Vehicle control method, vehicle and storage medium

Abstract

The embodiment of the invention provides a control method of a vehicle, the vehicle and a storage medium, and the method comprises the steps that in the moving process of the vehicle, multiple initial sensing results of the vehicle for a target object in the environment where the vehicle is located are obtained, and the different initial sensing results are obtained by sensing the target object under different sensing dimensions of the vehicle; determining a sensing performance index of the plurality of initial sensing results; in response to a sensed performance indicator of the initial sensing results being below a performance indicator threshold, determining risk data of the vehicle based on the plurality of initial sensing results; determining a to-be-triggered target control mode of the vehicle based on the risk data; in the target control mode, the vehicle is controlled to execute sensing operation on the target object, a target sensing result is obtained, and the sensing performance index corresponding to the target sensing result is larger than or equal to the performance index threshold value. The technical problem of low vehicle control accuracy is solved.

Description

Technical Field

[0001] This application relates to the field of intelligent driving technology, and more specifically, to a vehicle control method, a vehicle, and a storage medium. Background Technology

[0002] Currently, intelligent driving technology for vehicles widely adopts a sequential architecture, meaning that perception, decision-making, and path planning are performed in sequence. However, this sequential architecture suffers from limitations in complex and changing environments (such as intense sunlight, heavy rain, and dense fog), where the uncertainty in target object recognition is not fully considered. This directly leads to deviations in the vehicle's path planning; for example, lane lines may be misidentified as obstacles, causing unnecessary detours and increasing safety risks. Therefore, the technical problem of low vehicle control accuracy still exists.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a vehicle control method, a vehicle, and a storage medium to at least solve the technical problem of low vehicle control accuracy.

[0005] According to one aspect of the embodiments of this application, a vehicle control method is provided, wherein the method may include: acquiring multiple initial sensing results of the vehicle on a target object in its environment during vehicle movement, wherein different initial sensing results are obtained by the vehicle sensing the target object in different sensing dimensions; determining a sensing performance index of the multiple initial sensing results, wherein the sensing performance index corresponding to the initial sensing results is used to represent the sensing effect of the target object in the initial sensing results; in response to a sensing performance index of the initial sensing results being lower than a performance index threshold, determining risk data of the vehicle based on the multiple initial sensing results, wherein the risk data is used to represent the degree of risk of the vehicle in the environment; determining a target control mode to be triggered by the vehicle based on the risk data, wherein different target control modes correspond to different risk data of the vehicle; and controlling the vehicle to perform a sensing operation on the target object in the target control mode to obtain a target sensing result, wherein the sensing performance index corresponding to the target sensing result is greater than or equal to the performance index threshold, and the sensing performance index corresponding to the target sensing result is used to represent the sensing effect of the target object in the target sensing result.

[0006] Furthermore, determining the sensing performance metrics of multiple initial sensing results includes: performing spatiotemporal alignment processing on multiple initial sensing results to obtain spatiotemporal alignment results, wherein the spatiotemporal alignment results are used to represent different initial sensing results of the same target object at the same time and in the same space; determining the confidence level of the target object based on the spatiotemporal alignment results, wherein the confidence level is used to represent the degree of difference between different initial sensing results of the same target object at the same time and in the same space; and determining the sensing performance metrics of multiple initial sensing results based on the confidence level, wherein the confidence level and the sensing performance metrics of the initial sensing results are positively correlated.

[0007] Furthermore, the confidence level includes location confidence, type confidence, and movement confidence. Location confidence represents the degree of difference in the location of the same target object in different initial sensing results. Type confidence represents the degree of difference in the object type of the same target object in different initial sensing results. Movement confidence represents the degree of difference in the movement state of the same target object in different initial sensing results. Based on the spatiotemporal alignment results, the location confidence of the target object is determined, including: determining the regional overlap between the sensing areas of different initial sensing results based on the spatiotemporal alignment results; and determining the location based on the regional overlap. The confidence level is determined based on the spatiotemporal alignment results, where the region overlap is positively correlated with the location confidence level. The type confidence level of the target object is determined based on the spatiotemporal alignment results, including: determining the object type of the target object in different initial sensing results based on the spatiotemporal alignment results; determining the type confidence level based on the object types corresponding to different initial sensing results. The movement confidence level of the target object is determined based on the spatiotemporal alignment results, including: determining the movement parameters of the target object in different initial sensing results based on the spatiotemporal alignment results, where the movement parameters represent the movement state of the target object; determining the movement confidence level based on the movement parameters corresponding to different initial sensing results.

[0008] Furthermore, based on confidence levels, a sensing performance index for multiple initial sensing results is determined, including: in response to a confidence level difference greater than a confidence level difference threshold between at least two of the three confidence levels (location confidence, type confidence, and movement confidence), a first weight is determined for each initial sensing result; based on the first weight and the location confidence, type confidence, and movement confidence levels corresponding to each initial sensing result, a sensing performance index for each initial sensing result is determined.

[0009] Furthermore, in response to the sensing performance index of the initial sensing result being lower than the performance index threshold, risk data of the vehicle is determined based on multiple initial sensing results, including: in response to the sensing performance index of the initial sensing result being lower than the performance index threshold, determining environmental state information of the environment based on multiple initial sensing results, wherein the environmental state information is used to represent the influencing factors corresponding to the initial sensing results in the environment; determining second weights corresponding to the multiple initial sensing results based on the environmental state information; and using the second weights, performing a weighted summation of the multiple initial sensing results to obtain risk data.

[0010] Furthermore, the target control mode includes a first control mode, a second control mode, and a third control mode. The degree of risk present in the vehicle under the first control mode is lower than that under the second control mode, and the degree of risk present in the vehicle under the second control mode is lower than that under the third control mode. Based on the risk data, the target control mode to be triggered for the vehicle is determined, including: determining the target control mode as the first control mode in response to the risk data being less than a first risk threshold; determining the target control mode as the second control mode in response to the risk data being greater than or equal to the first risk threshold and less than or equal to the second risk threshold; and determining the target control mode as the third control mode in response to the risk data being greater than the second risk threshold.

[0011] Further, the initial sensing results include a first sensing result, a second sensing result, and a third sensing result. The first sensing result represents the two-dimensional visual features of the target object, the second sensing result represents the three-dimensional appearance attributes of the target object, and the third sensing result represents the relative movement state between the vehicle and the target object. The vehicle includes at least one first sensing device, at least one second sensing device, and at least one third sensing device. During vehicle movement, multiple initial sensing results of the vehicle on the target object in its environment are acquired, including: during vehicle movement, sensing the target object using at least one first sensing device to obtain a first sensing result, sensing the target object using at least one second sensing device to obtain a second sensing result, and sensing the target object using at least one third sensing device to obtain a third sensing result; and / or, in target control mode, controlling the vehicle to perform a sensing operation on the target object to obtain target sensing results, including: In response to a target control mode being a first control mode, the method controls a first sensing device and a third sensing device to perform a first sensing operation on the target object, obtaining a target sensing result; in response to a target control mode being a second control mode, the method controls the first sensing device, the second sensing device, and the third sensing device to perform a second sensing operation on the target object, obtaining a target sensing result, wherein the execution frequency of the second sensing operation is greater than the execution frequency of the first sensing operation; in response to a target control mode being a third control mode, the method controls the first sensing device, the second sensing device, and the third sensing device to perform a third sensing operation on the target object, obtaining a target sensing result, wherein the execution frequency of the third sensing operation is greater than the execution frequency of the second sensing operation; and / or, the method further includes: in response to a risk change rate corresponding to risk data being greater than a change rate threshold, adjusting the target control mode, and controlling the vehicle according to the adjusted target control mode to perform a sensing operation on the target object, obtaining a target sensing result.

[0012] Furthermore, the target control mode includes a preset sensing range and a preset sensing frequency. In the target control mode, the vehicle is controlled to perform a sensing operation on the target object to obtain a target sensing result. This includes: responding to the target control mode being a first control mode, controlling a first sensing device and a third sensing device to perform a first sensing operation on the target object within a preset sensing range at a preset sensing frequency corresponding to the first control mode, and obtaining a target sensing result; responding to the target control mode being a second control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a second sensing operation on the target object within a preset sensing range corresponding to the second control mode at a preset sensing frequency corresponding to the second control mode, and obtaining a target sensing result; and responding to the target control mode being a third control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a third sensing operation on the target object within a preset sensing range corresponding to the third control mode at a preset sensing frequency corresponding to the third control mode, and obtaining a target sensing result.

[0013] According to another aspect of the embodiments of this application, a vehicle control device is also provided. The device may include: an acquisition module, configured to acquire multiple initial sensing results of the vehicle on a target object in its environment during vehicle movement, wherein different initial sensing results are obtained by the vehicle sensing the target object in different sensing dimensions; a first determination module, configured to determine a sensing performance index of the multiple initial sensing results, wherein the sensing performance index corresponding to the initial sensing results is used to represent the sensing effect of the target object in the initial sensing results; a second determination module, configured to determine risk data of the vehicle based on the multiple initial sensing results in response to the sensing performance index of the initial sensing results being lower than a performance index threshold, wherein the risk data is used to represent the degree of risk of the vehicle in the environment; a third determination module, configured to determine a target control mode to be triggered by the vehicle based on the risk data, wherein different target control modes correspond to different risk data of the vehicle; and a control module, configured to control the vehicle to perform a sensing operation on the target object in the target control mode to obtain a target sensing result, wherein the sensing performance index corresponding to the target sensing result is greater than or equal to the performance index threshold, and the sensing performance index corresponding to the target sensing result is used to represent the sensing effect of the target object in the target sensing result.

