Vehicle wading detection method, device, storage medium and program product

Abstract

The application provides a vehicle wading detection method, device, storage medium and program product. It relates to the technical field of data processing. The method comprises: acquiring first water level data of vehicle driving in a preset scene and first visual data of the environment around the vehicle; determining a wading parameter according to the first water level data and the first visual data, the wading parameter being used to indicate the reliability of the vehicle passing through a waterlogged area; generating a wading control strategy according to the wading parameter and the first water level data, the wading control strategy being used to control the vehicle in stages; and controlling the vehicle to perform a corresponding driving operation according to the wading control strategy, and sending wading detection information to a cloud server when a preset condition is met. The reliability of vehicle wading detection is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, device, storage medium and program product for detecting vehicle wading. Background Technology

[0002] In autonomous driving scenarios, vehicles need to navigate through flooded areas, such as in rainy or foggy weather, ensuring smooth driving and operational safety.

[0003] In related technologies, water wading detection is usually carried out using vehicle-side visual perception, which involves using cameras to identify water bodies on the road surface, combining this with lidar point clouds to estimate the water depth, and controlling vehicle passage based on the detection results.

[0004] However, in the above process, vehicle-mounted visual perception is easily affected by environmental factors such as rain, fog, and strong light reflection from the water surface, which may lead to problems such as missed detection of water accumulation and excessive deviation in water depth. It is difficult to accurately identify the water depth and other traffic conditions, resulting in low reliability of vehicle wading detection. Summary of the Invention

[0005] This application provides a method, equipment, storage medium, and program product for detecting vehicle wading, in order to solve the problem of low reliability in vehicle wading detection.

[0006] Firstly, this application provides a method for detecting vehicle wading depth, comprising:

[0007] Acquire the first water level data of the vehicle in the preset scenario, as well as the first visual data of the vehicle's surrounding environment;

[0008] Based on the first water level data and the first visual data, the wading parameters are determined. The wading parameters are used to indicate the reliability of the vehicle passing through the waterlogged area.

[0009] Based on the wading parameters and the first water level data, a wading control strategy is generated, which is used to perform graded control of the vehicle.

[0010] According to the water wading control strategy, the vehicle is controlled to perform corresponding driving operations, and water wading detection information is sent to the cloud server when preset conditions are met.

[0011] In one possible implementation, water-related parameters are determined based on first water level data and first visual data, including:

[0012] Preprocessing the first water level data and the first visual data yields fused data.

[0013] The fused data is input into the first model to obtain the water-related parameters.

[0014] In one possible implementation, the first water level data and the first visual data are preprocessed to obtain fused data, including:

[0015] Align the first water level data and the first visual data to obtain the second water level data and the second visual data.

[0016] The second water level data and the second visual data are fused to obtain fused data.

[0017] In one possible implementation, the first water level data and the first visual data are aligned to obtain second water level data and second visual data, including:

[0018] The first water level data and the first visual data are synchronized in time to obtain the third water level data and the third visual data.

[0019] Spatial registration is performed on the third water level data and the third visual data to obtain the second water level data and the second visual data.

[0020] In one possible implementation, the third water level data and the third visual data are obtained, including:

[0021] Obtain the first timestamp corresponding to the first water level data, and the second timestamp corresponding to the first visual data;

[0022] Based on the first and second timestamps, the first water level data and the first visual data are synchronized to the same time base to obtain the third water level data and the third visual data.

[0023] In one possible implementation, the third water level data includes water level depth and detection coordinates; second water level data and second visual data are obtained, including:

[0024] According to the preset calibration parameters, the detection coordinates in the third water level data are projected onto the pixel coordinate system corresponding to the third visual data;

[0025] The second visual data is obtained by matching the detected coordinates after projection with the pixel positions of the water accumulation area in the third visual data. The water level depth of the second water level data is the same as that of the third water level data.

[0026] In one possible implementation, generating a water conservation control strategy includes:

[0027] When the wading parameters are greater than or equal to the first threshold and the water level depth in the first water level data is less than or equal to the second threshold, the wading control strategy is determined to be: control the vehicle to pass through the corresponding waterlogged area.

[0028] In one possible implementation, generating a water-related control strategy includes:

[0029] When the wading parameters are greater than or equal to the first threshold, and the water depth is greater than the second threshold and less than or equal to the third threshold, the wading control strategy is determined as follows: control the vehicle to slow down and pass through the waterlogged area, and send a vehicle takeover request to the cloud server. The vehicle takeover request includes wading detection information.

[0030] In one possible implementation, generating a water-related control strategy includes:

[0031] When the water wading parameters are greater than or equal to the first threshold and the water depth is greater than the third threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0032] In one possible implementation, generating a water-related control strategy includes:

[0033] When the water wading parameters are less than the first threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0034] In one possible implementation, the wading detection information includes wading detection reports and video data; the method further includes, before sending a vehicle takeover request to the cloud server:

[0035] The first water level data and water-related parameters are input into the second model to obtain the passage parameters for the waterlogged area.

[0036] A water erosion detection report is generated based on the passage parameters and the first water level data.

[0037] In one possible implementation, a water wading detection report is generated based on passage parameters and first water level data, including:

[0038] Based on traffic parameters, determine the vehicle movement data corresponding to the waterlogged area;

[0039] A water wading test report is generated based on the first water level data and vehicle movement data.

[0040] Secondly, this application provides a method for detecting vehicle wading depth, including:

[0041] Receive vehicle takeover requests sent by vehicles, which include water wading test reports and video data;

[0042] Based on the water immersion test report and video data, the vehicle control method was determined to be either contactless control or contact control.

[0043] Based on the mode of transportation, control commands are generated and issued to vehicles to ensure their safe passage through flooded areas.

[0044] In one possible implementation, when the control method is contactless control, determining the vehicle's control method includes:

[0045] Based on the vehicle movement data in the water wading test report and the road condition information in the video data, vehicles are guided to pass through the waterlogged area in accordance with the corresponding control commands and traffic methods.

[0046] In one possible implementation, when the control method is contact control, determining the vehicle control method includes:

[0047] In response to the user's driving input based on the water wading test report and video data, the vehicle will pass through the waterlogged area with the corresponding control command.

[0048] Thirdly, this application provides a vehicle wading detection device, wherein:

[0049] The acquisition module is used to acquire the first water level data of the vehicle driving in the preset scenario, as well as the first visual data of the vehicle's surrounding environment.

[0050] The first determining module is used to determine wading parameters based on the first water level data and the first visual data. The wading parameters are used to indicate the reliability of a vehicle passing through a waterlogged area.

[0051] The generation module is used to generate a wading control strategy based on the wading parameters and the first water level data. The wading control strategy is used to perform graded control of the vehicle.

[0052] The control module is used to control the vehicle to perform corresponding driving operations according to the wading control strategy, and to send wading detection information to the cloud server when preset conditions are met.

[0053] Fourthly, this application provides a vehicle wading detection device, wherein:

[0054] The receiving module is used to receive vehicle takeover requests sent by the vehicle, which include water wading test reports and video data;

[0055] The second determining module is used to determine the vehicle's control method based on the water wading test report and video data. The control method is either contactless control or contact control.

[0056] The sending module is used to generate control commands and send them to the vehicle according to the control method, so that the vehicle can safely pass through the flooded area.

[0057] Fifthly, embodiments of this application provide a vehicle wading detection device, comprising: at least one processor and a memory; the memory storing computer execution instructions; the at least one processor executing the computer execution instructions stored in the memory, causing the at least one processor to perform the vehicle wading detection method as described in the first aspect and any one of the first aspects.

[0058] In a sixth aspect, embodiments of this application provide a vehicle wading detection device, comprising: at least one processor and a memory; the memory storing computer execution instructions; the at least one processor executing the computer execution instructions stored in the memory, causing the at least one processor to perform the vehicle wading detection method as described in the second aspect above and any one of the second aspects.

[0059] In a seventh aspect, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the vehicle wading detection method as described in the first aspect and any one of the first aspects.

[0060] Eighthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the vehicle wading detection method as described in the second aspect above and any one of the second aspects.

[0061] The vehicle wading detection method, equipment, storage medium, and program product provided in this application acquire water level data and visual data of the surrounding environment during vehicle operation. The reliability of the water level data is determined by combining the visual data to obtain wading parameters. Based on these parameters and water level data, a hierarchical control strategy is generated to control the vehicle to perform corresponding driving operations. When conditions are met, wading detection information is reported to the cloud. This method combines visual perception results to determine the reliability of wading detection and executes differentiated hierarchical control strategies based on the level of reliability. When the vehicle-side perception results are unreliable or the wading risk is high, remote backup decisions are made via the cloud. This effectively avoids the problem of insufficient detection accuracy of single visual perception in adverse environments such as rain, fog, and strong light reflection, thus improving the reliability of wading detection. Simultaneously, through the hierarchical control strategy and cloud collaboration mechanism, vehicle interruptions due to insufficient decision-making ability in complex waterlogged scenarios are avoided. This method is applicable to autonomous driving scenarios with different water depths and various environmental conditions, improving the efficiency and continuity of autonomous vehicles in wading scenarios while ensuring traffic safety. Attached Figure Description

[0062] 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.

