Automobile whistling volume control method and device, electronic equipment and storage medium

Abstract

The invention discloses an automobile whistling volume control method and device, electronic equipment and a storage medium, and relates to the field of vehicle control. Obtaining a whistling rule of the target area; uploading the acquired data, information and target area whistling rules to a cloud for fusion, and then generating a vehicle whistling label; historical vehicle whistling information in the target area and environmental data in the target area are marked, and a training set is generated; training a target detection algorithm based on SPPNet, and obtaining whistling features and whistling types of the current vehicle passing through the target area; training an ANN-based algorithm, and predicting a whistling probability of a vehicle passing through the target area; judging whether the whistling probability value is greater than a preset whistling probability threshold value or not, and generating a list of predicted whistling volumes; and based on the vehicle whistling probability and the predicted whistling volume, generating a whistling volume target value according to the whistling characteristics, the whistling type and the whistling volume list of the current vehicle passing through the target area, and outputting the target value to an execution mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vehicle control, and particularly relates to a horn volume control method for a vehicle, a horn volume control device for a vehicle, an electronic device, a storage medium and a vehicle. BACKGROUND

[0002] Most traditional car horns are set to a single volume output, and the volume is fixed regardless of the environment, which may lead to improper use in different scenarios.

[0003] In urban environments, especially in densely populated areas, high-volume honking can exacerbate noise pollution and affect the quality of life of residents.

[0004] Excessive honking can scare pedestrians, especially causing unnecessary fright to the elderly and children, while too small a sound may not be enough to attract attention, especially in emergency situations.

[0005] Many areas have strict honking regulations, such as prohibiting honking in certain areas or time periods to reduce noise disturbance, and fixed-volume horns are difficult to comply with such agreements.

[0006] With the development of smart cars, users expect vehicles to be more intelligent in adapting to the surrounding environment, including automatically adjusting the honking volume based on environmental conditions.

[0007] Fixed-volume honking may not meet the actual needs of the driver in different road conditions, such as a slight reminder, warning or emergency, thereby affecting the driving experience.

[0008] Related patent documents include:

[0009] Patent document 1 (CN117087534A) discloses a control method, device, electronic device and readable storage medium for vehicle honking, which, compared with the prior art, acquires environmental behavior information of the vehicle, generates a target voltage based on the environmental behavior information of the vehicle, and controls the volume of the horn of the vehicle based on the target voltage. This patent relies solely on a single or small amount of data to determine whether to honk and control honking, which may lead to inaccurate honking determination in complex traffic scenarios.

[0010] Patent Document 2 (CN117002376A) discloses an intelligent control method and device for a vehicle horn system and a vehicle. The method includes obtaining category information of obstacles around the vehicle, wherein the category information includes whether the obstacle is a pedestrian or a vehicle; and controlling the horn volume of the horn system and playing a voice prompt according to the category information. However, this patent lacks deep fusion analysis of spatio-temporal context and regulation data, and it is difficult to flexibly adjust the horn control strategy in different time periods (day / night), different working day states (working day / rest day), and different regional traffic scenes (city / country / special region), which cannot meet the actual needs of diversified scenes. SUMMARY

[0011] Therefore, the present application provides a car horn sound control method, a car horn sound control device, an electronic device, and a storage medium. The present application predicts the horn probability and horn sound volume through a big data model, realizes horn volume control of vehicles in different regions, forms a total feature vector through the fusion of image data around the vehicle, vehicle GPS data, vehicle passing time of the section, local horn regulations, and a horn volume control regression model, reduces the response speed of the vehicle decision system, occupies small memory through a linear model, and avoids occupying too much vehicle chip computing power.

