Driving behavior model adaptive updating method based on cloud platform

Abstract

The invention relates to a driving behavior model adaptive updating method based on a cloud platform, and the method is applied to vehicle equipment, and comprises the steps: obtaining the parameter information of each sensor in at least one sensor; sending parameter information of each sensor in the at least one sensor to a cloud platform; receiving a downloading link of a target preset driving model returned by the cloud platform, wherein the target preset driving model is determined by the cloud platform based on parameter information of each sensor in the at least one sensor; downloading a target preset driving model based on the downloading link; and updating a local driving model of the vehicle equipment according to the target preset driving model. Visibly, the driving model on the vehicle equipment is determined based on parameter information of a sensor carried by the vehicle equipment. The mechanism can ensure that the driving model is matched with the data acquisition precision of the vehicle equipment sensor, effectively avoids the occurrence of event false triggering caused by the fact that the driving model is difficult to adapt to the precision difference of different sensors, and improves the driving experience of a user.

Description

Technical Field

[0001] This application relates to the field of intelligent vehicles, and in particular to an adaptive update method for driving behavior models based on a cloud platform. Background Technology

[0002] With the rapid development of vehicle-to-everything (V2X) technology, intelligent driving has been widely applied in the field of intelligent vehicles. The core support for realizing intelligent driving lies in driving models. These models not only assist users in driving and reduce operational difficulty, but also provide real-time alerts to help users avoid risks and reduce accidents.

[0003] Currently, traditional intelligent driving mostly adopts a "train-and-deploy" model. Specifically, a cloud platform trains a driving model based on massive amounts of user data, and then deploys the trained driving model to various vehicle devices for their use.

[0004] However, the same type of sensors in different vehicles often have significant individual differences, resulting in inconsistent accuracy of the collected data. This makes it difficult for driving models to make accurate judgments, which in turn leads to false event triggering and seriously affects the user's driving experience. Summary of the Invention

[0005] This application provides a cloud-based adaptive update method for driving behavior models, aiming to solve the technical problem of false event triggering caused by the difficulty of adapting driving models to the accuracy differences of different sensors.

[0006] In a first aspect, embodiments of this application provide an adaptive update method for a driving behavior model based on a cloud platform. The method is applied to vehicle equipment, which includes at least one sensor, comprising:

[0007] Obtain parameter information and sampling data for each of the at least one sensor;

[0008] Statistical features are extracted from the sampling data of each of the at least one sensor;

[0009] Send the parameter information and statistical characteristics of each of the at least one sensor to the cloud platform;

[0010] Receive a download link for a target preset driving model returned by the cloud platform, wherein the target preset driving model is determined by the cloud platform based on the parameter information and statistical characteristics of each of the at least one sensor;

[0011] Download the target preset driving model based on the download link;

[0012] The local driving model of the vehicle equipment is updated according to the target preset driving model.

[0013] Optionally, the method further includes:

[0014] Historical driving data is reported to the cloud platform so that the cloud platform can determine the correction value of the target parameter in the target preset driving model based on the historical driving data;

[0015] Receive the corrected value of the target parameter returned from the cloud platform;

[0016] The target parameters of the target preset driving model are adjusted based on the correction values ​​of the target parameters.

[0017] Optionally, the historical driving data includes: vehicle status data, driving behavior data, and environmental perception data, and the target parameters include: rapid deceleration threshold and collision threshold.

[0018] Secondly, embodiments of this application provide an adaptive update method for a driving behavior model based on a cloud platform, which includes:

[0019] The method is applied to a cloud platform, and the method includes:

[0020] Receive parameter information and statistical characteristics of each sensor from at least one sensor in the vehicle equipment;

[0021] Features are extracted from the parameter information and statistical characteristics of each of the at least one sensor to obtain multiple feature information;

[0022] Based on the aforementioned multiple feature information, a target preset driving model is selected from multiple preset driving models;

[0023] Return a download link for the target preset driving model to the vehicle device, so that the vehicle device can download the target preset driving model based on the download link.

[0024] Optionally, each of the plurality of preset driving models corresponds to a plurality of label information, and the plurality of label information corresponds one-to-one with the plurality of feature information. The step of selecting a target preset driving model from the plurality of preset driving models based on the plurality of feature information includes:

[0025] Select a preset driving model that meets the first condition from the plurality of preset driving models, and use the preset driving model as the target preset driving model. The first condition includes matching each label information in the plurality of label information corresponding to the preset driving model with the feature information corresponding to the label information.

[0026] Optionally, the method further includes:

[0027] Receive historical driving data reported from the vehicle equipment;

[0028] Based on the historical driving data, the correction values ​​of the target parameters in the target preset driving model are determined;

[0029] The vehicle equipment returns a correction value for the target parameter to the vehicle equipment so that the vehicle equipment adjusts the target parameters of the target preset driving model based on the correction value of the target parameter.

[0030] Optionally, the historical driving data includes vehicle state data, driving behavior data, and environmental perception data. The step of determining the correction value of the target parameter in the target preset driving model based on the historical driving data includes:

[0031] Acceleration response features, bump features, and roll features are extracted from the vehicle status data.

[0032] Extract driving behavior features from the driving behavior data;

[0033] Extract road condition features from the environmental perception data;

[0034] The correction value of the target parameter is determined based on the acceleration response characteristics, the bump characteristics, the roll characteristics, the driving behavior characteristics, and the road condition characteristics.

[0035] Optionally, determining the correction value of the target parameter based on the acceleration response characteristics, the bump characteristics, the roll characteristics, the driving behavior characteristics, and the road condition characteristics includes:

[0036] The vehicle type characteristics of the equipment are determined based on the acceleration response characteristics, the bump characteristics, and the roll characteristics;

[0037] The correction value of the target parameter is determined based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics.

