Control method, control device, vehicle and computer readable storage medium
By obtaining vehicle environment data and window attachment information, determining the current window scene and performing corresponding operations, the problem that window clarity affects driving safety is solved, and accurate identification and effective cleaning of window status is achieved.
Patent Information
- Application Number
- CN202510164849.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-13
- Publication Date
- 2025-05-30
AI Technical Summary
The clarity of the vehicle window affects the driver's driving safety, and it is difficult for the prior art to effectively identify and deal with changes in the state of the vehicle window in different environments.
By obtaining the current environment data of the vehicle and the attachment information of the window, combining the multi-dimensional information to determine the current scene where the window is located, and controlling the vehicle to perform corresponding operations based on the scene, such as activating the wiper, window heating or dehumidification functions to clean the window.
It realizes accurate identification and effective cleaning of the window status, improves the clarity of the windows, and ensures the safety and comfort of the driver.
Smart Images
Figure CN120056915A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of vehicles, and more specifically, to a control method, a control device, a vehicle, and a non-volatile computer-readable storage medium. Background Art
[0002] The clarity of the vehicle window affects the viewing effect of the user looking outside. Especially for the front windshield, the driver needs to observe the road conditions through the front windshield so as to determine the driving strategy according to the road conditions. If the clarity of the front windshield is low, it will be difficult for the driver to accurately determine the road conditions, thus affecting the driving safety of the driver. Therefore, how to ensure a high clarity of the vehicle window has become an urgent problem to be solved. Summary of the Invention
[0003] Embodiments of the present application provide a control method, a control device, a vehicle, and a non-volatile computer-readable storage medium.
[0004] The control method of the embodiment of the present application is used for a vehicle, and the vehicle includes a vehicle window. The method includes: obtaining current environmental data of the vehicle and attachment information of the vehicle window to determine a current scene where the vehicle window is located; and controlling the vehicle to perform a corresponding operation based on the current scene to clean the vehicle window.
[0005] The control device of the embodiment of the present application is used for a vehicle, and the vehicle includes a vehicle window. The device includes a determination module and a cleaning module. The determination module is configured to obtain current environmental data of the vehicle and attachment information of the vehicle window to determine a current scene where the vehicle window is located. The cleaning module is configured to control the vehicle to perform a corresponding operation based on the current scene to clean the vehicle window.
[0006] The vehicle of the embodiment of the present application includes a vehicle window, a processor, a memory, and a computer program. The computer program is stored in the memory and executed by the processor. The computer program includes instructions for executing the control method. The control method is used for a vehicle, and the vehicle includes a vehicle window. The method includes: obtaining current environmental data of the vehicle and attachment information of the vehicle window to determine a current scene where the vehicle window is located; and controlling the vehicle to perform a corresponding operation based on the current scene to clean the vehicle window.
[0007] The non - volatile computer - readable storage medium of the embodiment of the present application includes a computer program. When the computer program is executed by a processor, the processor is caused to execute a control method. The control method is used for a vehicle, and the vehicle includes a window. The method includes: obtaining the current environmental data of the vehicle and the attachment information of the window to determine the current scene where the window is located; and controlling the vehicle to execute a corresponding operation based on the current scene to clean the window.
[0008] For the control method, control device, vehicle, and non - volatile computer - readable storage medium of the embodiment of the present application, it is possible to first preset the operations corresponding to each scene when the window can be effectively cleaned. Then, based on the current environmental data of the vehicle and the attachment information of the window, the current scene where the window is located is determined. The current scene can be used to determine the environmental situation and the attachment situation of the window. Finally, the operation corresponding to the current scene is executed to clean the window. Since the operation corresponds to the current scene, the window can be effectively cleaned after the operation is completed, and the clarity is relatively high. In this way, the present application combines multi - dimensional information to accurately determine the current scene, and then executes the operation corresponding to the current scene to improve the recognition accuracy of the current scene and clean the window in a targeted manner, thereby ensuring that the clarity of the window can be effectively improved.
[0009] Additional aspects and advantages of the embodiments of the present application will be given in part in the following description, become apparent in part from the following description, or be understood through the practice of the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The above - mentioned and / or additional aspects and advantages of the present application will become apparent and be readily understood from the description of the embodiments in conjunction with the following drawings, wherein:
[0011] Figure 1 is a flowchart of the control method of some embodiments of the present application;
[0012] Figure 2 is a flowchart of the control method of some embodiments of the present application;
[0013] Figure 3 is a flowchart of the control method of some embodiments of the present application;
[0014] Figure 4 is a flowchart of the control method of some embodiments of the present application;
[0015] Figure 5 is a flowchart of the control method of some embodiments of the present application;
[0016] Figure 6 is a flowchart of the control method of some embodiments of the present application;
[0017] Figure 7 is a schematic flow chart of a control method according to some embodiments of the present application;
[0018] Figure 8 is a schematic flow chart of a control method according to some embodiments of the present application;
[0019] Figure 9 is a schematic diagram of modules of a control device according to some embodiments of the present application;
[0020] Figure 10 is a schematic diagram of modules of a vehicle according to some embodiments of the present application;
[0021] Figure 11 is a schematic diagram of the connection state of a non - volatile computer - readable storage medium and a processor according to some embodiments of the present application. Specific Embodiments
[0022] The following details the embodiments of the present application. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the embodiments of the present application and should not be construed as a limitation of the embodiments of the present application.