[0063] Figure 1 This is a schematic diagram illustrating an application scenario provided in the embodiments of this application;

[0064] Figure 2 A schematic flowchart of a vehicle wading detection method provided in an embodiment of this application;

[0065] Figure 3 This is a schematic diagram of the structure of the vehicle-end wading detection device provided in the embodiments of this application;

[0066] Figure 4 A schematic diagram illustrating the process of determining water-related parameters provided in the embodiments of this application;

[0067] Figure 5 A schematic diagram illustrating the process of determining the water-related control strategy provided in this application embodiment;

[0068] Figure 6 A schematic flowchart of another vehicle wading detection method provided in this application embodiment;

[0069] Figure 7 A schematic flowchart illustrating another vehicle wading detection method provided in this application embodiment;

[0070] Figure 8 A schematic diagram illustrating the interaction between the vehicle-mounted terminal and the cloud server provided in an embodiment of this application;

[0071] Figure 9 A schematic flowchart illustrating another vehicle wading detection method provided in this application embodiment;

[0072] Figure 10 This is a schematic diagram of the structure of a vehicle wading detection device provided in an embodiment of this application;

[0073] Figure 11 This is a schematic diagram of another vehicle wading detection device provided in an embodiment of this application;

[0074] Figure 12 This is a schematic diagram of the vehicle wading detection equipment provided in an embodiment of this application.

[0075] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0076] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0077] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0078] Level 4 autonomous driving water safety and vehicle-cloud collaboration technologies are widely used in the passage control of fully autonomous vehicles in complex waterlogged scenarios. During the vehicle's operation, it needs to cope with waterlogged sections with different road conditions, such as flat roads and slopes, under adverse conditions such as rain, fog, strong light reflection from the water surface, and turbid water. It needs to accurately detect the depth of the water on the road surface and pass safely to ensure driving safety and operational efficiency.

[0079] In related technologies, water wading detection typically employs a vehicle-based visual perception approach. This involves using cameras to identify water bodies on the road surface, combining this with lidar point cloud data to estimate water depth, and then controlling vehicle passage based on the detection results. While this solution has a certain detection capability under normal conditions such as sunny days and good lighting, its accuracy drops significantly in adverse environments such as rain, fog, strong light reflection from the water surface, and turbid water. This can lead to problems such as missed water wading detection and excessively inaccurate water depth estimation, making it difficult to accurately identify water depth and underwater obstacles, thus increasing the risk of vehicles misjudging and passing through.

[0080] Furthermore, when using ultrasonic sensors to detect water depth, the system is not integrated with the visual perception system, resulting in low reliability of the detection results. This makes it unsuitable for detecting water accumulation on different road conditions such as flat roads and slopes. Moreover, the lack of a vehicle-cloud collaborative confirmation mechanism means that there is no remote backup decision-making capability when the vehicle-side perception results are unreliable, leading to operational interruptions and low traffic efficiency.

[0081] Furthermore, existing solutions can only provide simple local responses such as pulling over and waiting for rescue in the event of flooding. They lack a tiered response strategy based on water depth and confidence level. The various stages of water detection, risk assessment, response execution, and remote confirmation are independent of each other, lacking unified time-series control and logical verification. This makes it impossible to form a complete closed-loop management system, which can easily lead to problems such as response delays and mishandling, further increasing operating costs and safety risks.

[0082] In summary, existing technologies face core challenges in water wading detection and handling, namely insufficient perception reliability and a lack of vehicle-cloud collaborative closed-loop management. Visual perception solutions suffer from insufficient detection accuracy due to harsh environments, while single ultrasonic solutions lack multi-source fusion and remote fallback mechanisms, resulting in low decision-making reliability. Furthermore, existing solutions generally lack tiered handling strategies and closed-loop management systems, further exacerbating the technical problems of insufficient reliability in water wading safety handling in autonomous driving scenarios.

[0083] Therefore, this application provides a vehicle wading detection method, which acquires water level data and visual data of the surrounding environment during vehicle driving, combines the water level data and visual data for fusion processing to evaluate the reliability of wading detection, obtains wading parameters, and generates a hierarchical control strategy based on the wading parameters and water level data to control the vehicle to perform corresponding driving operations, and reports wading detection information to the cloud when the conditions are met. This method introduces visual data to determine the credibility of wading detection results, achieving a fusion judgment between sensor detection and visual semantic perception. This effectively overcomes the problem of insufficient detection accuracy of a single sensor in harsh environments, improving the reliability of wading detection. Furthermore, through a hierarchical control strategy based on wading parameters and water level data, as well as a collaborative confirmation mechanism between the vehicle and the cloud, a cloud-based safety officer can make remote backup decisions when the vehicle's perception results are unreliable or the wading risk is high. This improves the traffic efficiency and operational continuity of autonomous vehicles in wading scenarios while ensuring traffic safety. It is applicable to autonomous driving operation scenarios with different water depths and various environmental conditions, reducing the frequency of human intervention and operating costs, and providing technical support for the large-scale commercial operation of L4 autonomous vehicles.

[0084] To facilitate understanding, the following will be combined with... Figure 1 The application scenarios applicable to the embodiments of this application will be described.

[0085] Figure 1 This is a schematic diagram illustrating an application scenario provided in an embodiment of this application. Please refer to [link / reference]. Figure 1 This includes vehicles, vehicle wading detection equipment, and cloud servers. The vehicle wading detection equipment can be deployed in vehicles.

[0086] When a vehicle travels to a flooded section of road, it acquires water level data and visual data of the surrounding environment through a vehicle water wading detection device. Based on the water level data and visual data, it generates a corresponding water wading control strategy, controls the vehicle to perform corresponding driving operations, and sends water wading detection information to the cloud server when preset conditions are met, so as to realize the safe passage of unmanned vehicles in complex waterlogged scenarios and vehicle-cloud collaborative management.

[0087] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0088] Figure 2 This is a schematic flowchart illustrating a vehicle wading detection method provided in an embodiment of this application. Please refer to... Figure 2 As shown, the method may include the following steps:

[0089] S201. Obtain the first water level data of the vehicle driving in the preset scenario, as well as the first visual data of the vehicle's surrounding environment.

[0090] The execution subject of this application embodiment can be an electronic device or a vehicle wading detection device installed in an electronic device. The vehicle wading detection device can be implemented by software or by a combination of software and hardware.

[0091] The preset scenarios can refer to water crossing scenarios encountered by L4 level fully autonomous vehicles during driving, including but not limited to flat road water accumulation scenarios, slope water accumulation scenarios, water accumulation scenarios in rainy or foggy weather, water accumulation scenarios with strong light reflection on the water surface, and complex water accumulation scenarios with turbid water or hidden obstacles underwater.

[0092] The first water level data can refer to the road surface water depth data collected by an ultrasonic wading detection system installed on the bottom or sides of the vehicle. In some embodiments, the first water level data can be a water depth value, or it can include the detection location information corresponding to the water level data.

[0093] First-vision data can refer to environmental perception data of the vehicle's surroundings collected by vision sensors (including but not limited to cameras, LiDAR, etc.) placed around the vehicle. In some embodiments, first-vision data may include images and / or video data of the road surface ahead collected by cameras, and environmental point cloud data scanned by LiDAR.

[0094] Understandably, first-vision data can be the raw perception data of the vehicle's surrounding environment collected in real time by the vehicle's vision sensors. This includes environmental information such as road conditions, the extent of water accumulation, road obstacles, and driving environment videos, which can intuitively reflect the real environmental conditions of the water-related scenarios around the vehicle.

[0095] Below, in conjunction with Figure 3 The structure of the vehicle-end water wading detection device is described in detail.

[0096] Figure 3 This is a schematic diagram of the vehicle-end wading detection device provided in an embodiment of this application. Please refer to... Figure 3 The vehicle-mounted water wading detection device includes a main ultrasonic sensor, a slave ultrasonic sensor, and a water depth detection controller.

[0097] The main ultrasonic sensor and the slave ultrasonic sensor are respectively installed at the rearview mirror positions on both sides of the vehicle, at a height h above the ground, for real-time detection of the water depth on the road surface. The main sensor and the slave sensor are electrically connected, and the main sensor integrates a water depth detection controller. The water depth detection controller is connected to the signal output terminals of both the main sensor and the slave sensor, and is used to receive the detection signals from the main sensor and the slave sensor, and calculate the current wading depth of the vehicle based on vehicle parameters (e.g., vehicle attitude information, sensor installation height, etc.) and the time difference between the detection signals from the main sensor and the slave sensor.

[0098] In some embodiments, when acquiring the first water level data, two wading sensors (ultrasonic sensors) are respectively arranged on both sides of the vehicle body to detect the road surface water depth. The main sensor is equipped with a water depth detection controller, which calculates the current wading depth based on vehicle parameters and the detection signals from the main and slave sensors, and sends the water depth and alarm information to the autonomous driving domain controller via CAN / CANFD signals. The sensors can detect different road conditions (e.g., flat roads, slopes, etc.) and send the detection results to the autonomous driving controller. After receiving the water depth data, the autonomous driving domain controller combines it with environmental video data collected by the visual sensor to determine subsequent wading parameters and generate strategies.

