CN112874519B

CN112874519B - Control method and system for adaptive cruise, storage medium and electronic device

Info

Publication number: CN112874519B
Application number: CN202110138377.3A
Authority: CN
Inventors: 熊胜健; 李飘; 盛凯; 周伟光; 谢金晶
Original assignee: Dongfeng Motor Corp
Current assignee: Dongfeng Motor Corp
Priority date: 2021-02-01
Filing date: 2021-02-01
Publication date: 2022-02-15
Anticipated expiration: 2041-02-01
Also published as: CN112874519A

Abstract

The invention discloses a control method, a control system, a storage medium and electronic equipment of adaptive cruise, wherein the method comprises the following steps: acquiring environment perception information and vehicle operation parameter information, wherein the vehicle operation parameter information comprises a plurality of vehicle state parameters and corresponding parameter values; determining the current scene category of the vehicle according to the environment perception information; when the vehicle is detected to be in the self-adaptive cruise control state, acquiring driver identity information; and determining an adaptive cruise parameter according to the driver identity information, the scene type and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter. The method and the device realize that the ACC system performance is basically consistent with the driving process of an actual driver in different scenes, and realize the customization of the driving styles of different drivers.

Description

Control method and system for adaptive cruise, storage medium and electronic device

Technical Field

The invention relates to the field of automobile control, in particular to a control method and system for adaptive cruise, a storage medium and electronic equipment.

Background

In recent years, rapid development and widespread use of internet technology, high-precision maps, and artificial intelligence have promoted the development of automated driving technology. Relevant reports show that the automatic driving technology can reduce 90% of traffic accidents and 70% of travel cost, and is expected to become a next generation computing platform. The automatic driving technology mainly comprises three major links of perception, decision and execution. Firstly, acquiring and processing environmental information and in-vehicle information through sensor equipment such as a camera, a laser radar or a millimeter wave radar; then, decision judgment is carried out according to the acquired information and the intention of the driver, and a corresponding control strategy is made; and finally, the execution system controls the mechanical energy of the vehicle and feeds the mechanical energy back to the bottom layer module to execute tasks, including acceleration by wire, braking by wire, steering by wire and the like.

At present, a driver manually sets fixed expected vehicle speed and driving time distance parameters after turning on an ACC (Adaptive Cruise Control) function, and cannot adopt a customized Control mode for the driver. However, in the actual driving process, the driver has great difference in the parameters of the vehicle such as the expected speed, the driving time interval, the acceleration change rate and the like under different scenes. If the vehicle runs under the congested road condition, the expected driving time interval of an aggressive driver is small so as to prevent congestion, the expected driving time interval of a conservative driver is large so as to ensure the safety of the congested road section and prevent rear-end collision, and the expected driving time interval of a general driver is between the aggressive driver and the conservative driver.

Disclosure of Invention

The present invention is directed to overcome the drawbacks of the background art, and provides a method, a system, a storage medium, and an electronic device for adaptive cruise control, so as to implement that the ACC system performance is substantially consistent with the driving process of an actual driver in different scenes, and to customize the driving styles of different drivers.

In a first aspect, a control method of adaptive cruise is provided, which includes the following steps:

acquiring environment perception information and vehicle operation parameter information, wherein the vehicle operation parameter information comprises a plurality of vehicle state parameters and corresponding parameter values;

determining the current scene category of the vehicle according to the environment perception information;

when the vehicle is detected to be in the self-adaptive cruise control state, acquiring driver identity information;

and determining an adaptive cruise parameter according to the driver identity information, the scene type and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter.

According to the first aspect, in a first possible implementation manner of the first aspect, after the step of "determining a current scene category of the vehicle according to the environment perception information", the method includes the following steps:

when the vehicle is detected to be in a driver driving state, acquiring driver identity information;

obtaining a plurality of vehicle operation parameter set samples according to the driver identity information and the scene category, wherein the vehicle operation parameter set samples comprise the vehicle operation parameter information;

and performing deep learning training on a plurality of vehicle manipulation parameter set samples to determine a target vehicle manipulation parameter set.