[0023] The embodiments of the present application provide a control method, and the control method of the present application will be elaborated in detail below:
[0024] Please refer to Figure 1 , an embodiment of the present application provides a control method for a vehicle. The vehicle includes a window, and the control method includes:
[0025] Step 01: Obtain the current environmental data of the vehicle and the attachment information of the window to determine the current scene where the window is located;
[0026] Specifically, the current scene where the window is located refers to the specific environment or situation where the window is currently located, which includes the environment where the window is currently located and the attachment situation of the window. For example, the scene is a rainy day entering a tunnel scene, a sunny day entering a tunnel scene, or a scene where the window is contaminated, resulting in a blurred window. The attachment situation of the window includes whether there are attachments on the window and the type of attachments. Whether there are attachments on the window, such as whether there are sundries on the window, whether the window is blurred, frosted, or frozen. Obviously, these situations will cause a decrease in the clarity of the window. The type of attachments includes water, fog, frost, haze, or other sundries, such as paper. The cleaning measures corresponding to different types of attachments may be different. For example, starting the windshield wipers is required to clean rainwater, and heating the window is required to clean frost. The environmental situation where the window is located includes the temperature and humidity inside and outside the window, as well as the light intensity and light transmittance. The temperature and humidity inside and outside the window will also affect the clarity of the window. For example, fog may occur when the temperature difference or humidity difference between the inside and outside of the window is too large. Especially in a low-temperature environment, if the temperature difference or humidity difference between the inside and outside of the window is too large, fog will be caused. Attachments such as frost and fog will affect the light intensity and light transmittance.
[0027] In some embodiments, the current environmental data includes environmental data inside and outside the vehicle, such as at least one of the light transmittance of the window, the light intensity, the indoor temperature of the vehicle, the indoor humidity of the vehicle, the environmental temperature (i.e., the temperature of the environment where the vehicle is located), and the environmental humidity (i.e., the humidity of the environment where the vehicle is located). The light transmittance of the window refers to the percentage of the light flux passing through the window glass to the light flux incident on the window glass. The light transmittance of the window can be obtained through an optical sensor, which can be used to identify a decrease in light transmittance caused by fog, frost, or pollution. The light intensity describes the magnitude of the visible light radiation energy received per unit area, specifically referring to the light intensity of the external light after passing through the glass. The light intensity can be obtained according to an optical sensor, which can be used to identify a decrease in light intensity caused by fog, frost, or pollution. The indoor temperature and environmental temperature can be obtained according to the temperature sensor of the vehicle, and the indoor humidity and environmental humidity can be obtained through the humidity sensor of the vehicle, or a temperature and humidity sensor can be directly set to obtain the temperature and humidity simultaneously.
[0028] In some embodiments, the attachment information includes at least one of a captured image of the window and a thermal image of the window. The captured image of the window refers to a real-time image of the window directly captured by a vehicle camera device. The captured image can be used for image blurriness analysis to determine whether there is an attachment on the window, and at the same time, the type of the attachment can be determined according to the form of the attachment. The thermal image of the window can be obtained by a thermal imaging sensor. The thermal image of the window can be used to obtain the heat of the attachment to accurately distinguish between water mist and frost. Therefore, it is possible to determine whether there is an attachment on the window and the type of the attachment, that is, to determine the specific situation of the attachment. For example, it is possible to determine whether there is an attachment on the window according to the captured image, and when there is an attachment on the window, obtain the form of the attachment on the window, then obtain the heat of the attachment according to the thermal image of the vehicle, and then determine the type of the attachment based on the heat and the form.
[0029] The attachment information refers to the attachment information obtained at the current moment. It can be understood that the current environmental data and the attachment information of the vehicle can be used to determine the environmental information of the current state of the window and the specific situation of the attachment currently attached to the window. Therefore, it is possible to obtain the current environmental data and the attachment information of the vehicle to determine the current scene in which the vehicle is located, so as to facilitate determining whether the window needs to be cleaned in the scene where the vehicle is located. For example, multiple preset scenes can be preset according to the actual situation, and the environmental data and attachment conditions corresponding to each preset scene are determined. Then, according to the currently obtained current environmental data and attachment information, and the environmental data and attachment conditions corresponding to each preset scene, the current scene in which the vehicle is located is determined. For example, it can be determined according to the attachment information that there is rain attached to the window, and the environmental data includes the light intensity. If the light intensity suddenly drops a lot at a certain moment, it can be considered that the current scene is entering a tunnel on a rainy day.
[0030] In some embodiments, the in-vehicle host in the intelligent cockpit of the vehicle 100 includes a variety of sensors, which can collect information and transmit it to a system-on-chip (SoC). The air quality sensor is mainly used to monitor the air quality index (AQI) inside and outside the vehicle, including the concentrations of pollutants such as fine particulate matter (PM2.5), volatile organic compounds (VOCs), and carbon monoxide (CO). The thermal imaging sensor uses infrared technology to capture the heat difference emitted by an object and generate a thermal image. The optical sensor is used to collect the light transmittance of the glass. The vehicle camera device captures a real-time captured image of the window. The temperature and humidity sensor is mainly used to monitor the temperature and humidity inside and outside the vehicle in real time and analyze the influence of the temperature difference change on the window state.
[0031] The above sensors are closely connected through communication protocols and data fusion algorithms to form a highly integrated sensing system. The preprocessed data is encapsulated into data frames compliant with the CAN protocol. Among them, the CAN protocol (Controller Area Network) is a communication protocol used in embedded systems, mainly for high-speed data transmission between microcontrollers and devices. The data frames are transmitted to the CAN controller of the in-vehicle system through the CAN bus. The CAN driver in the in-vehicle system receives the data frames and parses the data of each sensor transmitted to obtain the current environmental data and attachment information.
[0032] In this way, it is possible to ensure high-frequency acquisition of real-time multi-dimensional environmental data and attachment information, providing reliable input for subsequent analysis and execution operations.