[0099] In some embodiments, when acquiring the first visual data, environmental videos (e.g., road conditions, obstacles, etc.) of the vehicle's surroundings are captured by cameras positioned around the vehicle to provide general visual perception data for autonomous driving; and / or, three-dimensional point cloud data of the vehicle's surrounding environment is acquired by LiDAR. The visual sensors send the acquired environmental videos and / or point cloud data to the autonomous driving domain controller for subsequent fusion processing.

[0100] S202. Determine the water-related parameters based on the first water level data and the first visual data.

[0101] Among them, the wading parameter is used to indicate the reliability of the vehicle's passage through waterlogged areas. Understandably, the wading parameter is a confidence level parameter used to indicate whether the vehicle's wading detection results are reliable; its value can be used to determine whether the vehicle makes an autonomous decision to pass or needs to request cloud intervention.

[0102] In some embodiments, after receiving the first water level data and the first visual data, the reliability of the wading detection results can be evaluated by combining the algorithm itself. Specifically, the water level data and visual data can be fused to obtain fused data, which is then input into a pre-trained neural network model. After calculation by the neural network model, a wading parameter (confidence level C) is output. This wading parameter is used to indicate the reliability of the current water depth detection result, providing a decision-making basis for the generation of subsequent graded control strategies.

[0103] S203. Generate a water-related control strategy based on the water-related parameters and the first water level data.

[0104] The wading control strategy is used to implement tiered control of the vehicle. It can be understood that the wading control strategy is a three-tiered safety response strategy determined based on wading parameters and initial water level data, including but not limited to a safe passage strategy, a cautious passage strategy, and a no-passage strategy, used to implement differentiated driving control of the vehicle in different wading scenarios.

[0105] In some embodiments, the vehicle-mounted terminal can obtain preset confidence thresholds, safe water depth thresholds, and risk water depth thresholds, and compare the wading parameters with the corresponding thresholds. If the wading parameter is greater than or equal to the confidence threshold, it indicates that the vehicle-mounted perception result is reliable, and the water depth is further compared with the safe water depth threshold and the risk water depth threshold. Based on the comparison result, a corresponding hierarchical control strategy (safe passage / cautious passage / no passage) is generated. If the wading parameter is less than the confidence threshold, it indicates that the vehicle-mounted perception result is unreliable, and the water level threshold comparison is skipped, generating a cloud takeover strategy. Through the above method, dynamic allocation of vehicle-cloud decision-making power based on perception reliability is achieved.

[0106] S204. According to the water wading control strategy, control the vehicle to perform corresponding driving operations, and send water wading detection information to the cloud server when the preset conditions are met.

[0107] Preset conditions refer to pre-defined trigger conditions that require sending wading detection information to the cloud server. In some embodiments, preset conditions include, but are not limited to: requiring remote monitoring by a cloud-based safety operator when the wading control policy is a cautious approach policy, and requiring remote takeover decisions by a cloud-based safety operator when the wading control policy is a prohibition policy or a cloud takeover policy. It is understood that when the reliability of the vehicle-side perception result is low or the wading risk is high, the vehicle-cloud collaboration mechanism is triggered, and a cloud-based safety operator performs remote backup confirmation.

[0108] The wading detection information may refer to relevant information sent from a cloud server for remote safety officers to make monitoring or takeover decisions. In some embodiments, the wading detection information includes, but is not limited to: first water level data (water depth, etc.), wading parameters, multi-source perception data of the vehicle's surrounding environment (video data, point cloud data, etc.), vehicle location information, and vehicle speed information.

[0109] In some embodiments, the vehicle can be controlled to perform corresponding driving operations according to the generated wading control strategy. When the wading control strategy indicates that cloud intervention is required (e.g., in cases requiring remote monitoring or remote takeover), the vehicle sends wading detection information to the cloud server via the vehicle communication terminal. The cloud receives the data uploaded by the vehicle and automatically schedules and forwards it to the designated remote control cabin according to the busy status of the vehicle and the safety operator. The remote safety operator manually confirms the information by viewing environmental videos and issues decision instructions, which are then transmitted back to the autonomous driving domain controller via the cloud scheduling system and the vehicle communication terminal. The autonomous driving domain controller then controls the vehicle to perform the corresponding driving operations according to the instructions, realizing a closed-loop collaborative management and control system between the vehicle and the cloud.

[0110] In this embodiment, by acquiring water level data and visual data of the surrounding environment during vehicle operation, wading parameters are determined based on the fusion processing of water level data and visual data. A hierarchical control strategy is then generated based on the wading parameters and water level data to control the vehicle to perform corresponding driving operations. When conditions are met, wading detection information is sent to the cloud, achieving an organic combination of vehicle passage strategy and vehicle-cloud collaboration in wading scenarios. In this method, the electronic device can accurately assess the reliability of wading detection results through the fusion of water level data and visual data, effectively avoiding control deviations caused by inaccurate detection by a single sensor in adverse environments such as rain, fog, and strong light reflection, thus improving the reliability of wading detection. Simultaneously, through the hierarchical control strategy based on wading parameters and water level data, combined with the collaborative confirmation mechanism between the vehicle and the cloud, the cloud makes remote backup decisions when the vehicle's perception results are unreliable or the wading risk is high, avoiding operational interruptions due to insufficient vehicle-side decision-making capabilities. This improves the passage efficiency and operational continuity of autonomous vehicles in wading scenarios while ensuring traffic safety.

[0111] Based on any of the above embodiments, the following, in conjunction with Figure 3 The process of determining water-related parameters based on the first water level data and the first visual data is explained in detail.

[0112] Figure 4 This is a schematic diagram illustrating the process of determining water-related parameters provided in an embodiment of this application. Please refer to... Figure 4 The method may include:

[0113] S401. Preprocess the first water level data and the first visual data to obtain fused data.

[0114] Fusion data refers to structured data obtained by aligning and fusing first water level data with first visual data, and used as input to the first model. Fusion data includes water level depth information and its corresponding spatial location information, as well as spatial distribution information of water accumulation areas in the visual data.

[0115] In some embodiments, the first water level data and the first visual data can be preprocessed to obtain fused data in the following manner: the first water level data and the first visual data are aligned to obtain second water level data and second visual data; the second water level data and the second visual data are fused to obtain fused data.

[0116] The second water level data can refer to the water level data obtained after aligning the first water level data. Its data content is the same as the first water level data (both contain water level depth information). However, after alignment, the second water level data is on the same time reference as the visual data in the time dimension, and has a format for coordinate matching with the visual data in the spatial dimension.

[0117] Second visual data can refer to visual data obtained after aligning with first visual data. Its data content is the same as that of first visual data (both contain image / video data). However, after alignment, second visual data is on the same time reference as water level data in the time dimension and has a coordinate mapping relationship with detection coordinates in the spatial dimension.

[0118] In some embodiments, ultrasonic sensors and visual sensors collect water level data and visual data, respectively. These two data differ in their acquisition time, coordinate system definition, data format, and physical meaning. Directly fusing unaligned water level data and visual data can lead to information misalignment due to inconsistent spatiotemporal references, thus reducing the fusion quality. By preprocessing the first water level data and the first visual data (including time synchronization and spatial registration), they can be unified to the same time reference and spatial reference system. This ensures that the data pairs used in subsequent fusion processing describe the same scene at the same time in time and point to the same physical location in space, improving the accuracy and consistency of the fused data.

[0119] By aligning and fusing water level data and visual data in the above manner, an accurate correspondence is formed between the two heterogeneous data sources in time and space, providing a high-quality unified input for the first model and ensuring the reliability of the water-related parameter calculation results from the data source.

[0120] In some embodiments, the first water level data and the first visual data can be aligned to obtain the second water level data and the second visual data in the following manner: the first water level data and the first visual data are time-synchronized to obtain the third water level data and the third visual data; the third water level data and the third visual data are spatially registered to obtain the second water level data and the second visual data.

[0121] The third water level data can refer to the water level data obtained after time synchronization processing of the first water level data. It is based on the same time reference as the visual data in the time dimension, but has not yet been integrated with the spatial coordinate system.

[0122] Third visual data can refer to visual data obtained after time synchronization processing of first visual data. It is based on the same time reference as water level data in the time dimension, but has not yet been integrated with the spatial coordinate system.

[0123] In some embodiments, the first water level data is acquired by an ultrasonic sensor, and the first visual data is acquired by a camera and / or LiDAR. The different sensors have different sampling frequencies (e.g., ultrasonic sensors at 20Hz-50Hz, cameras at 25Hz-30Hz, and LiDAR at 10Hz-20Hz), and their trigger times have phase differences. If water level data and visual data acquired at different times are fused, the scene described by the data will be inconsistent in time (e.g., the water level data corresponds to the vehicle's position at time t1, while the visual data corresponds to the vehicle's position at time t2, which may differ by hundreds of milliseconds), leading to deviations in the fusion result. Time synchronization processing aligns the water level data and visual data to the same time reference, ensuring that the data involved in the fusion describes the wading scene at the same time, providing a temporally consistent input for subsequent spatial registration and fusion.

