Checking compliance of light beam in vehicle

Abstract

The invention relates to a method for evaluating the compliance of a light beam of a lighting module of a motor vehicle, comprising a step of storing (203) a model in the motor vehicle. The model is capable of determining information about compliance of the light beam based on the image received as a model input. The method comprises: checking (221) whether a lighting module of the motor vehicle is activated; -obtaining (223) a first image based on data from an image acquisition device of the vehicle; -applying (225) the model to the first image in order to determine first information about the compliance of the light beam; -evaluating (226) the compliance of the light beam of the at least one lighting module as a function of the at least one item of first information about the compliance of the light beam.

Description

[0001] This invention relates to the field of evaluating the compliance of light beams projected by motor vehicle lighting modules. In particular, this invention relates to a method and apparatus for evaluating the light beams of motor vehicle lighting modules.

[0002] This invention is particularly useful for correcting the focus of the lighting module of a motor vehicle or for correcting the luminous intensity distribution in the projected beam.

[0003] Motor vehicles include lighting modules located at the front, right, and left sides of the vehicle to perform several lighting functions, such as the "low beam" (also known as LB) function and the "high beam" (also known as HB) function.

[0004] The lighting module may be able to perform both of the above functions using the same light source. As a variation, several different modules are provided: one pair of modules is dedicated to the LB function, while another pair of modules is dedicated to the HB function.

[0005] The term "focus" should be understood to refer to the direction and shape of the beam of light projected by the vehicle's lighting module. This shape and direction may be subject to regulatory restrictions, especially to ensure that the driver has good visibility of the scene ahead of the vehicle without dazzling other road users.

[0006] Furthermore, regulatory restrictions may limit the distribution of luminous intensity projected within a beam. Notably, it may be stipulated that, for a given function, the projection of the left-side lighting module onto the ground should be symmetrical, and the luminous intensity distribution within each beam should be uniform.

[0007] The focal point of an illumination module can be observed or characterized by its projection onto the ground. Specifically, the focal points of two illumination modules within the same pair can be evaluated by their respective projections into a reference area. Therefore, during technical control of the vehicle, a professional can check whether the focal point of each illumination module complies with regulatory limits, or whether the focal points of the illumination modules within the same pair are indeed symmetrical. Furthermore, such a professional can also check whether the luminous intensity is correctly distributed within each beam of the illumination module and between several beams of a pair of illumination modules.

[0008] However, such technical control over vehicles is not frequent, and improper modifications to the focus or intensity distribution of the lighting module may go undetected for months or even years. Meanwhile, such improper focus modifications can reduce the driver's visibility of the scene and / or dazzle other road users, potentially leading to accidents.

[0009] Therefore, the lack of regular assessments of whether the beam characteristics of the lighting module meet the intended characteristics (e.g., regulatory or technical features) poses safety risks to vehicle occupants and other road users.

[0010] Therefore, it is necessary to periodically assess the compliance of the beams of motor vehicle lighting modules without requiring professional intervention.

[0011] Therefore, a first aspect of the present invention relates to a method for evaluating the compliance of a beam of at least one lighting module of a motor vehicle, the method comprising a preliminary step of storing a model in the memory of the motor vehicle, the model being capable of determining at least one piece of information regarding the compliance of the beam based on at least one image received as input to the model, the method further comprising the following present steps: - Check if at least one lighting module of the motor vehicle is activated; - Obtain at least one first image from data originating from an image acquisition device of the vehicle, wherein the image acquisition device is arranged to obtain data representing the scene in front of the vehicle; - Apply the model to the at least one first image to determine at least one piece of first information regarding the compliance of the beam; - Assess the compliance of the beam of the at least one lighting module based on the at least one piece of first compliance information.

[0012] Therefore, this invention enables the automated assessment of the compliance of the beams of a vehicle's lighting module based on a model stored in the vehicle and by applying that model to a first image obtained from data from an image acquisition device. Since such image acquisition devices are typically integrated into the latest vehicle models, this invention does not require additional sensors specifically designed for it, and thus can be implemented at a lower cost.

[0013] According to some embodiments, the method may further include a preliminary step of obtaining a model by performing machine learning based on a training dataset.

[0014] Therefore, this model can be obtained through machine learning, thus enabling reliable predictions of primary compliance information.

[0015] In addition, the model can be obtained through supervised learning, and the training dataset can include the association between at least one training image and at least one piece of information about the compliance of the beam.

