Range extender control method and vehicle

Abstract

The invention provides a range extender control method and a vehicle, and the method comprises the steps: obtaining a scene of the vehicle and a closed evaluation value, the scene of the vehicle being determined based on a front view image and a panoramic image of the vehicle, and the closed evaluation value being determined based on at least one of positioning data, radar data and the panoramic image of the vehicle; and determining a range extender control strategy of the vehicle based on at least one of the scene where the vehicle is located and the closed evaluation value, so as to control the range extender of the vehicle through the control strategy, determining a corresponding range extender control strategy for different scenes where the vehicle is located, the closed evaluation value or a combination of the scenes and the closed evaluation values, and further controlling the range extender. The method can meet actual requirements of users and improve user experience.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a range extender control method and a vehicle. Background Technology

[0002] With technological advancements, people are increasingly focusing on personalized control of vehicles, going beyond mere vehicle handling. The aim is to provide users with a superior driving experience by offering more thoughtful and personalized control strategies. Many personalized control strategies rely on recognizing the vehicle's environment. By providing corresponding control strategies based on different scenarios, user needs can be further met, enhancing the user experience.

[0003] In related technologies, the control of the operating state of the range extender in range-extended electric vehicles often only considers the vehicle's fuel situation or the user's choice of energy strategy, focusing more on the vehicle itself. This results in the actual selected range extender control strategy not meeting user needs, which may lead to a poor user experience. Summary of the Invention

[0004] This application provides a range extender control method and a vehicle to address the technical problem in the related art where the formulation and selection of range extender control strategies focuses more on the vehicle itself, resulting in the actual selected range extender control strategy failing to meet user needs and potentially leading to a poor user experience.

[0005] In a first aspect, this application provides a range extender control method, the method comprising: acquiring the vehicle's current scene and a closure evaluation value, wherein the vehicle's current scene is determined based on a front view image and a panoramic image of the vehicle, and the closure evaluation value is determined based on at least one of the vehicle's positioning data, radar data, and the panoramic image; determining a range extender control strategy for the vehicle based on at least one of the vehicle's current scene and the closure evaluation value, so as to control the vehicle's range extender through the control strategy.

[0006] Secondly, embodiments of this application also provide an electronic device, including: a memory storing a computer program thereon; and a processor for executing the computer program in the memory to implement the steps of the method described in any of the above embodiments.

[0007] Thirdly, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of the method described in any of the above embodiments.

[0008] Fourthly, embodiments of this application also provide a vehicle for performing the steps of the method described in any of the above embodiments.

[0009] Beneficial Effects: This application proposes a range extender control method and vehicle. The method obtains the vehicle's current scenario and closure evaluation value. The current scenario is determined based on the vehicle's front view image and panoramic image, and the closure evaluation value is determined based on at least one of the vehicle's positioning data, radar data, and panoramic image. Based on at least one of the current scenario and closure evaluation value, the method determines the vehicle's range extender control strategy to control the vehicle's range extender. The method determines the corresponding range extender control strategy for different vehicle scenarios and closure evaluation values, individually or in combination, thereby controlling the range extender. This approach can closely meet the actual needs of users and improve the user experience. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0011] In the attached diagram: Figure 1 This is a schematic diagram illustrating an application scenario of a range extender control method provided in an embodiment of this application. Figure 2 A schematic flowchart of a range extender control method provided in an embodiment of this application; Figure 3 A flowchart illustrating a method for determining the scene where a vehicle is located, provided in an embodiment of this application; Figure 4 A schematic flowchart illustrating a method for determining the vehicle's location according to an embodiment of this application; Figure 5 A schematic diagram of a specific process for providing a closure evaluation value according to an embodiment of this application; Figure 6 A schematic flowchart illustrating a specific embodiment of the range extender control method provided in this application; Figure 7 A schematic diagram of a range extender control device provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0013] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0014] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0015] The inventors discovered that the accuracy of vehicle scenario determination methods in related technologies is poor and cannot meet user needs. This leads to the implementation of vehicle control strategies based on scenarios with questionable accuracy, which may not satisfy user requirements and could even reduce user experience. Furthermore, in related technologies, the control of the range extender's operating status in range-extended vehicles often only considers the vehicle's fuel situation or the user's energy strategy selection, focusing more on the vehicle itself and neglecting other factors in the formulation and selection of range extender control strategies.

[0016] The inventors also discovered that enclosed environments (such as underground garages, tunnels, and parking lots inside buildings) have limited space and poor ventilation, making it difficult for the exhaust fumes generated after the range extender starts to disperse quickly, leading to a sharp deterioration in local air quality. This results in: Increased air pollution: Due to poor air circulation in enclosed spaces, harmful gases produced by range extenders cannot be expelled in time and tend to accumulate, posing a threat to human health, especially for people who are in such environments for extended periods (such as parking lot workers and nearby residents). Noise pollution: Enclosed spaces reflect and amplify noise, significantly increasing noise levels and affecting people's normal lives and work, potentially leading to hearing damage. Safety hazards: High concentrations of exhaust fumes may pose a fire or explosion risk, especially in environments containing flammable materials, increasing the safety risk.

[0017] In related technologies, range-extended electric vehicles often lack the ability to intelligently recognize enclosed environments, and are even less able to automatically adjust the operating status of the range extender according to actual scenarios. This deficiency leads to the following problems: Environmental pollution: Indiscriminately starting the range extender in enclosed environments may cause unnecessary air and noise pollution, potentially harming public health. Degraded user experience: Frequent noise and odors can affect the comfort of passengers, especially in underground parking garages or tunnels, where users may experience discomfort. Increased safety hazards: Prolonged operation of the range extender in enclosed spaces not only affects air quality but may also trigger safety accidents such as fires or explosions.

[0018] In view of this, a range extender control method based on intelligent driving system perception is proposed. Based on relevant data from the vehicle's sensors, the scene in which the vehicle is located can be determined, as well as the closure evaluation value of the environment in which the vehicle is located, providing a more accurate way to determine the vehicle's environment. Based on the closure evaluation value of the scene in which the vehicle is located determined by the aforementioned method and at least one of the scenes in which the vehicle is located, the control strategy of the range extender is determined, and then the range extender is controlled. This makes the control of the range extender not only focus on the state of the vehicle itself, but also take into account the surrounding environment. This can avoid unnecessary environmental pollution, improve user experience, and reduce safety hazards.