[0124] By employing the above methods, through dual alignment of time synchronization and spatial registration, water level data and visual data achieve dual unification in both time and space. This ensures that the data used in subsequent fusion processing corresponds precisely in both time and space, avoiding information deviations caused by temporal and spatial misalignment.

[0125] In some embodiments, the third water level data and the third visual data can be obtained by: acquiring the first timestamp corresponding to the first water level data and the second timestamp corresponding to the first visual data; synchronizing the first water level data and the first visual data to the same time base according to the first timestamp and the second timestamp to obtain the third water level data and the third visual data.

[0126] The first timestamp refers to the time information corresponding to the moment when the first water level data was collected. When the ultrasonic sensor collects water level data, the depth detection controller records the collection time corresponding to that frame of data.

[0127] The second timestamp can refer to the time information corresponding to the moment the first visual data was acquired. When a camera acquires image data or a LiDAR acquires point cloud data, the visual sensor records the acquisition time corresponding to that frame of data.

[0128] In some embodiments, the first water level data is water depth data collected by an ultrasonic sensor. When collecting water level data, the water depth detection controller simultaneously records the acquisition time of this data frame and encapsulates this time as a timestamp along with the water level depth data, sending it to the autonomous driving domain controller via a CAN / CANFD signal. The autonomous driving domain controller parses the first timestamp from the received water level data frame.

[0129] The first visual data consists of image data captured by a camera and / or point cloud data captured by a LiDAR sensor. When acquiring each frame of image or point cloud data, the visual sensor simultaneously records the acquisition time and encapsulates this time as a timestamp along with the image or point cloud data before sending it to the autonomous driving domain controller. The autonomous driving domain controller then parses the second timestamp from the received visual data frames.

[0130] In some embodiments, the first timestamp of the first water level data and the second timestamp of the first visual data can be compared with the time reference, respectively, with a fixed time reference (e.g., the acquisition time of the visual data or the system synchronization clock).

[0131] For water level data, when the deviation between the first timestamp and the reference time exceeds a preset threshold (e.g., 50ms), the water level data sequence is resampled using a linear interpolation method to calculate the water level depth value corresponding to the reference time point. For visual data, when the deviation between the second timestamp and the reference time exceeds a preset threshold, the visual data frame with the timestamp closest to the reference time is selected as the synchronization frame, or the visual information corresponding to the reference time is calculated through motion compensation between multiple frames. After the above processing, the time-aligned water level data is the third water level data, and the time-aligned visual data is the third visual data. Both describe the wading scene at the same moment in the time dimension.

[0132] By employing the above method, timestamp comparison and interpolation synchronization are used to align asynchronously acquired water level data and visual data to the same time reference, eliminating data timing deviations caused by differences in sensor sampling frequencies and trigger times.

[0133] In some embodiments, the second water level data and the second visual data can be obtained in the following manner: the third water level data includes water level depth and detection coordinates; according to preset calibration parameters, the detection coordinates in the third water level data are projected onto the pixel coordinate system corresponding to the third visual data; the projected detection coordinates are matched with the pixel positions of the water accumulation area in the third visual data to obtain the second visual data, wherein the water level depth of the second water level data is the same as the water level depth of the third water level data.

[0134] Water level depth can refer to the core numerical information in the third water level data, namely the height of water on the road surface detected by the ultrasonic sensor, which represents the vertical distance between the water surface and the road surface at the sensor detection location.

[0135] The detection coordinates refer to the coordinate information of the detection position corresponding to the third water level data in three-dimensional space. The detection coordinates are determined by the installation position of the ultrasonic sensor. The sensor is fixed at a specific position on the vehicle body (for example, below the rearview mirror), and its detection beam points to a specific area on the bottom of the vehicle. The three-dimensional coordinates of this area in the vehicle coordinate system are the detection coordinates.

[0136] Preset calibration parameters can refer to a set of fixed parameters used to describe the intrinsic and extrinsic parameters of vision sensors such as cameras or lidar, including sensor intrinsic parameters (e.g., focal length, principal point coordinates, distortion coefficients, etc.) and sensor extrinsic parameters (e.g., the rotation matrix and translation vector of the sensor relative to the vehicle coordinate system).

[0137] The calibration parameters are usually obtained offline before the vehicle leaves the factory and remain unchanged during use. They are used to project coordinate points in three-dimensional space onto the image pixel coordinate system to realize the mapping between visual data and physical space.

[0138] In some embodiments, the detection coordinates in the third water level data are three-dimensional spatial point coordinates described in the vehicle coordinate system, while the third visual data are two-dimensional image data captured by a camera. In order to correspond the detection position of the water level data with the pixel position in the visual image, it is necessary to project the three-dimensional detection coordinates onto the two-dimensional image plane using preset camera calibration parameters.

[0139] Specifically, firstly, based on the camera's extrinsic parameters (rotation matrix R and translation vector t), the detection coordinates are transformed from the vehicle coordinate system to the camera coordinate system. Then, based on the camera's intrinsic parameters, the 3D points in the camera coordinate system are transformed using perspective projection to calculate their projected positions on the image normalization plane. Furthermore, the radial distortion of the vehicle-mounted lens is considered to correct the projected positions, ultimately obtaining the precise projected positions of the detection coordinates in the image pixel coordinate system. This projected position is the pixel coordinate corresponding to the ultrasonic sensor's detection position in the image.

[0140] In some embodiments, the water accumulation area in the third visual data (image) is first identified by a semantic segmentation algorithm or an object detection algorithm, and the pixel areas in the image that belong to the water accumulation area are marked as water accumulation area masks (i.e., the set of pixels corresponding to the water accumulation area).

[0141] Then, the projected detection coordinates are matched with the water accumulation area mask: if the projected detection coordinates fall within the water accumulation area mask, it indicates that the detection position corresponding to the detection coordinates corresponds to the water accumulation area in the image, and the water level data and the image frame point to the same physical location in space, the spatial registration is successful, and the matching result is consistent; if the projected detection coordinates do not fall within the water accumulation area mask, it indicates that the detection position corresponds to a non-water accumulation area (e.g., road surface, curb, etc.) in the image, and the matching result is inconsistent.

[0142] For cases where the matching results are consistent, the pixel positions of the water accumulation area corresponding to the detection coordinates in the current third visual data are labeled, and the resulting visual data becomes the second visual data. The water level depth in the second water level data is consistent with the water level depth in the third water level data, completing the coordinate mapping relationship with the visual data only in space. Through the above spatial registration process, a precise correspondence is achieved between the detection position of the water level data and the pixel positions of the water accumulation area in the visual data.

[0143] By using the above method, the detection coordinates of the water level data are projected onto the image pixel coordinate system through calibration parameters and matched with the pixel position of the water accumulation area. This achieves a precise spatial correspondence between the ultrasonic detection position and the water accumulation area in the image, ensuring the consistency between the water level data and the visual data from a spatial perspective.

[0144] S402. Input the fused data into the first model to obtain the water-related parameters.

[0145] The first model can refer to a pre-trained machine learning model used to output water crossing parameters based on fused data. The first model is capable of extracting features and assessing the confidence level of the input fused data, and outputting a confidence value (i.e., the water crossing parameter).

[0146] Optionally, the first model can be a regression or classification model built on a deep neural network, whose network structure may include a feature extraction layer, a feature fusion layer, and an output layer. The first model has been pre-trained and its parameters have been fixed before being deployed to the autonomous driving domain controller, and it only performs forward inference during vehicle operation.

[0147] It should be noted that the specific network structure of the model is not limited in the embodiments of this application. Any model that can output a credibility evaluation value based on the fusion result of water level data and visual data can be used as the implementation method of the first model.

[0148] In some embodiments, the fused data obtained in step S401 can be standardized according to the input format preset by the first model to form a model input tensor. The fused data includes water level depth information, detection coordinate information, and spatial distribution characteristics of the water accumulation area in the visual data. The model input tensor is input into the first model, which performs feature extraction and mapping on the input data through forward computation, outputting a continuous value between 0 and 1. This value is the wading parameter (confidence level C), used to indicate the reliability of the current wading detection result.

[0149] The closer the confidence level is to 1, the more reliable the vehicle-side perception results are; the closer the confidence level is to 0, the less reliable the vehicle-side perception results are, and the cloud-based collaboration mechanism needs to be triggered.

[0150] For example, taking the first model as a multimodal fusion evaluation model based on a Transformer encoder, the specific reasoning process is as follows: The model adopts a dual-branch input architecture, with water level features and visual features entering independent feature encoding branches. The two features first pass through a location encoding layer to supplement spatial and temporal location information, and then each pass through a multi-layer Transformer encoder to extract deep abstract features. Subsequently, the two deep features are input into a cross-modal attention module, and the attention weights are used to calculate and mine the corresponding association between the water level detection value and the visual water accumulation area, realizing the interactive fusion of heterogeneous features. Finally, the fused features pass through a fully connected layer and a normalization layer to output water wading parameters, completing the reliable quantitative evaluation.