[0016] Therefore, by forming a training database that includes both training images and corresponding associated compliance information, an accurate model for predicting compliance information can be obtained, which can be stored in the vehicle and then used to evaluate the compliance of the beam at the current stage.

[0017] According to some embodiments, the model may be able to determine a binary compliance category from compliance and non-compliance categories.

[0018] Therefore, the model can be a simple classifier with two possible outputs, which reduces the cost associated with training the model and the runtime when applying the model.

[0019] As a variant, the model may be able to identify one non-compliance type from at least two different non-compliance types.

[0020] Therefore, by distinguishing between several types of non-compliance, richer compliance information than binary information can be obtained. This distinction improves the assessment results provided when the method is completed and helps operators implement corrective actions in cases of beam non-compliance.

[0021] Additionally, when the model is applied to the at least one first image, a first non-compliance type can be determined. This assessment may include instructions to correct the lighting module, wherein the correction depends on the determined first non-compliance type.

[0022] Therefore, in cases where the beam is non-compliant, it can help the operator to perform corrective actions.

[0023] According to one embodiment, the image acquisition device may be a camera capable of acquiring color images or grayscale images.

[0024] Therefore, cameras present in most motor vehicles can be used, such as those for driving assistance functions or vehicle parking assistance functions for lateral or longitudinal vehicle trajectory. Consequently, the costs associated with implementing the method according to the invention are reduced.

[0025] Additionally, the model may be able to determine at least one piece of information regarding the compliance of the beam based on at least one color or grayscale image received as input, and the at least one first image obtained may be color or grayscale and may originate from an image acquisition device.

[0026] Therefore, the camera images can be directly used as input for the model, which speeds up the implementation of the method according to the invention.

[0027] As a variant, the model may be a first model capable of determining at least one piece of information regarding beam compliance based on at least one brightness image. The method may further include a preliminary step of storing a second model in the vehicle, wherein the second model is capable of determining at least one brightness value of a pixel in the brightness image based on at least one value of a pixel in a color or grayscale image. The at least one first acquired image may be a first brightness image. Obtaining the at least one first brightness image may include applying the second model to the pixels of at least one first color or grayscale image obtained by an image acquisition device. The first model may be applied to the at least one first brightness image to determine the at least one piece of information regarding beam compliance.

[0028] By incorporating brightness images as input to the model, predictions of compliance information regarding the beam can be improved. In fact, brightness images provide better contrast, leading to better identification not only of the beam's geometry but also of the distribution of luminous intensity within the beam. Furthermore, the brightness image is determined by a second model, avoiding the use of expensive and slow brightness cameras.

[0029] According to another variant, the image acquisition device is an infrared camera capable of acquiring infrared images, and this model may be able to determine at least one piece of information regarding the compliance of the beam based on at least one infrared image received as input.

[0030] Considering infrared images as input to the model, compared to using grayscale or color images, makes it possible to better predict information about the compliance of the beam.

[0031] According to some embodiments, the at least one activated lighting module may include a right-side low beam module and a left-side low beam module of a motor vehicle.

[0032] Therefore, compliance with low beams can be assessed to detect non-compliance that is particularly likely to cause glare to other users on the road where the vehicle is traveling and / or impair the driver's visibility due to insufficient illumination of the scene in front of the vehicle.

[0033] According to some embodiments, a first model may be able to determine at least one piece of information about the compliance of a beam based on a series of images received as input, wherein the at least one first image may be a series of first images, and the model may be applied to the first series of images to obtain the at least one piece of first information.

[0034] Using a series of images as input to the model provides greater accuracy in predicting information about beam compliance.

[0035] A second aspect of the invention relates to a computer program comprising instructions for implementing the method according to the first aspect of the invention when executed by a processor.

[0036] A third aspect of the invention relates to an apparatus for a vehicle, the apparatus comprising: - A memory that stores a model capable of determining at least one piece of information about beam compliance based on at least one image received as input to the model; or an access interface for accessing the first model; - Processor, which is configured as: Check that at least one lighting module on the vehicle is activated; At least one first image is obtained from data originating from an image acquisition device of a vehicle, wherein the image acquisition device is arranged to obtain data representing a scene in front of the vehicle; The model is applied to the at least one first image to determine at least one piece of first information regarding the compliance of the beam; The compliance of the beam of the at least one lighting module is assessed based on the at least one piece of first compliance information.