[0038] Optionally, the driving behavior characteristics include: frequency of rapid deceleration and average intensity of rapid deceleration; the target parameter includes: a rapid deceleration threshold; and determining the correction value of the target parameter based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics includes:

[0039] The correction value for the rapid deceleration threshold is determined based on the vehicle characteristics, the frequency of rapid deceleration, the average intensity of rapid deceleration, and the road condition characteristics.

[0040] Optionally, the driving behavior characteristics further include: collision frequency; the target parameters further include: collision threshold and average collision intensity; and determining the correction value of the target parameters based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics includes:

[0041] The correction value for the collision threshold is determined based on the vehicle characteristics, the collision frequency, the average collision intensity, and the road condition characteristics.

[0042] Thirdly, embodiments of this application also provide a cloud-based adaptive update device for a driving behavior model, which includes a unit for performing the above-described method.

[0043] Fourthly, embodiments of this application also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0044] Fifthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method.

[0045] This application provides a cloud platform-based adaptive update method for a driving behavior model. The method is applied to a vehicle device, which includes at least one sensor. The method includes: acquiring parameter information of each of the at least one sensor; sending the parameter information of each of the at least one sensor to a cloud platform; receiving a download link for a target preset driving model returned by the cloud platform, the target preset driving model being determined by the cloud platform based on the parameter information of each of the at least one sensor; downloading the target preset driving model based on the download link; and updating the local driving model of the vehicle device according to the target preset driving model. Therefore, the technical solution of this application acquires parameter information of each of the at least one sensor. Then, it sends the parameter information of each of the at least one sensor to the cloud platform. Furthermore, it receives a download link for a target preset driving model returned by the cloud platform. The target preset driving model is determined by the cloud platform based on the parameter information of each of the at least one sensor. Finally, it downloads the target preset driving model based on the download link and updates the local driving model of the vehicle device according to the target preset driving model. Thus, the driving model on the vehicle device is determined based on the parameter information of the sensors carried by the vehicle device itself. This mechanism ensures that the data acquisition accuracy of the driving model matches that of the vehicle's equipment sensors, effectively avoiding false triggering of events caused by the driving model's inability to adapt to the accuracy differences of different sensors, thereby improving the user's driving experience. Attached Figure Description

[0046] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0049] Figure 1a One of the flowcharts of an adaptive update method for a driving behavior model based on a cloud platform provided in this application embodiment;

[0050] Figure 1b A system architecture diagram provided for an embodiment of this application;

[0051] Figure 2 A second schematic flowchart illustrating a cloud-based adaptive update method for a driving behavior model, provided in an embodiment of this application.

[0052] Figure 3 One of the schematic block diagrams of a cloud-based adaptive update device for driving behavior model provided in this application embodiment;

[0053] Figure 4 A second schematic block diagram of a cloud-based adaptive update device for driving behavior models provided in this application embodiment;

[0054] Figure 5 A computer device provided in an embodiment of this application. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0056] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0057] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0058] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0059] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0060] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0061] To address the technical problem of false event triggering caused by the difficulty of adapting driving models to the accuracy differences of different sensors in existing technologies, this application provides a cloud-based adaptive update device for driving behavior models, which enables driving models to make accurate judgments by adapting to the accuracy of different sensors.

[0062] Figure 1a This is one of the flowcharts illustrating a cloud-based adaptive update method for a driving behavior model, provided in an embodiment of this application. In one embodiment, the method is applied to a vehicle device, the vehicle device including at least one sensor, and the method includes steps S101-S105.

[0063] S101. Obtain parameter information for each of at least one sensor.

[0064] The sensor's parameter information includes, but is not limited to, sampling frequency and sensor type.

[0065] S102. Send parameter information of at least one sensor to the cloud platform.

[0066] A cloud platform refers to a cloud computing service platform that has the functions of data reception, data processing, model training, model matching, and information feedback. A cloud platform includes, but is not limited to, a distributed computing system composed of one or more servers, computing resource clusters, storage modules, and software operating environments.

[0067] S103. Receive the download link for the target preset driving model returned from the cloud platform.

[0068] The target preset driving model is determined by the cloud platform based on the parameter information of each sensor in at least one sensor.

[0069] It should be noted that the cloud platform has multiple built-in preset driving models. This application will describe in detail in the later embodiments how to select the target preset driving model from the multiple preset driving models based on the parameter information of each sensor. Here, this application will not repeat the details.

[0070] S104. Download the target preset driving model based on the download link.

[0071] In this embodiment, the target preset driving model is retrieved from the cloud platform according to the download link and stored in a local read / write partition.

[0072] S105. Update the local driving model of the vehicle equipment according to the target preset driving model.

[0073] It should be noted that the vehicle equipment employs a dual-partition redundancy design. Specifically, the downloaded target preset driving model is stored in the read-write partition, and the default driving model is stored in the read-only partition. Upon startup, the vehicle equipment first checks if a driving model exists in the read-write partition. If it does, that driving model is loaded directly. If the driving model does not exist in the read-write partition, the default driving model from the read-only partition is loaded. Additionally, the local driving model is stored in the read-write partition. In other words, the vehicle equipment can store a maximum of two driving models at any given time: one default driving model in the read-only partition and another in the read-write partition. The default driving model contains default parameters, including but not limited to rapid deceleration thresholds and collision thresholds.

[0074] It should be noted that if the vehicle device fails to download the target preset driving model or cannot successfully load the target preset driving model, the vehicle device will load the default driving model on the read-only partition. The fault recovery time is usually less than 100ms.