According to the first possible implementation manner of the first aspect, in a second possible implementation manner of the first aspect, the step of performing deep learning training on a plurality of vehicle operation parameter set samples to determine a target vehicle operation parameter set includes the following steps:

selecting calibrated vehicle operation parameter set samples from a plurality of vehicle operation parameter set samples to obtain a calibrated number of sample clusters;

distributing the rest vehicle operation parameter set samples to the sample clusters with the calibrated number one by one, and calculating the target vehicle operation parameter set of each sample cluster.

According to a second possible implementation manner of the first aspect, in a third possible implementation manner of the first aspect, the step of "assigning the remaining vehicle handling parameter set samples one by one to a calibrated number of sample clusters, and calculating the target vehicle handling parameter set of each sample cluster" includes the following steps:

calculating the clustering center of each sample cluster, selecting any one vehicle operation parameter set sample from the rest vehicle operation parameter set samples, and calculating the distance from the selected vehicle operation parameter set sample to the clustering center of each sample cluster;

distributing the selected vehicle operation parameter set samples to the closest sample cluster, and recalculating the cluster center of a new sample cluster;

selecting another vehicle operation parameter set sample from the rest vehicle operation parameter set samples to distribute until the rest vehicle operation parameter set samples are distributed into the sample clusters with the calibrated quantity;

calculating the set of target vehicle handling parameters for each of the sample clusters.

According to a third possible implementation manner of the first aspect, in a fourth possible implementation manner of the first aspect, the step of "determining an adaptive cruise parameter according to the driver identity information, the scene type, and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter" includes the steps of:

determining the corresponding calibrated sample cluster and the corresponding target vehicle operation parameter set according to the driver identity information and the scene category;

calculating the distance from the vehicle handling parameter information to the cluster center of each sample cluster;

and selecting parameters in the target vehicle operation parameter set corresponding to the sample cluster of the minimum distance as the adaptive cruise parameters, and controlling the vehicle to run according to the adaptive cruise parameters.

According to a third possible implementation manner of the first aspect, in a fifth possible implementation manner of the first aspect, after the step of "determining an adaptive cruise parameter according to the driver identity information, the scene type, and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter", the method includes the following steps:

when the situation that the vehicle is switched from the self-adaptive cruise control state to the driver driving state is detected, acquiring new vehicle operation parameter information after switching;

a new target vehicle handling parameter set is determined again by training new vehicle handling parameter set samples through deep learning, the new vehicle handling parameter set samples including the vehicle handling parameter set samples of adaptive cruise control states and the new vehicle handling parameter information.

In a second aspect, there is provided an adaptive cruise control system comprising:

the system comprises a parameter acquisition module, a parameter processing module and a parameter processing module, wherein the parameter acquisition module is used for acquiring environment perception information and vehicle operation parameter information, and the vehicle operation parameter information comprises a plurality of vehicle state parameters and corresponding parameter values;

the scene analysis module is in communication connection with the parameter acquisition module and is used for determining the current scene category of the vehicle according to the environment perception information;

the information acquisition module is used for acquiring the identity information of a driver when the vehicle is detected to be in the self-adaptive cruise control state;

and the parameter analysis module is in communication connection with the parameter acquisition module, the scene analysis module and the information acquisition module and is used for determining an adaptive cruise parameter according to the driver identity information, the scene type and the vehicle operation parameter information, controlling the vehicle to run according to the adaptive cruise parameter and controlling the vehicle to run according to the adaptive cruise parameter.

According to the second aspect, in a first possible implementation manner of the second aspect, the method further includes:

the information acquisition module is further used for acquiring the identity information of the driver when the vehicle is detected to be in the driving state of the driver;

the deep learning module is in communication connection with the parameter acquisition module, the scene analysis module and the information acquisition module, and is used for acquiring a plurality of vehicle operation parameter set samples according to the driver identity information and the scene category, wherein the vehicle operation parameter set samples comprise the vehicle operation parameter information; deep learning training is carried out on a plurality of vehicle operation parameter set samples, and a target vehicle operation parameter set is determined; the method specifically comprises the following steps: selecting calibrated vehicle operation parameter set samples from a plurality of vehicle operation parameter set samples to obtain a calibrated number of sample clusters; calculating the clustering center of each sample cluster, selecting any one vehicle operation parameter set sample from the rest vehicle operation parameter set samples, and calculating the distance from the selected vehicle operation parameter set sample to the clustering center of each sample cluster; distributing the selected vehicle control parameter set samples to clusters with the nearest distance, and recalculating the clustering centers of new sample clusters; selecting another vehicle operation parameter set sample from the rest vehicle operation parameter set samples to distribute until the rest vehicle operation parameter set samples are distributed into the sample clusters with the calibrated quantity; calculating the set of target vehicle handling parameters for each of the sample clusters.