[0037] Further features and advantages of the invention will become apparent from the following detailed description and accompanying drawings, in which: [ Figure 1 The image shows a motor vehicle according to some embodiments of the present invention; [ Figure 2 The steps of a method according to some embodiments of the present invention are illustrated; [ Figure 3a The image illustrates the beam of a motor vehicle's lighting module when the beam is compliant, according to some embodiments of the invention. [ Figure 3b The image illustrates the beam of a motor vehicle's lighting module when the beam is non-compliant, according to some embodiments of the invention. [ Figure 4 This invention illustrates an apparatus for calibrating the lighting module of a motor vehicle according to some embodiments thereof.

[0038] This specification focuses on features that distinguish the methods and apparatus from those known in the prior art.

[0039] Figure 1 A vehicle 100 according to some embodiments of the present invention is shown.

[0040] Vehicle 100 includes a device 101 according to some embodiments of the invention, wherein the device 101 is capable of assessing the compliance of the beam of at least one lighting module of vehicle 100. It should be noted that compliance is assessed relative to a set of predefined standards and / or technical standards derived from the manufacturer of vehicle 100, which may be regulatory standards applicable to all vehicles operating within a given area.

[0041] The vehicle 100 actually includes a pair of lighting devices 103, which include at least one lighting module capable of performing at least one lighting function of the vehicle 100.

[0042] exist Figure 1In the example, vehicle 100 is shown as a side view, making a single lighting device 103 (such as the left-side lighting device) visible in the figure. However, it should be understood that vehicle 100 further includes a lighting device located on the right side of the vehicle, which is arranged symmetrically with respect to the front and rear axles of the vehicle with respect to the lighting device 103.

[0043] exist Figure 1 In one example, the lighting device 103 includes a first lighting module 104 dedicated to a first light-emitting function and a second lighting module 105 dedicated to a second light-emitting function, which is different from the first light-emitting function. Alternatively, a single lighting module may be capable of implementing both the first and second lighting functions.

[0044] The first lighting function can be the low beam function called LB, while the second lighting function can be the high beam function called HB.

[0045] The vehicle 100 further includes a control module 102 for controlling the lighting device 103, which is capable of activating / deactivating a first function and / or a second function of the vehicle. When the light sources of the lighting modules 104 and 105 are matrix-type with multiple individually controllable light-emitting elements, the control module 102 can also transmit pixelated illumination luminance data, and the lighting device 103 can selectively activate / deactivate the light-emitting elements of the light source of any lighting module according to the pixelated illumination luminance data.

[0046] The control module 102 can be dedicated to controlling the lighting device, or it can control other functions of the vehicle. In particular, the control module 102 can be the central control module of the vehicle 100.

[0047] The vehicle 100 according to the present invention may further include an image acquisition device 106.

[0048] The image acquisition device 106 is capable of acquiring data representing a vehicle scene located in the opposite direction of vehicle 100 (i.e., in front of vehicle 100 in the direction of travel of vehicle 100). It is worth noting that the image acquisition device 106 may be located at the front of the vehicle, such as at the top of the windshield of vehicle 100 or on the bumper of vehicle 100.

[0049] The image acquisition device 106 can be an RGB type camera. An RGB (red-green-blue) type camera can acquire color image data, which can be processed in real time to measure luminous parameters related to the vehicle environment.

[0050] Therefore, the data acquired by the image acquisition device 106 can be a color image, meaning each image includes pixels arranged in a matrix, and each pixel is encoded to correspond to one of exactly two colors. For example, each pixel can be encoded using red-green-blue or RGB, meaning all colors are represented by three coordinates (red, green, and blue). Thus, each pixel can be encoded using several bits; for example, one byte can encode 256 colors.

[0051] As a variant, the data acquired by the image acquisition device 106 can be a grayscale image, i.e., each image includes pixels arranged in a matrix, wherein each pixel is encoded by strictly more than two gray levels, and there is at least one intermediate gray level between black and white. Therefore, each pixel can be encoded by several bits, for example, one byte encodes 256 gray levels.

[0052] As another variation, the image acquisition device 106 may be an infrared camera capable of acquiring data as infrared images, which include pixels arranged in a matrix, wherein each pixel is associated with an infrared value.

[0053] Image acquisition device 106 may be able to periodically acquire still images at a given frequency (e.g., once per second). However, preferably, in order to enable the data derived from image acquisition device 106 to be used for the real-time functions of the vehicle, image acquisition device 106 may be able to acquire images constituting video frames at a frequency of several frames per second (or fps) (e.g., at a frequency greater than 10 fps, especially equal to 30 fps).

[0054] Vehicle 100 may include a memory 107 capable of storing data (such as the first model and / or the second model described below). As a variant, memory 107 may be integrated into control module 102 or device 101.