[0019] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating an application scenario of a range extender control method provided in an embodiment of this application. For example... Figure 1 As shown, the vehicle includes at least a forward-facing camera 110 and a fisheye camera 120, wherein there can be multiple fisheye cameras. Multiple fisheye cameras acquire images of the vehicle's surroundings, and the acquired surrounding images are stitched together to obtain a panoramic image. The forward-facing camera acquires images of the vehicle's forward direction to obtain an initial image. As an example, the vehicle may also have sensors such as ultrasonic radar to acquire radar data, and a positioning device such as a GPS (Global Positioning System) positioning device to obtain positioning data. Subsequently, image processing is performed on the initial image and the panoramic image to obtain fused image features for scene recognition. A pre-trained scene recognition classifier is then used to determine the vehicle's current scene. By processing at least one of the panoramic image, positioning data, and radar data, a closure evaluation value for the vehicle's environment can be obtained. Furthermore, the current range extender control strategy can be determined based on the closure evaluation value and at least one of the factors in the vehicle's environment, and the vehicle's range extender is controlled based on this strategy. It should be noted that the position of the fisheye camera... Figure 1 This is just one example and does not mean that its location can only be set at the rearview mirror.

[0020] It should be noted that the above scenario is merely an example of an application scenario provided by the embodiments of this application. The embodiments of this application do not limit the actual form of the various devices, components, etc., included in this scenario. In the specific application of the solution, it can be set according to actual needs. It should also be noted that when the vehicle is used to implement the range extender control method, the vehicle must have a range extender.

[0021] The above method can be implemented locally in the vehicle, through interaction between the vehicle and a cloud server, or through interaction between the vehicle and other devices. The above is just one example and does not limit the method to be implemented only locally in the vehicle.

[0022] Please see Figure 2 , Figure 2 A schematic flowchart of a range extender control method provided in an embodiment of this application is shown below. Figure 2 As shown, the method includes the following steps: Step S210: Obtain the vehicle's current scene and closure evaluation value.

[0023] The vehicle's location is determined based on the vehicle's front view image and panoramic image, and the closure evaluation value is determined based on at least one of the vehicle's positioning data, radar data, and panoramic image.

[0024] Please see Figure 3 , Figure 3 This is a flowchart illustrating a method for determining the scene where a vehicle is located, as provided in an embodiment of this application. Figure 3 As shown, the method includes the following steps: Step S310: Obtain the front view image and panoramic image of the vehicle.

[0025] In one embodiment, acquiring a panoramic image of a vehicle includes: acquiring multiple surrounding images of the vehicle, wherein the multiple surrounding images cover a 360-degree view around the vehicle; performing radial distortion correction on each surrounding image to obtain a corrected surrounding image; and generating a panoramic image based on all the corrected surrounding images.

[0026] For example, surrounding images can be captured using four fisheye cameras positioned around the vehicle, providing a 360-degree panoramic view. Fisheye cameras can capture close-up environmental information, including but not limited to obstacles, walls, and road edges. The complete 360-degree view, or panoramic image, is then stitched together from the images from the four fisheye cameras.

[0027] For example, wide-angle images from four fisheye cameras in the front, back, left, and right directions can be acquired as surrounding images, which contain significant radial distortion. This is because the wide-angle characteristics of fisheye cameras cause significant radial distortion in the images they acquire. Of course, the number and distribution of fisheye cameras can be set by those skilled in the art as needed; this is merely an example.

[0028] Following the above embodiments, the radial distortion correction of the surrounding image is performed as follows: Formula (1), Formula (2), Here, x and y represent the pixel coordinates in the surrounding image, with the top-left corner of the surrounding image as the origin (0, 0), the horizontal direction as the x-axis, and the vertical direction as the y-axis. The position of each pixel can be represented by a pair of integer coordinates (x, y), k1 and k2 are distortion coefficients, and r is the radial distance. and For the corrected pixel coordinates, the LaTeX code is as follows: , .

[0029] As an example, k1 and k2 are distortion coefficients, which can be obtained through a calibration board (such as a checkerboard).

[0030] As an example, the radial distance is determined as follows: Formula (3), Where r is the radial distance, and x and y represent the pixel coordinates in the surrounding image, with the top-left corner of the surrounding image as the origin (0, 0), the horizontal direction as the x-axis, and the vertical direction as the y-axis. The LaTeX code is as follows: .

[0031] The method of generating a panoramic image based on all the corrected surrounding images can be implemented in a manner known to those skilled in the art, and is not limited here.

[0032] As an example, when performing image stitching, the brightness, contrast, etc. of the images are not adjusted in order to determine the subsequent closure evaluation value.

[0033] In one embodiment, obtaining a front view image of a vehicle includes: obtaining an initial image of the front of the vehicle, and performing histogram equalization processing on the initial image to obtain a front view image.

[0034] As an example, the initial image can be acquired using a forward-looking integrated camera mounted in front of the vehicle for long-range target detection (such as pedestrians and other vehicles), or it can be acquired using other image acquisition devices installed on the vehicle. For instance, a long-range image from the forward-looking integrated camera can be acquired first, which may be affected by factors such as lighting and weather, and used as the raw image. Since the forward-looking integrated camera is mainly used for long-range target detection, the focus of data preprocessing is to improve target detection accuracy and reduce background interference. Histogram equalization can be used to enhance image contrast, making the target more clearly visible. The resulting forward-looking image has higher contrast and less background interference.

[0035] Step S320: Extract features from the front view image and the panoramic image respectively, encode the extracted features to obtain a position encoding matrix, and combine the extracted features with the position encoding matrix to obtain the position front view image features and the position panoramic image features.

[0036] In one embodiment, feature extraction is performed on the front view image and the panoramic image respectively to obtain initial front view image features and initial panoramic image features.

[0037] As an example, features can be extracted from front-view and panoramic images using convolutional neural networks. One example is as follows: Formula (4), Formula (5), in, Features of the initial front view image, The initial panoramic image features are denoted by ∗, where ∗ represents the convolution operation and ϕ is the activation function. and This is the weight matrix. and For bias, For front view image, For panoramic images, the LaTeX code is as follows: , .

[0038] As an example, the extracted features are combined with the location encoding matrix in the following way: Formula (6), Formula (7), in, For the location of the front view image features, For location panoramic image features, Features of the initial front view image, Features of the initial panoramic image and For the position encoding matrix, the LaTeX code is as follows: , .

[0039] As an example, position encoding can be generated based on initial front view image features and initial panoramic image features, representing the position of the features in a preset coordinate system (such as the vehicle body coordinate system).

[0040] Step S330: Align the features of the front view image with the features of the panoramic image to obtain aligned front view image features and aligned panoramic image features.