[0033] Step 02: Based on the current scenario, control the vehicle to perform corresponding operations to clean the window.
[0034] Specifically, the operations performed by the vehicle include at least one of starting the windshield wiper, starting the window heating function, or starting the dehumidification function. After starting the windshield wiper, the rainwater or contaminants on the window can be cleared, such as certain debris or water mist. Some vehicles have a window heating function. After starting the window heating function, the temperature of the window rises to eliminate frost or ice on the window. Some vehicles have a dehumidification function. For example, the air conditioner of the window has a dehumidification function. After the dehumidification function is eliminated, the fog on the window can be cleared to solve the fogging situation.
[0035] The operations corresponding to each scenario can be preset according to the problems that the window actually needs to be cleaned in each scenario, so as to achieve window cleaning in a targeted manner and ensure that the clarity of the window is relatively high after the corresponding operations are completed. If the window does not need to be cleaned in a certain scenario, then the corresponding operation can be for the vehicle to maintain its current state. In this way, the mapping relationship between each scenario and its corresponding operation can be preset, and at this time, it can also be set as a mapping table to facilitate querying the operation corresponding to the current scenario. Therefore, after determining the current scenario, the operation corresponding to the current scenario can be executed according to the preset mapping relationship between the scenario and the operation, so as to improve the clarity of the window.
[0036] In this way, the vehicle can accurately determine the current scenario where the window is located based on multi-dimensional information, realizing precise identification of the state change of the window. And whenever a scenario change is detected, this application will automatically re-acquire and analyze the latest data to ensure the timeliness and accuracy of subsequent operations, thereby realizing intelligent management and optimization of the in-vehicle environment, and further enabling the vehicle to flexibly respond to various external condition changes and always maintain the best driving experience.
[0037] For example, when the vehicle detects an attachment on the window, the type of the attachment can be determined in combination with the thermal image to determine the current scene where the window is located, so as to facilitate determining the subsequent execution strategy according to the current scene. For example, when it is determined that the attachment is water based on the camera image and the thermal image, the current scene can be considered as a rainy day scene, and the windshield wiper can be activated at this time. When it is determined that the attachment is fog, the current scene can be considered as a foggy scene, and the windshield wiper can be activated or the dehumidification function can be activated. When it is determined that the attachment is frost, the current scene can be considered as a frosting scene, and the window heating function can be activated.
[0038] In the prior art, there are techniques that simply use camera images to determine the execution strategy. It is difficult to accurately determine the type of the attachment solely by using camera images, making it difficult to accurately determine the current scene where the window is located. For example, a frosting scene may be misidentified as a rainy day scene, resulting in the inability to quickly melt the frost on the window. However, the present application can accurately identify the current scene based on multi-dimensional information, so that after performing the operation corresponding to the current scene, the cleaning effect of the window is better.
[0039] At the same time, the present application determines the operations to be performed according to the current scene. Therefore, the scene can be refined based on the actual usage situation, and corresponding scenes can be set for complex situations, and then the operations corresponding to each scene can be set to ensure that the operations performed by the vehicle can highly match the current scene, so as to ensure that the clarity of the window is relatively high after performing the corresponding operations.
[0040] For example, after the vehicle enters the tunnel, the window is likely to fog up. The scene of entering the tunnel can be subdivided into entering the tunnel on a rainy day and entering the tunnel on a sunny day. In the case of entering the tunnel on a rainy day, since the vehicle itself is already activating the windshield wiper, maintaining the activation of the windshield wiper at this time can avoid the obstruction of the fog, so the operation corresponding to entering the tunnel on a rainy day is to continuously activate the windshield wiper. In the case of entering the tunnel on a sunny day, the corresponding operation can be to activate the windshield wiper to ensure that the fog can be quickly cleared after it appears, or to activate the dehumidification function to prevent the fog from appearing in advance. For another example, the foggy scene can be subdivided into a high-humidity foggy scene and a low-humidity foggy scene. The operation corresponding to the low-humidity foggy scene is to activate either the windshield wiper or the dehumidification function, and the operation corresponding to the high-humidity foggy scene is to activate both the windshield wiper and the dehumidification function.
[0041] In addition, performing corresponding operations based on the scene can also predict potential fogging or frosting situations. For example, after determining that the current scene is the scene of entering the tunnel, it can be determined that the probability of the window fogging up is relatively high. Then, the windshield wiper or the dehumidification function can be directly activated at this time to prevent the fog from generating and obstructing the user's line of sight. Therefore, performing corresponding operations based on the scene can predict potential risks and pre-adjust the operation of the device to avoid obstructing the user's line of sight, thereby improving the user's experience.
[0042] Thus, the present application has the ability to adapt to scenarios. In different environments such as tunnels, highways, and urban roads, it can dynamically adjust the control strategy of vehicle operations to specifically clean the vehicle window, thereby ensuring the best cleaning effect of the vehicle window.
[0043] The control method of the embodiment of the present application can first preset the operations corresponding to each scenario when the vehicle window can be effectively cleaned. Then, based on the current environmental data of the vehicle and the attachment information of the vehicle window, the current scenario where the vehicle window is located is determined. The current scenario can be used to determine the environmental situation and the attachment situation of the vehicle window. Finally, the operation corresponding to the current scenario is executed to clean the vehicle window. Since the operation corresponds to the current scenario, the vehicle window can be effectively cleaned and has a high clarity after the operation is completed. Thus, the present application combines multi-dimensional information to accurately determine the current scenario, and then executes the operation corresponding to the current scenario to improve the recognition accuracy of the current scenario and specifically clean the vehicle window, thereby ensuring that the clarity of the vehicle window can be effectively improved.