[0055] Vehicle 100 may include a human-machine interface 108 or HMI 108. There are no limitations on the HMI 108, which includes any interface element capable of receiving commands from the driver or occupants of the vehicle (whether the command pertains to the control of the vehicle or to the control of in-vehicle or external devices) and / or transmitting information to the driver and / or occupants of the vehicle. For this purpose, HMI 108 may include a screen (such as a touchscreen), a set of one or more buttons, a speaker, a microphone, an instrument panel capable of displaying one or more illuminated pictograms of predefined shapes, a system for vibrating the steering wheel, etc.

[0056] Figure 2 This is a simplified diagram illustrating the steps of a method for checking the compliance of a beam of at least one lighting module according to some embodiments of the present invention.

[0057] The method may include a preliminary stage 200 and a current stage 220. The preliminary stage includes steps 201 to 203 described below, and may further include steps 211 to 213 in a particular embodiment.

[0058] A first training database is formed during step 201, wherein the first training database includes first training data. The first training data includes first training images, which are obtained by a camera installed in the corresponding motor vehicle at a position close to the camera 106 described above, i.e.: -If camera 106 is positioned in the upper section of the vehicle windshield, the training images are acquired by the camera positioned in the corresponding upper section of the vehicle windshield. - If camera 106 is positioned at the bottom of the front of the vehicle body (especially for the purpose of enabling parking assistance), then training images are acquired by cameras positioned at the corresponding bottom of the vehicle body.

[0059] The training images represent the scene in front of a given vehicle when at least one lighting module of the given vehicle is turned on.

[0060] The first training image can be: - Grayscale image; -Color image; -Infrared image; or - A luminance image, also known as a luminance map.

[0061] There are no restrictions on how the initial training data is obtained. For example, it can be collected, in particular, from a fleet of vehicles in operation during technical control.

[0062] In one embodiment, the first training database includes time series of training images.

[0063] In the training database, each first training image or each training image time series can be associated with information about beam compliance. Therefore, the first training data further includes compliance information for each first training image or training image time series.

[0064] Each piece of compliance information associated with one or more training images is called reference compliance information. To some extent, this reference compliance information is not a prediction but real information, which is different from the compliance information obtained (or predicted) as the output of the model, as described later.

[0065] Compliance information indicates the compliance or non-compliance of at least one beam present in the associated first training image (or a series of associated training images). Compliance information can also indicate the compliance or non-compliance of all projected beams (e.g., two beams projected by the low beam illumination module).

[0066] Therefore, compliance information is used to label the first training image (or a series of training images) in the first database.

[0067] There are no restrictions on standards related to compliance: they can be manufacturer technical standards or regulatory standards.

[0068] In addition, compliance may involve: - The shape and / or orientation of one or more beams, the compliance of which can be assessed based on their projection onto the ground; and / or - The distribution of luminous intensity in one beam and / or two beams.

[0069] According to one embodiment, compliance information can be a binary compliance category, that is, compliance information can take two values: a "compliant" category and a "non-compliant" category.

[0070] As a variant, compliance information can take at least three values: a "compliant" category, a "non-compliant (first non-compliant type)" category, and a "non-compliant (second non-compliant type)" category. More generally, compliance information can identify N non-compliant types, where N is an integer greater than or equal to 2. Non-compliant types can be: - Indicates one of several levels of non-compliance (e.g., "minor," "moderate," and "severe"); and / or - Describes non-compliance. Non-compliance types can specifically indicate whether the right or left beam is too high, too low, too far to the right, or too far to the left. Additionally, this type can indicate a deviation value relative to the non-compliance condition, where the value could be, for example, a distance or angle value.

[0071] During step 202, a first model is trained by performing machine learning on first training data from the training database obtained in step 201. The model trained in this way is able to: - Receives at least one first image as input. As indicated above, depending on the first training image used (grayscale, color, brightness, or infrared scale), the model is able to: - Receive a first image of the same type as the first training image; - Determine or predict at least one piece of compliance information based on the first image received as input.

[0072] This model can be obtained through supervised learning, that is, by comparing the reference compliance information associated with the first training image with the compliance information predictions output from the model. Therefore, the model minimizes its error in determining the at least one piece of compliance information.

[0073] The principles of supervised learning are well known and will not be described in further detail in this application.

[0074] Thus, the first model trained is able to predict at least one compliance information based on the first image received as input or the first series of images received as input.