[0151] In some embodiments, the first model can be trained using supervised learning, which includes four core parts: model structure, training dataset, loss function, and data association, as detailed below:

[0152] At the model structure level, a mature fusion architecture combining a dual-branch Transformer encoder and cross-modal attention can be adopted. The two branches are adapted to numerical water level features and image-based visual features respectively, and the end outputs the credibility result through a fully connected layer. The whole is an existing general multimodal fusion network without any special custom structure.

[0153] The training data for the first model comes from historical water-related scenario data. Specifically, during vehicle road tests or actual operation, a large amount of water level data, visual data, and actual vehicle passage results are collected through onboard sensors. Cloud-based safety personnel or data labelers then label the reliability of the detection results for each set of data, forming a labeled training sample set. The training sample set should cover various water-related scenario types, including different water depths (shallow, medium, and deep water), different environmental conditions (sunny days, rain and fog, strong light reflection, and turbid water), and different road surface conditions (flat roads and slopes), to ensure that the model has good scenario generalization ability.

[0154] In this way, each training sample contains spatiotemporally synchronized water level data, visual image data, and manually labeled credibility tags, ensuring the consistency of the distribution of training data with real vehicle application scenarios.

[0155] The first model can be trained using a loss function suitable for regression tasks. For example, the mean squared error loss function can be used, calculating the mean squared error between the model's predicted confidence values ​​and the manually labeled true confidence labels as the training loss. Other loss functions suitable for regression tasks, such as the mean absolute error loss function, can also be used. During training, the network parameters of the model are continuously adjusted using the backpropagation algorithm, causing the loss function value to gradually decrease and converge, ultimately resulting in the trained first model.

[0156] The inherent correlation between the data is reflected in the following: during the training process, the model automatically learns the correspondence between water level depth values ​​and visual water accumulation area morphology and environmental interference conditions under different working conditions, mines the matching features between ultrasonic detection results and visual recognition results, establishes the mapping relationship between water level values, visual scenes and credibility, and supports automatic credibility assessment in the inference stage.

[0157] During actual vehicle operation, the autonomous driving domain controller loads the parameters of the first model that has been trained. When the vehicle reaches a flooded section and completes the collection and fusion of water level and visual data, the fused data is input into the first model for forward inference. Based on the trained parameters, the first model automatically extracts features and performs mapping calculations on the input data, outputting flooding parameters.

[0158] It should be noted that the above training and inference processes are merely illustrative examples. Those skilled in the art can select different network structures, loss functions, and training strategies to implement the first model according to actual needs. This application does not impose any specific limitations on these aspects.

[0159] In this embodiment, by performing time synchronization and spatial registration processing on the first water level data and the first visual data, the detection coordinates in the third water level data are projected onto the pixel coordinate system corresponding to the third visual data, and matched with the pixel positions of the water accumulation area in the third visual data, thus achieving alignment between the water level data and the visual data. Then, the aligned second water level data and the second visual data are fused, and the resulting fused data is input into the first model to obtain the wading parameters. This method, based on the spatiotemporal alignment and fusion processing of water level data and visual data, accurately associates and matches the ultrasonic detection location with the water accumulation area in the image. This effectively avoids conflicts in time and space between multiple sources of information and the misjudgment caused by the inaccuracy of a single sensor in harsh environments such as rain, fog, and strong light reflection from the water surface, thereby improving the accuracy of wading parameter determination.

[0160] Based on any of the above embodiments, the following, in conjunction with Figure 5 The process of generating a water control strategy based on water-related parameters and first water level data is explained in detail.

[0161] Figure 5 This is a schematic diagram illustrating the process of determining the water-related control strategy provided in an embodiment of this application. Please refer to... Figure 5 The method may include:

[0162] S501. When the wading parameter is greater than or equal to the first threshold and the water level depth in the first water level data is less than or equal to the second threshold, the wading control strategy is determined to be: control the vehicle to pass through the corresponding waterlogged area.

[0163] The first threshold can refer to the confidence limit for determining the reliability of the vehicle-side wading test results, denoted as... The first threshold is a pre-calibrated constant with a value ranging from 0 to 1.

[0164] When the wading parameter (confidence level C) is greater than or equal to the first threshold, it indicates that the vehicle-side perception result is reliable; when the wading parameter is less than the first threshold, it indicates that the vehicle-side perception result is unreliable. The first threshold can be calibrated according to the system's safety requirements; the higher the safety requirements, the larger the value of the first threshold.

[0165] The second threshold can refer to the safe water depth threshold, denoted as... Its value is related to the vehicle's own physical parameters such as chassis height and wading capability. When the water depth is less than or equal to the second threshold, it indicates that the current water depth is within the vehicle's safe wading range, and the vehicle can pass normally.

[0166] In some embodiments, after obtaining a preset first threshold (confidence threshold) and a second threshold (safe water depth threshold), and reading the wading parameter C and water level depth h, if This indicates that the vehicle-side water wading test results are reliable; further judgment is needed. This indicates that the current water depth is within a safe range, and the vehicle can pass safely without slowing down or detouring. At this time, the autonomous driving domain controller determines the wading control strategy as a safe passage strategy, that is, controlling the vehicle to maintain its current driving state and pass through the waterlogged area.

[0167] By using the above methods, when the water level parameters are high (the perception results are reliable) and the water depth is within a safe range, the normal passage of vehicles can be controlled, and the operation can be maintained without interruption while ensuring safety. This avoids unnecessary stops caused by excessive conservatism and improves traffic efficiency.

[0168] S502. When the wading parameters are greater than or equal to the first threshold and the water depth is greater than the second threshold and less than or equal to the third threshold, the wading control strategy is determined as follows: control the vehicle to slow down and pass through the waterlogged area, and send a vehicle takeover request to the cloud server. The vehicle takeover request includes wading detection information.

[0169] The third threshold can refer to the risk water depth threshold, denoted as... When the water depth is greater than the second threshold but less than or equal to the third threshold, it indicates that the current water depth is within the critical range, and vehicles can pass at low speed but there is a certain risk. When the water depth is greater than the third threshold, it indicates that the current water depth exceeds the vehicle's safe wading capability, and vehicles are prohibited from passing.

[0170] In some embodiments, after obtaining preset first threshold, second threshold, and third threshold, and reading the wading parameter C and water level depth h, if This indicates that the vehicle-side water wading test results are reliable; further judgment is needed. This indicates that the current water depth has exceeded the safe range but has not yet reached the risk limit. Vehicles can pass through at low speed, but there is a certain risk involved.

[0171] At this point, the autonomous driving domain controller determines the wading control strategy to be a cautious passage strategy, which means controlling the vehicle to turn on the hazard warning lights and reduce the vehicle speed. At the same time, it sends a vehicle takeover request to the cloud server. The takeover request includes wading detection information (water level data, wading parameters, environmental video, etc.) for the cloud safety operator to remotely monitor and issue intervention instructions when necessary.

[0172] By using the above methods, when the water depth is within the critical range, the vehicle speed is reduced to decrease the risk of wading through water. At the same time, a takeover request is sent to the cloud to achieve remote monitoring. This ensures safe passage and avoids operational interruptions caused by immediate stopping, thus balancing safety and efficiency.

[0173] S503. When the water wading parameters are greater than or equal to the first threshold and the water depth is greater than the third threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0174] In some embodiments, if This indicates that the vehicle-side water wading test results are reliable; further judgment is needed. This indicates that the current water depth exceeds the vehicle's safe wading capability, and forcibly passing through may lead to serious safety risks such as water entering the vehicle, short circuit of the power battery, or even vehicle stalling.

[0175] At this point, the water control strategy is determined to be a no-passage strategy, which means controlling the vehicle to turn on its hazard lights and pull over to the side of the road (entering the minimum risk state). At the same time, a vehicle takeover request is sent to the cloud server. The takeover request includes water detection information (water level data, water parameters, environmental video, etc.) so that the cloud safety officer can remotely confirm and issue decision instructions.

[0176] Using the above methods, when the water depth exceeds the vehicle's safe wading capability, the vehicle is controlled to stop and cloud takeover is requested to prevent safety accidents caused by the vehicle forcibly wading through the water. The safety of complex scenarios is ensured by remote confirmation by a cloud-based safety officer.

[0177] S504. When the water wading parameter is less than the first threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0178] In some embodiments, if This indicates that the vehicle-side water wading detection results are unreliable (e.g., ultrasonic sensors are affected by water surface waves, and visual sensors are affected by rain, fog, or strong light reflection, leading to a decrease in perception accuracy). In this case, the value of the water depth h itself no longer provides a reliable basis for decision-making. Regardless of the water depth value, the autonomous driving domain controller skips the hierarchical control logic based on water depth and directly determines the water wading control strategy as a cloud-based takeover strategy. This means controlling the vehicle to activate its hazard lights and pull over (entering a minimum risk state), while simultaneously sending a vehicle takeover request to the cloud server. The takeover request includes the water wading detection information, allowing the cloud-based safety operator to manually confirm the information by reviewing environmental video and issue decision instructions.

[0179] By using the above method, when the water level parameters are low, the vehicle can be stopped directly and the cloud can be requested to take over, bypassing the water level classification. This effectively avoids misjudgment or missed judgment caused by insufficient sensor perception accuracy, and realizes a safety backup in the case of perception failure.

