Advanced driver assistance systems and advanced driver assistance methods

Abstract

This application relates to an advanced driver assistance system, comprising: a first acquisition module for acquiring and outputting signals of the driving environment of the vehicle; a second acquisition module activated when the driver operates the vehicle to acquire and output signals of the driver's driving behavior; a training data generation module for acquiring the output driving environment and the output driving behavior of the driver, and generating training data; an information processing module for training and storing a driving style model for the driver, the driving style model describing the correspondence between the driving environment and the driving behavior of a single driver, wherein the information processing center trains and / or updates the driving style model upon obtaining the training data; and a decision module for downloading the driving style model corresponding to the driver of the current vehicle from the information processing center, and outputting driving behavior corresponding to the driving environment based on the driving environment output by the first acquisition module and the driving style model.

Description

Technical Field

[0001] This application relates to the field of autonomous driving of vehicles, and more specifically, to an advanced driver assistance system and an advanced driver assistance method. Background Technology

[0002] The current trend in the automotive industry is the continuous development and improvement of increasingly advanced driver assistance systems (ADAS). During this period, driving comfort and personalization are becoming increasingly important for high-end models, making driver assistance systems tailored to individual drivers a current research hotspot. Summary of the Invention

[0003] The purpose of this application is to realize or at least promote the technological hotspots proposed in the prior art.

[0004] According to a first aspect of this application, an advanced driver assistance system is proposed, comprising:

[0005] The first acquisition module acquires and outputs signals of the vehicle's driving environment.

[0006] The second acquisition module is activated when the driver operates the vehicle to acquire and output signals of the driver's driving behavior.

[0007] The training data generation module acquires the driving environment output by the first acquisition module and the driver's driving behavior output by the second acquisition module, and generates training data.

[0008] An information processing module trains and stores a driving style model for a driver, the driving style model describing the correspondence between the driving environment and the driving behavior of an individual driver, wherein the information processing module trains and / or updates the driving style model upon obtaining the training data; and

[0009] The decision module downloads a driving style model corresponding to the driver of the current vehicle from the information processing module, and outputs driving behavior corresponding to the driving environment based on the driving environment output by the first acquisition module through the driving style model.

[0010] According to a second aspect of this application, an advanced driver assistance method is proposed, comprising:

[0011] Acquire and output the driving environment of the vehicle;

[0012] The driver's driving behavior is acquired and output when the driver operates the vehicle;

[0013] The driving environment and the driver's driving behavior are acquired, and training data is generated.

[0014] Training and storing a driving style model for a driver, the driving style model describing the correspondence between the driving environment and the driving behavior of an individual driver, and training and / or updating the driving style model with the training data available; and

[0015] Download a driving style model corresponding to the driver of the current vehicle, and output driving behavior corresponding to the driving environment based on the driving style model.

[0016] According to a third aspect of this application, a vehicle is proposed that includes an advanced driver assistance system disclosed in any embodiment of this application.

[0017] According to a fourth aspect of this application, a computer-readable storage medium is provided, wherein a computer program is stored therein, wherein the computer program is capable of being executed by a processor of the steps of the advanced driver assistance method described in any embodiment of this application.

[0018] According to a fourth aspect of this application, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor is capable of performing the steps of the advanced driver assistance method described in any embodiment of this application. Attached Figure Description

[0019] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein:

[0020] Figure 1 An exemplary block diagram of an advanced driver assistance system disclosed in this application is shown; and

[0021] Figure 2 An exemplary flowchart of the steps of the advanced driver assistance method disclosed in this application is shown. Detailed Implementation

[0022] The advanced driver assistance system 10 disclosed in this application can "customize" the vehicle's autonomous driving behavior according to the driving style habitually adopted by the driver. (See reference...) Figure 1As can be seen, the advanced driver assistance system 10 disclosed in this application includes a first acquisition module 1, a second acquisition module 2, a training data generation module 3, an information processing module 4, and a decision-making module 5. Specifically, the advanced driver assistance system 10 uses a driving style model trained by the information processing module 4 that conforms to the current driver's driving style to determine the driving behavior that meets the driver's expectations for the current driving environment. Within the scope of this application, the driving style model is understood as being able to map a driver's driving behavior in different driving environments. In other words, the driving style model describes a one-to-one correspondence between the driving environment and a driver's driving behavior.