[0075] This application provides a cloud platform-based adaptive update method for a driving behavior model. The method is applied to a vehicle device, which includes at least one sensor. The method includes: acquiring parameter information of each of the at least one sensor; sending the parameter information of each of the at least one sensor to a cloud platform; receiving a download link for a target preset driving model returned by the cloud platform, the target preset driving model being determined by the cloud platform based on the parameter information of each of the at least one sensor; downloading the target preset driving model based on the download link; and updating the local driving model of the vehicle device according to the target preset driving model. Therefore, the technical solution of this application acquires parameter information of each of the at least one sensor. Then, it sends the parameter information of each of the at least one sensor to the cloud platform. Furthermore, it receives a download link for a target preset driving model returned by the cloud platform. The target preset driving model is determined by the cloud platform based on the parameter information of each of the at least one sensor. Finally, it downloads the target preset driving model based on the download link and updates the local driving model of the vehicle device according to the target preset driving model. Thus, the driving model on the vehicle device is determined based on the parameter information of the sensors carried by the vehicle device itself. This mechanism ensures that the data acquisition accuracy of the driving model matches that of the vehicle's equipment sensors, effectively avoiding false triggering of events caused by the driving model's inability to adapt to the accuracy differences of different sensors, thereby improving the user's driving experience.

[0076] In another embodiment, the vehicle equipment periodically acquires data from each of at least one sensor and extracts features from the data to obtain multiple noise distributions and extreme value statistics. Then, the vehicle equipment reports these multiple noise distributions and extreme value statistics to a cloud platform, which determines a target preset driving model based on the multiple noise distributions, extreme value statistics, and parameter information.

[0077] In another embodiment, the vehicle equipment periodically acquires data from each of at least one sensor. This data is then sent to an edge node. Upon receiving the data from each sensor, the edge node extracts features from the data to obtain multiple noise distributions and extreme value statistics. Finally, the edge node reports these noise distributions and extreme value statistics to a cloud platform. The cloud platform determines a target preset driving model based on the multiple noise distributions, extreme value statistics, and parameter information.

[0078] In yet another embodiment, please refer to Figure 1b , Figure 1b This is a system architecture diagram provided for an embodiment of this application. The end-side device is a vehicle device. The cloud platform is a cloud platform. The vehicle device periodically acquires data from each of at least one sensor, extracts features from the data, and obtains multiple noise distributions and extreme value statistics. Then, the vehicle device reports the multiple noise distributions and extreme value statistics to the edge nodes. Furthermore, the edge nodes locally cache the multiple noise distributions and extreme value statistics and report them to the cloud platform. Finally, the cloud platform determines a target preset driving model based on the multiple noise distributions, extreme value statistics, and parameter information, and notifies the vehicle device.

[0079] It should be noted that the vehicle equipment only needs to report the parameter information of each sensor to the cloud platform once. This parameter information is considered the fingerprint data of each sensor. Throughout the entire lifecycle of the vehicle equipment, the parameter information of each sensor only needs to be reported once and stored in the vehicle file on the cloud platform. Furthermore, the cloud platform periodically determines the target preset driving model and returns this determined target preset driving model to the vehicle equipment in real time, enabling the vehicle equipment to update its local driving model in real time.

[0080] It should be noted that the vehicle equipment also reports information about the driving model on the current read / write partition to the cloud platform to indicate whether the cloud platform needs to return the confirmed driving model. If the target preset driving model determined by the cloud platform is the same as the driving model stored on the vehicle equipment, the cloud platform does not need to return the target preset driving model to the vehicle equipment.

[0081] It should be noted that if the vehicle equipment cannot correctly load the driving model on the read-write partition, the vehicle equipment will load the default driving model on the read-only partition.

[0082] In one embodiment, the method further includes: S106-S108.

[0083] S106. Report historical driving data to the cloud platform so that the cloud platform can determine the correction value of the target parameter in the target preset driving model based on the historical driving data.

[0084] Historical driving data includes, but is not limited to, vehicle status data, driving behavior data, and environmental perception data. Vehicle status data includes, but is not limited to, vehicle acceleration and acceleration time. Vehicle acceleration includes horizontal and vertical acceleration. Vertical acceleration includes the left and right vertical acceleration of the vehicle's equipment. Driving behavior data includes, but is not limited to, data on rapid acceleration, rapid deceleration, and sharp turns. Environmental perception data includes, but is not limited to, road condition types and the percentage of mileage for each road condition type in the vehicle's travel route. Road condition types include, but are not limited to, construction roads, rugged mountain roads, low-speed roads, urban roads, and highways.

[0085] The target parameters include, but are not limited to, the rapid deceleration threshold and the collision threshold.

[0086] It should be noted that the vehicle equipment needs to periodically report historical driving data to the cloud platform in order to adjust the correction values ​​of the target parameters in the target preset driving model.

[0087] S107. Receive the correction value of the target parameter returned from the cloud platform.

[0088] S108. Adjust the target parameters of the target preset driving model based on the correction value of the target parameters.

[0089] It should be noted that S107-S108 will be explained in detail below.

[0090] In this embodiment, the corrected value of the target parameter returned from the cloud platform is received. Then, the value of the target parameter of the target preset driving model on the vehicle equipment's read / write partition is adjusted according to the corrected value of the target parameter.

[0091] Please see Figure 2 , Figure 2 This application provides a second method for adaptive updating of a driving behavior model based on a cloud platform. In one embodiment, the method is applied to a cloud platform and includes steps S201-S204.

[0092] S201, Receive parameter information from at least one sensor in the vehicle equipment for each sensor.

[0093] It should be noted that the sensor parameter information has been described in detail in S101 above. Therefore, this application will not repeat it here.