[0075] Step 202 can be performed by Figure 1 The training module, not shown, is used for implementation. This training module can be located outside the vehicle 100 and has access to the first training database.

[0076] During step 203, the first model is stored in vehicle 100 so that device 101 can access it. The first model may be stored, for example, in vehicle memory 107 or in device 101's internal memory.

[0077] There are no restrictions on the type of the first model; it can be a support vector machine (SVM), a type classifier, a convolutional neural network, or any other model that can be built through machine learning and can receive images as input.

[0078] When the first model is trained to receive grayscale or color images, according to some embodiments, the input to the first model may be able to directly receive images from an image acquisition device 106, which is in this case a grayscale camera or a color camera. Such cameras are available in most current vehicles, especially for lateral / longitudinal driving assistance functions (camera at the top of the windshield) or parking assistance functions (camera at the bottom of the vehicle body). Therefore, the costs associated with these embodiments are reduced.

[0079] According to another embodiment, the first model is trained to receive infrared images. In this case, the input to the first model may be able to directly receive images from an image acquisition device 106, which is in this case an infrared camera. Since infrared images have better contrast compared to grayscale or color images, the use of infrared images enables higher accuracy in predicting the at least one compliance information.

[0080] According to yet another embodiment, the first model is trained to receive a brightness image.

[0081] Luminance refers to the amount of light flux originating from an illuminated surface and reflected into the eye, expressed in candela per square meter or cd / m². 2 Therefore, brightness represents the visual perception of the lightness or darkness of a surface.

[0082] More specifically, brightness is the power of visible light transmitted through or reflected by a surface element in a given direction, per unit area, and per unit solid angle.

[0083] In external lighting, luminance measurements are normalized based on factors such as distance from the surface, observer position and height, and viewing angle.

[0084] Therefore, a luminance map is different from a grayscale image. In a grayscale image, the gray level of a pixel represents the brightness perceived by the camera's sensor at a given location, but this brightness does not correspond to the luminance defined above.

[0085] Compliance information based on brightness images can provide higher accuracy for predictions, especially when the scene in front of the vehicle is illuminated by street lighting or indoor lighting (parking lot), making it difficult to distinguish between ambient lighting and lighting from beams originating from the vehicle's lighting module.

[0086] In this embodiment, the input to the first model may be able to directly receive an image from an image acquisition device 106, which is in this case a luminance camera. However, such a luminance camera is expensive and requires a significant amount of processing time to acquire a luminance image. Therefore, it is not suitable for real-time applications.

[0087] Preferably, according to this embodiment of the invention, the second model is trained to predict a brightness image based on a grayscale image or a color image received as model input. Therefore, the second model can be placed between the image acquisition device 106 and the first model implemented by the device 101, and in this case, the image acquisition device 106 can be a color or grayscale camera, which has the aforementioned advantages.

[0088] The second model can be obtained from steps 211 to 213 described below.

[0089] The second training database can be formed during step 211.

[0090] The second training database may include second training data, which includes second grayscale or color training images derived from a camera or several different cameras. Specifically, the second training images in the second training database may originate from different cameras, particularly different types of cameras, wherein the two different types of cameras have at least one different technical feature, or come from different manufacturers. Forming the second database using second training images from different types of cameras enables the training of a general model, which can then be integrated into any vehicle, regardless of the image acquisition device 106 of vehicle 100 (capable of acquiring grayscale or color images).

[0091] As a variant, all images are derived from the same type of camera as the image acquisition device 106, in which case the second model trained based on the second training data is specific to the image acquisition device 106 of the vehicle 100.

[0092] Preferably, the second training data in the second database formed in step 211 includes the association between the second training images and the corresponding training luminance maps. Each training luminance map or image associated with the second training image can be acquired by a luminance camera placed next to the camera that acquired the second training image, so as to acquire visual data related to the same scene as the scene captured by the camera.