[0014] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0016] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0017] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

[0018] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0019] In this embodiment, during vehicle movement, target objects in the vehicle's environment can be sensed in real time across different dimensions, yielding multiple initial sensing results. The sensing performance index of these initial sensing results can be determined. If the sensing performance index of the initial sensing results is lower than a performance index threshold, risk data for the vehicle can be determined based on these initial sensing results. Based on the risk data, a target control mode to be triggered for the vehicle is determined. The vehicle is then controlled to enter the target control mode, and within this mode, the vehicle performs sensing operations on the target object, obtaining a target sensing result with a sensing performance index greater than or equal to the performance index threshold. In other words, this application dynamically evaluates the sensing performance index of multimodal initial sensing data to identify and quantify the risks in the vehicle's environment, thereby intelligently adjusting the vehicle's control mode. Unlike the serial architecture in related technologies that ignores the impact of environmental factors on sensing accuracy, this embodiment proactively analyzes the potential risks faced by the vehicle when the sensing performance index is lower than the performance index threshold, and selects an appropriate control mode based on these risks. This ensures that even in complex environments, the vehicle can make accurate and safe driving decisions based on target sensing results with high sensing performance indexes, effectively improving the vehicle's control accuracy and driving safety in complex environments. This achieves the technical effect of improving vehicle control accuracy and solves the technical problem of low vehicle control accuracy. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0021] Figure 1 This is a flowchart of a vehicle control method according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of an intelligent driving multimodal perception and decision-making collaboration system according to an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of a software system architecture according to an embodiment of this application;

[0024] Figure 4 This is a flowchart of a multimodal arbitration mechanism according to an embodiment of this application;

[0025] Figure 5 This is a flowchart of a dynamic mode scheduling according to an embodiment of this application;

[0026] Figure 6 This is a schematic diagram of a vehicle control device according to an embodiment of this application. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] According to an embodiment of this application, a vehicle control method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] This embodiment provides a vehicle control method. Figure 1 This is a flowchart of a vehicle control method according to an embodiment of this application, such as... Figure 1 As shown, it may include vehicle A, network B, and sensing device C. The method includes the following steps:

[0031] Step S102: During the vehicle's movement, acquire multiple initial sensing results of the vehicle on target objects in the environment.

[0032] In the technical solution provided by step S102 of the embodiments of this application, during the vehicle's movement, target objects in the vehicle's environment can be sensed in real time under different sensing dimensions to obtain multiple initial sensing results. These different initial sensing results are obtained by the vehicle sensing the target objects under different sensing dimensions. The target objects can be objects in the vehicle's environment, such as people, vehicles, or other obstacles. The initial sensing results can be obtained by different sensors in the vehicle identifying the target objects in the vehicle's environment, such as target images, target motion matrices, point cloud data, etc. The sensing dimensions can be two-dimensional, three-dimensional, etc.

[0033] In this embodiment, the target object can be any object (environmental element) that needs to be detected and identified in the environment in which the vehicle is located during driving, including but not limited to pedestrians, other vehicles, bicycles, animals, static obstacles (such as roadblocks, curbs, and buildings), dynamic elements (such as traffic lights and signs), and road surface features (such as lane lines and road markings). Identifying the target object can be used for vehicle path planning, obstacle avoidance, and compliance with traffic rules.

[0034] Optionally, the initial sensing results may originate from various sensors mounted on the vehicle, including but not limited to cameras, LiDAR, and millimeter-wave radar, without specific limitations. The initial sensing results can be the sensing results collected by the aforementioned sensors mounted on the vehicle. For example, if the sensor is a camera, the initial sensing result is the target image captured by the camera. If the sensor is LiDAR, the initial sensing result is the point cloud data collected by the LiDAR. If the sensor data is from millimeter-wave radar, the initial sensing result is from the millimeter-wave radar.

[0035] For example, if the initial sensing result is a target image, it can contain information such as the appearance, shape, and texture of the target object, facilitating its identification and classification. If the initial sensing result is a target motion matrix, it can include dynamic information such as the target object's velocity, acceleration, and direction, making it easier to predict the target object's movement trend. If the initial sensing result is point cloud data, it can contain the target object's three-dimensional position information and the spatial structure of its surrounding environment, facilitating precise target object localization.

[0036] It should be noted that the sensors for obtaining the initial sensing results and the initial sensing results mentioned above are only illustrative examples and are not specifically limited here. As long as the sensing results and the sensors that collect the sensing results are able to facilitate the vehicle to accurately sense the surrounding environment and thus plan the vehicle's driving decisions, they are all within the protection scope of the embodiments of this application.

[0037] Optionally, the sensing dimension can correspond to the sensor that acquires the initial sensing result. For example, if the sensor is a camera, the sensing dimension can be two-dimensional, which can be used to sense the planar features and visual information of the target object. If the sensor is a LiDAR, the sensing dimension can be three-dimensional, which can be used to sense the position, size, and shape of the target object in three-dimensional space.

[0038] Optionally, after the vehicle starts, sensors related to sensing the vehicle's surroundings can be automatically activated. After activation, the sensors can perform real-time sensing of the vehicle's environment to obtain initial sensing results. For example, millimeter-wave radar can emit millimeter-wave signals and receive echoes, analyzing the echo signals to obtain information such as the distance, speed, and angle of the target object. LiDAR can emit laser beams and measure reflection time to construct point cloud data of the target object. Cameras can capture image data corresponding to the target object.

[0039] Step S104: Determine the sensing performance metrics of multiple initial sensing results.

[0040] In the technical solution provided by step S104 of this application embodiment, after sensing the target object under different sensing dimensions and obtaining multiple initial sensing results, the sensing performance index corresponding to the multiple initial sensing results can be determined. The sensing performance index corresponding to the initial sensing results is used to represent the sensing effect of the target object in the initial sensing results. The sensing performance index may include a sensing accuracy index corresponding to the initial sensing results, a sensing efficiency index corresponding to the initial sensing results, and other comprehensive indicators, such as confidence scores and environmental adaptability scores. The confidence score can represent the degree of trust in the initial sensing results sensed by each sensor. The environmental adaptability score can be used to evaluate the performance stability of the sensor in different environments. The sensing performance index can also be a comprehensive index of the indicators corresponding to the various sensing effects mentioned above; that is, considering the sensing effects from various aspects, it comprehensively measures whether the initial sensing results can be used for subsequent vehicle driving decisions.

[0041] In this embodiment, the aforementioned sensing accuracy metrics may include the target object's position sensing accuracy, target object type recognition rate, target object motion state accuracy, and target object detection rate. Position accuracy measures the accuracy of the sensor in locating the target object based on the initial sensing results; it represents the deviation between the target object's position in the initial sensing results and its actual position in the vehicle's environment. Type recognition rate assesses the probability that the sensor correctly identifies the type of the target object. Motion state accuracy assesses the accuracy of the sensor's measurement of the target object's dynamic characteristics, such as speed, acceleration, and direction. Target detection rate assesses the proportion of target objects detected by the sensor, reflecting the sensor's detection coverage capability in complex environments.

[0042] Optionally, the aforementioned sensing efficiency metrics may include data processing latency metrics, computing resource utilization metrics, and energy consumption metrics. Among these, data processing latency metrics represent the time required from the sensor acquiring initial sensing results to completing target object recognition, which can affect the real-time performance of driving decisions. Computing resource utilization metrics assess the occupancy of vehicle computing resources by sensor data processing. Energy consumption metrics represent the electricity consumed by sensor operation and data processing.

[0043] It should be noted that the above sensing performance indicators are only illustrative examples and are not specifically limited here. As long as the sensing performance indicators can ensure that the sensors in the vehicle can accurately and quickly sense target objects in the surrounding environment and ensure the safety of vehicle driving, they are all within the protection scope of the embodiments of this application.

[0044] In this embodiment, determining the sensing performance index of the initial sensing results not only assesses the sensing quality of the corresponding sensors but also provides a valid basis for subsequent driving decisions. Through precise calculation of the sensing performance index, it is possible to determine which initial sensing results for which sensing dimensions are reliable, which require further driving decisions, and which initial sensing results for which sensing dimensions are unreliable and can be re-sensed. This allows for accurate and timely responses in complex and changing environments, ensuring safe and smooth driving.

[0045] Step S106: In response to the initial sensing results having a sensing performance index lower than a performance index threshold, risk data for the vehicle is determined based on multiple initial sensing results.

[0046] In the technical solution provided by step S106 of this application embodiment, after determining the sensing performance index corresponding to multiple initial sensing results, the magnitude between the sensing performance index and the performance index threshold can be judged. If the sensing performance index is lower than the performance index threshold, the vehicle's risk data can be determined based on the multiple initial sensing results. The risk data represents the degree of risk the vehicle faces in the environment, and the risk data can be the risk entropy R. The performance index threshold can be a pre-set sensing performance score that measures whether the sensing performance index meets the standard; for example, it can be pre-set to 0.98. No specific limitation is placed on the size of the performance index threshold here; it can be set according to the vehicle's driving safety requirements.

[0047] In this embodiment, the performance index threshold can serve as a benchmark for judging the quality of the sensing results obtained by the sensor. It can be a pre-set value based on the vehicle's driving safety requirements and the sensor's technical specifications, used to determine whether the sensing results are sufficiently reliable for the vehicle's driving decisions. The principles for setting the performance index threshold can consider safety standards, application scenario adjustments, and different sensor characteristics. Regarding safety standards, the performance index threshold setting ensures that even under adverse conditions, the sensing results obtained by the sensor can meet the basic requirements for safe driving. Regarding application scenario adjustments, different environments have different requirements for the accuracy and timeliness of the sensing results; the performance index threshold can be flexibly adjusted according to road type (highway, urban road) and weather conditions (sunny, rainy / foggy). Regarding sensor characteristics, the performance index threshold can consider the sensor's own performance indicators, such as resolution, refresh rate, and coverage, to reflect the sensor's good working condition.