[0094] S202. Extract features from the parameter information of each of at least one sensor to obtain multiple feature information.

[0095] Specifically, sampling rate feature information can be extracted from the sampling rate, and hardware type feature information can be extracted from the sensor type. The hardware type feature information characterizes the type of hardware used by the sensor. For example, the hardware type feature information can be a 3-axis sensor, a 6-axis sensor, or a 9-axis sensor. If at least one sensor includes a positioning sensor, then embodiments of this application extract positioning type feature information from the parameters of the positioning sensor. The positioning type feature information characterizes the positioning type of the positioning sensor. The positioning type can be GPS, BeiDou, or multi-system fusion, etc.

[0096] S203. Select the target preset driving model from multiple preset driving models based on multiple feature information.

[0097] It should be noted that multiple preset driving models are built into the cloud platform, and these preset driving models were trained by the applicant based on different scenarios and different sensor parameter data.

[0098] In one embodiment, each of the plurality of preset driving models corresponds to a plurality of label information, and the plurality of label information corresponds one-to-one with the plurality of feature information. S203 specifically includes the following steps:

[0099] S2031. Select the preset driving model that meets the first condition from multiple preset driving models, and use the preset driving model as the target preset driving model.

[0100] The first condition includes matching each label information and the feature information corresponding to the label information among multiple label information corresponding to the preset driving model.

[0101] Specifically, this application embodiment uses a feature similarity calculation method to calculate the similarity between label information and feature information. When the similarity between label information A and feature information a exceeds a preset similarity threshold, it is determined that label information A and feature information a match. Furthermore, the feature similarity calculation method includes, but is not limited to, the Euclidean distance calculation method.

[0102] It should be noted that in this embodiment, at least one tag information is set for each preset driving model. For example, the tag information for preset driving model A includes: low-frequency IMU, 3-axis IMU, and low-frequency GPS. Here, IMU stands for Inertial Measurement Unit.

[0103] In this embodiment, if the multiple feature information includes: IMU low frequency, IMU 3-axis, and low frequency GPS, then the multiple tag information corresponding to the preset driving model A matches the multiple feature information. In this case, the preset driving model A is the target preset driving model.

[0104] S204. Return a download link for the target preset driving model to the vehicle equipment, so that the vehicle equipment can download the target preset driving model based on the download link.

[0105] It should be noted that, in this embodiment of the application, the cloud platform generates a separate download link for each preset driving model, so that the vehicle equipment can download the corresponding preset driving model through the download link.

[0106] In one embodiment, the cloud platform periodically receives multiple noise distributions and extreme value statistics reported by vehicle devices. Then, features are extracted from the parameter information of each sensor (at least one sensor), the multiple noise distributions, and the extreme value statistics to obtain multiple feature information. Finally, a target preset driving model is selected from multiple preset driving models based on the multiple feature information.

[0107] In another embodiment, the cloud platform periodically receives multiple noise distributions and extreme value statistics reported from edge nodes. Then, features are extracted from the parameter information of each sensor, the multiple noise distributions, and the extreme value statistics of at least one sensor to obtain multiple feature information. Finally, a target preset driving model is selected from multiple preset driving models based on the multiple feature information.

[0108] It should be noted that the multiple labels corresponding to each preset driving model include noise distribution labels and extreme value statistics labels. The noise distribution labels are used to match the noise distribution, and the extreme value statistics labels are used to match the extreme value statistics. For example, the labels for preset driving model A include: noise distribution labels of 0.1-0.5, and extreme value statistics labels of 0-100. If the multiple feature information includes: noise distribution values ​​of 0.1-0.2, and extreme value statistics values ​​of 2-10, then each label in preset driving model A matches the feature information corresponding to that label. Therefore, preset driving model A is the target preset driving model.

[0109] It should be noted that, in this embodiment of the application, by accessing edge nodes, the computing power pressure on vehicle equipment can be effectively alleviated on the one hand, and the storage space of vehicle equipment can be reduced on the other hand.

[0110] In one embodiment, the method further includes: S205-S207.

[0111] S205, Receive historical driving data reported from vehicle equipment.

[0112] It should be noted that the historical driving data has already been described in detail in S106 above. Therefore, this application will not repeat it here.

[0113] S206. Determine the correction values ​​of the target parameters in the target preset driving model based on historical driving data.

[0114] In one embodiment, the historical driving data includes vehicle status data, driving behavior data, and environmental perception data, and S206 further includes: S2061-S2064.

[0115] S2061. Extract acceleration response features, bump features, and roll features from vehicle status data.

[0116] The vehicle status data includes, but is not limited to, the acceleration of the vehicle equipment, the acceleration time of the vehicle equipment, and the time when the roll reaches its peak. The acceleration of the vehicle equipment includes horizontal acceleration and vertical acceleration. Vertical acceleration includes the left-side vertical acceleration and the right-side vertical acceleration of the vehicle equipment. Embodiments of this application may use the vehicle equipment's acceleration time constant τ. accel Characterizing acceleration response features. Embodiments of this application may use the standard deviation of vertical acceleration. Characterizing bumpy ride features. Embodiments of this application can use the maximum lateral roll angle. and the rate of change of roll angle Characterizes tilting features.

[0117] In one embodiment, this application uses two inertial measurement units to estimate the roll characteristics in real time. The formula for calculating the maximum roll angle is as follows:

[0118]

[0119] in, The vertical acceleration on the left side of the vehicle equipment. Let g be the vertical acceleration on the right side of the vehicle equipment, and g be the acceleration due to gravity. The unit is degrees.

[0120] The formula for calculating the rate of change of roll angle is:

[0121]

[0122] in, The time it takes for the roll to reach its peak. The unit is degrees per second.

