Method for managing image data and motor vehicle lighting arrangement

Abstract

The present invention provides a method for manufacturing a motor vehicle lighting arrangement, the method comprising the steps of: training a depth autoencoder (1) to process image data (6); installing an encoder block (2) in a motor vehicle control unit (4) of a motor vehicle (10); and installing a decoder block (3) in a motor vehicle lighting module (5) of the motor vehicle (10). The present invention also provides a motor vehicle lighting arrangement for performing the steps of this method and a method for operating such a motor vehicle lighting arrangement.

Description

[0001] This application is a divisional application of a patent application filed on October 28, 2021, with national application number 202180066018.X and invention title "Method for Managing Image Data and Motor Vehicle Lighting Arrangement". Technical Field

[0002] This invention relates to the field of motor vehicle lighting devices, and more specifically, to the management of electronic data derived from the control of lighting sources. Background Technology

[0003] Current lighting fixtures include an ever-increasing number of light sources that must be controlled to provide adaptive lighting functionality.

[0004] Such a large number of light sources involves a significant amount of data, which must be managed by the control unit. The CAN protocol is frequently used, and in some variations (CAN-FD being one of the most common), to transmit data between the PCM and the optical module. However, some automakers have decided to limit the bandwidth of the CAN protocol, which impacts management operations that typically require approximately 5 Mbps.

[0005] Current compression methods are not very effective for light patterns, and this affects the bandwidth reduction required by automakers. Higher compression rates always involve data loss that may be unacceptable under motor vehicle regulations. Summary of the Invention

[0006] Seeking solutions to this problem.

[0007] The present invention provides a solution to these problems by means of a method for manufacturing a motor vehicle lighting arrangement, a motor vehicle lighting arrangement, and a method for operating the motor vehicle lighting arrangement.

[0008] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall be interpreted in accordance with the conventions of the field. It should also be understood that common terms should be interpreted as those used in the relevant field, rather than as idealized or overly formalized meanings, unless expressly stated otherwise herein.

[0009] In this document, the term “comprising” and its derivatives (such as “including”, etc.) should not be interpreted in an exclusive sense, that is, these terms should not be construed as excluding the possibility that the described and defined content may include other elements, steps, etc.

[0010] In a first aspect, the present invention provides a method for manufacturing a motor vehicle lighting arrangement, the method comprising the following steps:

[0011] - Train a deep autoencoder to process image data, the autoencoder comprising at least one encoder block, at least one decoder block and a loss function unit;

[0012] - The encoder block is installed in the vehicle control unit of the motor vehicle; and

[0013] - Install the decoder block in the vehicle's lighting module.

[0014] This method is used to manage image data exchanged between the control unit and the optical module. Instead of using methods that linearize values, a convolutional neural network (e.g., a depth autoencoder) is used to generate compressed data in the encoder block based on the image data. The same autoencoder is then able to recover the original version of the image data in the decoder block.

[0015] The main advantage of this approach is its ability to flexibly limit data loss, thereby improving a given compression ratio. Training can reduce data loss and can limit the compression ratio for the neural network.

[0016] In some specific embodiments, the encoder block includes a convolutional layer, a corrected linear unit layer, and a normalization layer.

[0017] Having multiple layers helps reduce the image size of the training data used in the learning function and improves the compression ratio.

[0018] In some specific embodiments, the method further includes the step of selecting a ratio between the size of the convolutional layer of the encoder block and the size of the normalized layer of the encoder block.

[0019] This choice is directly related to the compression ratio and also affects the processing speed.

[0020] In some specific embodiments, the decoder block includes a no-sampling convolutional layer, a corrected linear unit layer, and a normalization layer.

[0021] The presence of multiple layers of decoder blocks helps to increase the amount of data compressed by the encoder in order to find the decompressed image needed for learning functions and improve the compression ratio.

[0022] In a second aspect of the invention, the present invention provides a motor vehicle lighting arrangement manufactured by a method according to a first aspect of the invention, the motor vehicle lighting arrangement comprising:

[0023] - A vehicle control unit including an encoder block configured to process image data; and

[0024] - An illumination module including a decoder block configured to receive processed image data to recover the original version of the image data, wherein the illumination module is further configured to project a light pattern based on the image data recovered by the decoder block;

[0025] The encoder block and the decoder block, as part of the same depth autoencoder, have undergone a common training process.

[0026] This vehicle lighting arrangement can be installed in a motor vehicle for better operation of the lighting process. Since the encoder and decoder blocks have been trained as part of the same depth autoencoder, they can work together perfectly to obtain an accurate copy of the original image data.

[0027] In some specific embodiments, the lighting module includes a solid-state light source, such as an LED.