[0048] It should be noted that the above-mentioned method and principle for setting performance index thresholds are only illustrative examples and are not specifically limited here. As long as the performance index thresholds can measure the quality of the sensing effect of the sensor to ensure the safe driving of the vehicle, they are all within the protection scope of the embodiments of this application.

[0049] Optionally, risk data can be a quantitative indicator that comprehensively assesses the risk level of the vehicle's current environment. Risk data can be calculated based on sensing performance indicators from multiple initial sensing results and environmental factors. Risk data can be a mathematical model that integrates multiple risk factors, such as traffic density, weather conditions, and road curvature in the vehicle's environment. The weights of these risk factors can be preset according to the importance of driving safety or flexibly set according to real-time environmental conditions. For example, traffic density has a higher impact on risk assessment in urban driving environments, so a larger weight can be assigned to traffic density in urban driving environments; conversely, road curvature has a higher impact in mountainous road environments, so a larger weight can be assigned to road curvature in mountainous road environments.

[0050] Optionally, after acquiring the sensing performance indicators of multiple initial sensing results, the relationship between the sensing performance indicators of the initial sensing results and the performance indicator threshold can be analyzed. If the sensing performance indicator of the initial sensing results is greater than or equal to the performance indicator threshold, it indicates that the sensor has achieved good sensing performance of the initial sensing results and can be used for driving strategy planning during the current vehicle movement.

[0051] Optionally, if the sensing performance index of the initial sensing result obtained by any sensor is lower than the performance index threshold, it indicates that the sensor's sensing effect of the initial sensing result is poor. Insisting on using the initial sensing result for driving strategy planning would affect vehicle driving safety. In this case, the initial sensing result can be used to analyze risk factors in the current vehicle environment, such as traffic density, weather conditions, and road conditions. The weight of each risk factor may be dynamically adjusted according to the vehicle's environment and location to reflect the impact of different risk factors on vehicle driving safety.

[0052] Optionally, based on the aforementioned risk factors and the weights corresponding to each risk factor, predefined mathematical models or algorithms (such as weighted averages or risk entropy models) can be used to calculate vehicle risk data.

[0053] In this embodiment, the method not only focuses on the sensing effect of the initial sensing results obtained by the sensor, but also considers the overall risk assessment of the environment. By setting performance index thresholds and calculating risk data, it can respond to sensor performance fluctuations in real time, ensuring the safety and reliability of the vehicle when driving in complex and ever-changing environments.

[0054] Step S108: Based on the risk data, determine the target control mode to be triggered for the vehicle.

[0055] In the technical solution provided by step S108 of the embodiments of this application, after determining the vehicle's risk data based on multiple initial sensing data, the target control mode to be triggered by the vehicle can be determined based on the risk data. Different target control modes correspond to different risk data of the vehicle.

[0056] In this embodiment, the target control mode can be a series of preset autonomous driving strategies and behavior modes based on risk data of the vehicle's environment. The aforementioned autonomous driving strategies can be used to control the vehicle's movement within its environment. The behavior modes can include at least a sensor control mode. The sensor control mode can be used to control the sensors in the vehicle to re-sensitize target objects in the vehicle's environment, ensuring vehicle safety. Risk data is positively correlated with the degree of risk present in the vehicle's environment. For example, the higher the risk data of the environment, the greater the degree of risk to the vehicle, and in this case, a safer target control mode is needed to provide more cautious autonomous driving and a more comprehensive behavior mode.

[0057] Optionally, the risk data is directly related to the selection of the target control mode. The higher the risk data, the greater the threat to vehicle safety posed by the current environment. In this case, a more conservative driving strategy and a more comprehensive sensor control mode can be activated to enhance the vehicle's perception capabilities and reaction speed, thereby ensuring driving safety.

[0058] Optionally, the vehicle's target control mode can be automatically adjusted based on the magnitude of the risk data. This includes, but is not limited to, automatically adjusting the autonomous driving mode, such as adjusting vehicle speed, increasing safe distances from vehicles in front and behind, and reducing turning speed. The vehicle's sensor control mode can be dynamically optimized, for example, by increasing the operating frequency of sensors, expanding the scanning angle of the LiDAR, and enhancing the computational resources for image processing and target recognition algorithms.

[0059] It should be noted that the above-described methods and processes for automatically adjusting the target control mode based on risk data are merely illustrative examples and are not subject to specific limitations.

[0060] In this embodiment, by associating risk data with the vehicle's target control mode, the vehicle's driver assistance system achieves intelligent response and adaptation to complex environments. This mechanism ensures vehicle safety and enhances the driving experience.

[0061] Step S110: In target control mode, control the vehicle to perform a sensing operation on the target object and obtain the target sensing result.

[0062] In the technical solution provided by step S110 of the embodiments of this application, after determining the target control mode to be triggered by the vehicle based on risk data, the vehicle can be controlled to enter the target control mode. In this target control mode, the vehicle is controlled to perform a sensing operation on the target object to obtain the target sensing result. The sensing performance index corresponding to the target sensing result is greater than or equal to a performance index threshold, and this index is used to represent the sensing effect of the target object in the target sensing result.

[0063] In this embodiment, based on the needs of the target control mode, the assisted driving system can optimize the sensor operation strategy, such as increasing the scanning frequency of the LiDAR, improving the image resolution and frame rate of the camera, and ensuring more accurate sensing results in high-risk environments. By adjusting the sensor configuration and algorithm parameters, the goal is to make the performance indicators of the obtained target sensing results reach or exceed preset performance indicator thresholds, thereby improving the accuracy and real-time performance of target object recognition.

[0064] Optionally, according to the requirements of the target control mode, the resources of the sensors are intelligently scheduled. For example, in the target control mode corresponding to a high-risk environment, various available sensors in the vehicle are activated to sense the surrounding environment, while in the target control mode corresponding to a low-risk environment, some necessary sensors in the vehicle are activated to save energy and computing resources.

[0065] Optionally, the performance index of the target sensing result can be compared with a preset performance index threshold. If the performance index of the target sensing result is greater than or equal to the performance index threshold, it indicates that the sensing operation is effective and can meet the safety and performance requirements of driving in the current environment.

[0066] In this embodiment, based on environmental risk and target control mode, the sensor operation and sensing process are optimized to ensure the acquisition of target sensing data with higher sensing performance indicators. Through this mechanism, environments with varying levels of risk can be effectively addressed, improving driving safety while optimizing resource utilization.

[0067] In steps S102 to S110 of this embodiment, during vehicle movement, target objects in the vehicle's environment can be sensed in real time using different sensing dimensions, resulting in multiple initial sensing results. The sensing performance index of these initial sensing results can be determined. If the sensing performance index of the initial sensing results is lower than a performance index threshold, the vehicle's risk data can be determined based on the initial sensing results. Based on the risk data, a target control mode to be triggered for the vehicle is determined. The vehicle is then controlled to enter the target control mode, and within this mode, the vehicle performs sensing operations on the target object, obtaining a target sensing result with a sensing performance index greater than or equal to the performance index threshold. In other words, this application dynamically evaluates the sensing performance index of multimodal initial sensing data to identify and quantify the risks in the vehicle's environment, thereby intelligently adjusting the vehicle's control mode. Unlike the serial architecture in related technologies that ignores the impact of environmental factors on sensing accuracy, this embodiment actively analyzes the potential risks faced by the vehicle when the sensing performance index is lower than the performance index threshold, and selects an appropriate control mode based on these risks. This ensures that even in complex environments, the vehicle can make accurate and safe driving decisions based on target sensing results with high sensing performance indexes, effectively improving the vehicle's control accuracy and driving safety in complex environments. This achieves the technical effect of improving vehicle control accuracy and solves the technical problem of low vehicle control accuracy.

[0068] The embodiments of this application will be described in detail below with reference to the steps described above.

[0069] As an optional implementation, step S104, determining the sensing performance index of multiple initial sensing results, includes: performing spatiotemporal alignment processing on the multiple initial sensing results to obtain spatiotemporal alignment results, wherein the spatiotemporal alignment results are used to represent different initial sensing results of the same target object at the same time and in the same space; determining the confidence level of the target object based on the spatiotemporal alignment results, wherein the confidence level is used to represent the degree of difference between different initial sensing results of the same target object at the same time and in the same space; and determining the sensing performance index of multiple initial sensing results based on the confidence level, wherein the confidence level and the sensing performance index of the initial sensing results are positively correlated.

[0070] In this embodiment, during the process of determining the sensing performance indicators of multiple initial sensing results, spatiotemporal alignment processing can be performed on the multiple initial sensing results to obtain spatiotemporal alignment results. Based on the spatiotemporal alignment results, the confidence level of the target object is determined, and thus, the sensing performance indicators of the multiple initial sensing results are determined according to the confidence level. The spatiotemporal alignment results can be used to represent different initial sensing results of the same target object at the same time and in the same space. The confidence level can be used to represent the degree of difference between different initial sensing results of the same target object at the same time and in the same space, which can be the degree of difference in attributes such as position, type, and motion state. There is a positive correlation between the confidence level and the sensing performance indicators of the initial sensing results. The initial sensing results can be sensor data collected by the sensor. Initial sensing results under different sensing dimensions can constitute multimodal sensor data.