[0023] Within the scope of this application, it should be noted that "self-vehicle" refers to the vehicle currently under study, i.e., this vehicle.

[0024] Here, the current driving environment of the vehicle is obtained by the first acquisition module 1. Specifically, the first acquisition module 1 can acquire information such as environmental conditions, i.e., current weather, visibility, type of road surface, curvature, height, relative distance and relative speed between the vehicle and surrounding obstacles and targets, detected by the vehicle's sensors, such as radar, camera, laser, high-precision map and IMU, using sensor fusion technology. In this way, it can determine the driving environment information of the vehicle under what environmental conditions and on what road surface.

[0025] The second acquisition module 2 of the advanced driver assistance system 10 is only activated when the driver operates the vehicle. That is, when the vehicle is not being driven by the driver, the second acquisition module 2 does not acquire the vehicle's driving behavior. Within the scope of this application, driving behavior can be understood as including information such as the vehicle's acceleration and rate of change of acceleration. It should be noted that acceleration and rate of change of acceleration refer to vector acceleration and rate of change of acceleration, respectively; that is, they include not only acceleration along the longitudinal, lateral, and vertical axes of the vehicle's coordinate system, but also angular acceleration around the longitudinal, lateral, and vertical axes of the vehicle's coordinate system.

[0026] When the second acquisition module 2 is activated and can obtain the driver's driving behavior, the training data generation module 3 receives the vehicle's driving environment obtained by the first acquisition module 1 and the driver's driving behavior under the above driving environment obtained by the second acquisition module 2. Based on the above driving environment and driving behavior information, it generates training data and outputs the training data to train a driving style model. That is, the training data includes driving environment information and the driver's driving behavior information corresponding to different driving environments.

[0027] The training data is then transmitted to the information processing module 4. Within the scope of this application, the information processing module 4 needs to train, store, and update the driving style model using the training data; that is, it requires a significant amount of data processing and machine learning work. Therefore, the information processing module 4 can typically be configured as a cloud server and / or processor, enabling it to rapidly process large amounts of training data.

[0028] In some embodiments of this application, a portion of the training data, such as 70%, is used to input into a training model (a machine learning model for training a driving style model) to train a driving style model, while another portion, such as 30%, is used to validate and adjust the trained driving style model, thereby making the driving style model more reliable and adapting to the driver's driving style more safely and comfortably.

[0029] In some embodiments of this application, the information processing module 4 further includes a training data classification unit, which classifies the training data based on the characteristics of different driving environments. That is, for example, the training data classification unit can divide the training data into multiple categories based on different driving environment types, such as for curved driving environments and congested road driving environments. For example, the training data classification unit can classify the training data into a category for curved road driving environments based on the driving environment type of the curved road, i.e., the magnitude of variables related to curves in the driving environment. Optionally, within the scope of this application, for different driving environment types, different machine learning training models can be used to train a driving style model corresponding to that driving environment type using the training data classified above. For example, the information processing module 4 uses the driving environment and driving behavior in the training data for curved driving environments as input data for a machine learning training model, and for the training data of this curved category, a specific machine learning training model, such as a Long Short-Term Memory model for curved scenarios, can be used to train the driving style model. This approach, which uses appropriate machine learning training models to target different parameters of driving behavior, such as acceleration and vehicle speed, for different types of driving environments, can improve the accuracy and reliability of driving style models in simulating real-world driving behavior. For example, in a driving environment involving curves, the parameter of interest could be the continuity of speed in the time domain, thus allowing for the use of appropriate training models, such as long short-term memory models.

