Self-evolutionary learning autonomous driving assistance system and method based on edge scenarios

Abstract

The present invention discloses a self-evolutionary learning type automatic driving assistance system and method based on edge scenarios, which belongs to the field of automatic driving technology. The system includes a user identity recognition management module, a driving data collection and analysis module, a self-evolutionary intelligent learning module and a personalized intelligent control module; the user identity recognition management module is used to obtain user permissions, identify the user identity before the user drives, and retrieve the information database corresponding to the user; the driving data collection and analysis module is used to collect the vehicle control behavior data of the user during the vehicle driving process, and collect and determine the vehicle surrounding environment perception data in the same time series as the user's vehicle control behavior data; the self-evolutionary intelligent learning module is used to establish a neural network model for analyzing the user's vehicle control behavior data, and recommend personalized vehicle automatic driving paths for users; the personalized intelligent control module is used to adjust the driving path of the current vehicle.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and specifically to a self-evolving learning autonomous driving assistance system and method based on edge scenarios. Background Art

[0002] Automated driving assistance uses various sensors, advanced algorithms and artificial intelligence technologies to help drivers better control the vehicle and automatically perform driving tasks in some cases. It can reduce the driver's driving burden to a certain extent; but in some special scenarios, the driver still needs to understand the vehicle's driving status and be ready to take over the vehicle at any time; and due to the defects of artificial intelligence technology, it is difficult to handle some rare events with a low probability of occurrence; the existing autonomous driving technology cannot meet the user's personalized needs based on the user's driving habits, resulting in reduced user trust in the autonomous driving system. Summary of the invention

[0003] The purpose of the present invention is to provide a self-evolutionary learning type automatic driving assistance system and method based on edge scenarios to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solution: a self-evolutionary learning type automatic driving assistance method based on edge scenarios, the method comprising the following steps:

[0005] S10, obtaining user authority, identifying the user before the user drives, and retrieving the information database corresponding to the user according to the identified user identity;

[0006] S20, collecting vehicle control behavior data of the user during vehicle driving, and collecting and determining vehicle surrounding environment perception data in the same time series as the vehicle control behavior data of the user;

[0007] S30, establishing a neural network model for analyzing the user's vehicle control behavior data, and recommending a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data;

[0008] S40: Adjust the current vehicle's driving path according to the recommended personalized vehicle automatic driving path.

[0009] Furthermore, the information database is used to store historical vehicle control behavior data and historical vehicle surrounding environment perception data of the user; the identified user identity is the user in the main driving position of the vehicle, and when there is a new user driving the vehicle, an information database of the new user is established, and the collected and analyzed user vehicle control behavior data and vehicle surrounding environment perception data are stored as historical data in the information database corresponding to the new user;

[0010] By identifying user identities and analyzing historical data of different users, we can meet users' personalized needs and improve their car-using experience.

[0011] Furthermore, a time series management model is established to analyze and process the user's vehicle control behavior data and the vehicle's surrounding environment perception data under the same time series; based on the user's vehicle control behavior data, the user's acceleration control data for the vehicle is determined to obtain According to the vehicle's surrounding environment perception data, the vehicle's driving speed change data is determined to obtain Determine the straight-line distance between the vehicle and the nearest obstruction around it, and get L t ; Based on the moving direction and speed of the current vehicle and the surrounding obstructions, a simulation prediction is performed. The nearest obstruction should meet the following conditions: at the time t+1 of the simulation prediction, the straight-line distance L between the vehicle and the nearest obstruction is t+1 ≤L t ;

[0012] L t+1 ≤L t , so that the distance between the vehicle and the surrounding obstructions is getting closer and closer, so that the analyzed data is the behavior of the user adjusting the vehicle when the vehicle is about to collide with the surrounding obstructions, thereby improving the accuracy of data analysis;

[0013] The acceleration control data includes the user's control over the vehicle's driving acceleration and direction angle; Indicates the magnitude and direction of the vehicle's acceleration at time t; Indicates the speed and direction of the vehicle at time t; L t It indicates the straight-line distance between the vehicle and the nearest obstruction at time t; t indicates the timestamp.

