines that the to-be-tested instrument fails the test otherwise. Since the deep learning model is not sensitive to environmental factors, the requirement for the test environment is not high, so that the test result is more reliable. Moreover, since the test does not need to calibrate the test bench or the fixture, the test efficiency is improved.

[0055] It can be understood that the automobile instrument test method in the embodiments of the present application can also be applied to a chip, a computer readable storage medium or a computer program product.

[0056] The chip includes a processor and an interface circuit, and the processor is electrically connected with the interface circuit. The interface circuit can read computer instructions stored in a storage and send the computer instructions to the processor. When the processor executes the computer instructions, the automobile instrument test method in the embodiments of the present application is implemented.

[0057] The computer readable storage medium stores computer instructions, and when the processor executes the computer instructions, the automobile instrument test method in the embodiments of the present application is implemented.

[0058] The computer readable storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer readable instructions, data structures, program modules or other data. The computer readable storage medium includes, but is not limited to, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device, or any other medium that can be used to store desired information and can be accessed by a computer.

[0059] The computer program product includes computer instructions, and when the processor executes the computer instructions, the automobile instrument test method in the embodiments of the present application is implemented.

[0060] The above detailed description of the application has been made in conjunction with the accompanying drawings, but the application is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the purpose of the application.

Claims

1. A method of testing an automobile instrument, characterized by, The method comprises: controlling each target alarm light on the instrument under test to light up or turn off according to a pre-configured test case; wherein the test case comprises a first type and a first position of each target alarm light; obtaining an image of the instrument under test; inputting the image of the instrument under test into a pre-trained deep learning model to obtain a second type and a second position of each target alarm light; determining that the instrument under test passes the test when the second type of each target alarm light is the same as the first type and the second position is the same as the first position.

2. The method of claim 1, wherein, The second type of each target alarm light is determined based on a confidence level of the target alarm light belonging to various types of alarm lights, and the second position of each target alarm light is determined based on a position of a target element frame corresponding to the type of the target alarm light.

3. The method of claim 2, wherein, The confidence level of the target alarm light belonging to various types of alarm lights is represented by a 1xN matrix, where N is the number of types of alarm lights.

4. The method of claim 2 or 3, wherein, The second type of the target alarm light is the type of alarm light corresponding to the maximum confidence level.

5. The method of claim 2, wherein, The position of the target element frame corresponding to the type of the target alarm light is represented by a 2x2xN matrix, where N is the number of types of alarm lights, and the 2x2 matrix is the position of the element frame corresponding to various types of alarm lights.

6. A test bench, characterized in that It comprises an instrument fixing mechanism, a bus signal trigger, a camera and a control terminal. The instrument fixing mechanism is used to fix the instrument under test. The bus signal trigger is electrically connected to the instrument under test and the control terminal, and is used to control each target alarm light on the instrument under test to light up or turn off in response to a trigger instruction from the control terminal. The camera is electrically connected to the control terminal, and is used to collect an image of the instrument under test in response to an image collection instruction from the control terminal, and transmit the image of the instrument under test to the control terminal. The control terminal is used to generate the trigger instruction according to a pre-configured test case, transmit the trigger instruction to the bus signal trigger; wherein the test case comprises a first type and a first position of each target alarm light; transmit the image collection instruction to the camera; input the image of the instrument under test from the camera into a pre-trained deep learning model to obtain a second type and a second position of each target alarm light; and determine that the instrument under test passes the test when the second type of each target alarm light is the same as the first type and the second position is the same as the first position.

7. The test bed of claim 6, wherein, The second type of each target alarm light is determined based on a confidence level of the target alarm light belonging to various types of alarm lights, and the second position of each target alarm light is determined based on a position of a target element frame corresponding to the type of the target alarm light.

8. The test bed of claim 7, wherein, The confidence level of the target alarm light belonging to various types of alarm lights is represented by a 1xN matrix, where N is the number of types of alarm lights.

9. Test bench according to claim 7 or 8, characterized in that The second type of the target alarm light is the type of alarm light corresponding to the maximum confidence level. The position of the target element frame corresponding to the type of the target alarm light is represented by a 2x2xN matrix, where N is the number of types of alarm lights, and the 2x2 matrix is the position of the element frame corresponding to various types of alarm lights.

10. The test bed of claim 7, wherein, The position of the target element frame corresponding to the type of the target alarm lamp is represented by a 2x2xN matrix, where N is the number of types of alarm lamps, and the 2x2 matrix is the position of the element frame corresponding to various types of alarm lamps.

Images