[0012] The present application provides the following solutions:

[0013] According to one aspect of the present application, a car horn sound control method is provided, comprising:

[0014] obtaining image data and location information around the vehicle;

[0015] obtaining horn rules of the target area;

[0016] uploading the obtained image data, location information, and target area horn rules to the cloud, generating a vehicle horn label after matrix feature fusion;

[0017] obtaining historical vehicle horn information in the target area and environmental data in the target area;

[0018] annotating according to the historical vehicle horn information in the target area and the environmental data in the target area to generate a training set;

[0019] training a target detection algorithm based on SPPNet according to the horn label and the training set, obtaining horn features and horn types of the vehicle passing through the target area;

[0020] training an algorithm based on ANN neural network according to the horn label and the training set to predict the horn probability of the vehicle passing through the target area;

[0021] determining whether the honking probability value is greater than a preset honking probability threshold value;

[0022] If greater, in response to honking volume calculation, a list of predicted honking volume is generated;

[0023] Based on the vehicle honking probability and the predicted honking volume, according to the honking characteristics, honking type and honking volume list of the current vehicle passing through the target area, a honking volume target value is generated;

[0024] According to the generated honking volume target value, output to the actuator.

[0025] Further, comprising:

[0026] The way of collecting image data around the vehicle includes collecting through the camera sensor arranged on the vehicle;

[0027] The way of obtaining vehicle location information includes obtaining current vehicle location information through GPS;

[0028] The labeled content includes visual feature vector, space-time feature vector and honking label.

[0029] Further, further comprising:

[0030] After uploading the obtained image data, location information and target area honking rules to the cloud and fusing the matrix features, a feature vector is generated;

[0031] Including: X=(FI, FG, FT, FR);

[0032] Image data extraction feature FI, using convolutional neural network CNN to process image data, extracting features of pedestrians, vehicles and obstacles;

[0033] Output feature vector FI=(npedestrians, nvehicles, nobstacles), where n represents the number of each category, pedestrians represents pedestrians, vehicles represents motor vehicles or non-motor vehicles, and obstacles represents obstacles;

[0034] GPS data feature extraction FG, using geocoding service to convert latitude and longitude into specific location information;

[0035] Output feature vector FG=(location, zone), where location represents specific location name and zone represents area type;

[0036] The time data feature extraction FT uses periodic encoding to transform the time data to capture the periodic patterns of time variation, outputting a feature vector;

[0037] FT = (sin(hour), cos(hour), sin(minute), cos(minute)), where sin(hour) and cos(hour) represent mapping the hour to a unit circle on a two-dimensional plane, forming a periodic representation; sin(minute) and cos(minute) map the minute to a unit circle, capturing the periodicity at the minute level;

[0038] The local siren regulation feature extraction FR queries the siren regulation database according to the location information, outputting a feature vector FR = (allowed_times, restricted_zones), where allowed_times represents the time period during which the vehicle is allowed to legally sound the siren according to the local siren regulation, and restricted_zones represents the area in which the vehicle is restricted to sound the siren according to the local siren regulation.

[0039] Further comprising:

[0040] The list of siren volume values of the vehicle passing through the target area includes:

[0041] P(D|X) = σ(β0 + β1XI + β2XG + β3XT + β4XR);

[0042] σ is a sigmoid function, ;

[0043] βi is a parameter of the model, i = 0, 1, 2, 3, 4, where β0 is the intercept term;

[0044] XI, XG, XT, XR are features extracted from image data, GPS data, time data, and local siren regulation feature extraction, respectively.

[0045] Further comprising:

[0046] The generated siren volume target value includes:

[0047] V(X) = α0 + α1XI + α2XG + α3XT + α4XR;

[0048] αi is a parameter of the model, i = 0, 1, 2, 3, 4, where α0 is the intercept term;

[0049] XI, XG, XT, XR are features extracted from image data, GPS data, time data, and local siren regulation feature extraction, respectively.

[0050] Further comprising:

[0051] The target detection algorithm training process based on SPPNet includes feature extraction on the training set, multi-task model construction, training optimization strategy, and real-time inference process;

[0052] The recommendation algorithm training process based on ANN includes data preprocessing, model construction, model training, and prediction result evaluation.

[0053] According to the two aspects of the present application, a vehicle horn volume control device is provided, which comprises a collection module, a scale module, a label module, an environmental data module, a training set module, a horn type module, a horn probability module, a threshold judgment module, a list generation module, a target value module and an output execution module.