[0123] In another embodiment, this application uses a single inertial measurement unit and combines it with the quaternion method to estimate the roll characteristics in real time. The formula for calculating the real-time roll angle of the vehicle equipment is as follows:

[0124]

[0125] Where, q oq1, q3, and q4 are attitude quaternions. In this application, acceleration data of the vehicle equipment is collected in real time based on IMU3 (a...). x ,a y ,a z The attitude quaternion q = (q0, q1, q2, q3) is updated using quaternion differential equations. In the initial state of the vehicle equipment, q corresponds to the value (1, 0, 0, 0), which represents the horizontal attitude and eliminates the Euler angle gimbal lock problem. The formula for calculating the attitude quaternion q is as follows:

[0126]

[0127] Where g is the acceleration due to gravity. Let q0 be the derivative of q0 with respect to time, representing the real-time rate of change of q0. Let q1 be the derivative of q1 with respect to time, representing the real-time rate of change of q1. Let q2 be the derivative of q2 with respect to time, representing the real-time rate of change of q2. The derivative of q3 with respect to time represents the real-time rate of change of q3. In this embodiment, a first-order low-pass filter is used to smooth the acceleration data to reduce noise interference. The lateral roll angle of the vehicle equipment can be calculated based on the updated quaternion. It should be noted that the sampling period for the vehicle equipment acceleration data does not exceed 10ms.

[0128] The formula for calculating the real-time roll angle change rate of vehicle equipment is as follows:

[0129]

[0130] Where T is the acceleration data sampling period of the vehicle equipment, and θ y (t) represents the roll angle of the vehicle equipment at the current moment, θ y (tT) represents the tilt angle of the vehicle equipment in the previous sampling period at the current moment.

[0131] It should be noted that in this embodiment, θ is selected within a continuous 500ms sliding window. y The maximum value is used as the maximum lateral roll angle of the vehicle equipment.

[0132] S2062. Extract driving behavior features from driving behavior data.

[0133] It should be noted that S106 above has already described driving behavior data in detail. Therefore, this application will not repeat it here. Driving behavior characteristics include, but are not limited to, frequency of rapid acceleration, frequency of rapid deceleration, frequency of sharp turns, average intensity of rapid acceleration, average intensity of rapid deceleration, average intensity of sharp turns, and average intensity of collisions. Specifically, the frequency of rapid acceleration is the number of times the vehicle's equipment triggers a rapid acceleration event per 100 kilometers. The frequency of rapid deceleration is the number of times the vehicle's equipment triggers a rapid deceleration event per 100 kilometers. The frequency of sharp turns is the number of times the vehicle's equipment triggers a sharp turn per 100 kilometers. The average intensity of rapid acceleration refers to the average of the extreme values ​​of acceleration in its direction of travel when each rapid acceleration event is triggered within each 100 kilometers. The average intensity of rapid deceleration refers to the average of the extreme values ​​of acceleration in its direction of travel when each rapid deceleration event is triggered within each 100 kilometers. The average intensity of sharp turns refers to the average of the extreme values ​​of acceleration in its direction of travel when each sharp turn event is triggered within each 100 kilometers. The average intensity of collisions refers to the average value of the values ​​corresponding to the extreme values ​​of acceleration in its direction of collision when each collision event is triggered within each 100 kilometers. In this embodiment, when a collision event is triggered on the vehicle equipment and the extreme value of acceleration in the collision direction is greater than 5g, the corresponding value is 0. When a collision event is triggered on the vehicle equipment and the extreme value of acceleration in the collision direction is greater than 3g and less than or equal to 5g, the corresponding value is 0.5. When a collision event is triggered on the vehicle equipment and the extreme value of acceleration in the collision direction is less than or equal to 3g, the corresponding value is 1. For example, if the vehicle equipment triggers 3 collision events within 100 kilometers, and the acceleration in the collision direction is 2g, 4g, and 6g respectively at the time of each collision event, then the average collision intensity of the vehicle equipment is (1+0.5+0) / 3=0.5.

[0134] S2063. Extract road condition features from environmental perception data.

[0135] It should be noted that environmental perception data has been described in detail in S106 above. Therefore, this application will not repeat it here.

[0136] This application embodiment employs a "mileage-weighted" method to calculate the real-time comprehensive traffic condition value of the vehicle equipment. The real-time comprehensive traffic condition value of the vehicle equipment is used to characterize traffic conditions. The calculation formula for the real-time comprehensive traffic condition value of the vehicle equipment is as follows:

[0137]

[0138] Where, r i s represents the base weight for road condition type. iThis represents the percentage of mileage covered by this road condition within the route traveled by the vehicle / equipment. In the formula for calculating the real-time comprehensive road condition value, the value 5 represents the five different road condition types described in S106 above. It should be noted that in this embodiment, a basic weight for each road condition type is assigned based on practical experience. Specifically, the more unstable the road represented by a road condition type, the greater its corresponding basic weight value.

[0139] S2064. Determine the correction values ​​for the target parameters based on acceleration response characteristics, bump characteristics, roll characteristics, driving behavior characteristics, and road condition characteristics.

[0140] In one embodiment, S2064 specifically includes the following steps: S20641-S20642.

[0141] S20641. Determine the vehicle type characteristics based on acceleration response characteristics, bump characteristics, and roll characteristics.