[0014] Furthermore, the method steps of step S30 are as follows:

[0015] S301, based on historical vehicle surrounding environment perception data, determine the vehicle's conventional automatic driving path under different vehicle surrounding environment perception data, and simulate the vehicle's conventional automatic driving at a speed v at time t+1 t The shortest straight-line distance threshold between the vehicle and the nearest obstruction around it

[0016] S302, based on the user's historical vehicle control behavior data, taking the time when the user starts to adjust the vehicle's steering angle as the starting point and the time when the steering wheel is straightened as the ending point, taking the time period from the starting point to the ending point as a cycle, determining the number of times the user adjusts the vehicle's steering angle within the cycle, and determining the angle between the vehicle's steering angle adjusted by the user each time and the vehicle's steering angle during normal automatic driving, to obtain a set C; C = {C1, C2, ..., C n};

[0017] Among them, C i represents the user's behavior of adjusting the vehicle's direction angle in the i-th cycle based on the analysis of the user's historical vehicle control behavior data; Indicates the change in acceleration magnitude and direction during normal automatic driving caused by the jth adjustment of the vehicle by the user in the i-th cycle; i = 1, 2, ..., n; j = 1, 2, ..., k i ;

[0018] S303, according to set C, respectively calculate the speed and direction of the vehicle when it is traveling: At the next moment, under normal automatic driving and user control of the vehicle, the time when the vehicle collides with the nearest surrounding obstruction is calculated according to the formula:

[0019]

[0020] in, represents the time when the vehicle collides with the nearest obstruction in the surrounding area during normal automatic driving at time t+1; T t+1 Indicates the time when the vehicle collides with the nearest obstruction after the user adjusts the vehicle; Indicates the speed and direction of the nearest obstruction at time t; Indicates the magnitude and direction of the vehicle's acceleration at time t during normal automatic driving;

[0021] in, and The moving direction is the projection of the distance line between the two in the horizontal direction; the time when the vehicle collides with the nearest obstruction around is the collision time under the distance line when the acceleration of the two does not change in the future time period at a certain speed; the speed directions of the two are divided into the same speed direction on the straight line and the opposite speed direction;

[0022] S304, construct a neural network model, adopt RNN algorithm, take the speed and direction of the vehicle, the speed and direction of the nearest obstruction, the time when the vehicle collides with the nearest obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest obstruction as output, train the model, and calculate according to the formula:

[0023]

[0024] Among them, x t Represents the input vector at time t, which contains and L t ; f represents the activation function; W yy Indicates that The weight of xy For x t The weight of ; ε represents the bias term;

[0025] in, Indicates the change in acceleration magnitude and direction of the vehicle during normal automatic driving when the user adjusts it at time t;

[0026] According to set C, the model is trained and the model parameters are adjusted. The behavior data of users adjusting the vehicle direction angle in set C is used as The input terminal enables the model to simulate and predict the time when the vehicle collides with the nearest surrounding obstruction after the user adjusts the vehicle;

[0027] By building a neural network model, according to the user's adjustment data for the vehicle in set C, the network structure and hyperparameters are adjusted to optimize the performance, so as to train the user's change value of the time when the vehicle collides with the nearest obstruction during regular automatic driving, and according to the change value T after training t+1 Reverse inference is performed to determine the change in acceleration magnitude and direction that the user has applied to the vehicle during normal automatic driving, and the current vehicle driving path is adjusted based on the change.

[0028] Furthermore, based on the constructed neural network model, the moving speed and direction of the current vehicle and the nearest surrounding obstruction are determined, and the straight-line distance between the current vehicle and the nearest surrounding obstruction is determined. Based on the data stored in the user information database, the perceived time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle is determined, and based on the perceived time, the user's change in the magnitude and direction of the acceleration of the vehicle during normal automatic driving is inferred, and the driving path of the current vehicle is adjusted.

[0029] A self-evolving learning autonomous driving assistance system based on edge scenarios, which includes a user identity recognition management module, a driving data collection and analysis module, a self-evolving intelligent learning module, and a personalized intelligent control module;

[0030] The user identity recognition management module is used to obtain user permissions, identify the user identity before the user drives, and retrieve the information database corresponding to the user according to the identified user identity; the driving data collection and analysis module is used to collect the vehicle control behavior data of the user during the vehicle driving process, and collect and determine the vehicle surrounding environment perception data in the same time series as the user's vehicle control behavior data; the self-evolution intelligent learning module is used to establish a neural network model for analyzing the user's vehicle control behavior data, and recommend a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data; the personalized intelligent control module is used to adjust the driving path of the current vehicle according to the recommended personalized vehicle automatic driving path.