[0054] The collection module is used to obtain image data and position information around the vehicle;

[0055] The scale module is used to obtain the horn rules of the target area;

[0056] The label module is used to upload the obtained image data, position information and target area horn rules to the cloud, and generate vehicle horn labels after matrix feature fusion;

[0057] The environmental data module is used to obtain historical vehicle horn information in the target area and environmental data in the target area;

[0058] The training set module is used to label and generate a training set according to the historical vehicle horn information in the target area and the environmental data in the target area;

[0059] The horn type module is used to obtain the horn features and horn types of the current vehicle passing through the target area based on the horn labels and training set training, and the target detection algorithm based on SPPNet;

[0060] The horn probability module is used to predict the horn probability of the vehicle passing through the target area based on the horn labels and training set training based on the ANN neural network algorithm;

[0061] The threshold judgment module is used to determine whether the horn probability value is greater than the preset horn probability threshold;

[0062] The list generation module is used to generate a list of predicted horn volumes in response to the horn volume calculation if the horn probability value is greater than the preset horn probability threshold;

[0063] The target value module is used to generate a horn volume target value based on the vehicle horn probability and the predicted horn volume, according to the horn features, horn types and horn volume list of the current vehicle passing through the target area.

[0064] An output execution module is configured to output the generated horn volume target value to an execution mechanism.

[0065] According to the three aspects of the present application, an electronic device is provided, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus.

[0066] The memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the horn volume control method for a vehicle.

[0067] According to the four aspects of the present application, a computer readable storage medium is provided, which stores a computer program executable by an electronic device, and when the computer program runs on the electronic device, the electronic device executes the steps of the horn volume control method for a vehicle.

[0068] According to the five aspects of the present application, a vehicle is provided, comprising:

[0069] An electronic device is configured to implement the steps of the horn volume control method for a vehicle.

[0070] A processor is configured to run a program, and when the program runs, the processor executes the steps of the horn volume control method for a vehicle from data output by the electronic device.

[0071] A storage medium is configured to store a program, and when the program runs, the storage medium executes the steps of the horn volume control method for a vehicle from data output by the electronic device.

[0072] Through the above scheme, the following beneficial technical effects are obtained:

[0073] The present application controls the horn volume of a vehicle based on image recognition, big data and machine learning. The influence on the surrounding environment is reduced, and noise pollution is avoided. By adjusting the horn volume of the vehicle, the driver can avoid startling pedestrians due to excessive sound volume, and can issue a loud enough alarm when necessary. Through controllable horn volume, the driver can choose a more gentle or more intense reminder signal according to the need, thereby improving the communication effect between the two parties.

[0074] The present application can accurately help the driver respond to the horn policy of the relevant area by obtaining the noise control regulations of the area and the image around the vehicle and the GPS data of the vehicle, which helps to avoid fines and other legal issues. Based on multi-source data, through feature fusion, the multi-source heterogeneous features (image, GPS, time, regulations) are integrated into a unified representation key technology, so that heterogeneous data can be processed by a machine learning model, and the horn decision logic in a complex scene can be captured, which significantly improves the accuracy, robustness and efficiency of the horn control system.

[0075] This application allows drivers to adjust the horn volume according to personal preferences and specific situations through control, which can enhance the overall driving experience and reduce unnecessary tension and conflict. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 1 is a flow chart of a method for controlling the volume of a car horn provided by one or more embodiments of the present invention.

[0077] Figure 2 1 is a structural diagram of a device for controlling the volume of a car horn provided by one or more embodiments of the present invention.

[0078] Figure 3 A schematic diagram of a method for controlling the volume of a car horn according to a specific embodiment of the present invention.

[0079] Figure 4 A block diagram of an electronic device structure for a method for controlling the volume of a car horn provided by one or more embodiments of the present invention. DETAILED DESCRIPTION

[0080] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0081] Figure 1 1 is a flow chart of a method for controlling the volume of a car horn provided by one or more embodiments of the present invention.

[0082] like Figure 1 The method for controlling the volume of a car horn shown includes:

[0083] Step S1, acquiring image data and location information around the vehicle;

[0084] Get the horn rules of the target area;

[0085] The acquired image data, location information, and target area horn rules are uploaded to the cloud. After matrix feature fusion, a vehicle horn label is generated.