[0071] Optionally, since different sensors have different sampling frequencies and data formats, spatiotemporal alignment processing ensures that initial sensing results from different sensors can be compared at the same moment (time point) and in the same spatial coordinate system. Transforming initial sensing results from different sensing dimensions into a unified spatiotemporal reference frame allows for more accurate identification and localization of target objects.

[0072] Optionally, the degree of difference in attributes such as the position, type, and motion state of the target object in the spatiotemporal alignment results is evaluated, and the corresponding confidence level is determined. The higher the confidence level, the better the consistency of the initial sensing results of different sensors for the same target object, and the better the sensor's sensing performance.

[0073] Optionally, in determining the sensing performance indicators, the calculated confidence level is combined with the inherent performance parameters of the sensor to determine the sensing performance indicators of each sensor, such as position accuracy, target recognition rate, motion parameter accuracy, etc.

[0074] In this embodiment, a quantitative evaluation of the sensing performance of multimodal sensors is achieved through spatiotemporal alignment, confidence level calculation, and performance index determination. This process not only improves the accuracy and reliability of the vehicle's environmental perception but also enables intelligent adjustment of sensor operating strategies through refined sensing performance index evaluation, thus adapting to various driving environments and ensuring passenger safety and driving efficiency.

[0075] As an optional implementation, the confidence level includes location confidence, type confidence, and movement confidence. Location confidence represents the degree of difference in the location of the same target object across different initial sensing results. Type confidence represents the degree of difference in the object type of the same target object across different initial sensing results. Movement confidence represents the degree of difference in the movement state of the same target object across different initial sensing results. Based on the spatiotemporal alignment results, determining the location confidence of the target object includes: determining the regional overlap between the sensing areas of different initial sensing results based on the spatiotemporal alignment results; and determining the region overlap based on the region overlap. The system determines the confidence level of the target object's location, where the region overlap is positively correlated with the location confidence level. Based on the spatiotemporal alignment results, it determines the confidence level of the target object's type, including: determining the object type of the target object in different initial sensing results based on the spatiotemporal alignment results; determining the type confidence level based on the object types corresponding to different initial sensing results. Based on the spatiotemporal alignment results, it determines the confidence level of the target object's movement, including: determining the movement parameters of the target object in different initial sensing results based on the spatiotemporal alignment results, where the movement parameters represent the movement state of the target object; determining the movement confidence level based on the movement parameters corresponding to different initial sensing results.

[0076] In this embodiment, during the process of determining the location confidence of the target object based on the spatiotemporal alignment results, the overlap between the sensing areas of different initial sensing results can be determined based on the spatiotemporal alignment results, and the location confidence can be determined based on the overlap. During the process of determining the type confidence of the target object based on the spatiotemporal alignment results, the object type to which the target object belongs in different initial sensing results can be determined based on the spatiotemporal alignment results, thereby determining the type confidence. During the process of determining the movement confidence of the target object based on the spatiotemporal alignment results, the movement parameters of the target object in different initial sensing results can be determined based on the spatiotemporal alignment results, thereby determining the movement confidence. The confidence can include location confidence, type confidence, and movement confidence.

[0077] Optionally, in determining the location confidence level, the overlap between the sensing areas of the same target object detected by different sensors can be calculated based on the spatiotemporal alignment results. Higher overlap indicates more consistent target location identification by different sensors. The location confidence level is then determined based on the calculated overlap. There is a positive correlation between overlap and location confidence; that is, higher overlap corresponds to higher location confidence. High location confidence means a more reliable assessment of the target object's location.

[0078] Optionally, in determining type confidence, the type recognition results of the target object under different sensors can be determined based on the spatiotemporal alignment results. For example, a car may be identified as a "sedan" by a camera, while a LiDAR may identify it as a "truck." If different sensors consistently or similarly determine the object type of the same target object, the resulting type confidence is high. Conversely, if the differences in object type determination among sensors for the same target object are significant, the type confidence is low.

[0079] Optionally, during the determination of motion confidence, based on spatiotemporal alignment, the motion parameters of the target object under different sensors, such as velocity, direction, and acceleration, can be analyzed. Based on the motion parameters of the same target object from different sensors, the consistency of the motion parameter acquisition by different sensors can be evaluated. The more consistent the motion parameters acquired by different sensors, the higher the motion confidence, indicating a more reliable judgment of the target object's motion state.

[0080] Optionally, the combination of location confidence, type confidence, and movement confidence forms a multi-sensory dimension of perception of the target object. With high confidence, the driver assistance system can rely more confidently on the initial sensing results for path planning and obstacle avoidance decisions.

[0081] In the embodiments of this application, by precisely calculating the position confidence, type confidence, and movement confidence, the assisted driving system can quantitatively evaluate the perception effect and consistency of different sensors on the same target object, ensuring the safety, reliability, and efficiency of vehicle driving.

[0082] As an optional implementation, based on confidence levels, a sensing performance index for multiple initial sensing results is determined, including: in response to a confidence level difference greater than a confidence level difference threshold between at least two of the three—location confidence level, type confidence level, and movement confidence level—a first weight is determined for each different initial sensing result; based on the first weight and the location confidence level, type confidence level, and movement confidence level corresponding to each different initial sensing result, a sensing performance index for each initial sensing result is determined.

[0083] In this embodiment, during the process of determining the sensing performance index of multiple initial sensing results based on confidence levels, if the difference in confidence levels between at least two of the three types—location confidence, type confidence, and movement confidence—is greater than a confidence threshold, then a first weight corresponding to each initial sensing result can be determined. Based on the first weight and the location confidence, type confidence, and movement confidence, the sensing performance index of the initial sensing result is then determined. The confidence threshold can be a preset threshold, for example, 0.3; no specific limitation is placed on the size of the confidence threshold here.

[0084] Optionally, this embodiment mainly involves collision detection and confidence difference assessment, environmental adaptability weighting, and sensing performance indicators for determining initial sensing results.

[0085] Optionally, during the conflict detection and confidence difference assessment process, when there are significant inconsistencies in the initial sensing results of different sensors for the same target object, the driver assistance system will trigger a conflict arbitration mechanism. The core of this conflict arbitration mechanism lies in detecting whether the confidence differences between the target object's position confidence, type confidence, and movement confidence exceed a preset confidence difference threshold. For example, when a lidar identifies a target object as a static obstacle, while a camera identifies it as a moving car, the differences in position confidence, type confidence, and movement confidence between the lidar and camera can be compared. If the difference between position confidence and type confidence, or between movement confidence and position / type confidence, exceeds the confidence difference threshold, the driver assistance system determines that there is a sensing conflict between the different sensors for the same target object and can proceed with further arbitration.

[0086] Optionally, during the environmental adaptive weighting process, if the above judgment reveals a difference in confidence levels between at least two of the three—position confidence, type confidence, and movement confidence—that is, if different sensors experience sensing conflicts for different target objects, the assisted driving system can assign different first weights to different sensors based on the current vehicle environment to optimize the reliability of different initial sensing results. This weighting can be based on environmental adaptation principles. For example, in clear daytime conditions, the first weight of the initial sensing result corresponding to the camera is 70%, and the first weight of the initial sensing result corresponding to the LiDAR is 30%, emphasizing the priority of visual features. In rainy, foggy, or nighttime environments, the first weight of the initial sensing result corresponding to the LiDAR is 70%, and the first weight of the initial sensing result corresponding to the camera is 30%, highlighting the penetration capability of the LiDAR. In tunnel glare scenarios, the first weight of the initial sensing result corresponding to the LiDAR is increased to 80%, and the first weight of the initial sensing result corresponding to the camera is reduced to 20%.

[0087] It should be noted that the magnitude and setting method of the first weight mentioned above are only illustrative examples and are not intended to impose specific limitations. By assigning a first weight to different sensors, the initial sensing results obtained by reliable sensors can be intelligently selected based on environmental characteristics, thereby improving the accuracy and robustness of the target object sensing process.

[0088] Optionally, in determining the sensing performance index, the sensing performance index for each initial sensing result can be calculated using the first weight and the confidence level of each sensor. The formula for calculating the sensing performance index (final score) is: Σ(sensor weight × confidence level), where the sensor weight corresponds to the first weight mentioned above, and the confidence level corresponds to the position confidence level, type confidence level, and motion confidence level mentioned above.

[0089] In this embodiment, the process of determining sensing performance indicators demonstrates the close integration of multimodal sensing result fusion, conflict arbitration, and environmental adaptation mechanisms. By dynamically adjusting the first weight of the sensors, reliable sensors can be selected under various driving conditions, improving the accuracy and consistency of the vehicle's sensing of target objects in the surrounding environment. This ensures that the vehicle makes appropriate decisions based on environmental risks, achieving safe, efficient, and highly adaptable autonomous driving.

[0090] As an optional implementation, step S106, in response to the sensing performance index of the initial sensing result being lower than the performance index threshold, determines the vehicle's risk data based on multiple initial sensing results, including: in response to the sensing performance index of the initial sensing result being lower than the performance index threshold, determining environmental state information of the environment based on multiple initial sensing results, wherein the environmental state information is used to represent the influencing factors corresponding to the initial sensing results in the environment; determining second weights corresponding to the multiple initial sensing results respectively based on the environmental state information; and using the second weights to perform a weighted summation of the multiple initial sensing results to obtain the risk data.

[0091] In this embodiment, during the process of determining vehicle risk data based on multiple initial sensing results, if the sensing performance index of the initial sensing results is lower than a performance index threshold, environmental state information of the vehicle's environment is determined based on the multiple initial sensing results. The environmental state information is then used to determine the second weights corresponding to the multiple initial sensing results. These second weights can then be used to perform a weighted summation of the multiple initial sensing results to obtain the final risk data.