[0030] The decision module 5 of the advanced driver assistance system 10 according to this application can download a driving style model corresponding to the current driver of the vehicle from the information processing module 4, and output driving behavior corresponding to the driving environment based on the driving environment output by the first acquisition module 1 through the driving style model. Within the scope of this application, the decision module 5 is located in the vehicle controller, and the decision module 5 outputs the desired driving behavior based solely on the driving style model downloaded from the information processing module 4 and the current driving environment. That is, the decision module 5 itself does not train or update the driving style model, thus enabling the advanced driver assistance system 10 of this application to respond to different driving environments more quickly with driving behavior that conforms to the driver's driving style. Furthermore, the decision module 5 can be configured to send training data to the information processing module 4 for training and / or updating the driving style model.

[0031] In some embodiments of this application, it is possible to consider situations where the driving style model downloaded by the decision module 5 does not cover the driving environment type of the current driving environment; that is, the driver has not previously driven in the aforementioned driving environment type (the driver has not previously operated the vehicle in the driving environment type to which the current driving environment belongs); or where the training data for the current driving environment type is insufficient to form a driving style model for the driver in the current driving environment type. In such cases, the information processing module compares the stored driving style models with the driver's driving style models before the current driving environment and selects the driving style model that is closest to the previous driving style model as the recommended driving style model, which is then provided to the decision module 5 for download. The decision module 5 outputs the expected driving behavior in the current driving environment based on the recommended driving style model. In other words, the recommended driving style model selected by the information processing module 4 satisfies the following conditions: 1. It covers the driving environment type of the current driving environment; and 2. The driving style model is closest to the driver's driving style model before the current driving environment.

[0032] In some embodiments of this application, it can also be configured such that when the driver is operating the vehicle himself, the decision module 5 no longer outputs driving behavior corresponding to the driving environment; that is, the decision module 5 is placed in an "inactive" state. This ensures that the driver has the highest level of control over the vehicle.

[0033] Reference Figure 2 It can be seen that this application also discloses an advanced driver assistance method, including:

[0034] Acquire and output the driving environment signal S1 from the vehicle;

[0035] The driver's driving behavior signal S2 is acquired and output when the driver operates the vehicle;

[0036] The driving environment and the driver's driving behavior are obtained, and training data S3 is generated;

[0037] Training and storing a driving style model for a driver, the driving style model describing the correspondence between the driving environment and the driving behavior of an individual driver, wherein, in the case of obtaining the training data, the driving style model is trained and / or updated S4; and

[0038] Download the driving style model corresponding to the driver of the current vehicle, and output the driving behavior S5 corresponding to the driving environment based on the driving style model.

[0039] Within the scope of this application, a driving style model is understood as one that can map a driver's driving behavior in different driving environments. In other words, a driving style model describes a one-to-one correspondence between a driving environment and a driver's driving behavior.

[0040] Here, the current driving environment of the vehicle is derived using sensor fusion technology, such as radar, cameras, lasers, high-precision maps, and IMUs. This includes environmental conditions, such as current weather, visibility, road surface type, curvature, altitude, relative distance and speed between the vehicle and surrounding obstacles and targets. In other words, the driving environment indicates the environmental conditions and road surface conditions under which the vehicle is currently driving.

[0041] In step S2, where the driver's driving behavior is acquired and output only when the driver is operating the vehicle, it is understood that the vehicle's driving behavior signal is not acquired when the driver is not operating the vehicle. Within the scope of this application, driving behavior can be understood as including information related to the vehicle's speed, acceleration, and rate of change of acceleration. It should be noted that speed, acceleration, and rate of change of acceleration refer to vector speed, acceleration, and rate of change of acceleration, respectively. This includes not only the longitudinal, lateral, and vertical speeds and accelerations along the vehicle's coordinate system, but also the angular velocities and angular accelerations around the longitudinal, lateral, and vertical axes of the vehicle's coordinate system.

[0042] If the driver's driving behavior can be obtained in step S2, then in step S3, the obtained driving environment of the vehicle and the driver's driving behavior under the above driving environment are received. Training data is generated based on the above driving environment and driving behavior information, and the training data is output to train a driving style model. That is to say, the training data includes driving environment information and the driver's driving behavior information corresponding to different driving environments.