[0086] Step S2, obtaining historical vehicle horn information and environmental data within the target area;

[0087] Generate a training set based on the historical vehicle horn information and environmental data in the target area;

[0088] Step S3, according to the siren label and training set training, based on the target detection algorithm of SPPNet, the siren characteristics and siren type of the current vehicle passing through the target area are obtained;

[0089] Step S4, according to the siren label and training set training based on ANN neural network algorithm, the siren probability of the vehicle passing through the target area is predicted;

[0090] Determine whether the siren probability value is greater than the preset siren probability threshold;

[0091] If greater, in response to the siren volume calculation, a list of predicted siren volume is generated;

[0092] Step S5, based on the vehicle siren probability and the predicted siren volume, according to the siren characteristics, siren type and siren volume list of the current vehicle passing through the target area, the siren volume target value is generated;

[0093] Step S6, according to the generated siren volume target value, output to the actuator.

[0094] In this embodiment, it includes:

[0095] The way of collecting image data around the vehicle includes collecting through the camera sensor arranged on the vehicle;

[0096] The way of obtaining vehicle position information includes obtaining the current vehicle position information through GPS;

[0097] The content of the label includes visual feature vector, space-time feature vector and siren label.

[0098] Further, it also includes:

[0099] After uploading the obtained image data, position information and target area siren rules to the cloud and fusing the matrix features, the feature vector is generated;

[0100] It includes: X=(FI,FG,FT,FR);

[0101] The image data extraction feature FI is processed using convolutional neural network CNN to extract the features of pedestrians, vehicles and obstacles;

[0102] The output feature vector FI=(npedestrians,nvehicles,nobstacles), where n represents the number of each category, pedestrians represents pedestrians, vehicles represents motor vehicles or non-motor vehicles, and obstacles represents obstacles.

[0103] GPS data feature extraction FG, using geocoding service to convert latitude and longitude to specific location information;

[0104] Output feature vector FG=(location, zone), where location represents the specific location name, and zone represents the type of area;

[0105] Time data feature extraction FT uses periodic encoding to convert time data to capture the periodic pattern of time change, and outputs feature vector FT=(sin(hour), cos(hour), sin(minute), cos(minute));

[0106] FT=(sin(hour), cos(hour), sin(minute), cos(minute)), where sin(hour) and cos(hour) represent the unit circle mapping hours to a two-dimensional plane, forming a periodic representation; sin(minute) and cos(minute) map minutes to a unit circle to capture the periodicity at the minute level;

[0107] Local siren regulation feature extraction FR, according to the location information to query the siren regulation database, output feature vector FR=(allowed_times, restricted_zones), where allowed_times represents the time period allowed by local siren regulations for vehicles to legally sound the siren, and restricted_zones represents the area restricted by local siren regulations for vehicles to sound the siren.

[0108] Further, including:

[0109] Predicting the list of siren volume of vehicles passing through the target area, including:

[0110] P(D|X)=σ(β0+β1XI+β2XG+β3XT+β4XR);

[0111] σ is the sigmoid function, ;

[0112] βi is the parameter of the model, i=0,1,2,3,4, where β0 is the intercept term;

[0113] XI, XG, XT, XR are features extracted from image data, GPS data, time data and local siren regulation features, respectively.

[0114] Further, including:

[0115] The generated siren volume target value, including:

[0116] V(X)=α0+α1XI+α2XG+α3XT+α4XR;

[0117] ai is the parameter of the model, i = 0, 1, 2, 3, 4, where a0 is the intercept term;

[0118] XI, XG, XT, XR are respectively features extracted from image data, GPS data, time data and local horn regulation features.

[0119] Specifically, in this embodiment, by aligning the features of different data sources, the image feature FI (such as the number of pedestrians) is an integer vector, the GPS feature FG (such as the type of area) is a classification variable, the time feature FT (such as sin / cos coding) is a continuous value, and the local horn regulation feature extraction (FR) maps heterogeneous data to the same semantic space through fusion, for example, the image feature FI extracts the number of pedestrians, the GPS feature FG converts the type of area, and the local horn regulation feature extraction (FR) queries the horn prohibition period, so that different data sources represent the decision elements in the same dimension. It is solved that a single feature cannot clearly define whether the horn needs to be sounded,

[0120] For example, pedestrians are identified, but the distance is far; it is in a business district, but it is not a peak period.