[0092] Optionally, if the initial sensing performance index is lower than the performance index threshold, it indicates that the sensor's sensing capability for the current environment is uncertain or limited. In such cases, additional measures can be taken to improve the accuracy of the risk assessment. The environmental state information will be reconstructed based on multiple initial sensing results, not just those initial sensing results with performance indexes lower than the performance index threshold. This environmental state information may include, but is not limited to, traffic density, weather conditions, and road characteristics, etc., without specific limitations here.

[0093] Optionally, environmental state information can be used to assess factors in the environment that affect the sensing effect of the initial sensing results, such as the number of targets on the road where the vehicle is located (reflecting traffic density), visibility and precipitation (reflecting weather coefficients), curves and slopes (reflecting road curvature), etc., to gain a more comprehensive understanding of the complexity of the environment around the vehicle.

[0094] Optionally, a second weight is determined for each initial sensing result based on environmental state information. This second weight can represent the degree of influence of environmental factors on the reliability of the initial sensing result. For example, under high traffic density, the second weight of the lidar can be increased because lidar performs better in dense traffic.

[0095] Optionally, by utilizing the second weight of each initial sensing result, a weighted sum of the corresponding initial sensing results is obtained to arrive at the final risk data. This process ensures that even when individual sensors perform poorly in sensing the target object, accurate risk assessment can still be made with the assistance of other sensors besides the aforementioned individual sensors and environmental state information. The calculated risk entropy R can be used to determine the appropriate driving strategy and sensor control mode for the vehicle.

[0096] For example, the risk entropy R can be determined using the following formula: R = 0.4 × Traffic Density + 0.3 × Weather Coefficient + 0.3 × Road Curvature, with a real-time update cycle of 100ms for the R value calculation. Traffic density is obtained by counting the number of targets within a 100-meter range detected by LiDAR; the weather coefficient is obtained through camera image clarity analysis; and the road curvature is obtained through high-precision maps and IMU data. The 0.4, 0.3, and 0.3 in the above formula can be considered as secondary weights. Traffic density represents the number of target objects on the road currently being driven by the vehicle. The weather coefficient represents the current weather conditions (e.g., visibility, precipitation). Road curvature represents the degree of curvature and complexity of the road the vehicle is currently navigating, used to determine the vehicle's speed and stability.

[0097] It should be noted that the specific values ​​of the second weight setting mentioned above are only for illustrative purposes and are not subject to specific restrictions. They can be flexibly adjusted according to the actual vehicle environment and application scenario. For example, a higher second weight can be used for traffic density on busy urban roads, while the weight of the weather coefficient can be strengthened in rainy or foggy weather.

[0098] In this embodiment, by comprehensively considering all initial sensing results and environmental state information, the weights of sensor data can be dynamically adjusted to optimize the risk assessment process and ensure the safety and effectiveness of the driving strategy. This mechanism allows for customized risk assessment and decision control in complex driving environments, based on different environmental conditions and application scenarios, by customizing the value of the second weight.

[0099] As an optional implementation, the target control mode includes a first control mode, a second control mode, and a third control mode. The degree of risk of the vehicle under the first control mode is lower than that under the second control mode, and the degree of risk of the vehicle under the second control mode is lower than that under the third control mode. Step S108, based on the risk data, determines the target control mode to be triggered for the vehicle, including: determining the target control mode as the first control mode in response to the risk data being less than a first risk threshold; determining the target control mode as the second control mode in response to the risk data being greater than or equal to the first risk threshold and less than or equal to the second risk threshold; and determining the target control mode as the third control mode in response to the risk data being greater than the second risk threshold.

[0100] In this embodiment, the target control mode may include a first control mode, a second control mode, and a third control mode. The degree of risk posed by the vehicle under the first control mode is lower than that under the second control mode. The degree of risk posed by the vehicle under the second control mode is lower than that under the third control mode. During the process of determining the target control mode to be triggered based on risk data, if the risk data is less than a first risk threshold, the target control mode can be determined to be the first control mode. If the risk data is greater than or equal to the first risk threshold and less than or equal to a second risk threshold, the target control mode can be determined to be the second control mode. If the risk data is greater than the second risk threshold, the target control mode is the third control mode. The first control mode can be a safety mode, the second control mode can be a standard mode, and the third control mode can be an enhanced mode.

[0101] Optionally, based on the previously calculated risk data, the target control mode to be adopted by the vehicle is determined. The above process dynamically adjusts the vehicle's control mode to one of three levels according to the magnitude of the risk data: First Control Mode (Safety Mode), Second Control Mode (Standard Mode), and Third Control Mode (Enhanced Mode). Each of these target control modes corresponds to different driving and sensing strategies, aiming to provide the most suitable driving behavior for the driving strategy and the most suitable sensor configuration for the sensing strategy based on the risk level of the vehicle's environment.

[0102] Optionally, a first risk threshold (R1): When the risk data R is less than the first risk threshold R1, the target control mode is determined to be the first control mode (safe mode). This indicates that the current environment is considered low-risk, and the vehicle can drive in a more economical mode with limited sensor operation. The first risk threshold can be preset to 0.3, without specific restrictions here. A second risk threshold (R2): If the risk data R is greater than or equal to the first risk threshold R1 and less than or equal to the second risk threshold R2, the target control mode will be determined to be the second control mode (standard mode). At this time, the environmental risk level is moderate, and the vehicle can utilize more sensor resources and a higher computing frequency to ensure safe driving. A third control mode (enhanced mode): When the risk data R exceeds the second risk threshold R2, the target control mode can be determined to be the third control mode (enhanced mode). This indicates that the environmental risk is high, and the various sensors on the vehicle can be activated and the data processing capability of the sensing results can be enhanced to cope with potential driving challenges.

[0103] Optionally, the first control mode (safety mode) is suitable for simple, open environments, such as straight sections of highways. In this mode, the vehicle can operate with a lower sensor configuration and computing frequency to save energy and computing resources. The second control mode (standard mode) is mainly for more complex or changing environments, such as urban roads or mild inclement weather. It will activate more comprehensive sensor data and a higher computing frequency to cope with the challenges posed by traffic density and weather. The third control mode (enhanced mode) is used in high-risk environments, such as severe weather (rain, fog, snow), complex intersections, tunnels, etc. In this mode, all sensors in the vehicle are activated and operate at their highest performance level to ensure timely and accurate identification and response to various risks.

[0104] As an optional implementation, the initial sensing results include a first sensing result, a second sensing result, and a third sensing result. The first sensing result is used to represent the two-dimensional visual features of the target object, the second sensing result is used to represent the three-dimensional appearance attributes of the target object, and the third sensing result is used to represent the relative movement state between the vehicle and the target object. The vehicle includes at least one first sensing device, at least one second sensing device, and at least one third sensing device. Step S102 involves acquiring multiple initial sensing results of the vehicle on the target object in the environment during the vehicle's movement, including: during the vehicle's movement, using at least one first sensing device to sense the target object and obtain a first sensing result, using at least one second sensing device to sense the target object and obtain a second sensing result, and using at least one third sensing device to sense the target object and obtain a third sensing result.

[0105] In this embodiment, during the acquisition of multiple initial sensing results, a first sensing device can be used to sense the target object to obtain a first sensing result. A second sensing device can also be used to sense the target object to obtain a second sensing result. At least one third sensing device can also be used to sense the target object to obtain a third sensing result. The sensing devices can be sensors. The first sensing device can be an image acquisition device for the vehicle (e.g., a camera). The second sensing device can be a device for acquiring the three-dimensional appearance attributes of the target object, such as a lidar. The third sensing device can be a device for acquiring the relative movement state between the target object and the vehicle, such as a millimeter-wave radar.

[0106] Optionally, the first sensing result can be obtained by acquiring the two-dimensional visual features of the target object through image acquisition devices on the vehicle, such as forward-facing cameras and surround-view cameras. These features may include the target's color, shape, size, outline, and any visible markings or signs. When acquiring the first sensing result, the focus is on capturing and identifying the surface features of the target object. The second sensing result can be obtained by capturing the target's appearance attributes from a three-dimensional perspective using devices such as LiDAR. LiDAR, by emitting and receiving laser signals, can construct a three-dimensional point cloud model of the target, providing stereoscopic information such as the target's height, depth, and shape. The advantage of the second sensing result is that it can provide precise spatial location and structural information of the target object, used to determine the target object's distance and volume, perform obstacle detection, and construct a three-dimensional environmental model. The third sensing result can be obtained by acquiring the relative movement status information between the target and the vehicle using devices such as millimeter-wave radar. Millimeter-wave radar can measure the target's speed, acceleration, and direction, providing real-time perception of dynamic environmental changes, especially for fast-moving objects, such as overtaking or oncoming vehicles; vehicle speed and direction information is used for real-time obstacle avoidance and path planning.

[0107] Optionally, image acquisition devices (cameras) are primarily used to capture the visual features of target objects, featuring high resolution and excellent color recognition capabilities, suitable for target identification under daytime and good lighting conditions. 3D appearance acquisition devices (LiDAR) can provide 3D point cloud data of target objects, maintaining high accuracy in object detection and distance measurement even in low-light or nighttime environments, making them key devices for constructing 3D environmental models. Relative motion state acquisition devices (millimeter-wave radar) excel at measuring the speed and direction of targets, particularly effective for monitoring distant and high-speed moving objects, and are central to dynamic environment perception and motion prediction.