[0142] It should be noted that in this embodiment, vehicle characteristic values ​​are used to characterize the vehicle type features of the equipment. Specifically, in this embodiment, vehicle characteristic values ​​are calculated based on acceleration response characteristics, bump characteristics, and roll characteristics. The formula for the vehicle characteristic values ​​is:

[0143] m = [w1·S] accel +w2·S damp +w3·S roll ]×(m max -m min )+m min

[0144] Where w1 is the acceleration response feature weight of the vehicle equipment, S accel w1 represents the acceleration response characteristic value of the vehicle equipment, w2 represents the bump characteristic weight of the vehicle equipment, and S represents the acceleration response characteristic value of the vehicle equipment. damp w3 is the bump characteristic value of the vehicle equipment, w3 is the roll characteristic weight of the vehicle equipment, and S is the bump characteristic value of the vehicle equipment. roll Let m be the roll characteristic value of the vehicle equipment. max m is the maximum vehicle characteristic value. min This represents the minimum vehicle model characteristic value.

[0145] S accel The calculation formula is:

[0146]

[0147] Where, τ accel_ref τ is a reference value for the acceleration time constant of equipment in large heavy-duty trucks. accel_min τ is the reference value for the acceleration time constant of a typical car. accel τ represents the acceleration time constant of the current vehicle equipment. accel_refand τ accel_min This setting was based on the applicant's practical experience. Preferably, τ accel_ref The value is 2.5s, τ accel_min The value is 1.0s.

[0148] S damp The calculation formula is:

[0149]

[0150] in, The standard deviation of the vertical acceleration of the current vehicle equipment. This is a reference value for the standard deviation of vertical acceleration of large heavy-duty truck equipment. This is a reference value for the standard deviation of vertical acceleration of ordinary passenger car equipment. and This design was based on the applicant's practical experience. Therefore, this application makes no restrictions.

[0151] S roll The calculation formula is:

[0152]

[0153] in, This is a reference value for the maximum lateral roll angle of large heavy-duty truck equipment. This is a reference value for the maximum lateral roll angle of ordinary passenger car equipment. This is a reference value for the roll angle variation rate of large heavy-duty truck equipment. This is a reference value for the roll angle change rate of ordinary passenger car equipment. This represents the maximum lateral roll angle of the current vehicle equipment. This represents the rate of change of the roll angle of the current vehicle equipment. It should be noted that... and This design was based on the applicant's practical experience. Therefore, this application makes no restrictions.

[0154] It should be noted that acceleration response characteristics are a core indicator for vehicle type identification. For example, heavy-duty trucks have a slow acceleration response and a large acceleration time constant, resulting in higher corresponding vehicle type feature values. Weighting the bump characteristics of vehicle equipment can reduce the impact of road condition interference on vehicle type identification.

[0155] It should be noted that the maximum and minimum vehicle characteristic values ​​were set by the applicant based on practical experience. Preferably, the maximum vehicle characteristic value is 5.0 and the minimum vehicle characteristic value is 1.0.

[0156] S20642. Determine the correction value of the target parameter based on vehicle type characteristics, driving behavior characteristics, and road condition characteristics.

[0157] In one embodiment, the driving behavior characteristics include: rapid deceleration frequency and rapid deceleration average intensity, and the target parameter includes: rapid deceleration threshold. S20642 above also specifically includes the following step: A.

[0158] A. Determine the correction value for the rapid deceleration threshold based on vehicle characteristics, frequency of rapid deceleration, average intensity of rapid deceleration, and road condition characteristics.

[0159] It should be noted that the rapid deceleration threshold can resolve the false alarm problem caused by bumpy road conditions. In this embodiment, the rapid deceleration threshold is dynamically adjusted by measuring the current road conditions of the vehicle and equipment to avoid false alarms caused by bumpy road conditions.

[0160] The formula for calculating the correction value of the rapid deceleration threshold is as follows:

[0161]

[0162] Where k1 is the vehicle model characteristic coefficient, m is the vehicle model characteristic value, and m ref Here, k2 is the reference vehicle model characteristic value corresponding to the vehicle model, and f is the rapid deceleration frequency coefficient. brake For rapid deceleration frequency, F high_brake Here, l represents the upper limit of the ideal frequency for rapid deceleration, l is the real-time comprehensive value of road conditions for the vehicle and equipment, k3 is the average intensity coefficient of rapid deceleration, and A is the upper limit of the ideal frequency for rapid deceleration. brake For the average intensity of rapid deceleration, A ref_brake The average intensity of ideal rapid deceleration.

[0163] It should be noted that k1, k2, k3, and m ref F highbrake and A refbrake The embodiments of this application are set based on practical experience. This is for road condition compensation. Among them, The larger the value, the bumpier the road conditions, and the higher the threshold for rapid deceleration should be.

[0164] It should be noted that the embodiments of this application set different reference vehicle feature values ​​for each vehicle type. Specifically, please refer to Table 1 below, which is a reference table for vehicle feature values ​​for different vehicle types.

[0165] Table 1

[0166]

[0167] As shown in Table 1, the reference vehicle characteristic value for a regular sedan is 1.4. The reference vehicle characteristic value for an SUV is 2.3.

[0168] In another embodiment, the driving behavior characteristics further include: collision frequency and average collision intensity, and the target parameter further includes: collision threshold. S20642 above also specifically includes the following step: B.

[0169] B. Determine the correction value for the collision threshold based on vehicle characteristics, collision frequency, and road condition characteristics.

[0170] The formula for calculating the correction value of the collision threshold is as follows:

[0171]

[0172] Where k4 is the collision frequency coefficient, f impact Let F be the collision frequency. high_impact Here, l represents the upper limit of the ideal collision frequency, l is the real-time comprehensive value of road conditions for the vehicle and equipment, and k5 is the average collision intensity coefficient. denoted as the average collision intensity, and m as the vehicle model characteristic value.

[0173] It should be noted that, The "false alarm risk classification rule" represents the collision intensity of the vehicle and equipment. The greater the collision intensity, the lower the false alarm risk. The calculation process for the average collision intensity can be found in the example in S2062. It will not be elaborated upon here.