[0121] Through the logistic regression model, the probability is calculated, and the threshold (such as 0.5) is set to convert the continuous value into a binary classification label (horn / no horn), and the decision boundary is clear.

[0122] For example: embed the regulation feature (FR) into the classification model, and automatically learn the compliance decision logic through training, for example: when FG=“hospital” and FR=“horn prohibition”, the model tends to output the “no horn” label; school area (FG) + school time (FT) + a large number of pedestrians (FI) → correctly output the “horn” label.

[0123] When the FR feature is updated (such as temporary traffic control), the model can immediately adjust the output label, and the response time is <1 second.

[0124] In the image blur (add Gaussian noise) scene, the F1 score of the fusion model is 18% higher than that of the single feature model.

[0125] When a sensor fails (such as GPS signal loss), the model can still maintain an accuracy of more than 70% through other features.

[0126] When there are 3 vehicles honking within 100 meters, the cloud suppresses the unnecessary honking of subsequent vehicles.

[0127] Among them, the target detection algorithm training process of SPPNet mainly includes:

[0128] Collect vehicle-mounted camera images, GPS spatio-temporal data, and regional regulation data;

[0129] Preprocessing: Image normalization, spatio-temporal data encoding into feature vectors;

[0130] Convolutional layers: Extract multi-scale feature maps (edge, texture, semantic information);

[0131] SPP layer: Divide feature maps into 1x1, 2x2, 4x4 sub-regions, pool and concatenate fixed-length vectors.

[0132] Through multi-task model construction, existence detection branch: judge whether to sound the horn (binary classification);

[0133] Type classification branch: identify the type of the horn (emergency / reminder / violation); bounding box regression branch: locate the position of the horn target.

[0134] Labeled data: manually labeled image scenes, spatio-temporal context, horn labels;

[0135] Loss function: jointly optimize classification loss (cross-entropy) and positioning loss (smooth L1);

[0136] Data augmentation: image flipping / brightness adjustment / noise addition, to improve generalization ability.

[0137] Through input of real-time multi-modal data, model training, output: horn existence probability, type classification result, spatial position coordinates.

[0138] Horn signs include: warning pedestrians / non-motor vehicle labels;

[0139] Triggering scenarios: detection of pedestrians crossing the road, non-motor vehicles occupying the road, etc. (through image recognition of pedestrian / non-motor vehicle targets).

[0140] Features: presence of pedestrian / non-motor vehicle targets in image + GPS near pedestrian crossing + vehicle speed > 20 km / h.

[0141] Line-of-sight blind area warning label;

[0142] Triggering scenarios: blind road sections such as curves, tunnel entrances, steep slopes, etc. (through image recognition of curve signs / GPS matching terrain data).

[0143] Features: presence of curve signs in image + GPS slope > 5° + high accident rate area.

[0144] Emergency escape label;

[0145] Triggering scenarios: sudden obstacles, sudden braking of preceding vehicles, etc. (combined with radar data and driving behavior data).

[0146] Feature: Brake distance < safety threshold + steering wheel angle > 30° in a short time.

[0147] Horn regulation allowed label;

[0148] Trigger scenario: Non-banned horn area and compliance with local horn regulations (e.g. short honking allowed at night in suburban areas).

[0149] Feature: GPS area non-banned horn area + time non-night quiet period (e.g. 6:00-22:00).

[0150] Volume over-limit label;

[0151] Definition: Horn volume exceeds area regulation limit (e.g. urban area night limit ≤55dB).

[0152] Feature: Measured volume > regulation threshold + time period is night (22:00-6:00).

[0153] Safety volume label;

[0154] Definition: Volume within the range allowed by regulations and meets the needs of the scene (e.g. 60-80dB in suburban areas during the day).

[0155] Feature: Volume [regulation lower limit, regulation upper limit] + scene non-quiet sensitive area.

[0156] Through the above scheme, the technical problem of how to convert complex multi-source information into executable decision instructions while ensuring safety, compliance and real-time is solved. Through solving the data semantic gap, decision ambiguity and regulation adaptation difficulty, the technology enables the intelligent driving system to make rational horn decisions in various scenarios. The cloud collaborative feature fusion architecture will further improve decision accuracy and promote the development of intelligent transportation systems towards a safer and more efficient direction.