[0108] As an optional implementation, step S110, in the target control mode, controls the vehicle to perform a sensing operation on the target object to obtain a target sensing result, including: in response to the target control mode being a first control mode, controlling the first sensing device and the third sensing device to perform a first sensing operation on the target object to obtain a target sensing result; in response to the target control mode being a second control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a second sensing operation on the target object to obtain a target sensing result, wherein the execution frequency of the second sensing operation is greater than the execution frequency of the first sensing operation; in response to the target control mode being a third control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a third sensing operation on the target object to obtain a target sensing result, wherein the execution frequency of the third sensing operation is greater than the execution frequency of the second sensing operation.

[0109] In this embodiment, during the process of controlling the vehicle to perform sensing operations on the target object in the target control mode, if the target control mode is the first control mode, the first sensing device and the third sensing device can be controlled to perform the first sensing operation on the target object to obtain the target sensing result. If the target control mode is the second control mode, the first sensing device, the second sensing device, and the third sensing device can be controlled to perform the second sensing operation on the target object to obtain the target sensing result. If the target control mode is the third control mode, the second sensing device, the first sensing device, and the third sensing device can be controlled to perform the third sensing operation on the target object to obtain the target sensing result. The execution frequency of the third sensing operation is greater than the execution frequency of the second sensing operation.

[0110] Optionally, when the target control mode is the first control mode, only the first sensing device (such as a camera) and the third sensing device (such as millimeter-wave radar) are activated, while the second sensing device (such as lidar) may remain in a low-power state or be turned off. This mode is suitable for low-risk environments, such as open roads, where the system's main task is to maintain stable vehicle movement without excessive reliance on high-power, high-precision 3D perception devices. In standard mode, all sensing devices are activated, including the first, second, and third sensing devices. Compared to the first control mode, the second sensing operation is executed more frequently, meaning the system will capture perception data of the target object more often to cope with medium-risk environments, such as urban roads or mild inclement weather. The data acquisition strategy in this mode is a trade-off between risk detection and resource consumption. Enhanced mode is designed for high-risk environments, such as heavy rain, dense fog, and complex intersections. In this mode, not only are all sensing devices fully activated, but the execution frequency of the third sensing operation is further increased, exceeding that of the standard mode. This means the system will collect environmental information to the greatest extent possible to ensure fast and accurate decision-making even under the worst conditions. The scanning frequency of the LiDAR and the frame rate of the camera can be set to the highest permissible values ​​to provide the most detailed environmental perception data.

[0111] Optionally, adjusting the execution frequency can optimize the use of system resources, avoiding resource waste in simple environments while ensuring that the data acquisition density and accuracy in complex or high-risk environments meet the requirements for safe driving. In high-risk situations, increasing the sensing operation frequency can enhance the system's sensitivity to environmental changes, enabling timely detection of potential hazards, such as fast-moving objects or suddenly appearing obstacles, thus allowing for earlier responses and reducing the likelihood of accidents. High-frequency sensing data provides a more stable data stream, which is beneficial for the real-time performance and accuracy of subsequent data fusion and decision-making stages. Especially in dynamic environments, frequent perception updates are crucial for maintaining the system's environmental awareness.

[0112] In this embodiment, the operating status of the sensing device and the data acquisition strategy are dynamically adjusted according to the current risk level. This mechanism not only improves the sensing capability and decision-making efficiency of the target object but also optimizes resource consumption, ensuring that the vehicle can achieve the highest level of safety in a more economical way throughout the entire driving process. Whether on a low-risk open road or in high-risk severe weather or complex road conditions, intelligent sensing operation and data flow management can maintain stable vehicle operation while providing a safe and comfortable travel environment for the driver and passengers.

[0113] As an optional implementation, the method further includes: adjusting the target control mode in response to the risk change rate corresponding to the risk data being greater than the change rate threshold, and controlling the vehicle according to the adjusted target control mode to perform a sensing operation on the target object to obtain the target sensing result.

[0114] In this embodiment, if the risk change rate corresponding to the risk data is greater than the change rate threshold, the target control mode can be adjusted, and the vehicle can be controlled to perform sensing operations on the target object according to the adjusted target control mode to obtain the target sensing result.

[0115] Optionally, the risk change rate (ΔR) can be used to represent the rate at which risk data changes over time, i.e., the increment of risk data per unit time. The risk change rate can be calculated by comparing risk data at consecutive time points. A high risk change rate can indicate that the vehicle is entering or passing through an environment with a rapidly increasing risk level, such as entering an intersection from a straight road or encountering sudden severe weather. The change rate threshold can be a preset critical value for the risk change rate. If the risk change rate exceeds the change rate threshold, the control mode adjustment can be triggered. The change rate threshold can be set by balancing the vehicle's response speed to environmental changes with the risk of erroneous adjustment. For example, the change rate threshold can be determined through analysis of a large amount of experimental data and actual driving scenarios and can be preset to 0.2 / 100ms.

[0116] Optionally, when the risk change rate exceeds a threshold, the system will automatically trigger an adjustment to the target control mode, switching from the current mode to a higher-level mode to address the increased environmental risk. For example, switching from a safe mode (first control mode) to a standard mode (second control mode), or further upgrading from the standard mode to an enhanced mode (third control mode). According to the adjusted target control mode, the system will reconfigure the operating status of the sensing devices, such as increasing the operating frequency of LiDAR and cameras, and increasing the coverage of millimeter-wave radar, to enhance environmental perception and data acquisition speed. The adjusted sensing operation aims to capture key information about the target object more quickly and accurately, supporting subsequent decision-making. In the adjusted target control mode, the system will acquire the latest sensing results of the target object, which include richer and more detailed information collected by the optimized sensor configuration. The acquired target sensing results will be used by the system to reassess environmental risks and to adjust vehicle control strategies in real time to ensure driving safety.

[0117] For example, when a vehicle is detected entering an area where traffic density suddenly increases, the risk change rate may rise sharply, exceeding the change rate threshold. At this point, the system will automatically switch from standard mode to enhanced mode, activating various sensors in the vehicle and increasing the data collection frequency to more accurately identify and locate surrounding vehicles, pedestrians, and obstacles, thereby developing safer path planning and obstacle avoidance strategies.

[0118] In this embodiment, by implementing a mechanism to adjust the target control mode in response to the rate of change of risk, the intelligent driving system can maintain a high degree of adaptability and flexibility in the face of dynamic changes in driving environment risks. This real-time adjustment not only improves the system's response speed to environmental risks but also ensures that it operates with the most suitable configuration in various driving scenarios, providing the safest and most efficient driving support for the vehicle.

[0119] As an optional implementation, the target control mode includes a preset sensing range and a preset sensing frequency. In the target control mode, the vehicle is controlled to perform a sensing operation on the target object to obtain a target sensing result. This includes: responding to the target control mode being a first control mode, controlling a first sensing device and a third sensing device to perform a first sensing operation on the target object within a preset sensing range at a preset sensing frequency corresponding to the first control mode, and obtaining a target sensing result; responding to the target control mode being a second control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a second sensing operation on the target object within a preset sensing range corresponding to the second control mode at a preset sensing frequency corresponding to the second control mode, and obtaining a target sensing result; and responding to the target control mode being a third control mode, controlling the first sensing device, the second sensing device, and the third sensing device to perform a third sensing operation on the target object within a preset sensing range corresponding to the third control mode at a preset sensing frequency corresponding to the third control mode, and obtaining a target sensing result.

[0120] In this embodiment, during the process of controlling the vehicle to perform a sensing operation on a target object in the target control mode, if the target control mode is a first control mode, the first sensing device and the second sensing device can be controlled to perform a first sensing operation on the target object at a preset sensing frequency corresponding to the first control mode within a preset sensing range. If the target control mode is a second control mode, the first sensing device, the second sensing device, and the third sensing device can be controlled to perform a second sensing operation on the target object using a preset sensing frequency corresponding to the second control mode within a preset sensing range corresponding to the second control mode. If the target control mode is a third control mode, the first sensing device, the second sensing device, and the third sensing device can be controlled to perform a third sensing operation on the target object using a preset sensing frequency corresponding to the third control mode within a preset sensing range corresponding to the third control mode.

[0121] Optionally, the preset sensing frequency can be the frequency at which the sensing device collects data, set according to different target control modes. Lower sensing frequencies are suitable for low-risk environments, reducing energy consumption and data processing burden while ensuring safety; while higher sensing frequencies are more suitable for high-risk environments, providing more intensive and real-time environmental information to support faster decision-making processes. The preset sensing range can be the spatial range of the target object detected by the sensing device, set according to the target control mode. A larger sensing range is necessary in high-risk environments because it allows for earlier detection of distant targets, thus allowing more time and space to react. In low-risk environments, a smaller sensing range is sufficient to meet the needs while reducing resource waste.

[0122] Optionally, in safety mode, the operation of the first sensing device (such as a camera) and the third sensing device (such as millimeter-wave radar) can be controlled. Sensing operations will be performed at a lower preset sensing frequency within a smaller preset sensing range to acquire target sensing results. This configuration is suitable for low-risk environments, such as open areas and straight roads, effectively reducing energy consumption and computational load. In standard mode, the first, second, and third sensing devices can participate in the sensing work, operating at a higher preset sensing frequency than in safety mode and expanding the sensing range to adapt to more complex environments. This mode is suitable for medium-risk scenarios such as urban roads and mixed traffic environments, requiring more frequent updates to the perception of target objects while covering a larger spatial area to ensure timely response to potential emergencies.

[0123] Optionally, the enhanced mode is the most stringent and comprehensive perception configuration, where all sensing devices operate at the highest preset sensing frequency and cover the largest preset sensing range. In enhanced mode, the sensing of target objects achieves extreme density and breadth, suitable for high-risk environments such as severe weather conditions (rain, snow, fog), complex intersections, and emergency obstacle avoidance scenarios, ensuring that even in extreme situations, sufficient information can be obtained for safety decisions.