[0041] For example, feature alignment between two feature maps can be achieved through resampling, as illustrated in the following example: Formula (8), Formula (9), in, Here, H and W are the target height and width, respectively, and are the resampling functions. To align the features of the front view image, To align panoramic image features, For the location of the front view image features, For the location panoramic image features, the LaTeX code is as follows: , .

[0042] Step S340: Determine the feature similarity between the aligned front view image features and the aligned panoramic image features, and generate an attention weight matrix based on all feature similarities.

[0043] In one embodiment, determining the feature similarity between aligned front view image features and aligned panoramic image features includes: obtaining a preset weight coefficient matrix, wherein the preset weight coefficient matrix has the same size as the position encoding matrix, the element values ​​of the elements at the front view positions in the preset weight coefficient matrix are greater than preset values, the front view position is the position of the corresponding front view image feature in the position encoding matrix, and the element values ​​of the elements at non-front view positions in the preset weight coefficient matrix are preset values; adjusting the weights of the aligned front view image features according to the preset weight coefficient matrix to obtain adjusted front view image features; and determining the cosine similarity of each position based on the adjusted front view image features and the aligned panoramic image features to obtain feature similarity.

[0044] The method provided in the above embodiments, when determining feature similarity, not only considers the features themselves, but also combines the positional information of these features in the image for joint judgment. Furthermore, since the foreground image mainly focuses on the area in front, while the panoramic image covers the surrounding environment, the weights of the features in the foreground image area can be adjusted to make their effect greater. Therefore, a preset weight coefficient matrix can be designed to set a larger weight value for the features corresponding to the foreground image area.

[0045] As an example, cosine similarity is determined as follows: Formula (10), Formula (11), in, for Features of the front view image after position adjustment for Position alignment of front view image features, for Position alignment panoramic image features, It is a weighting coefficient (the element value of a preset weighting coefficient matrix). (representing its position in the matrix), which can be adjusted according to the actual situation; the default value is 1. Located in the forward-looking area (forward-looking position), a value greater than 1 can be set, meaning a higher value can be set in the forward-looking area (such as the upper part of the image). for Cosine similarity of positions.

[0046] Based on the cosine similarity of all positions, the feature similarity can be obtained. The LaTeX code is as follows: , .

[0047] In one embodiment, generating an attention weight matrix based on feature similarity includes performing probability normalization on the feature similarity to obtain probability values, and generating an attention weight matrix based on the probability values ​​of all positions.

[0048] As an example, the cosine similarity values ​​are converted into a probability distribution using the Softmax function, forming the attention weight matrix A. Let's assume... S Given an H×W matrix, Softmax is applied independently to each row of the similarity matrix S, such that the sum of the weights in each row is 1. Formula (12), in, for Location attention weights (probability values) for Cosine similarity of positions for The cosine similarity of positions, where W is the width of the attention weight matrix, is shown in the LaTeX code: .

[0049] Row normalization is employed, focusing on independent matching of forward-looking features. Softmax is calculated for all similarities S(i,1), S(i,2)...S(i,W) in the i-th row, and the output is... This represents the attention weight of the i-th query on the j-th key.

[0050] Step S350: Generate fused image features based on the attention weight matrix, aligned front view image features, and aligned panoramic image features.

[0051] In one embodiment, generating fused image features based on an attention weight matrix, aligned front-view image features, and aligned panoramic image features includes: determining an ignore weight matrix based on the attention weight matrix, wherein the size of the ignore weight matrix is ​​the same as the size of the attention weight matrix, and the sum of the elements in the ignore weight matrix and the elements at the same position in the attention weight matrix is ​​1; determining front-view fused image sub-features based on the attention weight matrix and aligned front-view image features; determining panoramic fused image sub-features based on the ignore weight matrix and aligned panoramic image features; and obtaining fused image features based on the front-view fused image sub-features and panoramic fused image sub-features.

[0052] As an example, the fused image features for a location are determined as follows: Formula (13), in, for Location-fused image features for Location attention weights (probability values) for Position alignment of front view image features, for Position alignment panoramic image features, To ignore the weight matrix The probability value of the location, in LaTeX code, is: .

[0053] Step S360: Input the fused image features into a pre-trained scene recognition classifier to obtain the scene in which the vehicle is located.

[0054] In one embodiment, inputting the fused image features into a pre-trained scene recognition classifier to obtain the scene in which the vehicle is located includes: inputting the fused image features into the pre-trained scene recognition classifier to obtain evaluation parameters for each preset scene corresponding to the fused image features; determining the prediction probability of each preset scene based on all the evaluation parameters; and determining the preset scene with the highest prediction probability as the scene in which the vehicle is located.

[0055] As an example, fusing image features It is a multidimensional tensor that can be directly input into a pre-trained scene recognition classifier such as a CNN classifier. The feature map output from the last convolutional layer is flattened into a one-dimensional vector and input into the fully connected layer. The fully connected layer has an output dimension equal to the number of classes K (e.g., 6 initial classes). To ensure that the output values ​​fall within a reasonable range, the Softmax function is used to convert the output into a probability distribution, thereby obtaining the confidence score for each class.

[0056] Taking the score as an example, one exemplary method for determining the evaluation parameter includes: Formula (14), in, To score points, The feature map output from the last convolutional layer is flattened into a one-dimensional vector. For bias, The weight matrix is ​​represented in LaTeX as follows: .

[0057] Due to the fusion of image features It is a multi-dimensional tensor, typically defined as [batch_size, height, width, channels]. Before being input into the fully connected layer, it needs to be flattened into a 2D tensor of [batch_size, height * width * channels] (i.e., the feature vector of each sample). The flattened vector is... .

[0058] As an example, the predicted probability is determined as follows: Formula (15), in, The uppercase K in the denominator indicates the number of categories. It is the first Category score Given features Category The probability is a value between 0 and 1, and the sum of the probabilities of all categories is 1. This is The exponential form ensures that all scores are positive and amplifies the differences between larger scores. It is the exponential sum of the scores of all categories, used for normalization, such that the probability of each category is between 0 and 1, and the sum of all probabilities is 1. To fuse image features, For a preset scene k, the LaTeX code is: .

[0059] As an example, the scenario in which the vehicle is located is determined as follows: Formula (16), in, The scene in which the vehicle is located. For a preset scenario k, To fuse image features, the LaTeX code is as follows: .

[0060] As an example, the preset scenarios include, but are not limited to, driving in tunnels, driving on restricted roads, open-air parking, indoor parking, obstacles obscuring the view, and vehicles covered by car covers.