[0157] Wherein, when the vehicle is running, real-time data is input into the trained ANN model, and through the feature extraction and pattern recognition ability, the horn probability of the vehicle in the current scene is predicted. If the horn probability exceeds the preset threshold, the system further responds to the horn volume calculation, combines the horn features, types, etc. to generate a reasonable horn volume target value, and outputs to the execution mechanism to adjust the horn operation.

[0158] Wherein, the simple linear regression model of horn volume control is as follows:

[0159] V(X)=α0+α1XI+α2XG+α3XT+α4XR;

[0160] αi is the parameter of the model, i=0,1,2,3,4, where α0 is the intercept term.

[0161] XI, XG, XT, XR are features extracted from image data, GPS data, time data and local horn regulation characteristics respectively.

[0162] The vehicle terminal is embedded in the T-box and connected with the Internet of Vehicles cloud. When the horn volume recommendation result is obtained, it is input to the vehicle horn controller and finally output to the horn device to complete the horn.

[0163] In another embodiment of the present application, when passing through a complex intersection;

[0164] Input features:

[0165] FI (3 pedestrians are passing through the zebra crossing);

[0166] FG (intersection);

[0167] FT (morning and evening peak);

[0168] FR (no special restrictions);

[0169] Output label: horn (probability 0.85);

[0170] Effect: early warning to pedestrians to avoid rush accidents.

[0171] In the identification of special vehicles to avoid,

[0172] Input features:

[0173] FI (flashing warning lights behind the ambulance);

[0174] FG (urban trunk road);

[0175] FT (Monday 10:00);

[0176] FR (special vehicle priority rule);

[0177] Output label: no horn (probability 0.12);

[0178] Effect: avoid interference with the ambulance, and send the yield information through V2X.

[0179] In the identification of adverse weather scenarios,

[0180] Input features:

[0181] FI (low visibility due to rain and fog);

[0182] FG (mountainous winding road);

[0183] FT (night);

[0184] FR (no special rules);

[0185] Output label: Siren (probability 0.92);

[0186] Effect: Fill in the visual blind area and give early warning to the oncoming vehicle.

[0187] Figure 2 A structural block diagram of a device for controlling the volume of a car siren is provided by one or more embodiments of the present application.

[0188] As Figure 2 shown, the device for controlling the volume of a car siren includes a collection module, a scale module, a label module, an environmental data module, a training set module, a siren type module, a siren probability module, a threshold judgment module, a list generation module, a target value module, and an output execution module.

[0189] The collection module is used to obtain image data and position information around the vehicle.

[0190] The scale module is used to obtain siren rules for the target area.

[0191] The label module is used to upload the obtained image data, position information, and target area siren rules to the cloud, generate a vehicle siren label after matrix feature fusion.

[0192] The environmental data module is used to obtain historical vehicle siren information in the target area and environmental data in the target area.

[0193] The training set module is used to label and generate a training set based on the historical vehicle siren information in the target area and the environmental data in the target area.

[0194] The siren type module is used to obtain siren features and siren types of the current vehicle passing through the target area based on the siren label and training set training, and a target detection algorithm based on SPPNet.

[0195] The siren probability module is used to predict the siren probability of the vehicle passing through the target area based on the siren label and training set training, and an algorithm based on ANN neural network.

[0196] The threshold judgment module is used to determine whether the siren probability value is greater than a preset siren probability threshold.

[0197] The list generation module is used to generate a list of predicted siren volumes in response to siren volume calculation if the siren probability value is greater than the preset siren probability threshold.

[0198] The target value module is used to generate a siren volume target value based on the vehicle siren probability and the predicted siren volume, the siren features, the siren type, and the siren volume list of the current vehicle passing through the target area.

[0199] An output execution module is configured to output to an execution mechanism according to the generated target value of the horn sound volume.

[0200] It is worth noting that, although the system only discloses the collection module, the scale module, the label module, the environmental data module, the training set module, the horn type module, the horn probability module, the threshold judgment module, the list generation module, the target value module and the output execution module, but it does not mean that the device is limited to the above basic function modules, relatively, the meaning expressed by the present application is that on the basis of the above basic function modules, the person skilled in the art can add one or more function modules according to the prior art to form infinite embodiments or technical solutions, that is, the system is open rather than closed, and the protection scope of the present application claimed in the present embodiment cannot be limited to the above disclosed basic function modules.