In a third aspect, a storage medium is provided, on which a computer program is stored, wherein the computer program is executed by a processor to implement the method for testing the solid state disk code.

In a fourth aspect, an electronic device is provided, which includes a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, and is characterized in that the processor implements the method for testing the solid state disk code when executing the computer program.

Compared with the prior art, the method and the device realize that the ACC system performance is basically consistent with the driving process of an actual driver in different scenes, and realize the customization of the driving styles of different drivers.

Drawings

FIG. 1 is a flow chart illustrating an adaptive cruise control method according to an embodiment of the present invention;

FIG. 2 is a functional block diagram of a self-learning parameter calibration system of the adaptive cruise system based on a scenario in accordance with an embodiment of the present invention;

FIG. 3 is a logic algorithm block diagram of a self-learning parameter calibration method of the scene-based adaptive cruise system in accordance with an embodiment of the present invention;

fig. 4 is a schematic structural diagram of an adaptive cruise control system according to an embodiment of the present invention.

Reference numerals:

10. a sensing device; 20. a self-learning parameter module; 30. an ACC system control module; 100. a control system for adaptive cruise; 110. a parameter acquisition module; 120. a scene analysis module; 130. an information acquisition module; 140. a parameter analysis module; 150. and a deep learning module.

Detailed Description

Reference will now be made in detail to the present embodiments of the invention, examples of which are illustrated in the accompanying drawings. While the invention will be described in conjunction with the specific embodiments, it will be understood that they are not intended to limit the invention to the embodiments described. On the contrary, it is intended to cover alternatives, modifications, and equivalents as may be included within the spirit and scope of the invention as defined by the appended claims. It should be noted that the method steps described herein may be implemented by any functional block or functional arrangement, and that any functional block or functional arrangement may be implemented as a physical entity or a logical entity, or a combination of both.

In order that those skilled in the art will better understand the present invention, the following detailed description of the invention is provided in conjunction with the accompanying drawings and the detailed description of the invention.

Note that: the example to be described next is only a specific example, and does not limit the embodiments of the present invention necessarily to the following specific steps, values, conditions, data, orders, and the like. Those skilled in the art can, upon reading this specification, utilize the concepts of the present invention to construct more embodiments than those specifically described herein.

Referring to fig. 1, an embodiment of the present invention provides an adaptive cruise control method, including the following steps:

s1000, acquiring environment perception information and vehicle operation parameter information, wherein the vehicle operation parameter information comprises a plurality of vehicle state parameters and corresponding parameter values;

s2000, determining the current scene type of the vehicle according to the environment perception information;

s3000, when the vehicle is detected to be in the self-adaptive cruise control state, acquiring driver identity information;

and S4000, determining an adaptive cruise parameter according to the driver identity information, the scene type and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter.

Specifically, in this embodiment, environment sensing information and Vehicle operation parameter information detected by the sensing device are obtained, where the environment sensing information includes the current environments of vehicles such as the number of surrounding vehicles, the number of surrounding pedestrians, the road curvature radius, and a ramp, and the Vehicle operation parameter information includes a plurality of Vehicle state parameters and corresponding parameter values, the Vehicle state parameters include Vehicle current driving related information such as lateral and longitudinal Vehicle speed, lateral and longitudinal distance, Vehicle speed, longitudinal acceleration, lateral acceleration, steering wheel angle, Vehicle accelerator pedal depth, brake pedal depth, and the like, where the information of surrounding vehicles may be obtained in a manner of V2V (Vehicle to Vehicle) and the like, and the road condition information such as the road curvature radius may be obtained in a manner of high-precision map, V2I (Vehicle to Infrastructure) and the like.