[0031] Furthermore, the user identity management module includes a user identity management unit and a database management unit;

[0032] The user identity management sheet is used to obtain user permissions and identify the user's identity before the user drives;

[0033] The database management unit is used to establish an information database for the new user when there is a new user driving a vehicle, and store the collected and analyzed user vehicle control behavior data and vehicle surrounding environment perception data as historical data in the information database corresponding to the new user; the information database is used to store the user's historical vehicle control behavior data and historical vehicle surrounding environment perception data.

[0034] Furthermore, the driving data collection and analysis module includes an environment perception data collection unit, a vehicle motion collection unit, a user control data collection unit and a time series management unit;

[0035] The environmental perception data acquisition unit includes a distance measuring sensor, which is used to perform simulation prediction based on the current moving direction and speed of the vehicle and the surrounding obstructions, and determine the straight-line distance between the vehicle and the nearest surrounding obstructions by edge computing;

[0036] The vehicle motion acquisition unit is used to determine the vehicle's driving speed change data based on the vehicle's surrounding environment perception data;

[0037] The user control data acquisition unit is used to determine the acceleration control data of the user on the vehicle according to the user's vehicle control behavior data;

[0038] The time series management unit is used to establish a time series management model to analyze and process the user's vehicle control behavior data and the vehicle surrounding environment perception data in the same time series.

[0039] Furthermore, the self-evolution intelligent learning module includes a simulation prediction unit, a data processing and analysis unit, a time calculation unit and a model training management unit;

[0040] The simulation prediction unit is used to determine the conventional automatic driving path of the vehicle under different vehicle surrounding environment perception data, and simulate the shortest threshold of the straight-line distance between the vehicle and the nearest surrounding obstruction at the next moment when the vehicle is in conventional automatic driving;

[0041] The data processing and analysis unit is used to determine the number of times the user adjusts the vehicle's direction angle within the period, taking the time when the user starts to adjust the vehicle's direction angle as the starting point, the time when the steering wheel is aligned with the vehicle's straight-line driving as the ending point, and taking the time period from the starting point to the ending point as a cycle, and determining the angle between the user's adjustment of the vehicle's direction angle each time and the vehicle's direction angle during normal automatic driving;

[0042] The time calculation unit is used to calculate the time when the vehicle collides with the nearest surrounding obstruction at the next moment under normal automatic driving and user control of the vehicle;

[0043] The model training management unit is used to construct a neural network model, adopts an RNN algorithm, takes the vehicle's speed and direction, the speed and direction of the nearest surrounding obstruction, and the time when the vehicle collides with the nearest surrounding obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest surrounding obstruction as output to train the model.

[0044] Furthermore, the personalized intelligent control module includes a user behavior learning unit and a path adjustment unit;

[0045] The user behavior learning unit is used to determine the moving speed and direction of the current vehicle and the nearest surrounding obstruction according to the constructed neural network model, and determine the straight-line distance between the current vehicle and the nearest surrounding obstruction, and determine the perceived time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle according to the stored data in the user information database, and reversely infer the user's change value of the acceleration magnitude and direction during normal automatic driving of the vehicle according to the perceived time;

[0046] The path adjustment unit is used to adjust the current vehicle's driving path according to the change value of the acceleration magnitude and direction during normal automatic driving of the vehicle set by the user.

[0047] Compared with the prior art, the beneficial effects achieved by the present invention are: providing a self-evolving learning type automatic driving assistance method and system based on edge scenarios, by establishing a self-evolving network learning model, analyzing the user's vehicle control behavior data during the vehicle's automatic driving process, thereby obtaining the user's driving habit data, and adjusting the vehicle's driving path according to the user's driving habits, thereby improving the user's driving comfort and helping to increase the user's trust in the automatic driving system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0049] Figure 1 It is a structural schematic diagram of the self-evolutionary learning type automatic driving assistance system based on edge scenarios of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] See also Figure 1 The present invention provides a technical solution: a self-evolutionary learning type automatic driving assistance method based on edge scenarios, the method comprising the following steps:

[0052] S10, obtaining user authority, identifying the user before the user drives, and retrieving the information database corresponding to the user according to the identified user identity;

[0053] S20, collecting vehicle control behavior data of the user during vehicle driving, and collecting and determining vehicle surrounding environment perception data in the same time series as the vehicle control behavior data of the user;

[0054] S30, establishing a neural network model for analyzing the user's vehicle control behavior data, and recommending a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data;

[0055] S40: Adjust the current vehicle's driving path according to the recommended personalized vehicle automatic driving path.