[0044] Please refer to Figure 2 , in some embodiments, step 01: Obtain the current environmental data of the vehicle and the attachment information of the vehicle window to determine the current scenario where the vehicle window is located, including:
[0045] Step 011: Obtain the current environmental data and the attachment information to determine the probability that the vehicle window is currently in each preset scenario;
[0046] Step 012: Determine the preset scenario with the highest corresponding probability as the current scenario.
[0047] Specifically, multiple preset scenarios can be preset in advance. Each preset scenario includes its corresponding environment and attachment information, and at least one of the environmental and attachment situations corresponding to each preset scenario is different. Therefore, the specific situations corresponding to each preset scenario are different.
[0048] Therefore, the current environmental data and the attachment information can be obtained, and then based on these two types of data and the environmental and attachment situations corresponding to different preset scenarios, the probability that the vehicle window is currently in each preset scenario is determined. For example, the corresponding features can be extracted from the current environmental data and the attachment information, and then combined with the environmental and attachment situations corresponding to different preset scenarios to determine the probability that the vehicle window is currently in each preset scenario. Then, the preset scenario with the highest corresponding probability is determined as the current scenario.
[0049] Thus, the scenario in which the vehicle window is currently most likely to be located, that is, the current scenario, can be accurately determined based on the current environmental data and the attachment information, thereby facilitating subsequent execution of the corresponding operation according to the current scenario, and further helping to ensure the clarity of the vehicle window.
[0050] Please refer toFigure 3 , in some embodiments, step 011: Obtain the current environmental data and attachment information to determine the probabilities of the window being in each preset scenario, including:
[0051] Step 0111: Obtain the current environmental data and attachment information;
[0052] Step 0112: Process the current environmental data and attachment information based on the neural network model to determine the probabilities of the window being in each preset scenario.
[0053] Specifically, the neural network model, especially the deep learning model, automatically discovers and extracts the features in the input data by learning the representation form of the data. Then the neural network model can classify the data into different categories according to the extracted features. Therefore, the historical environmental data and historical attachment information of each preset scenario can be obtained, and then the neural network model can be trained based on the historical environmental data and historical attachment information to obtain a neural network model that can well determine the probabilities of the window being in each preset scenario according to the current environmental data and attachment information.
[0054] Then, the current environmental data and attachment information can be obtained according to various sensors or camera devices of the vehicle. Then, the current environmental data and attachment information are processed based on the neural network model to determine the probabilities of the window being in each preset scenario.
[0055] For example, the neural network model includes a data processing model and a preset algorithm. The current environmental data and attachment information can be processed based on the corresponding data processing model to obtain environmental data features and attachment information features. That is, the data processing model corresponding to the current environmental data processes the current environmental data to obtain environmental data features. Then, the attachment information is processed according to the data processing model corresponding to the attachment information to obtain attachment information features. Then, the environmental data features and attachment information features are processed based on the preset algorithm to determine the probabilities of the window being in each preset scenario.
[0056] Among them, the data processing models corresponding to different data types in the current environmental data can be the same or different. Similarly, the data processing models corresponding to different data in the attachment information can be the same or different. The data processing models corresponding to different data can be determined according to the data processing models suitable for different data.
[0057] For example, for the information obtained by the imaging device and the thermal imaging sensor (i.e., the imaging image and the thermal image), a Convolutional Neural Network (CNN) model is used for feature extraction. The Convolutional Neural Network is a deep learning model that extracts local features through convolutional layers, reduces the data dimension through pooling layers, and finally performs classification or regression through fully connected layers. It can automatically learn features such as edges, shapes, and textures in images and is widely used in tasks such as image recognition and object detection. It performs excellently in image processing and can effectively capture local features. That is, the data processing models for the imaging image of the window and the thermal image of the window are both Convolutional Neural Networks.
[0058] For another example, for the data provided by the optical sensor, the temperature and humidity sensor, and the air quality sensor, a Transformer model is selected for feature extraction. The Transformer model is a deep learning architecture for Natural Language Processing (NLP) and other sequential data tasks. This model is particularly good at processing sequential data and can effectively capture long-range dependencies, making it very suitable for processing complex data structures such as text and time series. That is, the data processing models for the indoor temperature of the vehicle, the indoor humidity of the vehicle, the ambient temperature and humidity, the light transmittance of the window, and the light intensity are all Transformer models.
[0059] Then, based on a preset algorithm, the environmental data features and the attachment information features are processed to determine the probabilities of the window being in each preset scenario currently. For example, the preset algorithm is the softmax function. The softmax function, also known as the normalization function, is a generalization of the logistic function. It can "compress" a K-dimensional vector z containing arbitrary real numbers into another K-dimensional real vector σ(z) such that the range of each element is between (0, 1) and the sum of all elements is 1.
[0060] It should be noted that determining the current scenario based on the preset scenario with the maximum probability is actually also processed by the neural network model. The current environmental data and attachment information are the input data of the neural network model, and the output data of the neural network model is the current scenario.
[0061] In this way, the neural network model can accurately convert the current environmental data and attachment information into the probabilities of the window being in each preset scenario currently. Subsequently, the preset scenario with the maximum probability can be used as the prediction result, that is, as the current scenario, thus ensuring the scientificity and accuracy of the classification decision and providing a reliable basis for subsequent intelligent control.
[0062] Subsequent determination of the operation corresponding to the current scenario based on the current scenario can also be performed using a neural network model. After determining the current scenario, the current scenario is input into the neural network model to obtain the operation corresponding to the current scenario. Alternatively, a mapping table can be set up, which includes each preset scenario and its corresponding operation. After determining the current scenario, the operation corresponding to the current scenario can be directly obtained according to the mapping table.