And determining the current scene category of the vehicle according to the environment perception information, wherein the scene category comprises scenes such as cruising, conventional follow-up driving, congested road condition driving, follow-up stop, curve speed limit, crossroads, country roads, busy road sections, expressways and the like. Each scene type is provided with a set of corresponding main characteristic parameters and characteristic threshold values thereof, and the scenes are enumerated to obtain the scene type, wherein the characteristic threshold values can be calibrated manually or analyzed based on big data. Therefore, in the step, the environmental perception information is compared with the characteristic threshold values corresponding to the scene categories one by one, and the current scene category is decided. Assuming that the number of surrounding vehicles is greater than a certain threshold value and the number of surrounding pedestrians is greater than a certain threshold value, it may be preliminarily determined that the own vehicle is in the scene category of the busy road section, and further, if the sensors display that there are many tall buildings around, it may be further assisted to determine that the own vehicle is in the scene category of the busy road section, where the number of surrounding vehicles and the number of surrounding pedestrians are derived from the acquired environment perception information in this example. In addition, the current scene type of the vehicle can be analyzed by assisting the vehicle operation parameter information, for example, if the accelerator pedal and the brake pedal of the vehicle are frequently switched and the change rate of the depth of the accelerator pedal and the brake pedal is large, the vehicle is preliminarily judged to be in the traffic jam condition for driving.

Different drivers have different habits of driving the vehicle, and therefore, when it is detected that the vehicle is in the adaptive cruise control state, the driver identity information is acquired. And determining adaptive cruise parameters corresponding to the driver habits according to the identity information of the driver, the scene type and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameters.

The invention realizes that the ACC (Adaptive Cruise Control) system performance is basically consistent with the driving process of an actual driver under different scenes, and realizes the customization of the driving styles of different drivers.

Optionally, in another embodiment of the present invention, after the step of "S2000 determining the current scene category of the vehicle according to the environment perception information", the method includes the following steps:

s2100, when detecting that the vehicle is in a driver driving state, acquiring driver identity information;

s2200, according to the driver identity information and the scene category, obtaining a plurality of vehicle operation parameter set samples, wherein the vehicle operation parameter set samples comprise the vehicle operation parameter information;

s2300, deep learning training is carried out on a plurality of vehicle manipulation parameter set samples, and a target vehicle manipulation parameter set is determined.

Specifically, in this embodiment, when it is detected that the vehicle is in the driving state of the driver, that is, the vehicle is currently driven by the driver, the driver identity information is acquired, and the current vehicle operation parameter information is divided into corresponding driver categories to analyze the driving habits of the driver.

According to the driver identity information and the scene category, a plurality of corresponding vehicle operation parameter set samples are obtained, namely all vehicle operation parameter information of a driver corresponding to the driver identity information in the scene category is obtained, the vehicle operation parameter information comprises recorded historical information and real-time information collected in the current state, the vehicle operation parameter set is used as a deep learning sample for training, a target vehicle operation parameter set is determined, vehicle operation parameters of the corresponding driver in the corresponding driving habit in the scene category are obtained, and when the driver is detected to be in the adaptive cruise control state, the target vehicle operation parameter set of the corresponding driver is directly called to control vehicle driving, so that the ACC system performance in different scenes is basically consistent with the driving process of an actual driver. The deep learning method for deep learning and training a plurality of vehicle operation parameter set samples to obtain corresponding target vehicle operation parameter sets is not limited to the method in the invention.

Optionally, in another embodiment of the present invention, the step of "S2300 performing deep learning training on a plurality of samples of the vehicle operation parameter set to determine a target vehicle operation parameter set" includes the following steps:

s2310, selecting calibrated vehicle operation parameter set samples from the plurality of vehicle operation parameter set samples to obtain a calibrated number of sample clusters;

s2320, distributing the rest vehicle operation parameter set samples to the sample clusters with the calibrated number one by one, and calculating the target vehicle operation parameter set of each sample cluster.

Specifically, in this embodiment, calibrated vehicle operation parameter set samples are selected from all vehicle operation parameter set samples to obtain a calibrated number of sample clusters, where the calibrated number is preset, and for the initially selected calibrated number of sample clusters, each sample cluster only contains one vehicle operation parameter set sample, so that the cluster center of each sample cluster is the selected corresponding vehicle operation parameter set sample. Similarly, each vehicle operation parameter set sample comprises a plurality of vehicle state parameters and corresponding parameter values, the vehicle state parameters comprise vehicle current driving related information such as transverse and longitudinal vehicle speed, transverse and longitudinal distance, vehicle speed, longitudinal acceleration, lateral acceleration, steering wheel angle, vehicle accelerator pedal depth, brake pedal depth and the like, and the parameter values are the specific size of each vehicle state parameter.