[0056] The information database is used to store the user's historical vehicle control behavior data and historical vehicle surrounding environment perception data; the identified user identity is the user in the main driving position of the vehicle. When there is a new user driving the vehicle, an information database for the new user is established, and the collected and analyzed user vehicle control behavior data and vehicle surrounding environment perception data are stored as historical data in the information database corresponding to the new user.

[0057] Establish a time series management model to analyze and process the user's vehicle control behavior data and vehicle surrounding environment perception data in the same time series; determine the user's acceleration control data for the vehicle based on the user's vehicle control behavior data, and obtain According to the vehicle's surrounding environment perception data, the vehicle's driving speed change data is determined to obtain Determine the straight-line distance between the vehicle and the nearest obstruction around it, and get L t ; Based on the moving direction and speed of the current vehicle and the surrounding obstructions, a simulation prediction is performed. The nearest obstruction should meet the following conditions: at the time t+1 of the simulation prediction, the straight-line distance L between the vehicle and the nearest obstruction is t+1 ≤L t ;

[0058] The acceleration control data includes the user's control over the vehicle's driving acceleration and direction angle; Indicates the magnitude and direction of the vehicle's acceleration at time t; Indicates the speed and direction of the vehicle at time t; L t It indicates the straight-line distance between the vehicle and the nearest obstruction at time t; t indicates the timestamp.

[0059] The method steps of step S30 are:

[0060] S301, based on historical vehicle surrounding environment perception data, determine the vehicle's conventional automatic driving path under different vehicle surrounding environment perception data, and simulate the vehicle's conventional automatic driving at a speed v at time t+1 t The shortest straight-line distance threshold between the vehicle and the nearest obstruction around it

[0061] S302, based on the user's historical vehicle control behavior data, taking the time when the user starts to adjust the vehicle's steering angle as the starting point and the time when the steering wheel is straightened as the ending point, taking the time period from the starting point to the ending point as a cycle, determining the number of times the user adjusts the vehicle's steering angle within the cycle, and determining the angle between the vehicle's steering angle adjusted by the user each time and the vehicle's steering angle during normal automatic driving, to obtain a set C; C = {C1, C2, ..., C n};

[0062] Among them, C i represents the user's behavior of adjusting the vehicle's direction angle in the i-th cycle based on the analysis of the user's historical vehicle control behavior data; Indicates the change in acceleration magnitude and direction during normal automatic driving caused by the jth adjustment of the vehicle by the user in the i-th cycle; i = 1, 2, ..., n; j = 1, 2, ..., k i ;

[0063] S303, according to set C, respectively calculate the speed and direction of the vehicle when it is traveling: At the next moment, under normal automatic driving and user control of the vehicle, the time when the vehicle collides with the nearest surrounding obstruction is calculated according to the formula:

[0064]

[0065] in, represents the time when the vehicle collides with the nearest obstruction in the surrounding area during normal automatic driving at time t+1; T t+1 Indicates the time when the vehicle collides with the nearest obstruction after the user adjusts the vehicle; Indicates the speed and direction of the nearest obstruction at time t; Indicates the magnitude and direction of the vehicle's acceleration at time t during normal automatic driving;

[0066] in, and The moving direction is the projection of the distance line between the two in the horizontal direction; the time when the vehicle collides with the nearest obstruction around is the collision time under the distance line when the acceleration of the two does not change in the future time period at a certain speed;

[0067] S304, construct a neural network model, adopt RNN algorithm, take the speed and direction of the vehicle, the speed and direction of the nearest obstruction, the time when the vehicle collides with the nearest obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest obstruction as output, train the model, and calculate according to the formula:

[0068]

[0069] Among them, x t Represents the input vector at time t, which contains and L t ; f represents the activation function; W yy Indicates that The weight of xy For x t The weight of ; ε represents the bias term;

[0070] in, Indicates the change in acceleration magnitude and direction of the vehicle during normal automatic driving when the user adjusts it at time t;

[0071] According to set C, the model is trained and the model parameters are adjusted. The behavior data of users adjusting the vehicle direction angle in set C is used as The input of the model enables the model to simulate and predict the time when the vehicle collides with the nearest surrounding obstruction after the user adjusts the vehicle.