[0028] The term "solid-state" refers to light emitted by solid-state electroluminescent materials that use semiconductors to convert electricity into light. Compared to incandescent lighting, solid-state lighting produces visible light with reduced heat generation and lower energy consumption. Solid-state electronic lighting arrangements, which are typically less heavy than fragile glass tubes / bulbs and thin filaments, offer greater resistance to shock and vibration. They also eliminate filament evaporation, potentially increasing the lifetime of the lighting arrangement. Some examples of these types of lighting include semiconductor light-emitting diodes (LEDs), organic light-emitting diodes (OLEDs), or polymer light-emitting diodes (PLEDs) as the light source, rather than filaments, plasma, or gas.

[0029] In a third aspect of the invention, the present invention provides a method for operating a motor vehicle lighting arrangement according to the foregoing aspects of the invention, the method comprising the following steps:

[0030] - Manipulate the encoder block to reduce the data size of some image data, thereby producing processed image data;

[0031] - Transfer the processed image data to the decoder block;

[0032] - Operate the decoder block to process the processed image data to produce restored image data; and

[0033] - Operate the lighting module to project light patterns based on the recovered image data.

[0034] This method allows vehicle-mounted systems to operate with a lower communication bandwidth than conventional communication methods. The processed image data generated as the encoder's output represents a compressed image. The decoder then receives this compressed data as input to reconstruct an image that is as close as possible to the original image data.

[0035] In some specific embodiments, the method includes a step of normalizing the image data before operating the encoder block to reduce the data size of the image data. Specifically, the step of normalizing the image data includes converting each value of the image data into a transformed value ranging from 0 to 1.

[0036] Because these normalized values ​​are optimal for the operation of the autoencoder, this normalization is used to improve the operation of the autoencoder.

[0037] In some specific embodiments, the method includes the step of dividing the image data into data subarrays of the same format before manipulating the encoder block to reduce the data size of the image data.

[0038] Each type of lighting pattern (near beam, far beam, etc.) operates with specific characteristics. Isolating these characteristics in different subarrays and processing them independently will improve the compression and operation of the arrangement. Attached Figure Description

[0039] To complete the description and to provide a better understanding of the invention, a set of accompanying drawings is provided. These drawings form part of the specification and illustrate embodiments of the invention; they should not be construed as limiting the scope of the invention, but merely as examples of how the invention can be practiced. The drawings include the following figures:

[0040] [ Figure 1 The first step of a method for manufacturing a motor vehicle lighting arrangement according to the present invention is shown.

[0041] [ Figure 2 The image shows a motor vehicle in which two lighting arrangements are installed.

[0042] [ Figure 3 ] Provides detailed images of the operation of this lighting arrangement.

[0043] The following reference numerals are used in these figures:

[0044] 1 Depth Auto Encoder

[0045] 2 encoder blocks

[0046] 21-encoder block convolutional layer

[0047] Correction linear unit layer of 22 encoder blocks

[0048] 23. Normalization layer of encoder block

[0049] 3 decoder blocks

[0050] 31 decoder blocks of unsampling convolutional layers

[0051] Correction linear unit layer of 32 decoder blocks

[0052] Normalization layer of 33 decoder blocks

[0053] 4. Vehicle's central control unit

[0054] 5 lighting fixtures

[0055] 6 Image Patterns

[0056] 10 motor vehicles

[0057] The first part (flat part) of the image pattern 61

[0058] The second part (bend) of image pattern 62.

[0059] 71 First Vector

[0060] 72 Second Vector

[0061] 81 First restored image

[0062] 82 Second restored image Detailed Implementation

[0063] Example embodiments are described in sufficient detail to enable those skilled in the art to embodied and implement the systems and processes described herein. It is important to understand that embodiments may be provided in many alternative forms and should not be construed as limited to the examples set forth herein.

[0064] Accordingly, while embodiments may be modified in various ways and take various alternative forms, specific embodiments thereof are shown in the accompanying drawings and described in detail below as examples. It is not intended to limit one to the particular forms disclosed. Rather, all modifications, equivalents, and alternatives falling within the scope of the appended claims are to be included.

[0065] Figure 1 The first step of a method for manufacturing a motor vehicle lighting arrangement according to the present invention is shown.

[0066] In the first step, a deep autoencoder 1 is trained to process image data. The autoencoder 1 includes an encoder block 2, a decoder block 3, and a loss function unit for minimizing the error generated by the autoencoder 1.

[0067] The training can be performed using real light patterns provided by the vehicle manufacturer so that compression can be optimized and the autoencoder can provide the minimum possible data loss for a given compression ratio.