[0061] The training method for the scene recognition classifier can be adopted in a way known to those skilled in the art, and will not be elaborated here.

[0062] In one embodiment, the method for determining the closure evaluation value includes: acquiring closure evaluation data of the vehicle, wherein the closure evaluation data includes at least one of the positioning data, the radar data, and the panoramic image; quantifying the closure evaluation data; and determining the closure evaluation value of the environment in which the vehicle is located based on the quantized closure evaluation data.

[0063] Following the above embodiments, the method further includes: If the closed evaluation data includes positioning data, the quantification of the closed evaluation data includes determining a first positioning evaluation value based on vehicle location information, determining a second positioning evaluation value based on the number of received satellites and the average signal strength, and determining a positioning closed evaluation sub-value based on the first positioning evaluation value and the second positioning evaluation value, wherein the positioning data includes vehicle location information, the number of received satellites and the average signal strength; If the closure evaluation data includes radar data, the quantification of the closure evaluation data includes: determining the obstacle distance between the vehicle and surrounding obstacles using radar data; determining a first radar evaluation value based on the minimum obstacle distance and a preset safety distance; determining the probability of obstacles appearing in a preset distance interval based on all obstacle distances; determining the entropy of obstacle distribution based on the probability of obstacles appearing in all preset distance intervals to obtain a second radar evaluation value; and determining a radar closure evaluation sub-value based on the first radar evaluation value and the second radar evaluation value, wherein the radar data includes obstacle distances; If the closure evaluation data includes panoramic images, the quantification of the closure evaluation data includes: determining environmental visibility based on image contrast and average image brightness; identifying obstacles in the panoramic images to obtain the number of obstacles, and determining obstacle density based on the number of obstacles and a preset number threshold; detecting road edges in the panoramic images and determining road boundary clarity based on the edge evaluation values ​​of the detected road edges; detecting lane lines in the panoramic images and determining lane line visibility based on the confidence level of the detected lane lines; detecting the sky in the panoramic images to obtain the sky area proportion of the panoramic images, and determining the sky proportion evaluation value based on the sky area proportion and a preset proportion threshold; and determining a panoramic image closure evaluation sub-value based on at least one of environmental visibility, obstacle density, road boundary clarity, lane line visibility, and sky proportion evaluation value. Furthermore, at least one of the positioning closure evaluation sub-value, radar closure evaluation sub-value, and panoramic image closure evaluation sub-value is determined as the quantized closure evaluation data.

[0064] As an example, the values ​​for the location closure evaluation sub-value, radar closure evaluation sub-value, and panoramic image closure evaluation sub-value are all within the range of [0, 1]. In the closure evaluation data, more closed environments correspond to higher values.

[0065] In one embodiment, the method for determining the location closure evaluation sub-value is as follows: Taking GPS data as an example, the location information obtained from GPS and the satellite signal strength can be used to determine the degree of enclosure. The location information can be used to determine whether the vehicle is located in a known enclosed area (such as an underground parking garage or tunnel). The satellite signal strength can be used to characterize the number of satellites receiving the signal and the signal quality, helping to determine whether the vehicle is in a scenario with poor signal (such as an underground parking garage).

[0066] For identification of known enclosed areas: Geofencing technology is used to determine whether a vehicle is within a known enclosed area based on GPS coordinates. A confidence score is defined. The range is [0, 1], representing the probability that a vehicle is in a closed area.

[0067] Formula (17), in, The first positioning evaluation value is 1 if the GPS positioning data indicates that the current location is in a closed area, otherwise the value is 0.

[0068] Satellite signal strength assessment: Count the number of satellites received by the vehicle at the current moment. and average signal strength And standardize it to the range [0, 1].

[0069] Formula (18), in, This is the second positioning evaluation value. To receive the number of satellites, The LaTeX code for the average signal strength is: .

[0070] The final closure value of the GPS portion, i.e., the positioning closure evaluation sub-value, is: Formula (19), in, and These are weights, initialized to 0.5, 0.5. Engineers can adjust them according to the actual situation during implementation. This represents the confidence level (first location evaluation value) that the vehicle is located in a known closed area. The standardized value representing satellite signal strength (secondary positioning evaluation value) is closer to 1, indicating a stronger signal. Therefore, it is used... To indicate a more closed situation, To locate the closed evaluation subvalue, the LaTeX code is as follows: .

[0071] In one embodiment, the panoramic image closure evaluation sub-value is determined as follows: Taking the use of visibility (environmental visibility), obstacle density, road boundary clarity, lane line detection (lane line visibility), and sky ratio (sky ratio evaluation value) identified by a fisheye camera to determine the degree of closure as an example, where: Visibility: The brightness and contrast of an image are assessed in relation to the ambient lighting conditions.

[0072] Obstacle density: Detects static or dynamic obstacles such as walls, pillars, vehicles, water-filled barriers, and traffic cones in an image.

[0073] Road boundary clarity: Assess the continuity and clarity of road edges.

[0074] Lane line detection: Identifies whether there are obvious lane lines. Blurred or absent lane lines may indicate a closed environment.

[0075] Sky proportion: The proportion of the sky in an image; the smaller the proportion, the more enclosed the environment.

[0076] a. Visibility assessment: By calculating the average brightness of the panoramic image and contrast And it is standardized to the range [0, 1]. One example is determined as follows: Formula (20), in, For environmental visibility, This represents the average brightness level of pixel values ​​in an image. It indicates the degree of brightness difference between different regions in an image. Indicates brightness and contrast The smaller value in the range (because if either the brightness or contrast of the image is low, it will affect visibility). This represents the visibility threshold, a reference value used to standardize visibility scoring. The LaTeX code is: .

[0077] b. Obstacle density assessment: Use object detection algorithms to count the number of obstacles in an image. And normalize it to the range [0, 1]. Assume the maximum number of obstacles is... .

[0078] Formula (21), in, For obstacle density, The number of obstacles, To preset the quantity threshold, the LaTeX code is: .

[0079] c. Road boundary clarity assessment: The sharpness of road edges is detected using a fisheye camera and normalized to the range [0, 1]. Assume the maximum value of the sharpness index is... .

[0080] Formula (22), in, For the clarity of road boundaries, The edge evaluation value of the road edge. To preset the maximum value of the sharpness index, the LaTeX code is: .