[0072] Based on the constructed neural network model, the moving speed and direction of the current vehicle and the nearest surrounding obstruction are determined, and the straight-line distance between the current vehicle and the nearest surrounding obstruction is determined. Based on the data stored in the user information database, the perception time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle is determined, and based on the perception time, the user's change in the magnitude and direction of the acceleration of the vehicle during normal automatic driving is reversed, and the driving path of the current vehicle is adjusted.

[0073] A self-evolving learning autonomous driving assistance system based on edge scenarios, which includes a user identity recognition management module, a driving data collection and analysis module, a self-evolving intelligent learning module, and a personalized intelligent control module;

[0074] The user identity recognition management module is used to obtain user permissions, identify the user identity before the user drives, and retrieve the information database corresponding to the user according to the identified user identity; the driving data collection and analysis module is used to collect the vehicle control behavior data of the user during the vehicle driving process, and collect and determine the vehicle surrounding environment perception data in the same time series as the user's vehicle control behavior data; the self-evolution intelligent learning module is used to establish a neural network model for analyzing the user's vehicle control behavior data, and recommend a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data; the personalized intelligent control module is used to adjust the driving path of the current vehicle according to the recommended personalized vehicle automatic driving path.

[0075] The user identity management module includes a user identity management unit and a database management unit;

[0076] The user identity management sheet is used to obtain user permissions and identify the user's identity before the user drives;

[0077] The database management unit is used to establish an information database for the new user when there is a new user driving a vehicle, and store the collected and analyzed user vehicle control behavior data and vehicle surrounding environment perception data as historical data in the information database corresponding to the new user; the information database is used to store the user's historical vehicle control behavior data and historical vehicle surrounding environment perception data.

[0078] The driving data collection and analysis module includes an environment perception data collection unit, a vehicle motion collection unit, a user control data collection unit and a time series management unit;

[0079] The environmental perception data acquisition unit includes a distance measuring sensor, which is used to perform simulation prediction based on the current moving direction and speed of the vehicle and the surrounding obstructions, and determine the straight-line distance between the vehicle and the nearest surrounding obstructions by edge computing;

[0080] The vehicle motion acquisition unit is used to determine the vehicle's driving speed change data based on the vehicle's surrounding environment perception data;

[0081] The user control data acquisition unit is used to determine the acceleration control data of the user on the vehicle according to the user's vehicle control behavior data;

[0082] The time series management unit is used to establish a time series management model to analyze and process the user's vehicle control behavior data and the vehicle surrounding environment perception data in the same time series.

[0083] The self-evolutionary intelligent learning module includes a simulation prediction unit, a data processing and analysis unit, a time calculation unit and a model training management unit;

[0084] The simulation prediction unit is used to determine the conventional automatic driving path of the vehicle under different vehicle surrounding environment perception data, and simulate the shortest threshold of the straight-line distance between the vehicle and the nearest surrounding obstruction at the next moment when the vehicle is in conventional automatic driving;

[0085] The data processing and analysis unit is used to determine the number of times the user adjusts the vehicle's direction angle within the period, taking the time when the user starts to adjust the vehicle's direction angle as the starting point, the time when the steering wheel is aligned with the vehicle's straight-line driving as the ending point, and taking the time period from the starting point to the ending point as a cycle, and determining the angle between the user's adjustment of the vehicle's direction angle each time and the vehicle's direction angle during normal automatic driving;

[0086] The time calculation unit is used to calculate the time when the vehicle collides with the nearest surrounding obstruction at the next moment under normal automatic driving and user control of the vehicle;

[0087] The model training management unit is used to construct a neural network model, adopts an RNN algorithm, takes the vehicle's speed and direction, the speed and direction of the nearest surrounding obstruction, and the time when the vehicle collides with the nearest surrounding obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest surrounding obstruction as output to train the model.