[0124] In this embodiment, by dynamically adjusting the operating frequency and sensing range of the sensing device, the most suitable perception strategy can be adopted in driving environments with different risk levels, ensuring both driving safety and efficient resource utilization. This mechanism enables autonomous vehicles to not only drive energy-efficiently in simple road environments but also maintain high alertness and responsiveness in complex and changing environments, laying a solid foundation for the widespread applicability and safety of intelligent driving. By rationally configuring the operating parameters of the sensing device, the perception strategy can be flexibly adjusted based on real-time environmental risk assessment.

[0125] The technical solutions of the embodiments of this application will be illustrated below with reference to preferred embodiments.

[0126] Currently, in the field of intelligent driving, related autonomous driving systems typically adopt a serial architecture. This means that the perception layer collects data and transmits it to the decision-making layer for processing. The decision-making layer then makes decisions and plans routes based on the received information. While this architecture can meet the basic requirements of autonomous driving to a certain extent, its limitations become increasingly apparent when facing complex and ever-changing environments. The decision-making layer cannot obtain the confidence level of the raw perception data, leading to misjudgments in risky scenarios; it cannot dynamically adjust the sensor combination and computing frequency according to scenario risks; and it lacks an arbitration mechanism when a single sensor fails (such as a cascading decision error caused by camera overexposure). Therefore, the technical problem of low vehicle control accuracy still exists.

[0127] This application proposes a multimodal perception and decision-making collaboration method for intelligent driving. During vehicle movement, it can sense target objects in the vehicle's environment in real time across different sensing dimensions, obtaining multiple initial sensing results. The sensing performance index of these initial sensing results can be determined. If the sensing performance index of the initial sensing results is lower than a performance index threshold, risk data for the vehicle can be determined based on these initial sensing results. Based on the risk data, a target control mode to be triggered for the vehicle is determined. The vehicle is then controlled to enter the target control mode, and within this mode, the vehicle performs sensing operations on the target object, obtaining a target sensing result with a sensing performance index greater than or equal to the performance index threshold. In other words, this application dynamically evaluates the sensing performance index of the initial multimodal sensing data to identify and quantify the risks in the vehicle's environment, thereby intelligently adjusting the vehicle's control mode. Unlike the serial architecture in related technologies that ignores the impact of environmental factors on sensing accuracy, this embodiment actively analyzes the potential risks faced by the vehicle when the sensing performance index is lower than the performance index threshold, and selects an appropriate control mode based on these risks. This ensures that even in complex environments, the vehicle can make accurate and safe driving decisions based on target sensing results with high sensing performance indexes, effectively improving the control accuracy and driving safety of the vehicle in complex environments. This achieves the technical effect of improving vehicle control accuracy and solves the technical problem of low vehicle control accuracy.

[0128] The methods of the embodiments of this application will be further illustrated below.

[0129] In this embodiment, Figure 2 This is a schematic diagram of an intelligent driving multimodal perception and decision-making collaboration system according to an embodiment of this application, such as... Figure 2As shown, the system may include a millimeter-wave radar 201, a camera 202, a lidar 203, a collaborative perception module 204, a dynamic decision control module 205, and an execution unit 206. The camera 202 (a forward-facing binocular camera, mounted on the top inside the windshield, with a resolution of 8 megapixels and a frame rate of 30fps) is responsible for acquiring images of lane lines, traffic signs, and obstacles ahead. The lidar 203 is a 128-line rotating lidar deployed in the center of the vehicle roof, with a horizontal viewing angle of 360° and a vertical viewing angle of 40°, generating 320,000 point cloud data points per second for constructing a high-precision 3D environment model. The millimeter-wave radar 201 can be installed at each of the four corners of the vehicle, operating at a frequency of 77GHz, with a detection range of 300 meters, and can penetrate rain and fog to measure target motion parameters (speed, azimuth, etc.).

[0130] Optionally, such as Figure 2 As shown, the collaborative perception module 204 is built on a high-functional-safety-level microcontroller unit (MCU) chip, connecting all sensors via gigabit Ethernet to achieve spatiotemporal alignment of multi-source data; target confidence calculation; and a multimodal perception conflict arbitration module. The dynamic decision control module 205 uses a high-performance system-on-a-chip (SOC) chip for real-time calculation of scene risk entropy and generation of dynamic sensor scheduling instructions. The execution unit: the vehicle drive-by-wire system (steering / brake / throttle controller) receives instructions via the Controller Area Network Flexible Data-rate (CANFD) bus and executes vehicle motion control for autonomous driving.

[0131] Figure 3 This is a schematic diagram of a software system architecture according to an embodiment of this application, such as... Figure 3 As shown, the architecture may include sensing data 301, confidence calculation 302, spatiotemporal alignment 303, risk entropy calculation 304, and dynamic mode scheduling 305. Specifically, sensing data 301 may include millimeter-wave-target motion matrix 3011, camera-target image 3012, and LiDAR-point cloud data 3013. Confidence calculation 302 may include position confidence 3021, type confidence 3022, motion confidence 3023, and multimodal arbitration 3024. Spatiotemporal alignment 303 may include LiDAR point cloud mapping 3031 and target coordinate system 3032. Risk entropy calculation 304 may include traffic density 3041, weather coefficient 3042, and road curvature 3043. Motion mode scheduling 305 may include safe mode 3051, standard mode 3052, and enhanced mode 3053.

[0132] like Figure 3As shown, during the data acquisition phase (0–20ms), each sensor is synchronously activated via hardware triggering. The millimeter-wave radar analyzes the motion matrix of the target object. The binocular camera generates a stereo image of the target object. The lidar generates the raw point cloud data for the current frame. The generated data is transmitted to the collaborative perception layer. During the collaborative perception phase (20–50ms), the perceived data is spatiotemporally aligned, mapping the lidar point cloud to the camera coordinate system to verify the spatial consistency of the target. Confidence calculations include position confidence (3021Cp), type confidence (3022Ct), and motion confidence (3023Cm). Position confidence (Cp) is obtained by multi-sensor detection area overlap rate; type confidence (Ct) is obtained by fusing visual classification and lidar point cloud geometric features; and motion confidence (Cm) is obtained by comparing the deviation between radar velocity measurement and trajectory prediction.

[0133] Figure 4 This is a flowchart of a multimodal arbitration mechanism according to an embodiment of this application, such as... Figure 4 As shown, the method may include the following steps:

[0134] Step S401: Receive target data. The target data can be perception data. The vehicle can receive target data input from sensors such as LiDAR and cameras.

[0135] Step S402: Determine if there is a target conflict. This can also be understood as determining whether targets within the same space are different for the same target object. If there is a target conflict, proceed to step S403; otherwise, proceed to step S410. For example, compare the differences in the target's position / type / motion confidence level, setting a threshold of >0.3 for a conflict.

[0136] Step S403: Initiate the arbitration mechanism. In this embodiment, if the determination of target objects in the same space is inconsistent (e.g., the camera identifies it as a car, while the lidar identifies it as an obstacle), the arbitration mechanism is triggered.

[0137] Step S404, Feature Backtracking Analysis. In this embodiment, when the arbitration mechanism is activated simultaneously, the features of the target object are backtracked to the original physical feature layer of the sensor (such as point cloud geometry, image texture, etc.) to verify the target attributes from the essence of the data and improve the confidence of the target.

[0138] Step S405, Environmental Adaptive Weighting. In this embodiment, environmental adaptive weighting is performed on the target object, wherein: during sunny daytime, the camera weight is 70% and the LiDAR weight is 30%, with the processing principle being visual feature priority; during rainy, foggy, or nighttime conditions, the camera weight is 30% and the LiDAR weight is 70%, with the processing principle being LiDAR feature priority; during tunnel glare conditions, the camera weight is 20% and the LiDAR weight is 80%, with the processing principle being LiDAR point cloud dominance.

[0139] Step S406, final decision. In this embodiment, the final arbitration can be based on the formula: final score = Σ(sensor weight × confidence level).

[0140] Step S407: Calculate the weighted score. In this embodiment, the condition for adopting a target is: if Σ≥0.8, it is adopted as a valid target; if Σ<0.8, it is marked as a target to be verified (transferred to the decision-making level).

[0141] Step S408: Determine if the score is ≥0.8. If the score is ≥0.8, proceed to step S410; if the score is <0.8, proceed to step S409.

[0142] Step S410, Adopt the target. In this embodiment, the adopted target can be used to determine the driving strategy for subsequent vehicle driving processes.

[0143] Figure 5 This is a flowchart of a dynamic mode scheduling according to an embodiment of this application, such as... Figure 5 As shown, the scheduling process may include the following steps:

[0144] Step S501: Calculate the risk entropy. In this embodiment, the risk entropy R is calculated using the formula R = 0.4 × traffic density + 0.3 × weather coefficient + 0.3 × road curvature. The real-time update cycle for the R value is 100ms. Traffic density is obtained by counting the number of targets within a 100-meter range sensed by LiDAR; the weather coefficient is obtained through camera image clarity analysis; and the road curvature is obtained through high-precision map data and inertial measurement unit data.

[0145] Step S502, Dynamic mode adjustment.