[0043] Within the scope of this application, step S4 requires training, storing, and updating the driving style model using the training data; that is, it requires a significant amount of data processing and machine learning work. Therefore, step S4 can typically be specified to be performed on a cloud server and / or processor, thereby ensuring rapid processing of large amounts of training data.

[0044] In some embodiments of this application, a portion of the training data, such as 70%, is used to input into a training model (a machine learning model for training a driving style model) to train a driving style model, while another portion, such as 30%, is used to validate and adjust the trained driving style model, thereby making the driving style model more reliable and adapting to the driver's driving style more safely and comfortably.

[0045] In some embodiments of this application, step S4 further includes classifying the training data based on the characteristics of different driving environments. That is, the training data can be divided into multiple categories based on different driving environment types, such as for curved driving environments or congested road driving environments. For example, the training data can be classified into a category for curved driving environments based on the characteristics of the curved road driving environment type, such as the magnitude of variables related to curves in the driving environment. Optionally, within the scope of this application, for different driving environment types classified above, a driving style model corresponding to that driving environment type can be trained using the corresponding training data and different machine learning training models. For example, the driving environment and driving behavior in the training data for the curved driving environment type can be used as input data for the machine learning training model, while a specific machine learning training model, such as a Long Short-Term Memory model for curved scenarios, can be used to train the driving style model for the training data of this curved category. This approach, which uses appropriate machine learning training models to target different parameters of driving behavior, such as acceleration and vehicle speed, for different types of driving environments, can improve the accuracy and reliability of driving style models in simulating real-world driving behavior. For example, in a driving environment involving curves, the parameter of interest could be the continuity of speed in the time domain, thus allowing for the use of appropriate training models, such as long short-term memory models.

[0046] In step S5, a driving style model corresponding to the current driver of the vehicle can be downloaded, and driving behavior corresponding to the driving environment is output based on the obtained driving environment through the driving style model. Within the scope of this application, step S5 can be implemented in the vehicle's controller, and the desired driving behavior is output based solely on the downloaded driving style model and the current driving environment. That is, step S5 itself does not train or update the driving style model, thus enabling the vehicle to respond more quickly to different driving environments with driving behavior that matches the driver's driving style.

[0047] In some embodiments of this application, if the driving style model for the driver does not cover the driving environment type of the current driving environment (i.e., the driver has no prior driving behavior for the current driving environment type, or the driver has not previously operated the vehicle in the current driving environment type), or if the training data for the current driving environment type is insufficient to form a driving style model for the driver in the current driving environment, step S4 compares the stored driving style models with the driver's previous driving style models in the current driving environment, selects the driving style model closest to the previous driving style model as a recommended driving style model for download, and outputs the expected driving behavior in the current driving environment based on the recommended driving style model. In other words, the selected recommended driving style model satisfies the following conditions: 1. It covers the driving environment type of the current driving environment; and 2. The driving style model is closest to the driver's previous driving style model in the current driving environment.

[0048] In some embodiments of this application, it can also be configured such that when the driver is operating the vehicle himself, step S5 no longer outputs driving behavior corresponding to the driving environment. This ensures that the driver has the highest level of control over the vehicle.

[0049] Furthermore, this application also relates to a computer-readable storage medium and / or a computer device, wherein the computer program includes steps of an advanced driver assistance method according to any embodiment of this application that can be executed by a processor.