[0088] The personalized intelligent control module includes a user behavior learning unit and a path adjustment unit;

[0089] The user behavior learning unit is used to determine the moving speed and direction of the current vehicle and the nearest surrounding obstruction according to the constructed neural network model, and determine the straight-line distance between the current vehicle and the nearest surrounding obstruction, and determine the perceived time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle according to the stored data in the user information database, and reversely infer the user's change value of the acceleration magnitude and direction during normal automatic driving of the vehicle according to the perceived time;

[0090] The path adjustment unit is used to adjust the current vehicle's driving path according to the change value of the acceleration magnitude and direction during normal automatic driving of the vehicle set by the user.

[0091] In this embodiment:

[0092] The system is specifically a self-evolutionary learning automatic driving assistance system for private cars. In this system, distance measuring sensors are installed around the vehicle body to simulate and predict the movement direction and speed of the current vehicle and surrounding obstructions, and determine the straight-line distance between the vehicle and the nearest obstruction through edge computing.

[0093] The user identity is identified as the user in the main driving position of the vehicle, and the information database corresponding to the user is retrieved according to the identified user identity; wherein, when there is a new user driving the vehicle, the information database of the new user is established, and the collected and analyzed user vehicle control behavior data and vehicle surrounding environment perception data are stored as historical data in the information database corresponding to the new user;

[0094] Establish a time series management model to analyze and process the user's vehicle control behavior data and vehicle surrounding environment perception data in the same time series; determine the user's acceleration control data for the vehicle based on the user's vehicle control behavior data, and obtain According to the vehicle's surrounding environment perception data, the vehicle's driving speed change data is determined to obtain Determine the straight-line distance between the vehicle and the nearest obstruction around it, and get L t ; Based on the moving direction and speed of the current vehicle and the surrounding obstructions, a simulation prediction is performed. The nearest obstruction should meet the following conditions: at the time t+1 of the simulation prediction, the straight-line distance L between the vehicle and the nearest obstruction is t+1 ≤L t ;

[0095] Based on the historical vehicle surrounding environment perception data, the conventional automatic driving path of the vehicle under different vehicle surrounding environment perception data is determined, and the vehicle is simulated at the speed v at time t+1 during conventional automatic driving. t The shortest straight-line distance threshold between the vehicle and the nearest obstruction around it

[0096] Based on the user's historical vehicle control behavior data, the time when the user starts to adjust the vehicle's direction angle is taken as the starting point, and the time when the steering wheel is straightened and the vehicle is driving in a straight line is taken as the ending point. The time period from the starting point to the ending point is taken as a cycle, and the number of times the user adjusts the vehicle's direction angle within the cycle is determined, and the angle between the direction angle adjusted by the user each time and the direction angle of the vehicle during normal automatic driving is determined to obtain a set C; C = {C1, C2, ..., C n};

[0097] According to set C, the speed and direction of the vehicle are calculated as follows: At the next moment, under normal automatic driving and user control of the vehicle, the time when the vehicle collides with the nearest surrounding obstruction is calculated according to the formula:

[0098]

[0099] Construct a neural network model, use the RNN algorithm, take the vehicle speed and direction, the speed and direction of the nearest obstruction, and the time when the vehicle collides with the nearest obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest obstruction as output, train the model, and use the calculation formula:

[0100]

[0101] According to set C, the model is trained and the model parameters are adjusted. The behavior data of users adjusting the vehicle direction angle in set C is used as The input terminal enables the model to simulate and predict the time when the vehicle collides with the nearest surrounding obstruction after the user adjusts the vehicle;

[0102] Based on the constructed neural network model, the moving speed and direction of the current vehicle and the nearest surrounding obstruction are determined, and the straight-line distance between the current vehicle and the nearest surrounding obstruction is determined. Based on the data stored in the user information database, the perception time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle is determined, and based on the perception time, the user's change in the magnitude and direction of the acceleration of the vehicle during normal automatic driving is reversed, and the driving path of the current vehicle is adjusted.