[0068] Encoder block 2 includes a convolutional layer 21, a calibrated linear unit layer 22, and a normalization layer 23, while decoder block 3 includes an unsampled convolutional layer 31, a calibrated linear unit layer 32, and a normalization layer 33. The presence of multiple layers helps reduce the amount of training data required for learning functions and improves the compression ratio.

[0069] The compression ratio is given by the ratio between the size of the convolutional layer 21 and the size of the normalization layer 23 in encoder block 2, which will provide the size of the data transmitted to decoder block 3. This is relevant because once the autoencoder 1 is trained, decoder block 3 will be separate from encoder block 2, and each element will be installed in a different part of the motor vehicle.

[0070] Figure 2 A motor vehicle 10 is shown, in which these two components of an illumination arrangement are installed. The vehicle includes a central control unit 4 and an illumination device 5. An encoder block 2 is installed in the vehicle's control unit 4, and this encoder block 2 is used to generate image data that should be projected by the illumination device 5. An encoder 3 is therefore configured to process this image data and generate compressed data. A decoder 3 is installed in the illumination device 5 and is configured to receive the compressed data generated by the encoder and decompress the compressed data, thereby providing the illumination device with the information required for projecting the desired light pattern.

[0071] The lighting device 5 thus achieves good quality projection with improved transmission bandwidth.

[0072] Figure 3 Detailed images of the operation of this lighting arrangement are provided.

[0073] First, the vehicle's control unit generates image pattern 6. This image pattern is then sent to the encoder block, which divides the image into different parts based on its properties.

[0074] In this case, since the image is a near-beam pattern, the first part 61 includes a flat portion, and the second part 62 includes a kink portion. Both images are normalized so that the luminous intensity of each pixel is scaled to a range between 0 and 1. The normalized data then undergoes encoding, resulting in two vectors 71 and 72.

[0075] These two vectors 71 and 72, which are significantly smaller than the original pattern 6, are transmitted to the lighting device.

[0076] There, the decoder receives and processes vectors 71 and 72 to produce two restored images 81 and 82, one for the flat section and the other for the curved section. These are combined and sent to the lighting module for final projection.

Claims

1. A method for manufacturing a motor vehicle lighting device, the method comprising the following steps: - Train a deep autoencoder (1) to process image data (6), the deep autoencoder (1) including at least one encoder block (2), at least one decoder block (3) and loss function unit, wherein the encoder block (2) and the decoder block (3) have undergone a common training process as part of the same deep autoencoder (1); - The encoder block (2) is installed in the motor vehicle control unit (4) of the motor vehicle (10); and - The decoder block (3) is installed in the motor vehicle lighting module (5) of the motor vehicle (10).

2. The method according to claim 1, wherein, The encoder block (2) includes a convolutional layer (21), a corrected linear unit layer (22), and a normalization layer (23).

3. The method according to claim 2, wherein, The method further includes the step of selecting the ratio between the size of the convolutional layer (21) of the encoder block (2) and the size of the normalization layer (23) of the encoder block (2).

4. The method according to any one of the preceding claims, wherein, The decoder block (3) includes a no-sampling convolutional layer (31), a corrected linear unit layer (32), and a normalization layer (33).

5. A motor vehicle lighting device, said motor vehicle lighting device being manufactured by the method according to any one of the preceding claims, said motor vehicle lighting device comprising: - A vehicle control unit (4) including an encoder block (2) configured to process image data (6); and - An illumination module (5) including a decoder block (3) configured to receive processed image data (71, 72) to recover the original version (81, 82) of the image data, wherein the illumination module (5) is also configured to project a light pattern based on the image data (81, 82) recovered by the decoder block (3); The encoder block (2) and the decoder block (3), as parts of the same depth autoencoder (1), have undergone a common training process.

6. The motor vehicle lighting device (10) according to claim 5, wherein The lighting module (5) includes a solid-state light source, such as an LED.

7. A method for operating a motor vehicle lighting device according to any one of claims 5 or 6, the method comprising the following steps: - Operate encoder block (2) to reduce the data size of some image data (6), thereby producing processed image data (71, 72). - Transmit the processed image data (71, 72) to the decoder block (3); - Operate the decoder block to process the processed image data (71, 72) to produce restored image data (81, 82); and - Operate the lighting module (5) to project a light pattern based on the recovered image data (81, 82).

8. The method according to claim 7, the method comprising the step of normalizing the image data (6) before operating the encoder block (2) to reduce the data size of the image data.

9. The method of claim 8, wherein, The step of normalizing the image data (6) includes converting each value of the image data (6) into a converted value between 0 and 1.

10. The method according to any one of claims 7 to 9, the method comprising the step of dividing the image data (6) into data subarrays (61, 62) of the same format before operating the encoder block (2) to reduce the data size of the image data.

Images