[0174] S207. Return the correction value of the target parameter to the vehicle equipment so that the vehicle equipment can adjust the target parameter of the target preset driving model based on the correction value of the target parameter.

[0175] It should be noted that S107-S108 above have already described in detail how the vehicle equipment adjusts the target parameters of the target preset driving model based on the correction values ​​of the target parameters. This application will not repeat those details here.

[0176] It should be noted that the range of rapid deceleration threshold and collision threshold varies depending on the type of vehicle or equipment. Please refer to Table 2 for details; Table 2 shows the correspondence between vehicle type and the range of rapid deceleration threshold and collision threshold.

[0177] Table 2

[0178]

[0179]

[0180] As shown in Table 2, the rapid deceleration threshold for ordinary cars ranges from -1.0g to -0.3g, and the collision threshold ranges from 1.8g to 3.8g.

[0181] See Figure 3 , Figure 3This is a schematic block diagram of a cloud-based adaptive update device for a driving behavior model, provided in an embodiment of this application. Corresponding to the above-described cloud-based adaptive update method for a driving behavior model, this application also provides a cloud-based adaptive update device for a driving behavior model. This cloud-based adaptive update device includes a unit for executing the aforementioned cloud-based adaptive update method for a driving behavior model, and is applied to vehicle equipment. The vehicle equipment includes at least one sensor. Specifically, the cloud-based adaptive update device for a driving behavior model includes:

[0182] Acquisition unit 301 is used to acquire parameter information of each of the at least one sensor;

[0183] The sending unit 302 is used to send parameter information of each of the at least one sensor to the cloud platform;

[0184] The first receiving unit 303 is used to receive a download link for a target preset driving model returned by the cloud platform. The target preset driving model is determined by the cloud platform based on the parameter information of each of the at least one sensor.

[0185] Download unit 304 is used to download the target preset driving model based on the download link;

[0186] The update unit 305 is used to update the local driving model of the vehicle equipment according to the target preset driving model.

[0187] In one embodiment, the device further includes:

[0188] The reporting unit 306 is used to report historical driving data to the cloud platform so that the cloud platform can determine the correction value of the target parameter in the target preset driving model based on the historical driving data.

[0189] The first receiving unit 303 is further configured to receive the correction value of the target parameter returned from the cloud platform;

[0190] The adjustment unit 307 is used to adjust the target parameters of the target preset driving model based on the correction value of the target parameters.

[0191] In one embodiment, the historical driving data includes: vehicle status data, driving behavior data, and environmental perception data, and the target parameters include: abrupt deceleration threshold and a collision threshold.

[0192] See Figure 4 , Figure 4This is a schematic block diagram of a cloud-based adaptive update device for a driving behavior model, provided in an embodiment of this application. Corresponding to the above-described cloud-based adaptive update method for a driving behavior model, this application also provides a cloud-based adaptive update device for a driving behavior model. This cloud-based adaptive update device includes a unit for executing the above-described cloud-based adaptive update method for a driving behavior model, and can be configured in a cloud platform such as a desktop computer, tablet computer, or laptop computer. Specifically, the cloud-based adaptive update device for a driving behavior model includes:

[0193] The second receiving unit 401 is used to receive parameter information from at least one sensor in the vehicle equipment for each sensor.

[0194] Extraction unit 402 is used to extract features from the parameter information of each of the at least one sensor to obtain multiple feature information;

[0195] The filtering unit 403 is used to filter a target preset driving model from multiple preset driving models based on the multiple feature information;

[0196] The return unit 404 is used to return a download link for the target preset driving model to the vehicle device, so that the vehicle device can download the target preset driving model based on the download link.

[0197] In one embodiment, each of the plurality of feature information includes a first feature value; each of the plurality of preset driving models corresponds to a plurality of second feature values, and the plurality of second feature values ​​correspond one-to-one with the plurality of first feature values; the filtering unit 403 is specifically used for:

[0198] Select a preset driving model that meets the first condition from the plurality of preset driving models, and use the preset driving model as the target preset driving model. The first condition includes each of the plurality of second feature values ​​corresponding to the preset driving model matching the first feature value corresponding to the second feature value.

[0199] In one embodiment, the second receiving unit 401 is further configured to receive historical driving data reported from the vehicle equipment;

[0200] The device further includes:

[0201] The determining unit 405 is used to determine the correction value of the target parameter in the target preset driving model based on the historical driving data;

[0202] The return unit 404 is also used to return the correction value of the target parameter to the vehicle equipment, so that the vehicle equipment can adjust the target parameter of the target preset driving model based on the correction value of the target parameter.

[0203] In one embodiment, the historical driving data includes vehicle status data, driving behavior data, and environmental perception data, and the determining unit 405 is specifically used for:

[0204] Acceleration response features, bump features, and roll features are extracted from the vehicle status data.

[0205] Extract driving behavior features from the driving behavior data;

[0206] Extract road condition features from the environmental perception data;

[0207] The correction value of the target parameter is determined based on the acceleration response characteristics, the bump characteristics, the roll characteristics, the driving behavior characteristics, and the road condition characteristics.

[0208] In one embodiment, the determining unit 405 is further specifically used for:

[0209] The vehicle type characteristics of the equipment are determined based on the acceleration response characteristics, the bump characteristics, and the roll characteristics;

[0210] The correction value of the target parameter is determined based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics.

[0211] In one embodiment, the driving behavior characteristics include: frequency of rapid deceleration and average intensity of rapid deceleration, the target parameter includes: rapid deceleration threshold, and the determining unit 405 is further specifically used for:

[0212] The correction value for the rapid deceleration threshold is determined based on the vehicle characteristics, the frequency of rapid deceleration, the average intensity of rapid deceleration, and the road condition characteristics.