[0081] As an example, the determination of road boundary sharpness, or the edge evaluation value of a road edge, can be achieved by using a confidence score obtained through deep learning algorithms to represent the Clarity Index; alternatively, edge detection algorithms such as Canny and Sobel can be used to extract road edges, and then the gradient magnitude or continuity of the edges can be calculated to obtain the Clarity Index. Sharp road edges produce continuous, elongated, and high-contrast Canny edges; while blurry road edges produce broken, thick, and dull Canny edges. Therefore, edge detection operators are first used to calculate the gradient intensity and direction of each pixel in the image, and then the sharpness is evaluated based on indicators such as the number of edge pixels, edge continuity, or edge thickness; alternatively, engineers can design other methods for calculation in practical implementations.

[0082] d. Lane line detection: The presence of clear lane lines is identified using a fisheye camera, and these lines are standardized to the range [0, 1]. The maximum confidence level for a lane line is assumed to be... .

[0083] Formula (23), in, To improve lane line visibility, To detect the confidence level of the algorithm output, For the purpose of The maximum confidence level for lane lines is given in LaTeX code as follows: .

[0084] e. Sky ratio assessment: Calculate the proportion of the sky in the panoramic image based on the obtained fisheye camera image. And normalize it to the range [0,1]. Assume the maximum value of the sky ratio is... .

[0085] Formula (24), in, This is an assessment value for the proportion of sky. The percentage of the sky area. To preset the percentage threshold, the LaTeX code is: .

[0086] The method for determining the overall closure value (panoramic image closure evaluation sub-value) of a fisheye camera is as follows: Formula (25), in,w 3. w 4. w 5. w 6. w 7 represents the weight of each indicator, which can be evenly distributed among the components. For example, w3 = 0.2. w 3=0.2, w4=0.2 、 w5=0.2, w6=0.2, w7=0.2. Engineers can adjust these values ​​according to the actual situation during implementation. For panoramic image closed evaluation sub-values, For environmental visibility, For obstacle density, For the clarity of road boundaries, To improve lane line visibility, The LaTeX code for the sky scale assessment value is: .

[0087] In one embodiment, the radar closure evaluation sub-value is determined as follows: For example, distance measurements obtained from ultrasonic radar and obstacle distribution can be used to determine the degree of enclosure. Distance measurement involves determining the shortest distance between surrounding objects and the vehicle. Obstacle distribution involves counting the number of obstacles in different directions and their relative positions.

[0088] a. Distance measurement assessment: The excellent ranging characteristics of ultrasonic radar can be utilized to calculate the distance to the nearest obstacle in each direction. And normalize it to the range [0, 1]. Assume the minimum safe distance is... .

[0089] Formula (26), in, This is the first radar evaluation value. The minimum obstacle distance, To preset a safe distance, the LaTeX code is: .

[0090] b. Obstacle distribution assessment: Count the number of obstacles in different directions and their relative positions, and calculate the entropy of the obstacle distribution. This is then standardized to the range [0, 1]. Entropy is a statistical concept used to measure the degree of disorder or uncertainty in a system. Entropy is used to quantify the uniformity and complexity of the distribution of obstacles in the surrounding environment. Assume the maximum entropy is... .

[0091] Formula (27), in, This is the second radar evaluation value. The entropy of the obstacle distribution. To preset the maximum entropy value, the LaTeX code is as follows: .

[0092] in, It is the entropy of the obstacle distribution around the entire vehicle. The calculation method is as follows: (1) Data partitioning of ultrasonic radar. Twelve ultrasonic radars (or other numbers, this is just an example) are deployed on the vehicle, distributed in different directions around the vehicle. Each radar can detect obstacles within a certain range and return a distance value. These distance values ​​can be divided into several intervals (e.g., 0-1 meter, 1-2 meters, 2-3 meters, etc.), and the frequency of obstacles appearing in each interval is counted. (2) Calculation of obstacle distribution probability in each radar direction. For each radar direction... Calculate the probability of an obstacle appearing within each distance interval. Divide the distance into If there are intervals, then: Formula (28), in, Radar direction The probability of an obstacle appearing within a preset distance interval i. The number of obstacles within a preset distance interval i. The total number of obstacles is represented in LaTeX code as follows: .

[0093] (3) Calculate the entropy for each radar direction. For each radar direction Calculate its corresponding entropy : Formula (29), in, It is the number of preset distance intervals. Radar direction The probability of an obstacle appearing within a preset distance interval i. Radar direction The entropy of the obstacle distribution, in LaTeX code: .

[0094] (4) Combine all radar directions. Finally, average the entropy of all radar directions. Here, 12 represents the number of ultrasonic waves used (if the actual vehicle model only has 4 / 8 ultrasonic waves, the formula will change accordingly): Formula (30), in, 12 represents the entropy of obstacle distribution, and 12 represents the radar direction. The quantity can be set according to the actual situation; this is just one example. Radar direction The entropy of the obstacle distribution, in LaTeX code: .

[0095] The method for determining the comprehensive containment value (radar containment evaluation sub-value) of ultrasonic radar is as follows: Formula (31), in, w 8 and w9 represent the weights of each indicator; for example, setting the parameter values ​​to w8=0.5 and w9=0.5. ， The specific implementation can be adjusted according to the actual situation. This is the first radar evaluation value. This is the second radar evaluation value. The LaTeX code for the radar closure evaluation subvalue is: .

[0096] As an example, taking the quantified closed-loop evaluation data, including location closed-loop evaluation sub-values, radar closed-loop evaluation sub-values, and panoramic image closed-loop evaluation sub-values, the method for determining the closed-loop evaluation value of the vehicle's environment based on the quantified closed-loop evaluation data is as follows: Taking into account the closure values ​​obtained from the perspectives of the three sensors mentioned above, the final closure value F can be calculated by weighted averaging: Formula (32), in, , and As an example, the weights of each indicator are given. Because fisheye cameras have the widest field of view, capable of obtaining a 360° view and capturing detailed images, the degree of closure determined by fisheye cameras is given the highest weight. Because ultrasonic radar has the largest number of sensors and high reliability, but cannot provide specific images, the degree of enclosure determined by ultrasonic radar is given more weight. Because GPS provides relatively limited information, the closure degree determined by GPS is assigned a certain weight, where F is the closure value. For radar closed evaluation sub-values, For panoramic image closed evaluation sub-values, To locate the closed evaluation subvalue, the LaTeX code is as follows: .

[0097] Then, the closure evaluation value is determined based on the closure value. This can be obtained by normalizing the closure value using an activation function (Sigmoid function). The Sigmoid function has a smooth gradient, which is beneficial for backpropagation during neural network training and for the algorithm corresponding to the range extender strategy in subsequent processes. It can map values ​​of any range to the range [0, 1]. Formula (33), in, F is the closed-value evaluation value, and its LaTeX code is: .