[0103] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0104] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A self-evolutionary learning autonomous driving assistance method based on edge scenarios, characterized by: The method comprises the following steps: S10, obtaining user authority, identifying the user before the user drives, and retrieving the information database corresponding to the user according to the identified user identity; S20, collecting vehicle control behavior data of the user during vehicle driving, and collecting and determining vehicle surrounding environment perception data in the same time series as the vehicle control behavior data of the user; S30, establishing a neural network model for analyzing the user's vehicle control behavior data, and recommending a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data; S40, adjusting the current vehicle driving path according to the recommended personalized vehicle automatic driving path; The information database is used to store the user's historical vehicle control behavior data and historical vehicle surrounding environment perception data; the identified user identity is the user in the main driving position of the vehicle. When there is a new user driving the vehicle, an information database for the new user is established, and the collected and analyzed user's vehicle control behavior data and vehicle surrounding environment perception data are stored as historical data in the information database corresponding to the new user; Establish a time series management model to analyze and process the user's vehicle control behavior data and vehicle surrounding environment perception data in the same time series; determine the user's acceleration control data for the vehicle based on the user's vehicle control behavior data, and obtain ; According to the vehicle's surrounding environment perception data, determine the vehicle's driving speed change data and obtain , determine the straight-line distance between the vehicle and the nearest obstruction around it, and get ; Based on the moving direction and speed of the current vehicle and the surrounding obstructions, a simulation prediction is performed. The nearest obstruction should meet the following conditions: at the time t+1 of the simulation prediction, the straight-line distance between the vehicle and the nearest obstruction is ; The acceleration control data includes the user's control over the vehicle's driving acceleration and direction angle; Indicates the magnitude and direction of the vehicle's acceleration at time t; Indicates the speed and direction of the vehicle at time t; Indicates the straight-line distance between the vehicle and the nearest obstruction at time t; t represents the timestamp; The method steps of step S30 are: S301, based on historical vehicle surrounding environment perception data, determine the vehicle's conventional automatic driving path under different vehicle surrounding environment perception data, and simulate the vehicle's conventional automatic driving at a speed of 1 at time t+1. The shortest straight-line distance threshold between the vehicle and the nearest obstruction around it ; S302, based on the historical vehicle control behavior data of the user, taking the time when the user starts to adjust the direction angle of the vehicle as the starting point, and the time when the steering wheel is aligned with the vehicle driving in a straight line as the ending point, taking the time period from the starting point to the ending point as a cycle, determining the number of times the user adjusts the direction angle of the vehicle within the cycle, and determining the angle between the direction angle of the vehicle adjusted by the user each time and the direction angle of the vehicle during normal automatic driving, to obtain a set C; ; ; in, represents the user's behavior of adjusting the vehicle's direction angle in the i-th cycle based on the analysis of the user's historical vehicle control behavior data; Indicates the change in acceleration magnitude and direction during normal automatic driving caused by the jth adjustment made by the user in the ith cycle; i = 1, 2, ..., n; j = 1, 2, ..., k i ; S303, according to set C, respectively calculate the speed and direction of the vehicle when it is traveling: At the next moment, under normal automatic driving and user control of the vehicle, the time when the vehicle collides with the nearest surrounding obstruction is calculated according to the formula: ； ； in, Indicates the time when the vehicle collides with the nearest obstruction in the surrounding area during normal automatic driving at time t+1; Indicates the time when the vehicle collides with the nearest obstruction after the user adjusts the vehicle; Indicates the speed and direction of the nearest obstruction at time t; Indicates the magnitude and direction of the vehicle's acceleration at time t during normal automatic driving; in, , and The moving direction is the projection of the distance line between the two in the horizontal direction; the time when the vehicle collides with the nearest obstruction around is the collision time under the distance line when the acceleration of the two does not change in the future time period at a certain speed; S304, construct a neural network model, adopt RNN algorithm, take the speed and direction of the vehicle, the speed and direction of the nearest obstruction, the time when the vehicle collides with the nearest obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest obstruction as output, train the model, and calculate according to the formula: ； in, Represents the input vector at time t, which contains , , , and ; f represents the activation function; Indicates that The weight of Indicates that The weight of represents the bias term; in, ; Indicates the change in acceleration magnitude and direction of the vehicle during normal automatic driving when the user adjusts it at time t; According to set C, the model is trained and the model parameters are adjusted. The behavior data of users adjusting the vehicle direction angle in set C is used as The input of the model enables the model to simulate and predict the time when the vehicle collides with the nearest surrounding obstruction after the user adjusts the vehicle.

2. The self-evolutionary learning type automatic driving assistance method based on edge scenarios according to claim 1, characterized in that: Based on the constructed neural network model, the moving speed and direction of the current vehicle and the nearest surrounding obstruction are determined, and the straight-line distance between the current vehicle and the nearest surrounding obstruction is determined. Based on the data stored in the user information database, the perception time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle is determined, and based on the perception time, the user's change in the magnitude and direction of the acceleration of the vehicle during normal automatic driving is reversed, and the driving path of the current vehicle is adjusted.