Claims

1. An advanced driver assistance system (10), characterized by Comprising: a first acquisition module (1) that acquires and outputs signals of a driving environment of a vehicle; a second acquisition module (2) that is activated when a driver operates the vehicle to acquire and output signals of driving behaviors of the driver; a training data generation module (3) that acquires the driving environment output by the first acquisition module (1) and the driving behaviors of the driver output by the second acquisition module (2) and generates training data; an information processing module (4) that trains and stores a driving style model for the driver, the driving style model describing a correspondence between the driving environment and the driving behaviors of the individual driver, wherein the information processing module (4) trains and / or updates the driving style model upon obtaining the training data; and a decision module (5) that downloads the driving style model corresponding to the driver of the current vehicle from the information processing module (4) and outputs driving behaviors corresponding to the driving environment through the driving style model according to the driving environment output by the first acquisition module (1); wherein the information processing module (4) further comprises a training data classification unit that classifies the training data based on the characteristics of different driving environment types, thereby classifying the training data for different driving environment types; and wherein the information processing module (4) trains the driving style model corresponding to different driving environment types through different training models using the corresponding training data, and in the case that the driving style model does not cover the driving environment type to which the current driving environment belongs, the information processing module (4) selects a recommended driving style model from the stored driving style models that corresponds to the driving style model of the driver before the current driving environment, and transmits the recommended driving style model to the decision module (5).

2. The advanced driver assistance system (10) according to claim 1, characterized in that For the driving style model, the training data is used to train the driving style model, and then the trained driving style model is verified and / or adjusted through the training data.

3. The advanced driver assistance system (10) according to claim 1, characterized in that The decision module (5) does not output the driving behaviors corresponding to the driving environment when the driver operates the vehicle.

4. The advanced driver assistance system (10) according to claim 1, characterized in that The information processing module (4) is set in the cloud.

5. The advanced driver assistance system (10) according to claim 1, characterized in that The driving environment includes environmental conditions, types of driving surfaces, curvatures, heights, relative distances and relative speeds of the vehicle and surrounding obstacles.

6. The advanced driver assistance system (10) according to claim 1, characterized in that The driving behaviors include information related to the speed, acceleration, and acceleration change rate of the vehicle.

7. An advanced driver assistance method, characterized by, Comprising: acquiring and outputting a driving environment of a vehicle (S1); acquiring and outputting driving behaviors of a driver when the driver operates the vehicle (S2); acquiring the driving environment and the driving behaviors of the driver and generating training data (S3); training and storing a driving style model for a driver, the driving style model describing a correspondence between a driving environment and a driving behavior of an individual driver, the driving style model being trained and / or updated (S4) upon obtaining the training data; and downloading a driving style model corresponding to a driver of a current ego vehicle and outputting a driving behavior corresponding to the driving environment according to the driving style model and the driving environment (S5); wherein, in the step of training and storing a driving style model for a driver, further comprising: classifying the training data based on characteristics of different driving environment types, thereby classifying the training data for different driving environment types; and wherein, the driving style models respectively corresponding to different driving environment types are trained by different training models using corresponding training data for different driving environment types, and in the case that the driving style model does not cover a driving environment type to which a current driving environment belongs, in the step of training and storing a driving style model for a driver (S4), a recommended driving style model is selected from the stored driving style models, which correspondingly is closest to a driving style model of the driver before the current driving environment, and the recommended driving style model is provided for downloading.

8. The advanced driver assistance method according to claim 7, characterized by, For the driving style model, a part of the training data is used to train the driving style model, and another part of the training data is used to verify and / or adjust the trained driving style model.

9. The advanced driver assistance method according to claim 7, characterized by, The driving behavior corresponding to the driving environment is not output in the case that the driver is manipulating the ego vehicle.

10. The advanced driver assistance method according to claim 7, characterized by, The step of training and storing a driving style model for a driver (S4) is implemented in the cloud.

11. The advanced driver assistance method according to claim 7, characterized by, The driving environment includes environmental conditions, types of travel road surface, curvature, height, relative distance and relative speed of the ego vehicle and surrounding obstacles.

12. The advanced driver assistance method according to claim 7, characterized by, The driving behavior includes information related to speed, acceleration, and acceleration rate of change of the ego vehicle.

13. A vehicle comprising an advanced driver assistance system (10) according to any one of claims 1 to 6.

14. A computer readable storage medium having stored therein a computer program, wherein, The computer program is executable by a processor to perform the steps of the advanced driver assistance method according to any one of claims 7 to 12.

15. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein, The processor, when executing the program, implements the steps of the advanced driver assistance method according to any one of claims 7 to 12.

Images