[0098] The vehicle scene determination method provided in the range extender control method of the above embodiments acquires a front view image and a panoramic image of the vehicle; extracts features from the front view image and the panoramic image respectively; performs position encoding on the extracted features to obtain a position encoding matrix; combines the extracted features with the position encoding matrix to obtain position front view image features and position panoramic image features; aligns the position front view image features and position panoramic image features to obtain aligned front view image features and aligned panoramic image features; determines the feature similarity between the aligned front view image features and aligned panoramic image features, and generates an attention weight matrix based on the feature similarity; generates fused image features based on the attention weight matrix, aligned front view image features, and aligned panoramic image features; inputs the fused image features into a pre-trained scene recognition classifier to obtain the vehicle's scene. By using the vehicle's front view image and panoramic image, a more comprehensive understanding of the vehicle's scene can be obtained. The introduction of position data during similarity calculation makes the similarity determination more accurate, thus making the final fused image features closer to the actual scene, improving the accuracy of scene recognition, meeting user needs, and enhancing user experience.

[0099] In addition, before feature fusion, the features corresponding to the front view image are emphasized by pre-setting a weight coefficient matrix. This can further ensure that the features of the fused image receive more attention and avoid recognition errors due to the limitations of panoramic images.

[0100] By using perception data from intelligent driving systems to determine the closure evaluation value, the closure status of the vehicle's environment can be further evaluated, providing a quantitative value of the closure status. This further facilitates the subsequent application of the system and allows for more detailed formulation of relevant control strategies, rather than simply formulating different strategies based on different scenarios.

[0101] It should be noted that the determination of the closure evaluation value does not necessarily depend on the judgment of the vehicle's environment. In other words, the closure evaluation value can be determined solely based on the closure evaluation data. This closure evaluation value can characterize the degree of closure of the vehicle's environment and can also serve as a method for determining the vehicle's environment. For example, different closure evaluation values ​​can be classified into corresponding vehicle environments and used as a premise for subsequent control.

[0102] Step S220: Determine the range extender control strategy for the vehicle based on at least one of the vehicle's current scenario and closed evaluation value, so as to control the vehicle's range extender through the control strategy.

[0103] The corresponding range extender control strategy can be determined in advance for different vehicle scenarios, closed evaluation values ​​and their combinations. Then, based on the currently obtained vehicle scenario and / or closed evaluation value, the corresponding range extender control strategy can be matched to obtain the range extender control.

[0104] As an example, determining the range extender control strategy based on the vehicle's environment and closure evaluation value includes, if the vehicle is in a closed environment and the closure evaluation value is greater than a preset threshold, the range extender control strategy includes prohibiting the range extender from starting while ensuring normal vehicle operation for a preset period of time. It can be understood that if the vehicle can maintain normal operation through other power sources, such as providing the energy required for normal vehicle operation through a battery, then the range extender can be prohibited from starting to avoid noise pollution from its use in a confined space.

[0105] As another example, determining the vehicle's range extender control strategy based on the vehicle's environment and enclosure evaluation value includes, if the vehicle's environment is classified as a closed environment and the enclosure evaluation value is less than or equal to a preset threshold, the range extender control strategy includes limiting the range extender's operating power according to the power limitation range corresponding to the vehicle's environment. For environments with low enclosure evaluation values, the enclosure level is considered low, allowing for gas exchange with the outside air at a relatively fast rate. Furthermore, due to the low enclosure level, the range extender's operating noise is relatively low. Therefore, limiting the range extender's operating power can achieve a balance between maintaining normal vehicle operation and reducing noise. The power limitation range can be set by those skilled in the art as needed.

[0106] As another example, determining the vehicle's range extender control strategy based on the vehicle's environment includes, if the environment is a closed environment, limiting the range extender's operating power according to the power limitation range corresponding to that environment. Alternatively, range extender control can be performed through simple environment recognition; in this case, refer to the aforementioned embodiments, where limiting the range extender's operating power can also achieve a balance between maintaining normal vehicle operation and reducing noise. The power limitation range can be set by those skilled in the art as needed.

[0107] As another example, determining the vehicle's range extender control strategy based on the closure evaluation value includes, if the closure evaluation value is greater than a preset threshold, prohibiting the range extender from starting under normal vehicle operation conditions within a preset time period. The closure evaluation value can indicate the degree of closure of the vehicle's environment. If the closure evaluation value is too high, it means that the environment is relatively closure-oriented, and therefore the range extender can be "not started unless absolutely necessary."

[0108] As another example, determining the vehicle's range extender control strategy based on a closed-loop evaluation value includes, if the closed-loop evaluation value is less than or equal to a preset threshold, determining the maximum operating power of the range extender based on the closed-loop evaluation value, and limiting the range extender's operating power to the maximum operating power. If the closed-loop evaluation value is relatively small, it indicates that the environment is not very closed; therefore, by limiting a maximum operating power, the range extender's operating power can be kept below that maximum operating power, thus avoiding excessive noise.

[0109] It should be noted that the above control strategies are merely examples, and those skilled in the art can formulate other range extender control strategies as needed.

[0110] The range extender control method provided in the above embodiments can identify enclosed scenarios and determine the specific degree of enclosure, assisting subsequent hardware in determining the range extender strategy. It can improve energy efficiency by optimizing the start-stop timing of the range extender. Based on the enclosed scenario and degree of enclosure, the VCU can intelligently decide the range extender strategy, avoiding unnecessary energy consumption. It can enhance environmental protection by reducing exhaust emissions. In highly enclosed environments (such as driving in tunnels or indoor parking), the system prohibits the start of the range extender, thereby avoiding the generation of harmful gases and noise pollution. It reduces noise interference by precisely controlling the power and start-stop timing of the range extender, enabling the system to maintain low-noise operation in open and partially enclosed environments, reducing the impact on surrounding people. It can also improve user experience by providing a quieter and more comfortable riding environment, enhancing user satisfaction. Vehicle sensors can identify the surrounding environment in real time and intelligently adjust the range extender's operating strategy when an enclosed environment is detected. It reduces the safety risks of operating the range extender in enclosed spaces, preventing accidents such as fires and explosions, and ensuring the safety of personnel and property.

[0111] As an example, see Figure 4 , Figure 4 A schematic flowchart illustrating a method for determining the vehicle's location according to an embodiment of this application is shown below. Figure 4 As shown, in the process of recognizing the scene in which the vehicle is located, perception data is first acquired, including a front view image provided by the all-in-one machine and a panoramic image provided by the fisheye camera. The front view image and the panoramic image are then preprocessed separately. Next, the data is fused with the preprocessed data, and the fused data is input into a trained scene recognition classifier to obtain the scene recognition result of the vehicle's environment.