[0213] In one embodiment, the driving behavior feature further includes: collision frequency; the target parameter further includes: collision threshold; and the determining unit 405 is further specifically used for:

[0214] The correction value for the collision threshold is determined based on the vehicle characteristics, the collision frequency, and the road condition characteristics.

[0215] like Figure 5 As shown, this application provides a computer device including a processor 51, a communication interface 52, a memory 53, and a communication bus 54. The processor 51, the communication interface 52, and the memory 53 communicate with each other through the communication bus 54. The memory 53 is used to store computer programs.

[0216] In one embodiment of this application, when the processor 51 executes the program stored in the memory 53, it implements the control method for adaptive updating of the driving behavior model based on the cloud platform provided in any of the foregoing method embodiments.

[0217] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0218] Therefore, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps of the cloud-based adaptive update method for driving behavior models as provided in any of the foregoing method embodiments.

[0219] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing program code. The computer-readable storage medium can be non-volatile or volatile.

[0220] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0221] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0222] The steps in the methods of this application embodiment can be adjusted, merged, or deleted according to actual needs. The units in the apparatus of this application embodiment can be merged, divided, or deleted according to actual needs. 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.

[0223] 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 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, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0224] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0225] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Since these modifications and variations fall within the scope of the claims and their equivalents, this application also intends to include these modifications and variations.

[0226] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An adaptive update method for a driving behavior model based on a cloud platform, characterized in that, The method is applied to vehicle equipment, the vehicle equipment including at least one sensor, and the method includes: Obtain parameter information for each of the at least one sensor; Send parameter information of each of the at least one sensor to the cloud platform; Receive a download link for a target preset driving model returned by the cloud platform, wherein the target preset driving model is determined by the cloud platform based on parameter information of each of the at least one sensor; Download the target preset driving model using the download link; The local driving model of the vehicle equipment is updated according to the target preset driving model.

2. The method according to claim 1, characterized in that, The method further includes: Historical driving data is reported to the cloud platform so that the cloud platform can determine the correction value of the target parameter in the target preset driving model based on the historical driving data; Receive the corrected value of the target parameter returned from the cloud platform; The target parameters of the target preset driving model are adjusted based on the correction values ​​of the target parameters.

3. The method according to claim 2, characterized in that, The historical driving data includes vehicle status data, driving behavior data, and environmental perception data, and the target parameters include: rapid deceleration threshold and collision threshold.

4. An adaptive update method for a driving behavior model based on a cloud platform, characterized in that, The method is applied to a cloud platform, and the method includes: Receive parameter information from each of at least one sensor in the vehicle equipment; Features are extracted from the parameter information of each of the at least one sensor to obtain multiple feature information; Based on the aforementioned multiple feature information, a target preset driving model is selected from multiple preset driving models; Return a download link for the target preset driving model to the vehicle device, so that the vehicle device can download the target preset driving model based on the download link.

5. The method according to claim 4, characterized in that, Each of the plurality of preset driving models corresponds to a plurality of label information, and the plurality of label information corresponds one-to-one with the plurality of feature information. The step of selecting a target preset driving model from the plurality of preset driving models based on the plurality of feature information includes: Select a preset driving model that meets the first condition from the plurality of preset driving models, and use the preset driving model as the target preset driving model. The first condition includes matching each label information in the plurality of label information corresponding to the preset driving model with the feature information corresponding to the label information.

6. The method according to claim 4 or 5, characterized in that, The method further includes: Receive historical driving data reported from the vehicle equipment; Based on the historical driving data, the correction values ​​of the target parameters in the target preset driving model are determined; The vehicle equipment returns a correction value for the target parameter to the vehicle equipment so that the vehicle equipment adjusts the target parameters of the target preset driving model based on the correction value of the target parameter.

7. The method according to claim 6, characterized in that, The historical driving data includes vehicle status data, driving behavior data, and environmental perception data. Determining the correction values ​​of the target parameters in the target preset driving model based on the historical driving data includes: Acceleration response features, bump features, and roll features are extracted from the vehicle status data. Extract driving behavior features from the driving behavior data; Extract road condition features from the environmental perception data; The correction value of the target parameter is determined based on the acceleration response characteristics, the bump characteristics, the roll characteristics, the driving behavior characteristics, and the road condition characteristics.

8. The method according to claim 7, characterized in that, The step of determining the correction value of the target parameter based on the acceleration response characteristics, the bump characteristics, the roll characteristics, the driving behavior characteristics, and the road condition characteristics includes: The vehicle type characteristics of the equipment are determined based on the acceleration response characteristics, the bump characteristics, and the roll characteristics; The correction value of the target parameter is determined based on the vehicle type characteristics, driving behavior characteristics, and road condition characteristics.

9. The method according to claim 8, characterized in that, The driving behavior characteristics include: frequency of rapid deceleration and average intensity of rapid deceleration; the target parameter includes: rapid deceleration threshold; and determining the correction value of the target parameter based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics includes: The correction value for the rapid deceleration threshold is determined based on the vehicle characteristics, the frequency of rapid deceleration, the average intensity of rapid deceleration, and the road condition characteristics.

10. The method according to claim 9, characterized in that, The driving behavior characteristics further include: collision frequency and average collision intensity; the target parameter further includes: a collision threshold; and determining the correction value of the target parameter based on the vehicle type characteristics, the driving behavior characteristics, and the road condition characteristics includes: The correction value for the collision threshold is determined based on the vehicle characteristics, the collision frequency, the average collision intensity, and the road condition characteristics.

Images