[0180] In this embodiment, by acquiring wading parameters and water depth from the first water level data, the wading parameters are compared with a first threshold, and the water depth is compared with a second threshold and a third threshold, respectively. Based on the combined judgment conditions of the wading parameters and water depth, the corresponding wading control strategy is determined. In this method, a two-layer judgment based on wading parameters and water depth allows for multi-level differentiated control based on water depth when the perception result is reliable, and skips water depth grading to directly trigger cloud takeover when the perception result is unreliable. This effectively avoids misjudgment caused by relying solely on water depth thresholds when perception fails, improving the rationality and reliability of vehicle passage decisions in wading scenarios. Furthermore, this method is applicable to autonomous driving operation scenarios with different confidence levels and various water depth conditions, improving the passage efficiency and operational continuity of autonomous vehicles in wading scenarios while ensuring traffic safety.

[0181] Figure 6 This is a schematic flowchart illustrating another vehicle wading detection method provided in this application embodiment. Please refer to... Figure 6 The method may include:

[0182] S601. Input the first water level data and water-related parameters into the second model to obtain the passage parameters of the waterlogged area.

[0183] The second model can refer to a pre-trained machine learning model used to output passage parameters for flooded areas based on the first water level data and wading parameters. The second model can comprehensively evaluate the input water level data and wading parameters to predict the difficulty of vehicles passing through the current flooded area and recommend passage methods.

[0184] The second model can be the same as or different from the first model, and can be determined according to actual needs. The second model can be deployed on the vehicle side (such as in the autonomous driving domain controller).

[0185] Traffic parameters can refer to information indicating the recommended passage method and related control parameters for vehicles to pass through the current flooded area. Traffic parameters may include, but are not limited to: recommended passage method (e.g., normal passage, low-speed passage, not recommended passage), recommended passage speed, and vehicle motion data suggestions (e.g., acceleration limit, steering angle limit, torque output limit, etc.).

[0186] In some embodiments, by acquiring the first water level data (including water depth h) and wading parameters (confidence C) in the current wading scenario, the water level depth h and wading parameters C are combined into a model input feature vector, processed according to the input format preset by the second model, and input into the second model. The second model performs nonlinear mapping on the input features through forward inference and outputs the passage parameters for the current waterlogged area (e.g., recommended passage method, recommended vehicle speed, etc.). If the second model is deployed in the cloud, the vehicle uploads the water level data and wading parameters to the cloud server. The cloud server sends the input data into the second model for inference, obtains the passage parameters, and then sends them back to the vehicle via the vehicle-cloud link.

[0187] For example, taking the second model as a classification and regression model built on decision trees or lightweight neural networks: the second model can include an input layer, multiple fully connected layers, and an output layer. During inference, the water level depth h and the confidence score C are concatenated to form a two-dimensional input feature vector, which undergoes nonlinear transformation through multiple fully connected layers. Finally, two parallel output branches output the classification result and the regression result, respectively. The classification branch outputs the probability distribution of the three travel methods through the Softmax function, and the regression branch outputs the recommended travel speed value. The travel method with the highest classification probability is taken as the recommended travel method, and together with the recommended speed value output by the regression branch, they form the travel parameters.

[0188] Regarding the model structure and training process: The second model can adopt a classification and regression model based on deep neural networks. The model structure can include an input layer, a feature extraction layer, and an output layer. The feature extraction layer consists of multiple fully connected layers, used to extract water wading scene features from the input water level depth and confidence level. The output layer includes a classification head and a regression head. The classification head outputs the probability distribution of the travel mode, and the regression head outputs continuous values ​​such as the recommended speed.

[0189] The training data for the second model comes from a historical water crossing event database. During vehicle road tests or actual operation, water level depth, water crossing parameters, environmental information, and the final vehicle passage result are recorded for each water crossing event, forming labeled training samples. The training samples should cover various passage results under different water depths and confidence levels to ensure that the model can output accurate passage parameters in various scenarios. The loss function consists of two parts: classification loss and regression loss. The classification loss can use the cross-entropy loss function, and the regression loss can use the mean squared error loss function. The total loss is the weighted sum of the two losses. During training, the network parameters of the model are continuously adjusted through the backpropagation algorithm, so that the total loss function value gradually decreases and converges, finally obtaining the trained second model. It should be noted that the above model structure is only an illustrative example, and the specific network structure of the second model is not limited in this application embodiment.

[0190] S602. Generate a water-related inspection report based on the passage parameters and the first water level data.

[0191] A wading inspection report refers to structured information describing the inspection results of a current wading scenario, providing a reference for remote decision-making by cloud-based safety personnel. The report may include water level and depth information, passage parameters, environmental descriptions of the flooded area, risk assessment conclusions, and recommended passage methods. The wading inspection report can be sent to the cloud server along with a vehicle takeover request.

[0192] In some embodiments, a water wading detection report may be generated based on traffic parameters and first water level data in the following manner: determining vehicle movement data corresponding to passing through the waterlogged area based on traffic parameters; and generating a water wading detection report based on the first water level data and vehicle movement data.

[0193] Vehicle motion data refers to the specific motion control parameters used to guide a vehicle through a flooded area. Vehicle motion data may include, but is not limited to: recommended speed, acceleration limits, steering angle limits, and motor torque output limits.

[0194] In some embodiments, after obtaining the traffic parameters output by the second model, if the recommended traffic mode in the traffic parameters is normal traffic or low-speed traffic, then the suggested traffic speed value in the vehicle motion data is determined based on the recommended traffic speed; simultaneously, acceleration limit values ​​and steering angle limit values ​​are determined based on the water depth h (for example, limiting acceleration to suppress wheel slippage and limiting steering angle to prevent vehicle roll when the water level is deep). The above control parameters are then summarized to form the vehicle motion data.

[0195] In some embodiments, a wading test report is generated by combining information such as water depth and water accumulation range from the first water level data, and recommended travel speed, acceleration limit, and steering angle limit from the vehicle motion data. The wading test report may include, but is not limited to, the following: water depth information, travel parameters (recommended travel method), vehicle motion data (recommended travel speed, acceleration limit, steering limit), and vehicle location information. The wading test report can be sent to a cloud server along with a vehicle takeover request for remote safety operator reference.

[0196] Using the above method, the vehicle movement data corresponding to the water accumulation area is determined by the traffic parameters, and a water wading detection report is generated based on the water level data and vehicle movement data. This provides the cloud-based safety officer with a structured scenario description and traffic suggestions, helping the safety officer to quickly understand the scenario and make decisions.

[0197] In this embodiment, by inputting the first water level data and wading parameters into the second model, passage parameters for the flooded area are obtained, and a wading detection report is generated based on the passage parameters and the first water level data. In this method, the electronic device can automatically output recommended passage methods and speeds for the current flooded scenario through a comprehensive evaluation of water depth and confidence level using the second model. Based on this, a structured wading detection report containing vehicle motion data suggestions is generated, providing cloud-based safety operators with a clear and complete scenario description and passage recommendations. This effectively reduces the time required for safety operators to analyze videos and judge scenarios, improves the efficiency and consistency of remote decision-making, and reduces the risk of misjudgment due to human analysis oversight. It is applicable to various wading scenarios with different water depths and confidence levels, providing efficient technical support for vehicle-cloud collaborative decision-making in wading scenarios for autonomous vehicles.

[0198] Figure 7 This is a schematic flowchart illustrating another vehicle wading detection method provided in this application embodiment. Please refer to... Figure 7 The method may include:

[0199] S701, Receive vehicle takeover request sent by the vehicle.

[0200] The vehicle takeover request included water immersion test reports and video data.

[0201] Video data can refer to the image information of the vehicle's surrounding environment collected by the vehicle's vision sensors.

[0202] S702. Determine the vehicle control method based on the water wading test report and video data.

[0203] The control method can be either contactless control or contact control.

[0204] Contactless control refers to a control method where the system automatically generates control commands and issues them to the vehicle based on water wading test reports and video data. Under contactless control, the system comprehensively analyzes the reports and videos through preset decision logic to generate recommended control commands, eliminating the need for direct operation by a safety operator.

[0205] Contact control refers to a control method where a remote safety operator, through a remote control terminal, directly drives a vehicle through a flooded area based on water wading detection reports and video data. Under contact control, the safety operator controls the vehicle's acceleration, braking, steering, and other actions in real time through the control terminal, achieving remote driving of the vehicle.

[0206] In some embodiments, the vehicle control method can be determined as follows: when the control method is contactless control; based on the vehicle motion data in the water wading test report and the road condition information in the video data, the vehicle passes through the waterlogged area in the manner corresponding to the control command.

[0207] Specifically, after receiving the water wading detection report and video data, the system parses vehicle motion data from the report and uses video analysis algorithms to identify road condition information in the video data (e.g., identifying waterlogged area boundaries, detecting floating obstacles, and determining if the road surface is a slope). Then, based on preset safety rules, the system corrects the suggested speed in the vehicle motion data to obtain the final control speed, while keeping acceleration and steering angle limits unchanged, generating complete control commands.