[0112] As an example, see Figure 5 , Figure 5 A specific flowchart illustrating the closure evaluation value provided in one embodiment of this application is shown below. Figure 5 As shown, firstly, the perception data of the vehicle's environment is acquired, including at least the positioning data from the GPS sensor, the panoramic image generated from the surrounding images captured by the fisheye camera, and the radar data from the ultrasonic radar; then, the corresponding data processing is performed to obtain the corresponding closure evaluation sub-values, and finally, the closure value is obtained; finally, based on the closure value, the closure degree quantification value of the closed environment, i.e., the closure evaluation value, is obtained, so as to match the corresponding range extender control strategy according to the closure evaluation value.

[0113] As an example, see Figure 6 , Figure 6 A specific flowchart illustrating a range extender control method provided in an embodiment of this application is shown below. Figure 6 As shown, the ADS (Autonomous Driving System) comprehensively perceives the vehicle's surrounding environment, achieving real-time, high-precision perception. Through a closed-scene recognition method based on the ADS's perception (vehicle location scene recognition method), the scene category of the current vehicle location is obtained, with the corresponding CAN signal being ADS_ClosedScene (vehicle location scene). Through a closed-degree judgment method based on the ADS's perception (closed-degree evaluation value determination method), the closed-degree evaluation value of the current vehicle environment is obtained, with the corresponding CAN signal being ADS_ClosedValue (closed-degree evaluation value). The ADS transmits these signals to the VCU (Vehicle Controller Unit) in real time. The VCU then employs different range extender control strategies based on different scenes and closed-degree levels, thereby achieving intelligent range extension. Please refer to Table 1, which provides an example of specific signals and their corresponding 0xn signal values.

[0114] Table 1

[0115] In one embodiment, a range extender control device is provided for executing the range extender control method provided in any of the above embodiments. See also... Figure 7 , Figure 7 A schematic diagram of a range extender control device provided in an embodiment of this application is shown below. Figure 7 As shown, the range extender control device 700 includes an acquisition module 710 for acquiring the vehicle's current scene and closure evaluation value, wherein the vehicle's current scene is determined based on the vehicle's front view image and panoramic image, and the closure evaluation value is determined based on at least one of the vehicle's positioning data, radar data, and panoramic image; and a control module 720 for determining the vehicle's range extender control strategy based on at least one of the vehicle's current scene and closure evaluation value, so as to control the vehicle's range extender through the control strategy.

[0116] In one embodiment, the device further includes a vehicle scene determination module, comprising: an image acquisition module for acquiring a front view image and a panoramic image of the vehicle; a feature extraction module for extracting features from the front view image and the panoramic image respectively, performing position encoding on the extracted features to obtain a position encoding matrix, and combining the extracted features with the position encoding matrix to obtain position front view image features and position panoramic image features; an alignment module for aligning the position front view image features and position panoramic image features to obtain aligned front view image features and aligned panoramic image features; a matrix determination module for determining the feature similarity between the aligned front view image features and the aligned panoramic image features, and generating an attention weight matrix based on all feature similarities; a feature fusion module for generating fused image features based on the attention weight matrix, the aligned front view image features, and the aligned panoramic image features; and a scene recognition module for inputting the fused image features into a pre-trained scene recognition classifier to obtain the vehicle scene.

[0117] In one embodiment, the range extender control device further includes a closure evaluation value determination module, used to acquire closure evaluation data of the vehicle, the closure evaluation data including at least one of positioning data, radar data and panoramic images; quantify the closure evaluation data; and determine the closure evaluation value of the environment in which the vehicle is located based on the quantized closure evaluation data.

[0118] Specific limitations regarding the range extender control device can be found in the limitations of the range extender control method described above, and will not be repeated here. Each module in the aforementioned range extender control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the electronic device in hardware form or independently of it, or stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0119] In this embodiment, the range extender control device is essentially equipped with multiple modules to execute the range extender control method in any of the above embodiments. The specific functions and technical effects can be referred to in the above embodiments, and will not be repeated here.

[0120] See Figure 8 , Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown below. Figure 8 As shown, this embodiment of the invention also provides an electronic device 800, including a processor 801, a memory 802, and a communication bus 803; the communication bus 803 is used to connect the processor 801 and the memory 802; the processor 801 is used to execute a computer program stored in the memory 802 to implement the method provided in any of the above embodiments.

[0121] In one embodiment, a vehicle is provided, which includes the electronic equipment provided in any of the above embodiments, or performs the method provided in any of the above embodiments. The specific functions and technical effects of the vehicle can be referred to the above embodiments, and will not be repeated here.

[0122] This invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being used to cause a computer to perform the method described in any of the above embodiments.

[0123] This application also provides a computer-readable storage medium storing one or more modules (programs) that, when applied to a device, enable the device to execute the instructions included in the steps provided in this application.

[0124] This application also provides a computer program product, including a computer program or instructions, which, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0125] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0126] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0127] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0129] It should be understood that the terms "first," "second," etc., used in this application are used to distinguish similar objects and do not necessarily indicate a specific order or sequence. The technical features to which these terms are used can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0130] It should be understood that although the flowcharts provided in the embodiments of this application indicate the various steps with arrows, the order indicated by the arrows does not necessarily limit the implementation order of these steps. Those skilled in the art can perform these steps in other orders according to different implementation scenarios and requirements.

[0131] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A range extender control method, characterized in that, The method includes: The vehicle's location scene and closure evaluation value are obtained, wherein the vehicle's location scene is determined based on the vehicle's front view image and panoramic image, and the closure evaluation value is determined based on at least one of the vehicle's positioning data, radar data, and the panoramic image. The range extender control strategy of the vehicle is determined based on at least one of the vehicle's current scenario and the closed evaluation value, so as to control the vehicle's range extender through the control strategy.

2. The range extender control method as described in claim 1, characterized in that, The methods for determining the scenario in which the vehicle is located include: Acquire front view and panoramic images of the vehicle; Feature extraction is performed on the front view image and the panoramic image respectively. The extracted features are then encoded to obtain a position encoding matrix. The extracted features are then combined with the position encoding matrix to obtain the position front view image features and the position panoramic image features. Align the features of the front view image at the location with the features of the panoramic image at the location to obtain aligned front view image features and aligned panoramic image features; Determine the feature similarity between the aligned front view image features and the aligned panoramic image features, and generate an attention weight matrix based on all the feature similarities; Generate fused image features based on the attention weight matrix, aligned front view image features, and aligned panoramic image features; The fused image features are input into a pre-trained scene recognition classifier to obtain the scene in which the vehicle is located.