[0063] Please refer to Figure 4 , in some embodiments, the control method further includes:
[0064] Step 03: In the case where the user operation is different from the operation corresponding to the current scenario, train the cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation. The cloud model corresponds to the neural network model;
[0065] Step 04: Update the neural network model based on the cloud model.
[0066] Specifically, the neural network model can be deployed on the vehicle side and is used to determine the current scenario and the operation corresponding to the current scenario. After the current environmental data and the attachment information of the window are input into the neural network model, the neural network model processes the current environmental data and the attachment information of the window to obtain the current scenario. The operation corresponding to the current scenario is preset and fixed. Then, the vehicle can be controlled to execute the corresponding operation according to the current scenario output by the neural network model.
[0067] The neural network model can be updated and optimized in real time according to the user operation. The user operation refers to the operation of the user on the devices related to cleaning the window in the vehicle, such as the start or stop of the windshield wiper, the window heating function, and the dehumidification function by the user. After the vehicle executes the corresponding operation based on the current scenario, the user can still operate the devices related to cleaning the window in the vehicle. In the case where the user operation is different from the operation corresponding to the current scenario, it can be considered that the current scenario output by the neural network model is inappropriate and does not meet the user's requirements for the actual scenario of the current window. At this time, it is necessary to match the actual scenario of the current window according to the user operation.
[0068] First, in the case where the user operation is different from the operation corresponding to the current scenario, record the user operation to form an operation record, such as recording that the user starts the windshield wiper or turns off the dehumidification function. At the same time, record the environmental data and the attachment information when the user operation is executed to obtain the environmental data corresponding to the user operation and the attachment information corresponding to the user operation.
[0069] Meanwhile, a cloud model with the same parameters and structure as the neural network model deployed on the vehicle can be set up in the cloud. Then, based on the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation, the cloud model is trained. At this time, the actual scene where the current window is located can be inferred from the user operation, and then the actual scene where the current window is located obtained by inference, as well as the environmental data corresponding to the user operation and the attachment information corresponding to the user operation are used as the training set to train the cloud model, so as to improve the recognition accuracy of the cloud model for the current scene and ensure that the recognition ability of the cloud model for the scene matches the user's needs. Then, the neural network model is updated based on the cloud model, that is, each parameter of the neural network model is updated to the parameter of the cloud model, so as to improve the recognition accuracy of the neural network model for the current scene and ensure that the recognition ability of the neural network model for the scene matches the user's needs, so that when the control method of this application is executed based on the neural network model later, the operations performed by the vehicle can match the user's needs.
[0070] In this way, a cloud model corresponding to the neural network model can be deployed in the cloud, and the cloud model can be trained specifically using the user's real-time operations. Then, the neural network model is updated using the cloud model, so as to update the neural network model using historical data and the user's real-time operations, thereby ensuring the recognition accuracy of the neural network model deployed on the vehicle for the scene, and ensuring that the recognition ability of the neural network model for the scene matches the user's needs, and being able to better adapt to the user's needs in different scenarios. Furthermore, it is beneficial for this application to adapt to various complex environmental conditions and improve driving safety and user comfort.
[0071] In addition, the setting of the cloud model is also beneficial for technicians to update the neural network model. Technicians can also train the cloud model at any time. For example, when technicians find certain errors, they can correct the neural network model deployed on the vehicle by updating the parameters of the cloud model. Or, technicians can also collect data of cloud models corresponding to a large number of different vehicles, and then train a cloud model that can meet the requirements of most people, and update the neural network model based on this cloud model.
[0072] Please refer to Figure 5 , in some embodiments, step 03: Training the cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation includes:
[0073] Step 031: When the user operation is not an incorrect operation, train the cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation.
[0074] Specifically, in real life, a user may accidentally trigger the device related to cleaning the vehicle window in a vehicle. For example, accidentally trigger the dehumidification function. Therefore, it can be determined whether the user operation is a non-accidental operation. For example, when the duration of the user operation is greater than a preset duration threshold, it is determined that the user operation is a non-accidental operation. The preset duration threshold is the shortest duration for which the trigger state is maintained when the user actually needs to trigger the device related to cleaning the vehicle window. Once the duration of the user operation is greater than the preset duration threshold, it means that the user actually needs to trigger the device related to cleaning the vehicle window. When it is determined that the user operation is a non-accidental operation, a cloud model is trained based on the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation.
[0075] It can be understood that if a user quickly cancels a certain operation within a short period of time after performing it, this behavior is regarded as accidental touch and is not included in the dataset for model training. In this way, it is possible to avoid model training deviation caused by accidental operations, ensure the quality of training data and the accuracy of the model, thereby ensuring the training accuracy of the cloud model.
[0076] Please refer to Figure 6 , in some embodiments, step 04: updating the neural network model based on the cloud model, includes:
[0077] Step 041: When the difference between the parameters of the cloud model and the neural network model is greater than a preset difference threshold, update the neural network model based on the cloud model.
[0078] Specifically, the preset difference threshold is the minimum value of the difference between the parameters of the cloud model and the neural network model when the gap between the output results of the cloud model and the neural network model affects the user experience. The difference between the parameters of the cloud model and the neural network model can be determined. For example, subtract the corresponding parameters in the cloud model and the neural network model to obtain the corresponding difference. Obtain the square of each difference, and use the sum of the squares of each difference as the difference between the parameters of the cloud model and the neural network model. When the difference between the parameters of the cloud model and the neural network model is greater than the preset difference threshold, it can be considered that the output result of the current neural network model has a large gap with the user operation preference and will affect the user experience. At this time, the neural network model is updated based on the cloud model.