[0093] During step 212, the second model can be trained based on the second training data in the second database formed in step 211. There are no restrictions on the second model generated from the training in step 212; the second model can be: - A curve or function that maps pixel values ​​in a grayscale or color image to the luminance values ​​of corresponding pixels in a luminance map. This curve can be obtained using curve fitting analysis techniques based on associations stored in a second training database. In fact, based on these associations in the second training database (where each association relates a second training image to a training luminance map), the pixel association between a given pixel in the training image and its corresponding pixel in the training luminance map can be determined. These pixel associations provide a feed for curve fitting analysis to obtain a model in the form of a curve; - A Support Vector Machine (SVM) model, which determines the brightness value of a pixel in a brightness map by classification based on the values ​​of corresponding pixels in a grayscale or color image. The SVM model can be constructed through supervised learning based on associations in a second training database. In fact, based on these associations in the second training database (where each association associates a training image with a training brightness map), the pixel association between a given pixel in the second training image and its corresponding pixel in the training brightness map can be determined. These pixel associations provide a feed for the supervised learning of the SVM model. An advantage of the SVM model is that it can be formed based on a second training database containing a limited amount of data. - A neural network, such as a convolutional neural network, capable of determining the entire brightness map based on a grayscale or color image received as input. The neural network can be obtained through machine learning using second training data from a second training database formed in step 211. In the case of supervised learning, the neural network can be configured by modifying the parameters of specific neurons based on the measured deviation between the brightness map determined or predicted by the neural network when a second given training image is submitted to the neural network and the training brightness map associated with the second training image given in the second training database; - Any other model obtained through machine learning or data analysis techniques, which is able to determine brightness values ​​based on the values ​​of grayscale or color pixels, or is able to determine the entire brightness map based on grayscale or color images received as input.

[0094] Step 212 can be done by Figure 1 The training module, not shown, is used for implementation. This training module can be located outside the vehicle 100 and has access to a second training database.

[0095] After completing the training in step 212 (which is the analysis or machine learning described above), a second model according to the invention is obtained, wherein the model is able to determine the brightness value based on the value of grayscale or color pixels, or is able to determine the entire brightness map based on the grayscale image or color image received as input.

[0096] During step 213, the second model obtained from the training in step 212 is stored in the motor vehicle 100, either in the memory of device 101 or in a memory outside device 101 but accessible to device 101.

[0097] The method according to the invention further includes a current stage 220, which includes steps 221 to 227 described below.

[0098] During step 221, device 101 checks whether at least one lighting module of vehicle 100 is activated. Specifically, according to the invention, device 101 can check whether the right low beam lighting module and the left low beam lighting module are activated. This check can involve checking the activation state of the low beam lighting module by polling control module 102, or it can be based on identifying the light beam in the data received from the image acquisition device in step 222 (in which case, the lighting module activation check in step 221 is performed after step 222).

[0099] If at least one lighting module of vehicle 100 is not activated, device 101 may request activation of the at least one lighting module, specifically by sending a request to control module 102. Alternatively, the device may periodically poll the activation status of the at least one lighting module at a given frequency, and subsequent steps of the current phase 220 may only be performed when the at least one lighting module is activated.

[0100] During step 222, device 101 receives data from image acquisition device, wherein the data includes one or more images from image acquisition device 106.

[0101] During step 223, device 101 obtains at least one first image from the data received from image acquisition device 106 during step 222.

[0102] The at least one first image obtained in step 223 may be a single first image or a series of several first images.

[0103] When the data from the image acquisition device 106 includes a single image, that image forms the first single image obtained in step 223. When the data from the image acquisition device 106 includes several consecutively acquired images, the first single image can be: - An image selected from a plurality of images continuously acquired by the image acquisition device 106; - An image determined from a number of consecutively acquired images. The determined image may be a stabilized or averaged image, which makes it possible to obtain a clearer image than each image acquired by the image acquisition device 106.

[0104] As a variant, where the first model is capable of receiving a series of images as input, the at least one first image is a series of first images. In this case, the series of first images may be a series of images acquired by the image acquisition device 106.

[0105] According to the embodiment corresponding to the first model obtained during the preliminary stage 200, each first image 106 may be a grayscale image, a color image, an infrared image, or a brightness image.

[0106] If the first model is capable of receiving a brightness image as input, the method may include step 224, implemented by device 101 between steps 222 and 223, namely, applying the second model stored in step 213 to data originating from image acquisition device 106 in order to obtain the at least one first brightness image (a first single brightness image or a series of brightness images).

[0107] In order to obtain a series of first brightness images, device 101 sequentially applies the second model to a series of (grayscale or color) images originating from image acquisition device 106.

[0108] During step 225, following step 223 of obtaining the at least one first image, device 101 applies a first model stored in vehicle 100 to the at least one first image in order to obtain at least one piece of compliance information by means of prediction.

[0109] As indicated above, compliance information can be binary or take more than two values ​​to indicate the type of non-compliance in the event of non-compliance.

[0110] During step 226, device 101 assesses the compliance of the beam from at least one lighting module based on the at least one piece of first compliance information. Specifically, the assessment indicates whether one or more beams originating from the lighting module comply with technical and / or regulatory standards.