[0201] Figure 3 The schematic diagram of the control method of the automobile horn sound volume in one specific embodiment of the present application.

[0202] In one specific embodiment, a control method of automobile horn sound volume is disclosed, and the operation process of the method is as follows:

[0203] After the vehicle is started, the system collects the data around the vehicle in real time by relying on the vehicle terminal supporting the camera device;

[0204] The vehicle terminal is embedded with a T-box, and supports access to the Internet through a 4G / 5G mode;

[0205] The vehicle terminal supports uploading image data, GPS data, time and other related data to the cloud through the T-box, and supports the horn sound volume control signal value obtained by the cloud computing to be returned to the vehicle terminal.

[0206] Figure 4 The electronic device structure block diagram of the control device of the automobile horn sound volume provided by one or more embodiments of the present application.

[0207] As Figure 4 shown, the present application provides an electronic device, which comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete the communication among each other through the communication bus.

[0208] The memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the control method of the automobile horn sound volume.

[0209] The application further provides a computer readable storage medium storing a computer program executable by an electronic device, which causes the electronic device to perform the steps of the method for controlling the volume of a car horn sound when the computer program is run on the electronic device.

[0210] The application further provides a vehicle comprising:

[0211] An electronic device for implementing the steps of the method for controlling the volume of a car horn sound.

[0212] A processor for running a program, which performs the steps of the method for controlling the volume of a car horn sound when the program is run on data output from the electronic device.

[0213] A storage medium for storing a program, which performs the steps of the method for controlling the volume of a car horn sound when the program is run on data output from the electronic device.

[0214] For the method embodiments, in order to simply describe, they are all described as a series of action combinations, but those skilled in the art should know that the embodiments of the application are not limited to the action sequence described, because according to the embodiments of the application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions involved are not necessarily essential for the embodiments of the application.

[0215] From the above description of the embodiments, those skilled in the art can clearly understand that the application can be implemented by means of software plus necessary universal hardware platforms. Based on such understanding, the technical solutions of the application can be embodied in the form of a software product, which can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of the application.

[0216] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the application.

Claims

1. A method for controlling the volume of a car horn, characterized in that: The method comprises: Acquire image data and location information around the vehicle; Get the horn rules of the target area; The acquired image data, location information, and target area horn rules are uploaded to the cloud. After matrix feature fusion, a vehicle horn label is generated. Obtain historical vehicle horn information and environmental data within the target area; Generate a training set based on the historical vehicle horn information and environmental data in the target area; According to the horn label and training set training, the target detection algorithm based on SPPNet is used to obtain the horn characteristics and horn type of the current vehicle passing through the target area; Training an ANN-based neural network algorithm based on the horn labels and the training set to predict the probability of vehicles passing through the target area honking their horns; Determine whether the whistle probability value is greater than a preset whistle probability threshold; If,is greater than,then, in response to the whistle volume calculation, a list of predicted whistle volumes is generated; Based on the vehicle honking probability and predicted honking volume, a target honking volume value is generated according to the honking characteristics, honking type, and honking volume list of the current vehicle passing through the target area; The generated whistle volume target value is output to the actuator.

2. The method for controlling the volume of a car horn according to claim 1, characterized in that: include: The method of collecting image data around the vehicle includes collecting data through a camera sensor installed on the vehicle; The method of obtaining the vehicle location information includes obtaining the current vehicle location information through GPS; The labeled content includes visual feature vectors, spatiotemporal feature vectors, and whistle labels.