[0208] Through the above methods, when the control method is contactless, the cloud automatically generates and issues control commands based on the vehicle movement data and video data in the water wading detection report, realizing a closed loop of automatic system decision-making in water wading scenarios, reducing the operational burden on safety personnel, and improving response efficiency.

[0209] In some embodiments, the vehicle control method may be determined as follows: when the control method is contact control; in response to a user's driving operation input based on a water wading detection report and video data, the vehicle passes through the waterlogged area with a corresponding control command.

[0210] Specifically, the remote control cabin is equipped with a simulated driving terminal, including a steering wheel, accelerator pedal, and brake pedal. The remote safety operator generates driving control signals in real time through this terminal. After receiving the safety operator's driving signals, the remote control cabin system converts them into control commands that the vehicle can execute. If the safety operator determines that the current flooded area is passable, they control the vehicle to slowly accelerate forward via the terminal. If the safety operator observes underwater obstacles or rapid currents, they control the vehicle to decelerate or reverse back to a safe area via the terminal. If the safety operator determines that the vehicle cannot pass safely, they control the vehicle to come to a complete stop and maintain a parked state via the terminal, while simultaneously calling for roadside assistance.

[0211] In this way, when the control method is contact control, the remote safety officer can operate the vehicle in real time to pass through the water-crossing area based on the water wading detection report and video data. This fully utilizes the comprehensive judgment ability of humans to handle complex scenarios and ensures traffic safety under extreme conditions.

[0212] S703. Based on the control method, generate control commands and issue them to the vehicle to enable the vehicle to safely pass through the flooded area.

[0213] In some embodiments, if the control method is contactless control, the cloud server automatically generates control commands based on the analysis results of vehicle motion data and video data in the water wading detection report, and sends them to the vehicle via the T-BOX through the cloud dispatch system. If the control method is contact control, the remote cabin system generates control commands in real time based on the driving operations input by the remote safety operator through the operating terminal, and sends them to the vehicle via the cloud dispatch system and T-BOX. After receiving the control commands, the autonomous driving domain controller controls the vehicle's acceleration, braking, steering, and other actuators according to the command content to realize the corresponding driving operations of the vehicle and ensure that the vehicle safely passes through the waterlogged area.

[0214] In this embodiment, the system can automatically determine whether to use contactless control (automatic decision-making) or contact control (remote driving by a safety operator) based on water wading detection reports and video data. This enables flexible switching between automated machine decision-making and remote manual control in complex water wading scenarios. When using contactless control, the system can automatically generate and issue control commands, effectively improving response efficiency. When using contact control, a remote safety operator can directly control the vehicle to pass through, ensuring traffic safety under extreme conditions. Through these methods, machine and human collaboration can be achieved in different water wading scenarios, improving the flexibility and safety of remote decision-making in water wading situations.

[0215] Figure 8 This is a schematic diagram illustrating the interaction between the vehicle-mounted terminal and the cloud server provided in an embodiment of this application. Please refer to... Figure 8 The process begins with wading sensors detecting the wading depth of the road surface in real time and outputting the water depth signal to the autonomous driving domain controller. Simultaneously, visual sensors collect environmental video and road condition information (first-line visual data) around the vehicle and send it to the autonomous driving domain controller. After receiving the water depth and visual data, the autonomous driving domain controller uses a fusion algorithm to assess the reliability of the wading detection results, generates wading parameters, and executes vehicle-side hierarchical decisions. When the vehicle-side perception results are unreliable or the wading risk is high, the autonomous driving domain controller uploads vehicle status, takeover requests, and perception video data to the cloud via the T-BOX (vehicle communication terminal). The T-BOX acts as a data relay station between the vehicle and the cloud, sending vehicle-side data to the cloud dispatch system via an uplink. The cloud dispatch system automatically forwards the data to the designated remote control based on the vehicle and safety driver status. After reviewing the water wading test report and video data in the remote cabin, the safety operator makes a comprehensive judgment and generates a safety operator decision (e.g., whether to take over, how to adjust vehicle control). This decision is transmitted back to the autonomous driving domain controller via the downlink of T-BOX through the cloud scheduling system. Based on the decision, the controller controls the vehicle to perform the corresponding driving operation, completing the closed-loop management of the entire process of vehicle detection, cloud confirmation, and vehicle execution.

[0216] Figure 9This is a schematic flowchart illustrating another vehicle wading detection method provided in this application embodiment. Please refer to... Figure 9 The process begins with wading detection at the vehicle end. The wading sensor sends the detected water depth result to the autonomous driving domain controller. Simultaneously, the domain controller receives environmental perception data collected by the visual sensor, fuses the water depth detection result and visual perception data, and combines this with a model to output wading parameters (i.e., confidence level C). This confidence level describes whether the current water depth detection result is reliable. The domain controller then obtains a preset confidence level threshold and compares the confidence level with the threshold. If... This indicates that the vehicle-side perception results are reliable, and further multi-level hierarchical control is performed based on the water depth data: when At that time, the vehicle autonomously decides to pass safely; when At that time, the vehicle slows down and passes through, triggering remote monitoring in the cloud; when At that time, the vehicle executes a pullover maneuver and triggers a cloud-based takeover request. If This indicates that the vehicle's perception results are unreliable. At this point, regardless of the water depth value, the water depth classification logic is skipped, and the vehicle is directly pulled over to the side of the road and a takeover request is sent to the cloud for remote comprehensive decision-making by the cloud safety officer.

[0217] In this embodiment, by fusing master-slave dual ultrasonic sensors with vehicle-side visual perception data, the depth of wading water can be accurately detected in harsh environments such as rain, fog, strong light reflection from the water surface, and turbid water, effectively improving the accuracy and anti-interference capability of wading detection. Furthermore, a graded handling strategy based on wading depth and confidence level allows for flexible adjustment of traffic patterns according to different wading scenarios, ensuring safety while avoiding operational interruptions caused by blind parking. Moreover, a vehicle-cloud collaborative fallback mechanism allows for remote confirmation and decision-making by a cloud-based safety officer in situations where vehicle-side perception results are unreliable or complex wading scenarios, significantly reducing the frequency of on-site handling and lowering operating costs. Simultaneously, a complete closed-loop management system encompassing vehicle-side detection, fusion judgment, graded handling, cloud confirmation, and post-mortem optimization ensures that all wading-related data is uploaded to cloud storage, which can be used for event tracing and strategy optimization, continuously improving system reliability. In addition, a scientific collaborative control strategy ensures passenger safety and efficiency, effectively improving passenger satisfaction and providing technical support for the large-scale commercial operation of L4-level autonomous vehicles.

[0218] Figure 10 This is a schematic diagram of a vehicle wading detection device provided in an embodiment of this application. Please refer to... Figure 10 The vehicle wading detection device 1000 includes: an acquisition module 1001, a first determination module 1002, a generation module 1003, and a control module 1004, wherein:

[0219] The acquisition module 1001 is used to acquire the first water level data of the vehicle driving in the preset scenario, as well as the first visual data of the vehicle's surrounding environment.

[0220] The first determining module 1002 is used to determine wading parameters based on the first water level data and the first visual data. The wading parameters are used to indicate the reliability of a vehicle passing through a waterlogged area.

[0221] The generation module 1003 is used to generate a wading control strategy based on the wading parameters and the first water level data. The wading control strategy is used to perform graded control of the vehicle.

[0222] The control module 1004 is used to control the vehicle to perform corresponding driving operations according to the wading control strategy, and to send wading detection information to the cloud server when preset conditions are met.

[0223] In one possible implementation, the first determining module 1002 is specifically used for:

[0224] Preprocessing the first water level data and the first visual data yields fused data.

[0225] The fused data is input into the first model to obtain the water-related parameters.

[0226] In one possible implementation, the first determining module 1002 is specifically used for:

[0227] Align the first water level data and the first visual data to obtain the second water level data and the second visual data.

[0228] The second water level data and the second visual data are fused to obtain fused data.

[0229] In one possible implementation, the first determining module 1002 is specifically used for:

[0230] The first water level data and the first visual data are synchronized in time to obtain the third water level data and the third visual data.

[0231] Spatial registration is performed on the third water level data and the third visual data to obtain the second water level data and the second visual data.

[0232] In one possible implementation, the first determining module 1002 is specifically used for:

[0233] Obtain the first timestamp corresponding to the first water level data, and the second timestamp corresponding to the first visual data;

[0234] Based on the first and second timestamps, the first water level data and the first visual data are synchronized to the same time base to obtain the third water level data and the third visual data.

[0235] In one possible implementation, the first determining module 1002 is specifically used for:

[0236] According to the preset calibration parameters, the detection coordinates in the third water level data are projected onto the pixel coordinate system corresponding to the third visual data;

[0237] The second visual data is obtained by matching the detected coordinates after projection with the pixel positions of the water accumulation area in the third visual data. The water level depth of the second water level data is the same as that of the third water level data.

[0238] In one possible implementation, the generation module 1003 is specifically used for:

[0239] When the wading parameters are greater than or equal to the first threshold and the water level depth in the first water level data is less than or equal to the second threshold, the wading control strategy is determined to be: control the vehicle to pass through the corresponding waterlogged area.