[0111] The evaluation in step 226 can be obtained from the following: - When the first model is applied to the first image only once, at least one first compliance information derived from step 225; - Multiple compliance information derived from at least two iterations of steps 222 to 225 described above. Obtaining at least one piece of first compliance information for each iteration and incorporating the first compliance information into the assessment of the compliance of one or more beams improves the accuracy associated with the compliance assessment.

[0112] During step 227, device 101 transmits the compliance assessment obtained in step 226 to another entity of vehicle 100. This transmission may only be performed if the assessment indicates that at least one beam is non-compliant.

[0113] Therefore, the assessment can be transmitted to the driver or a third party, enabling corrective actions to be taken in cases of non-compliance. The assessment can also indicate corrective actions or adjustments to be taken, wherein the adjustments are determined by device 101 based on the at least one piece of first compliance information.

[0114] This assessment can be transmitted, for example, to: -HMI 108, to notify the driver that at least one beam has been detected as non-compliant, or to indicate that the beam is compliant; -Memory 107, which is used for storage and subsequent retrieval; - A wireless communication module, which is used to transmit the evaluation to a remote server accessible via a telecommunications network.

[0115] Figure 3a The lighting modules 103.1 and 103.2 of a motor vehicle 100 according to some embodiments of the present invention are shown as beams 301.1 and 301.2 when the beams are compliant.

[0116] Figure 3a The ground projection of the beams is illustrated by illumination modules 103.1 and 103.2, which are the low beam, right-side, and left-side illumination modules. The projections of beams 301.1 and 301.2 are shown as a top-down view above the road where they are projected. This view differs from the perspective of the image acquisition device 106 located in the vehicle. However, the top-down view facilitates a more intuitive understanding of the standards presented in an illustrative manner.

[0117] A standard relating to the compliance of beams 301.1 and 301.2 could be that beams 301.1 and 301.2 must not intersect within a reference area 302, which could be a predefined distance range in front of the vehicle. More specifically, the standard could specify that the two beams intersect at the end of the reference area 302.

[0118] Therefore, in Figure 3a In the example shown, beams 301.1 and 301.2 conform to the standard described above. With the first model trained to recognize only compliance information regarding this standard, the projection is similar... Figure 3a The training image obtained from the projected beam can be associated with compliance information indicating the compliance of the beam.

[0119] The standards defined in the reference area are provided in an illustrative manner. Alternatively, or as variations, other standards may be evaluated to characterize compliance.

[0120] Figure 3b The lighting modules 103.1 and 103.2 of a vehicle 100 according to some embodiments of the present invention are shown as beams 311.1 and 311.2 in cases where the beam is non-compliant with the standards described above.

[0121] exist Figure 3b In this case, beams 311.1 and 311.2 intersect in the reference region, causing them to not conform to the criteria described above.

[0122] With the first model trained to recognize only compliance information related to the standard, the projection is similar. Figure 3b The training image obtained from the projected beam can be associated with compliance information indicating non-compliance of the beam.

[0123] As described above, compliance information can be more accurate than a simple indication of non-compliance: it can also indicate the type of non-compliance. Figure 3b In the example, the compliance information could indicate that the non-compliance is due to the beams converging in reference area 302. More specifically, the compliance information could indicate that the position of the right-hand beam is too low.

[0124] Figure 4 The structure of a device 101 according to some embodiments of the present invention is shown.

[0125] Device 101 includes a processor 401 configured to communicate unidirectionally or bidirectionally with memory 402 (such as random access memory (RAM) or read-only memory (ROM) or any other type of memory (flash memory, EEPROM, etc.)) via one or more buses or via a direct wired connection. As a variant, memory 402 may include multiple memories of the aforementioned types.

[0126] Memory 402 can be permanently or temporarily stored in the implementation reference. Figure 2 The steps of the method according to the invention shown herein use and / or are derived from at least a portion of the data from these steps.

[0127] Specifically, memory 402 may be able to store the first model during step 203, and optionally store the second model during step 213.

[0128] Furthermore, the at least one first image obtained in step 223 may be stored at least temporarily in memory 402.

[0129] Processor 401 is capable of executing instructions stored in memory 402 in order to implement the reference Figure 2 Steps 221 to 227 of the current stage 220 are described. Alternatively, processor 401 may be designed and configured to execute reference... Figure 2 The current stage 220 described in steps 221 to 227 is replaced by a microcontroller.

[0130] Device 101 includes a first interface 403, which is capable of receiving data from image acquisition device 106.