3. The method for controlling the volume of a car horn according to claim 1, wherein: include: The acquired image data, location information, and target area horn rules are uploaded to the cloud and fused with matrix features to generate feature vectors. Includes: X=(FI,FG,FT,FR); Image data feature extraction FI, uses convolutional neural network CNN to process image data and extract features of pedestrians, vehicles and obstacles; Output feature vector FI=(npedestrians,nvehicles,nobstacles), where n represents the number of each category, pedestrians represents pedestrians, vehicles represents motor vehicles or non-motor vehicles, and obstacles represents obstacles; GPS data feature extraction FG, using geocoding service to convert longitude and latitude into specific location information; Output feature vector FG=(location,zone), where location represents the specific location name and zone represents the type of area; Temporal data feature extraction FT uses periodic coding to transform temporal data to capture the periodic pattern of temporal changes and outputs a feature vector; FT=(sin(hour), cos(hour), sin(minute), cos(minute)), where sin(hour) and cos(hour) represent the mapping of hours onto the unit circle on a two-dimensional plane, forming a periodic representation; sin(minute) and cos(minute) map minutes onto the unit circle, capturing minute-level periodicity. Extract the local horn regulations feature FR, query the horn regulations database based on the location information, and output the feature vector FR = (allowed_times, restricted_zones), where allowed_times represents the time period during which vehicles are allowed to legally honk according to the local horn regulations, and restricted_zones represents the areas where vehicles are restricted from honking according to the local horn regulations.

4. The method for controlling the volume of a car horn according to claim 1, wherein: A list of predicted horn volumes of vehicles passing through the target area, including: P(D∣X)=σ(β0+β1XI+β2XG+β3XT+β4XR); σ is the sigmoid function, ; βi is the parameter of the model, i=0,1,2,3,4, where β0 is the intercept term; XI, XG, XT, and XR are features extracted from image data, GPS data, time data, and local horn regulations, respectively.

5. The method for controlling the volume of a car horn according to claim 1, wherein: The generated whistle volume target value includes: V(X)=α0+α1XI+α2XG+α3XT+α4XR; αi is the parameter of the model, i=0,1,2,3,4, where α0 is the intercept term; XI, XG, XT, and XR are features extracted from image data, GPS data, time data, and local horn regulations, respectively.

6. The method for controlling the volume of a car horn according to claim 1, characterized in that ； The training process of the target detection algorithm based on SPPNet includes feature extraction of the training set, multi-task model construction, training optimization strategy and real-time inference process; The ANN-based recommendation algorithm training process includes data preprocessing, model construction, model training, and prediction result evaluation.

7. A device for controlling the volume of a car horn, characterized in that: The device comprises: An acquisition module, used to obtain image data and location information around the vehicle; The scale module is used to obtain the whistle rules of the target area; The labeling module is used to upload the acquired image data, location information, and target area horn rules to the cloud, and generate a vehicle horn label after matrix feature fusion; Environmental data module, used to obtain historical vehicle horn information and environmental data in the target area; The training set module is used to generate a training set based on the historical vehicle horn information and the environmental data in the target area; A horn type module is used to obtain the horn characteristics and horn type of the current vehicle passing through the target area based on the horn label and training set training, and the target detection algorithm based on SPPNet; A horn probability module is used to train an ANN-based algorithm based on the horn labels and a training set to predict the probability of honking the horn of vehicles passing through the target area; A threshold judgment module is used to judge whether the whistle probability value is greater than a preset whistle probability threshold; A list generation module is used to generate a list of predicted whistle volumes in response to the whistle volume calculation if , is greater than ; A target value module is used to generate a target horn volume value based on the vehicle horn probability and predicted horn volume, and according to the horn characteristics, horn type, and horn volume list of the current vehicle passing through the target area; The output execution module is used to output the generated whistle volume target value to the actuator.

8. An electronic device, characterized in that: include: A processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; The memory stores a computer program, and when the processor executes the computer program, the processor executes the steps of the method for controlling the volume of a car horn according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that It stores a computer program that can be executed by an electronic device. When the computer program runs on the electronic device, the electronic device executes the steps of the method for controlling the volume of a car horn according to any one of claims 1 to 6.

10. A vehicle, characterized in that: include: An electronic device for implementing the steps of the method for controlling the volume of a car horn according to any one of claims 1 to 6; a processor, wherein the processor runs a program, and when the program runs, the steps of the method for controlling the volume of a car horn according to any one of claims 1 to 6 are executed based on data outputted from the electronic device; A storage medium for storing a program, wherein when the program is run, the program executes the steps of the method for controlling the volume of a car horn according to any one of claims 1 to 6 for data output from an electronic device.

Images