[0240] In one possible implementation, the generation module 1003 is specifically used for:

[0241] When the wading parameters are greater than or equal to the first threshold, and the water depth is greater than the second threshold and less than or equal to the third threshold, the wading control strategy is determined as follows: control the vehicle to slow down and pass through the waterlogged area, and send a vehicle takeover request to the cloud server. The vehicle takeover request includes wading detection information.

[0242] In one possible implementation, the generation module 1003 is specifically used for:

[0243] When the water wading parameters are greater than or equal to the first threshold and the water depth is greater than the third threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0244] In one possible implementation, the generation module 1003 is specifically used for:

[0245] When the water wading parameters are less than the first threshold, the water wading control strategy is determined as follows: control the vehicle to stop passing through the waterlogged area and send a vehicle takeover request to the cloud server.

[0246] In one possible implementation, the device further includes a processing module, which is specifically used for:

[0247] The first water level data and water-related parameters are input into the second model to obtain the passage parameters for the waterlogged area.

[0248] A water erosion detection report is generated based on the passage parameters and the first water level data.

[0249] In one possible implementation, the processing module is specifically used for:

[0250] Based on traffic parameters, determine the vehicle movement data corresponding to the waterlogged area;

[0251] A water wading test report is generated based on the first water level data and vehicle movement data.

[0252] The vehicle wading detection device provided in this application embodiment can execute the technical solution shown in the above method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0253] Figure 11 A schematic diagram of another vehicle wading detection device provided in this application embodiment. Please refer to... Figure 11 The vehicle wading detection device 1100 includes: a receiving module 1101, a second determining module 1102, and a transmitting module 1103, wherein:

[0254] The receiving module 1101 is used to receive a vehicle takeover request sent by the vehicle, which includes a water wading test report and video data;

[0255] The second determining module 1102 is used to determine the vehicle control method based on the water wading test report and video data, wherein the control method is contactless control or contact control.

[0256] The sending module 1103 is used to generate control commands and send them to the vehicle according to the control method, so that the vehicle can safely pass through the flooded area.

[0257] In one possible implementation, the second determining module 1102 is specifically used for:

[0258] Based on the vehicle movement data in the water wading test report and the road condition information in the video data, vehicles are guided to pass through the waterlogged area in accordance with the corresponding control commands and traffic methods.

[0259] In one possible implementation, the second determining module 1102 is specifically used for:

[0260] In response to the user's driving input based on the water wading test report and video data, the vehicle will pass through the waterlogged area with the corresponding control command.

[0261] The vehicle wading detection device provided in this application embodiment can execute the technical solution shown in the above method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0262] Figure 12 This is a schematic diagram of the vehicle wading detection equipment provided in an embodiment of this application. Figure 12 As shown, the vehicle wading detection device 1200 may include: a transceiver 1201, a processor 1202, and a memory 1203.

[0263] Processor 1202 executes computer execution instructions stored in memory, causing processor 1202 to perform the scheme in the above embodiments. Processor 1202 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0264] The memory 1203 is connected to the processor 1202 via the system bus and completes communication between them. The memory 1203 is used to store computer program instructions.

[0265] Transceiver 1201 can be used to obtain the task to be run and its configuration information.

[0266] The system bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus. Transceivers are used to enable communication between database access devices and other computers (e.g., clients, read-write libraries, and read-only libraries). Memory may include random access memory (RAM) and may also include non-volatile memory.

[0267] This application also provides a chip for executing instructions, which is used to execute the technical solution of the vehicle wading detection method in the above embodiments.

[0268] This application also provides a computer-readable storage medium storing computer instructions. When the computer instructions are executed on a computer, the computer performs the technical solution of the vehicle wading detection method described in the above embodiments.

[0269] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When the at least one processor executes the computer program, it can implement the technical solution of the vehicle wading detection method in the above embodiments.

[0270] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0271] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs.

[0272] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit composed of the above modules can be implemented in hardware or in the form of hardware plus software functional units.

[0273] The integrated modules described above, implemented as software functional modules, can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application.

[0274] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0275] The memory may include high-speed RAM, and may also include non-volatile storage (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0276] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0277] The aforementioned storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0278] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor. The processor and storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and storage medium can exist as discrete components in an electronic control unit or main control device.

[0279] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0280] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for detecting vehicle wading through water, characterized in that, The method includes: Acquire the first water level data of the vehicle in the preset scenario, as well as the first visual data of the vehicle's surrounding environment; Based on the first water level data and the first visual data, wading parameters are determined, which are used to indicate the reliability of the vehicle passing through the waterlogged area; Based on the wading parameters and the first water level data, a wading control strategy is generated, which is used to perform graded control of the vehicle. According to the water wading control strategy, the vehicle is controlled to perform corresponding driving operations, and water wading detection information is sent to the cloud server when preset conditions are met.

2. The method according to claim 1, characterized in that, Based on the first water level data and the first visual data, the water wading parameters are determined, including: The first water level data and the first visual data are preprocessed to obtain fused data; The fused data is input into the first model to obtain the water-related parameters.

3. The method according to claim 2, characterized in that, The first water level data and the first visual data are preprocessed to obtain fused data, including: Align the first water level data and the first visual data to obtain the second water level data and the second visual data. The second water level data and the second visual data are fused together to obtain the fused data.

4. The method according to any one of claims 1-3, characterized in that, Aligning the first water level data and the first visual data yields second water level data and second visual data, including: The first water level data and the first visual data are time-synchronized to obtain the third water level data and the third visual data. Spatial registration processing is performed on the third water level data and the third visual data to obtain the second water level data and the second visual data.

5. The method according to claim 4, characterized in that, Obtain third water level data and third visual data, including: Obtain the first timestamp corresponding to the first water level data and the second timestamp corresponding to the first visual data; Based on the first timestamp and the second timestamp, the first water level data and the first visual data are synchronized to the same time base to obtain the third water level data and the third visual data.

6. The method according to claim 4, characterized in that, The third water level data includes water level depth and detection coordinates; obtaining the second water level data and the second visual data includes: According to the preset calibration parameters, the detection coordinates in the third water level data are projected onto the pixel coordinate system corresponding to the third visual data; The second visual data is obtained by matching the detected coordinates after projection with the pixel position of the water accumulation area in the third visual data. The water level depth of the second water level data is the same as the water level depth of the third water level data.

7. The method according to claim 1, characterized in that, Generate water-related control strategies, including: When the wading parameter is greater than or equal to a first threshold and the water level depth in the first water level data is less than or equal to a second threshold, the wading control strategy is determined to be: controlling the vehicle to pass through the corresponding waterlogged area.

8. The method according to claim 7, characterized in that, Generate water-related control strategies, including: When the wading parameter is greater than or equal to the first threshold, and the water level depth is greater than the second threshold and less than or equal to the third threshold, the wading control strategy is determined to be: controlling the vehicle to slow down and pass through the waterlogged area, and sending a vehicle takeover request to the cloud server, the vehicle takeover request including the wading detection information.

9. The method according to claim 8, characterized in that, Generate water-related control strategies, including: When the water wading parameter is greater than or equal to the first threshold and the water level depth is greater than the third threshold, the water wading control strategy is determined to be: control the vehicle to stop passing through the waterlogged area and send the vehicle takeover request to the cloud server.

10. The method according to claim 9, characterized in that, Generate water-related control strategies, including: When the water wading parameter is less than the first threshold, the water wading control strategy is determined to be: control the vehicle to stop passing through the waterlogged area and send the vehicle takeover request to the cloud server.

11. The method according to any one of claims 8-10, characterized in that, The water wading detection information includes water wading detection reports and video data; before sending the vehicle takeover request to the cloud server, the method further includes: The first water level data and the water-related parameters are input into the second model to obtain the passage parameters of the waterlogged area. A water wading detection report is generated based on the passage parameters and the first water level data.

12. The method according to claim 11, characterized in that, Based on the passage parameters and the first water level data, a water wading detection report is generated, including: Based on the traffic parameters, determine the vehicle movement data corresponding to the waterlogged area; The water wading detection report is generated based on the first water level data and the vehicle movement data.

13. A method for detecting vehicle wading through water, characterized in that, The method includes: Receive a vehicle takeover request sent by the vehicle, the vehicle takeover request including a water wading test report and video data; Based on the water wading test report and the video data, the control method of the vehicle is determined, which is either contactless control or contact control. According to the control method, control commands are generated and sent to the vehicle to enable the vehicle to safely pass through the flooded area.

14. The method according to claim 13, characterized in that, When the control method is contactless control, determining the vehicle's control method includes: Based on the vehicle movement data in the water wading test report and the road condition information in the video data, the vehicle is directed to pass through the waterlogged area using the corresponding control command and traffic method.

15. The method according to claim 13, characterized in that, When the control method is contact-based control, determining the vehicle's control method includes: In response to a user's driving operation input based on the water wading test report and the video data, the vehicle is made to pass through the waterlogged area with corresponding control commands.

16. A vehicle wading detection device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-15.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-15.

18. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-15.

Images