[0131] Device 101 includes a second interface 404 that enables the evaluation to be transmitted to an entity outside device 101 during step 227 described above.

[0132] Device 101 includes a third interface 405 that can access the first model and / or the second model when the first model and / or the second model are stored outside the device 101 in the vehicle 100.

[0133] This invention is not limited to the embodiments described above by way of example; this invention covers other variations.

Claims

1. A method for evaluating the compliance of a beam (301.1; 301.2; 311.1; 311.2) of at least one lighting module (105; 104; 104.1; 104.2) of a motor vehicle (100), the method comprising a preliminary step (203) of storing a model in a memory (107; 402) of the motor vehicle, the model being capable of determining at least one piece of information regarding the compliance of the beam based on at least one image received as input to the model, the method further comprising the following current step: - Check whether at least one lighting module of the motor vehicle (221) is activated; - Obtain (223) at least one first image from data originating from the image acquisition device (106) of the vehicle, wherein, The image acquisition device is arranged to obtain data representing the scene in front of the vehicle; - Apply the model (225) to the at least one first image to determine at least one piece of first information regarding the compliance of the beam; - Assess the compliance of the beam of the at least one lighting module based on the at least one first compliance information. (226) 2. The method of claim 1, further comprising a preliminary step (202) of obtaining the model by performing machine learning based on a training dataset.

3. The method as described in claim 2, wherein, The model is obtained through supervised learning (202), and the training dataset includes the association between at least one training image and at least one piece of information regarding the compliance of the beam.

4. The method as described in any of the preceding claims, wherein, The model is able to determine a binary compliance category from compliance and non-compliance categories.

5. The method according to any one of claims 1 to 3, wherein, The model is able to identify one type of non-compliance from at least two different types of non-compliance.

6. The method of claim 5, wherein, When the model is applied (225) to the at least one first image, a first non-compliance type is determined, wherein the evaluation includes an indication to correct the lighting module, wherein the correction depends on the determined first non-compliance type.

7. The method as described in any of the preceding claims, wherein, The image acquisition device (106) is a camera capable of acquiring color or grayscale images.

8. The method of claim 7, wherein, The model is capable of determining at least one piece of information regarding the compliance of the beam based on at least one color or grayscale image received as input, wherein the at least one first image obtained is color or grayscale and originates from the image acquisition device (106).

9. The method of claim 7, wherein, The model is a first model capable of determining at least one piece of information about the compliance of the beam based on at least one brightness image, wherein the method further includes a preliminary step (213) of storing a second model in the motor vehicle (100), wherein the second model is capable of determining at least one brightness value of a pixel in a brightness image based on at least one value of a pixel in a color or grayscale image; Wherein, the at least one first obtained image is a first brightness image; Wherein, obtaining the at least one first brightness image includes applying the second model (224) to the pixels of at least one first color or grayscale image obtained by the image acquisition device (106); The first model is then applied to the at least one first brightness image to determine the at least one piece of information regarding the compliance of the beam.

10. The method according to any one of claims 1 to 6, wherein, The image acquisition device (106) is an infrared camera capable of acquiring infrared images, wherein the model is capable of determining at least one piece of information regarding the compliance of the beam based on at least one infrared image received as input.

11. The method as described in any of the preceding claims, wherein, The at least one activated lighting module (104; 105; 104.1; 104.2) includes the right low beam module (104.2) and the left low beam module (104.1) of the motor vehicle (100).

12. The method as described in any of the preceding claims, wherein, The first model is capable of determining at least one piece of information regarding the compliance of the beam based on a series of images received as input, wherein the at least one first image is a series of first images, and wherein the model is applied (225) to the first series of images to obtain the at least one piece of first compliance information.

13. A computer program comprising instructions for implementing the method as described in any of the preceding claims when executed by a processor (401).

14. An apparatus (101) for a vehicle (100), the apparatus comprising: - A memory (402) storing a model capable of determining at least one piece of information regarding beam compliance based on at least one image received as input to the model; or an access interface (405) for accessing the model; - Processor (401), the processor being configured to: Check whether at least one lighting module (104; 105; 104.1; 104.2) of the vehicle is activated; At least one first image is obtained from data originating from an image acquisition device (106) of the vehicle, wherein the image acquisition device is arranged to obtain data representing a scene in front of the vehicle; The model is applied to the at least one first image to determine at least one piece of first information regarding the compliance of the beam; The compliance of the beam of the at least one lighting module is assessed based on the at least one piece of first compliance information.

Images