[0079] In this way, on the premise of ensuring the output result of the neural network model and the user experience, the number of unnecessary model updates can be effectively reduced, thereby saving resources such as network bandwidth and improving the overall efficiency of the system.
[0080] In summary, the present application records the user's manual operation preferences and continuously optimizes the response in combination with machine learning technology, enabling the parameters of the neural network model to automatically adjust operations according to the user's habits during use, providing a more intelligent and comfortable user experience. In addition, it can effectively avoid unnecessary activation of wipers and air conditioners, achieving energy conservation and consumption reduction while improving convenience, and bringing higher usage efficiency to users.
[0081] Please refer to Figure 7 , in some embodiments, step 02: controlling the vehicle to perform corresponding operations based on the current scenario, including:
[0082] Step 021: When the current scenario belongs to the first type of scenario, controlling the vehicle to start the wiper;
[0083] Step 022: When the current scenario belongs to the second type of scenario, controlling the vehicle to start the window heating function;
[0084] Step 023: When the current scenario belongs to the third type of scenario, controlling the vehicle to start the dehumidification function, where the current scenario belongs to at least one of the first type of scenario, the second type of scenario, and the third type of scenario.
[0085] Specifically, the first type of scenario is a scenario where the window can be effectively cleaned after starting the wiper. For example, the first type of scenario includes at least scenarios where there are sundries on the window, driving on a normal road on a rainy day, entering a tunnel on a rainy day, or a foggy scenario. It should be noted that the first type of scenario mainly refers to the window equipped with a corresponding wiper, that is, only when the window is equipped with a corresponding wiper, such as when the window is the front windshield or the rear windshield, the current scenario can belong to the first type of scenario. The second type of scenario is a scenario where the window can be effectively cleaned after the vehicle starts the window heating function. For example, the second type of scenario includes at least a frosting scenario. For example, the third type of scenario includes at least scenarios where entering a tunnel on a sunny day or a foggy scenario. The third type of scenario is a scenario where the window can be effectively cleaned after the vehicle starts the dehumidification function.
[0086] In the same scenario, there may be multiple operations that can effectively clean the window. For example, in a foggy scenario, starting the wiper and starting the dehumidification function can both eliminate the fog. The current scenario can belong to at least one of the first type of scenario, the second type of scenario, and the third type of scenario, that is, the measures corresponding to the current scenario include at least one of starting the wiper, starting the window heating function, and starting the dehumidification function. For example, a foggy scenario can belong to both the first type of scenario and the third type of scenario at the same time. Therefore, when the current scenario is a foggy scenario, the corresponding operations of the vehicle include starting the wiper and starting the dehumidification function.
[0087] Among them, when the current scene belongs to multiple scene types, the execution schemes of the corresponding multiple operations, such as the execution order and the execution conditions, can be set according to the needs. For example, when the current scene is a foggy scene, the corresponding operations of the vehicle include starting the wipers and starting the dehumidification function. At this time, the dehumidification function can be set to start first. After the dehumidification function has been running for a period of time, if the fog can no longer be detected, the wipers can be turned off and the dehumidification function can be kept started. If the dehumidification function is still detected after running for a period of time, the wipers are started to improve the ability to eliminate the fog. Alternatively, the dehumidification function and the wipers can be started at the same time to ensure that the fog can be cleared quickly.
[0088] In order to better illustrate the implementation methods of the present application, the following examples are given for different scenarios:
[0089] Scenario 1: Entering a tunnel on a rainy day, there is a large temperature difference between the inside and outside of the car, and the windows quickly fog up.
[0090] Linked sensor: optical sensor + humidity sensor + camera device
[0091] process:
[0092] When the vehicle is driving in the rain, the wipers are turned on and the optical sensor and humidity sensor continuously detect the external environment.
[0093] When the vehicle enters the tunnel, the light changes and the transmittance is temporarily restored, but the humidity sensor still detects that the outside humidity is high and the camera device continues to identify raindrops.
[0094] At this time, the system determines that it is still raining outside, so the wipers remain on, on the one hand to continue to clear the rain on the windows, and on the other hand to clear the fog when it is generated, thereby avoiding the situation where the wipers are accidentally turned off due to changes in light in the tunnel, causing the windows to be covered with fog, affecting the user's line of sight, and causing a safety accident.
[0095] Scenario 2: The temperature difference between inside and outside the car causes fogging
[0096] Linked sensor: temperature sensor + humidity sensor + optical sensor
[0097] process:
[0098] When the vehicle is driving in cold weather, the temperature sensor detects a large temperature difference between the inside and outside of the vehicle, and the optical sensor begins to detect a decrease in the transparency of the window glass.
[0099] The humidity sensor detects an increase in humidity inside the vehicle, which means the inside windows may be fogging up.
[0100] The system activates the air conditioner's defog function to prevent the wipers from starting by mistake, and adjusts the air conditioner outlet to the window position.
[0101] Scenario 3: Haze weather
[0102] Linked sensors: Air quality sensor + Optical sensor + Camera device
[0103] Process:
[0104] When the air quality sensor detects an increase in the external PM2.5 concentration, it indicates that the environment is haze weather, and the optical sensor also detects a decrease in the light transmittance of the vehicle window.
[0105] The camera device does not recognize raindrops or water mist, and the system determines that the current is haze weather rather than rainfall.
[0106] The system turns off the external circulation, turns on the internal circulation and the air purification system, and avoids starting the windshield wiper.