3. The range extender control method as described in claim 2, characterized in that, Determining the feature similarity between the aligned front view image features and the aligned panoramic image features includes: Obtain a preset weight coefficient matrix, the preset weight coefficient matrix having the same size as the position encoding matrix, the element values ​​of the elements at the forward view positions in the preset weight coefficient matrix being greater than a preset value, the forward view position being the position corresponding to the forward view image feature in the position encoding matrix, and the element values ​​of the elements at the non-forward view positions in the preset weight coefficient matrix being preset values. The aligned front view image features are weighted according to the preset weight coefficient matrix to obtain the adjusted front view image features. Based on the adjusted front view image features and the aligned panoramic image features, the cosine similarity of each position is determined to obtain the feature similarity.

4. The range extender control method as described in claim 2, characterized in that, Based on the attention weight matrix, aligned foreground image features, and aligned panoramic image features, fused image features are generated, including: An ignore weight matrix is ​​determined based on the attention weight matrix. The size of the ignore weight matrix is ​​the same as that of the attention weight matrix. The sum of the elements in the ignore weight matrix and the elements at the same position in the attention weight matrix is ​​1. The foreground fusion image sub-features are determined based on the attention weight matrix and the aligned foreground image features; Based on the ignored weight matrix and aligned panoramic image features, determine the sub-features of the panoramic fusion image; The fused image features are obtained based on the forward-looking fused image sub-features and the panoramic fused image sub-features.

5. The range extender control method according to any one of claims 2-4, characterized in that, The fused image features are input into a pre-trained scene recognition classifier to obtain the scene in which the vehicle is located, including: The fused image features are input into a pre-trained scene recognition classifier to obtain the evaluation parameters corresponding to each preset scene for the fused image features; The predicted probability for each preset scenario is determined based on all evaluation parameters. The preset scenario with the highest predicted probability is determined as the scenario in which the vehicle is located.

6. The range extender control method according to any one of claims 2-4, characterized in that, Acquiring front view and panoramic images of the vehicle includes: Multiple surrounding images of the vehicle are acquired, wherein the multiple surrounding images cover a 360-degree view around the vehicle; radial distortion correction is performed on each surrounding image to obtain a corrected surrounding image; the panoramic image is generated based on all the corrected surrounding images. An initial image of the front of the vehicle is acquired, and histogram equalization is performed on the initial image to obtain the front view image.

7. The range extender control method according to any one of claims 1-4, characterized in that, The methods for determining the closure evaluation value include: Obtain the vehicle's closure evaluation data, which includes at least one of the positioning data, the radar data, and the panoramic image; The closed evaluation data is quantified; The closed-loop evaluation value of the vehicle's environment is determined based on the quantified closed-loop evaluation data.

8. The range extender control method as described in claim 7, characterized in that, If the closed evaluation data includes positioning data, quantifying the closed evaluation data includes determining a first positioning evaluation value based on vehicle location information, determining a second positioning evaluation value based on the number of received satellites and the average signal strength, and determining a positioning closed evaluation sub-value based on the first positioning evaluation value and the second positioning evaluation value, wherein the positioning data includes vehicle location information, the number of received satellites, and the average signal strength; If the closure evaluation data includes radar data, quantifying the closure evaluation data includes: determining the obstacle distance between the vehicle and surrounding obstacles using the radar data; determining a first radar evaluation value based on the minimum obstacle distance and a preset safe distance; determining the probability of obstacles appearing in a preset distance interval based on all obstacle distances; determining the entropy of obstacle distribution based on the probability of obstacles appearing in all preset distance intervals; determining a second radar evaluation value based on the entropy of obstacle distribution and a preset maximum entropy value; and determining a radar closure evaluation sub-value based on the first radar evaluation value and the second radar evaluation value, wherein the radar data includes obstacle distances; If the closure evaluation data includes a panoramic image, quantifying the closure evaluation data includes: determining environmental visibility based on image contrast and average image brightness; identifying obstacles in the panoramic image to obtain the number of obstacles, and determining obstacle density based on the number of obstacles and a preset number threshold; detecting road edges in the panoramic image and determining road boundary clarity based on the edge evaluation value of the detected road edges; detecting lane lines in the panoramic image and determining lane line visibility based on the confidence level of the detected lane lines; detecting the sky in the panoramic image to obtain the sky area proportion of the panoramic image, and determining a sky proportion evaluation value based on the sky area proportion and a preset proportion threshold; and determining a panoramic image closure evaluation sub-value based on at least one of the environmental visibility, obstacle density, road boundary clarity, lane line visibility, and sky proportion evaluation value. Furthermore, at least one of the positioning closure evaluation sub-value, radar closure evaluation sub-value, and panoramic image closure evaluation sub-value is determined as the quantized closure evaluation data.

9. The range extender control method according to any one of claims 1-4, characterized in that, The range extender control strategy for the vehicle based on the vehicle's current scenario and the closure evaluation value includes, if the scenario category of the vehicle's current scenario is a closed scenario and the closure evaluation value is greater than a preset threshold, the range extender control strategy includes, under the condition that the vehicle is operating normally within a preset time period, prohibiting the start of the range extender. Determining the range extender control strategy for the vehicle based on the vehicle's current scenario and the closed evaluation value includes, if the scenario category of the vehicle's current scenario is a closed scenario and the closed evaluation value is less than or equal to the preset threshold, the range extender control strategy includes limiting the operating power of the range extender according to the power limitation range corresponding to the vehicle's current scenario. Determining the range extender control strategy based on the vehicle's location scenario includes, if the scenario category of the vehicle's location scenario is a closed scenario, the range extender control strategy includes limiting the operating power of the range extender according to the power limitation range corresponding to the vehicle's location scenario. Determining the range extender control strategy for the vehicle based on the closed evaluation value includes, if the closed evaluation value is greater than the preset threshold, the range extender control strategy includes, under the condition that the vehicle is operating normally within a preset time period, prohibiting the start of the range extender. Determining the range extender control strategy for the vehicle based on the closed evaluation value includes, if the closed evaluation value is less than or equal to the preset threshold, the range extender control strategy includes, based on the closed evaluation value, determining the maximum operating power of the range extender, and limiting the operating power of the range extender with the maximum operating power.

10. A vehicle, characterized in that, The vehicle is used to perform the steps of the method according to any one of claims 1 to 9.

Images