[0146] In this embodiment, the magnitude of the risk entropy R corresponds to different risk levels (safe mode, standard mode, and enhanced mode). When R < 0.3, the system enters safe mode, using only forward-facing camera data and millimeter-wave radar data, with a calculation frequency of 10Hz. This mode is suitable for open highways. When 0.3 ≤ R ≤ 0.7, the system uses a full configuration of forward-facing camera, millimeter-wave radar, and lidar data, with a calculation frequency of 20Hz. This mode is suitable for ordinary urban roads. When R > 0.7, the system uses all sensors, an increased lidar scanning frequency, and improved camera side focusing, increasing the calculation frequency to 40Hz. This mode is suitable for rainy days, nighttime, and complex intersections. To handle sudden changes in risk entropy, a mechanism is implemented: when ΔR > 0.2 / 100ms is detected, it is determined whether R is rising or falling sharply. If rising, the mode is upgraded to a higher level. If falling, a 3-cycle delay is used to confirm a subsequent downgrade.

[0147] Optionally, the scanning and activation strategies for LiDAR and millimeter-wave radar differ in safety mode, standard mode, and enhanced mode. For example, in safety mode, the LiDAR's horizontal x vertical angle needs to be 120° x 30°, the scanning frequency 10Hz, and the point cloud density minimum; in standard mode, the scanning frequency is the standard 20Hz, and the point cloud density is standard; while in enhanced mode, the scanning frequency of the region of interest (sensing area) is increased to 30Hz. The camera is set to 1080P@15 in safety mode and 4K@30fps in enhanced mode. The millimeter-wave radar is set to two forward-facing radars at 10Hz in safety mode and four radars at 40Hz in enhanced mode.

[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0149] According to another aspect of the embodiments of this application, corresponding to the embodiments of the above-described vehicle control method, this specification also provides a vehicle control device.

[0150] Figure 6 This is a schematic diagram of a vehicle control device according to an embodiment of this application, such as... Figure 6As shown, the vehicle control device 60 may include: an acquisition module 602, a first determination module 604, a second determination module 606, a third determination module 608, and a control module 610. The acquisition module 602 is used to acquire multiple initial sensing results of the vehicle on target objects in its environment during vehicle movement; the first determination module 604 is used to determine the sensing performance index of the multiple initial sensing results; the second determination module 606 is used to determine the vehicle's risk data based on the multiple initial sensing results in response to the sensing performance index of the initial sensing results being lower than a performance index threshold; the third determination module 608 is used to determine the target control mode to be triggered by the vehicle based on the risk data; and the control module 610 is used to control the vehicle to perform sensing operations on the target object in the target control mode to obtain target sensing results.

[0151] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0152] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0153] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0154] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0155] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0156] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0157] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0158] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0159] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0160] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0161] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for controlling a vehicle, characterized in that, include: During the movement of the vehicle, multiple initial sensing results of the vehicle on target objects in the environment are acquired, wherein the different initial sensing results are obtained by the vehicle sensing the target objects in different sensing dimensions; Determine a plurality of sensing performance indicators for the initial sensing results, wherein the sensing performance indicators corresponding to the initial sensing results are used to represent the sensing effect of the target object in the initial sensing results; In response to the initial sensing results having a sensing performance index below a performance index threshold, risk data for the vehicle is determined based on multiple initial sensing results, wherein the risk data is used to represent the degree to which the vehicle poses a risk in the environment; Based on the risk data, a target control mode to be triggered for the vehicle is determined, wherein different target control modes correspond to different risk data of the vehicle; In the target control mode, the vehicle is controlled to perform a sensing operation on the target object to obtain a target sensing result. The sensing performance index corresponding to the target sensing result is greater than or equal to the performance index threshold. The sensing performance index corresponding to the target sensing result is used to represent the sensing effect of the target object in the target sensing result.

2. The method according to claim 1, characterized in that, Determine the sensing performance metrics for the multiple initial sensing results, including: Spatiotemporal alignment processing is performed on multiple initial sensing results to obtain a spatiotemporal alignment result, wherein the spatiotemporal alignment result is used to represent different initial sensing results of the same target object at the same time and in the same space; Based on the spatiotemporal alignment results, the confidence level of the target object is determined, wherein the confidence level is used to represent the degree of difference between different initial sensing results for the same target object at the same time and in the same space; Based on the confidence level, a sensing performance index of multiple initial sensing results is determined, wherein the confidence level and the sensing performance index of the initial sensing results are positively correlated.

3. The method according to claim 2, characterized in that, The confidence level includes location confidence, type confidence, and movement confidence. The location confidence level represents the degree of difference in the location of the same target object in different initial sensing results. The type confidence level represents the degree of difference in the object type of the same target object in different initial sensing results. The movement confidence level represents the degree of difference in the movement state of the same target object in different initial sensing results. Determining the location confidence level of the target object based on the spatiotemporal alignment result includes: Based on the spatiotemporal alignment results, the degree of regional overlap between the sensing regions of different initial sensing results is determined; Based on the region overlap, the location confidence is determined, wherein the region overlap is positively correlated with the location confidence. Based on the spatiotemporal alignment results, the type confidence of the target object is determined, including: Based on the spatiotemporal alignment results, the object type to which the target object belongs in different initial sensing results is determined; Based on the object type corresponding to the different initial sensing results, the type confidence level is determined; Based on the spatiotemporal alignment result, determining the movement confidence of the target object includes: Based on the spatiotemporal alignment result, the movement parameters of the target object in different initial sensing results are determined, wherein the movement parameters are used to represent the movement state of the target object; The movement confidence level is determined based on the movement parameters corresponding to the different initial sensing results.

4. The method according to claim 3, characterized in that, Based on the confidence level, a plurality of sensing performance metrics for the initial sensing results are determined, including: In response to a confidence difference greater than a confidence difference threshold between at least two of the location confidence, type confidence, and movement confidence, a first weight corresponding to different initial sensing results is determined. Based on the first weight, and the location confidence, type confidence, and movement confidence corresponding to different initial sensing results, the sensing performance index of the initial sensing result is determined.

5. The method according to claim 1, characterized in that, In response to the initial sensing results showing that the sensing performance index is below a performance index threshold, risk data for the vehicle is determined based on multiple initial sensing results, including: In response to the sensing performance index of the initial sensing result being lower than the performance index threshold, environmental state information of the environment is determined based on multiple initial sensing results, wherein the environmental state information is used to represent the influencing factors corresponding to the initial sensing results in the environment; Based on the environmental state information, second weights are determined for each of the initial sensing results; Using the second weight, the multiple initial sensing results are weighted and summed to obtain the risk data.

6. The method according to claim 1, characterized in that, The target control mode includes a first control mode, a second control mode, and a third control mode. The degree of risk present in the vehicle under the first control mode is lower than the degree of risk present in the vehicle under the second control mode, and the degree of risk present in the vehicle under the second control mode is lower than the degree of risk present in the vehicle under the third control mode. Based on the risk data, the target control mode to be triggered for the vehicle is determined, including: In response to the risk data being less than a first risk threshold, the target control mode is determined to be the first control mode; In response to the risk data being greater than or equal to the first risk threshold and less than or equal to the second risk threshold, the target control mode is determined to be the second control mode; In response to the risk data being greater than the second risk threshold, the target control mode is determined to be the third control mode.

7. The method according to claim 6, characterized in that, The initial sensing results include a first sensing result, a second sensing result, and a third sensing result. The first sensing result represents the two-dimensional visual features of the target object, the second sensing result represents the three-dimensional appearance attributes of the target object, and the third sensing result represents the relative movement state between the vehicle and the target object. The vehicle includes at least one first sensing device, at least one second sensing device, and at least one third sensing device. During the vehicle's movement, multiple initial sensing results of the vehicle on the target object in its environment are acquired, including: During the movement of the vehicle, at least one of the first sensing devices is used to sense the target object to obtain the first sensing result, and at least one of the second sensing devices is used to sense the target object to obtain the second sensing result, and at least one of the third sensing devices is used to sense the target object to obtain the third sensing result; And / or, In the target control mode, the vehicle is controlled to perform sensing operations on the target object to obtain target sensing results, including: In response to the target control mode being the first control mode, the first sensing device and the third sensing device are controlled to perform a first sensing operation on the target object to obtain the target sensing result; In response to the target control mode being the second control mode, the first sensing device, the second sensing device, and the third sensing device are controlled to perform a second sensing operation on the target object to obtain the target sensing result, wherein the execution frequency of the second sensing operation is greater than the execution frequency of the first sensing operation; In response to the target control mode being the third control mode, the first sensing device, the second sensing device, and the third sensing device are controlled to perform a third sensing operation on the target object to obtain the target sensing result, wherein the execution frequency of the third sensing operation is greater than the execution frequency of the second sensing operation; And / or, The method further includes: adjusting the target control mode in response to the risk change rate corresponding to the risk data being greater than the change rate threshold, and controlling the vehicle according to the adjusted target control mode to perform the sensing operation on the target object to obtain the target sensing result.

8. The method according to claim 7, characterized in that, The target control mode includes a preset sensing range and a preset sensing frequency. Under this mode, the vehicle is controlled to perform sensing operations on the target object, obtaining the target sensing results, including: In response to the target control mode being the first control mode, the first sensing device and the third sensing device are controlled to perform the first sensing operation on the target object at the preset sensing frequency corresponding to the first control mode within the preset sensing range, so as to obtain the target sensing result; In response to the target control mode being the second control mode, the first sensing device, the second sensing device, and the third sensing device are controlled to perform the second sensing operation on the target object at the preset sensing frequency corresponding to the second control mode and within the preset sensing range corresponding to the second control mode, so as to obtain the target sensing result; In response to the target control mode being the third control mode, the first sensing device, the second sensing device, and the third sensing device are controlled to perform the third sensing operation on the target object at the preset sensing frequency corresponding to the third control mode and within the preset sensing range corresponding to the third control mode, so as to obtain the target sensing result.

9. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method according to any one of claims 1 to 8.

Images