3. A self-evolving learning automated driving assistance system based on edge scenarios, characterized by: The system includes a user identity recognition and management module, a driving data collection and analysis module, a self-evolutionary intelligent learning module, and a personalized intelligent control module; The user identity recognition management module is used to obtain user permissions, identify the user's identity before the user drives, and retrieve the information database corresponding to the user according to the identified user identity; The driving data collection and analysis module is used to collect the vehicle control behavior data of the user during the vehicle driving process, and collect and determine the vehicle surrounding environment perception data in the same time series as the user's vehicle control behavior data; the self-evolution intelligent learning module is used to establish a neural network model for analyzing the user's vehicle control behavior data, and recommend a personalized vehicle automatic driving path for the user based on the current vehicle surrounding environment perception data; the personalized intelligent control module is used to adjust the current vehicle's driving path based on the recommended personalized vehicle automatic driving path; The user identity management module includes a user identity management unit and a database management unit; The user identity management sheet is used to obtain user permissions and identify the user's identity before the user drives; The database management unit is used to establish an information database of the new user when a new user drives a vehicle, and store the collected and analyzed vehicle control behavior data and vehicle surrounding environment perception data of the new user as historical data in the information database corresponding to the new user; the information database is used to store the user's historical vehicle control behavior data and historical vehicle surrounding environment perception data; The driving data collection and analysis module includes an environment perception data collection unit, a vehicle motion collection unit, a user control data collection unit and a time series management unit; The environmental perception data acquisition unit includes a distance measuring sensor, which is used to perform simulation prediction based on the current moving direction and speed of the vehicle and the surrounding obstructions, and determine the straight-line distance between the vehicle and the nearest surrounding obstructions by edge computing; The vehicle motion acquisition unit is used to determine the vehicle's driving speed change data based on the vehicle's surrounding environment perception data; The user control data acquisition unit is used to determine the acceleration control data of the user on the vehicle according to the user's vehicle control behavior data; The time series management unit is used to establish a time series management model to analyze and process the user vehicle control behavior data and vehicle surrounding environment perception data in the same time series; The self-evolutionary intelligent learning module includes a simulation prediction unit, a data processing and analysis unit, a time calculation unit and a model training management unit; The simulation prediction unit is used to determine the conventional automatic driving path of the vehicle under different vehicle surrounding environment perception data, and simulate the shortest threshold of the straight-line distance between the vehicle and the nearest surrounding obstruction at the next moment when the vehicle is in conventional automatic driving; The data processing and analysis unit is used to determine the number of times the user adjusts the vehicle's direction angle within the period, taking the time when the user starts to adjust the vehicle's direction angle as the starting point, the time when the steering wheel is aligned with the vehicle's straight-line driving as the ending point, and taking the time period from the starting point to the ending point as a cycle, and determining the angle between the user's adjustment of the vehicle's direction angle each time and the vehicle's direction angle during normal automatic driving; The time calculation unit is used to calculate the time when the vehicle collides with the nearest surrounding obstruction at the next moment under normal automatic driving and user control of the vehicle; The model training management unit is used to construct a neural network model, adopts an RNN algorithm, takes the vehicle's speed and direction, the speed and direction of the nearest surrounding obstruction, and the time when the vehicle collides with the nearest surrounding obstruction during normal automatic driving as input, and the time when the user adjusts the vehicle and collides with the nearest surrounding obstruction as output to train the model.

4. The self-evolving learning automatic driving assistance system based on edge scenarios according to claim 3, characterized in that: The personalized intelligent control module includes a user behavior learning unit and a path adjustment unit; The user behavior learning unit is used to determine the moving speed and direction of the current vehicle and the nearest surrounding obstruction according to the constructed neural network model, and determine the straight-line distance between the current vehicle and the nearest surrounding obstruction, and determine the perceived time of the vehicle colliding with the nearest surrounding obstruction after the user adjusts the vehicle according to the stored data in the user information database, and reversely infer the user's change value of the acceleration magnitude and direction during normal automatic driving of the vehicle according to the perceived time; The path adjustment unit is used to adjust the current vehicle's driving path according to the change value of the acceleration magnitude and direction during normal automatic driving of the vehicle set by the user.

Images