[0107] In summary, through the integration of multiple sensors, this application can monitor the external environment of the vehicle, the light transmittance of the vehicle window, and the changes in temperature and humidity in real time, effectively identify the fogging, frosting, or icing conditions of the vehicle window, and realize the linkage control of the windshield wiper, window heating, and dehumidification functions, so as to ensure a clear view for the driver under various complex weather conditions and guarantee driving safety. Therefore, this application can solve the problem that the traditional windshield wiper control system cannot accurately identify and respond to the changes in the window state in tunnels, high-humidity environments, and rapid temperature difference changes.
[0108] In addition, the system can achieve precise control, effectively distinguish different conditions such as rain, fog, and frost by using multi-dimensional data fusion technology, and avoid the frequent start of the windshield wiper due to misidentification. At the same time, in tunnels or environments with large temperature differences, the system can automatically judge and start the defogging and heating functions, thus eliminating the hidden danger of blurred vision for the driver and greatly improving driving safety.
[0109] Please refer to Figure 8 , in some embodiments, the current environmental data further includes air quality, and the air quality is used to determine whether the current scenario is haze weather. The control method further includes:
[0110] Step 05: In the case where the current scenario is haze weather, turn off the external circulation, turn on the internal circulation, and turn on the air purification system;
[0111] Step 06: Control the windshield wiper of the vehicle to turn off.
[0112] Specifically, this application can control the start and stop of the windshield wiper, window heating function, and dehumidification function. Therefore, some scenarios that are prone to mis-triggering the windshield wiper, window heating function, or dehumidification function can also be specifically set, and the windshield wiper, window heating function, or dehumidification function can be controlled to stop running. For example, in haze weather, it is possible to trigger the start of the windshield wiper, but obviously, the clarity of the vehicle window cannot be improved after the windshield wiper starts. Therefore, in haze weather, the windshield wiper of the vehicle can be controlled to turn off.
[0113] The current environmental data and attachment information can also be used to determine whether the current scene is a haze weather. For example, the vehicle is also equipped with an air quality sensor, which is mainly used to monitor the air quality index (AQI) of the interior and exterior environments of the vehicle, including the concentrations of pollutants such as fine particulate matter (PM2.5), volatile organic compounds (VOCs), and carbon monoxide (CO), that is, to detect the air quality of the interior and exterior environments of the vehicle. Therefore, the current environmental data also includes air quality.
[0114] When the concentration of PM2.5 in the outside world rises to exceed the concentration threshold, it can be considered that there is more haze in the current environment. At this time, an optical sensor (such as a light transmittance sensor) can also detect a decrease in the light transmittance of the window. At the same time, based on the captured image, no raindrops or water mist are recognized. Therefore, it can be determined that the current scene is a haze weather rather than rainfall. Or, the present application can also use an optical sensor for multi-band optical detection (such as ultraviolet rays, infrared rays) to distinguish pollution particles and water droplets, improve the accurate recognition rate in haze and rainy days, and avoid unnecessary activation of the windshield wipers.
[0115] When the current scene is a haze scene, the external circulation can be turned off, the internal circulation can be turned on, and the air purification system can be turned on to prevent external pollutants from entering the vehicle interior, thereby ensuring the air quality inside the vehicle. At the same time, in haze weather, although there will be attachments on the window, obviously starting the windshield wipers at this time cannot effectively clean the attachments. Therefore, the windshield wipers of the vehicle can also be controlled to be turned off, that is, to avoid starting the windshield wipers, thereby reducing unnecessary activation of the windshield wipers and further reducing unnecessary energy consumption.
[0116] Please refer to Figure 9 , to facilitate better implementation of the control method of the embodiment of the present application, the embodiment of the present application also provides a control device 10. The control device 10 is used for a vehicle. The vehicle includes a window. The control device 10 may include a determination module 11 and a cleaning module 12. The determination module 11 is used to obtain the current environmental data of the vehicle and the attachment information of the window to determine the current scene where the window is located. The cleaning module 12 is used to control the vehicle to perform corresponding operations based on the current scene to clean the window.
[0117] Specifically, the determination module 11 is used to obtain the current environmental data and the attachment information to determine the probabilities of the window being in each preset scene currently; and determine the preset scene with the largest corresponding probability as the current scene.
[0118] Specifically, the determination module 11 is used to obtain the current environmental data and the attachment information; and process the current environmental data and the attachment information based on a neural network model to determine the probabilities of the window being in each preset scene currently.
[0119] The determination module 11 is specifically configured to process the current environmental data and the attachment information based on the corresponding data processing model to obtain the environmental data features and the attachment information features; and process the environmental data features and the attachment information features based on a preset algorithm to determine the probabilities of the vehicle window being in each preset scenario.
[0120] The cleaning module 12 is specifically configured to control the vehicle to start the windshield wiper when the current scenario belongs to the first type of scenario; control the vehicle to start the window heating function when the current scenario belongs to the second type of scenario; control the vehicle to start the dehumidification function when the current scenario belongs to the third type of scenario, and the current scenario belongs to at least one of the first type of scenario, the second type of scenario, and the third type of scenario.
[0121] The control device 10 further includes an updating module 13. The updating module 13 is configured to train a cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation when the user operation is different from the operation corresponding to the current scenario. The cloud model corresponds to the neural network model; and update the parameters of the neural network model based on the cloud model.
[0122] The updating module 13 is specifically configured to train a cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation when the user operation is not a misoperation.
[0123] The updating module 13 is specifically configured to determine that the user operation is not a misoperation when the duration of the user operation is greater than a preset duration threshold.
[0124] The updating module 13 is specifically configured to update the neural network model based on the cloud model when the difference between the parameters of the cloud model and the neural network model is greater than a preset difference threshold.
[0125] The control device 10 further includes a control module 14. The control module 14 is configured to turn off the external circulation, turn on the internal circulation, and turn on the air purification system when the current scenario is a haze weather; and turn off the windshield wiper of the vehicle.
[0126] Please refer to Figure 10 , the vehicle 100 according to the embodiment of the present application includes a vehicle window, a processor 20, a memory 30, and a computer program. The computer program is stored in the memory 30 and is executed by the processor 20. The computer program includes instructions for executing the control method according to any one of the above embodiments.
[0127] Please refer to Figure 11, Embodiments of the present application also provide a computer-readable storage medium 200, on which a computer program 210 is stored. When the computer program 210 is executed by a processor 220, the steps of the control method of any of the above embodiments are implemented. For the sake of brevity, they will not be elaborated here.
[0128] In the description of this specification, the descriptions with reference to terms such as "certain embodiments", "in an example", "exemplarily", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.
[0129] Any process or method description shown in the flowchart or described in other ways herein can be understood to represent a module, segment or part of code including one or more executable instructions for implementing a specific logical function or process. The scope of the preferred embodiments of the present application includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed. This should be understood by those skilled in the art to which the embodiments of the present application belong.
[0130] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present application.
Claims
1. A control method, characterized in that: For a vehicle, the vehicle comprising a vehicle window, the method comprising: Acquire current environment data of the vehicle and information about objects attached to the vehicle window to determine a current scene in which the vehicle window is located; The vehicle is controlled to perform a corresponding operation based on the current scene to clean the vehicle window.
2. The control method according to claim 1, characterized in that: The current environmental data includes at least one of the light transmittance of the vehicle window, the light intensity, the indoor temperature of the vehicle, the indoor humidity of the vehicle, the ambient temperature and the ambient humidity.
3. The control method according to claim 1, characterized in that: The attachment information includes at least one of a camera image of the vehicle window and a thermal image of the vehicle window.
4. The control method according to claim 1, characterized in that: The acquiring current environment data of the vehicle and the attachment information of the vehicle window to determine the current scene where the vehicle window is located includes: Acquiring the current environment data and the attachment information to determine the probability that the vehicle window is currently in each preset scene; Determine the preset scene with the highest corresponding probability as the current scene.
5. The control method according to claim 2, characterized in that: The acquiring the current environment data and the attachment information to determine the probability that the vehicle window is currently in each preset scene includes: Acquiring the current environment data and the attachment information; The current environment data and the attachment information are processed based on a neural network model to determine the probability that the vehicle window is currently in each preset scene.
6. The control method according to claim 5, characterized in that: The neural network model includes a data processing model and a preset algorithm, and the current environment data and the attachment information are processed based on the preset data processing model to determine the probability that the vehicle window is currently in each preset scene, including: Processing the current environment data and the attachment information based on the corresponding data processing model to obtain environment data features and attachment information features; The environmental data features and the attachment information features are processed based on the preset algorithm to determine the probability that the vehicle window is currently in each preset scene.
7. The control method according to claim 5, characterized in that: The method further comprises: In the case where the user operation is different from the operation corresponding to the current scene, a cloud model is trained according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation, and the cloud model corresponds to the neural network model; The parameters of the neural network model are updated based on the cloud model.
8. The control method according to claim 7, characterized in that: The training of the cloud model according to the user operation, the environmental data corresponding to the user operation, and the attachment information corresponding to the user operation includes: In the case that the user operation is not an erroneous operation, a cloud model is trained according to the user operation, environmental data corresponding to the user operation, and attachment information corresponding to the user operation.
9. The control method according to claim 8, characterized in that: The method comprises: When the duration of the user operation is greater than a preset duration threshold, it is determined that the user operation is not an erroneous operation.
10. The control method according to claim 9, characterized in that: The updating of the neural network model based on the cloud model includes: When the difference between the parameters of the cloud model and the neural network model is greater than a preset difference threshold, the neural network model is updated based on the cloud model.
11. The control method according to claim 1, characterized in that: The controlling the vehicle to perform a corresponding operation based on the current scenario includes: When the current scene belongs to the first type of scene, controlling the vehicle to start the windshield wipers; When the current scene belongs to the second type of scene, controlling the vehicle to start a window heating function; In the case that the current scene belongs to the third type of scene, the vehicle is controlled to start a dehumidification function, and the current scene belongs to at least one of the first type of scene, the second type of scene and the third type of scene.
12. The control method according to claim 11, characterized in that: The first category of scenes at least includes scenes where there are debris on the car window, scenes where the car is driving on a normal road in the rain, scenes where the car is entering a tunnel in the rain, or scenes where fog is generated; and / or The second type of scenes at least includes frosting scenes; and / or The third category of scenes at least includes a scene of entering a tunnel on a sunny day or a scene of fog.
13. The control method according to claim 1, characterized in that: The current environment data also includes air quality, and the air quality is used to determine whether the current scene is foggy or hazy. The method further includes: When the current scene is foggy and hazy, the external circulation is turned off, the internal circulation is turned on, and the air purification system is turned on; Controlling the windshield wipers of the vehicle to be turned off.
14. A control device, characterized in that: For a vehicle, the vehicle comprising a vehicle window, the device comprising: A determination module, used to obtain current environment data of the vehicle and information of objects attached to the vehicle window to determine a current scene in which the vehicle window is located; A cleaning module is used to control the vehicle to perform corresponding operations based on the current scene to clean the vehicle window.
15. A vehicle, characterized in that: include: Car windows; Processor, memory; and A computer program, wherein the computer program is stored in the memory and executed by the processor, and the computer program includes instructions for executing the control method according to any one of claims 1 to 13.
16. A non-volatile computer-readable storage medium containing a computer program, characterized in that: When the computer program is executed by a processor, the processor executes the control method according to any one of claims 1 to 13.