And distributing the rest vehicle operation parameter set samples except the vehicle operation parameter set samples selected as the sample clusters to the sample clusters with the calibrated number one by one, namely finally distributing all the vehicle operation parameter set samples to obtain the sample clusters with the calibrated number, and finally respectively calculating the target vehicle operation parameter set of each sample cluster.

Even if the same driver is in the same scene category, the driving modes of the same driver in different time periods may be different due to other factors, so all the clustering is performed, that is, all the vehicle operation parameter set samples are classified, so that in the subsequent self-adaptive cruise, the scenes are further subdivided based on the same driver and the same scene category, and the selected target vehicle operation parameter set is more suitable for the current actual driving situation of the driver.

Optionally, in another embodiment of the present invention, the step of "S2320 assigning the rest of the vehicle handling parameter set samples to a calibrated number of the sample clusters one by one, and calculating the target vehicle handling parameter set of each sample cluster" includes the following steps:

s2321, calculating the cluster center of each sample cluster, selecting any one vehicle operation parameter set sample from the rest vehicle operation parameter set samples, and calculating the distance from the selected vehicle operation parameter set sample to the cluster center of each sample cluster;

s2322, the selected vehicle operation parameter set samples are distributed to the sample cluster with the nearest distance, and the cluster center of a new sample cluster is calculated again;

s2323, another vehicle operation parameter set sample in the rest vehicle operation parameter set samples is selected to be distributed until the rest vehicle operation parameter set samples are distributed into the sample clusters with the calibrated number;

s2324 calculates the set of target vehicle handling parameters for each of the sample clusters.

Specifically, in this embodiment, the cluster centers of each sample cluster are first calculated, and for an initial sample cluster obtained by extracting a preset number of vehicle operation parameter set samples from all vehicle operation parameter set samples for the first time, the cluster centers are the extracted vehicle operation parameter set samples, respectively.

And selecting any one vehicle operation parameter set sample from the rest vehicle operation parameter set samples, and calculating the distance from the selected vehicle operation parameter set sample to the clustering center of each sample cluster, such as the Manhattan distance, namely the sum of the absolute values of the vehicle state parameter differences between all the vehicle state parameters in the selected vehicle operation parameter set sample and the clustering center corresponding to each sample cluster. And distributing the selected vehicle operation parameter set samples to the cluster with the closest distance, and recalculating the cluster center of the new sample cluster. That is, when a new vehicle operation parameter set sample is allocated to a certain sample cluster to form a new sample cluster, the cluster center of the new sample cluster is recalculated, the new sample cluster replaces the original sample cluster, and when another vehicle operation parameter set sample is selected again for allocation, the cluster with the cluster center of the new sample cluster is calculated, and the sample cluster to which the new vehicle operation parameter set sample is not allocated is kept intact.

Therefore, another vehicle operation parameter set sample in the rest vehicle operation parameter set samples is selected one by one to be distributed until the rest vehicle operation parameter set samples are distributed into the sample clusters with the calibrated number, and finally the sample clusters with the calibrated number are obtained. Wherein, each vehicle operation parameter set sample is distributed, and only one sample cluster is updated to obtain a new sample cluster.

And calculating the target vehicle operation parameter set of each sample cluster, namely finally obtaining the target vehicle operation parameter sets with the calibrated number. For each sample cluster, the calculation method of the target vehicle operation parameter set is not particularly limited, for example, any one vehicle operation parameter set sample (with the maximum or minimum parameter value or based on other rules) in each sample cluster may be selected as the target vehicle operation parameter set, or the target vehicle operation parameter set is calculated by using a mean method, that is, the parameter value of each vehicle operation parameter in the target vehicle operation parameter set is the mean value of the parameter values of the corresponding vehicle operation parameters of all vehicle operation parameter set samples in the sample cluster.

The method for deep learning based on the K-Means algorithm obtains the target vehicle operation parameter set, is simple in calculation mode, and further subdivides scenes based on the same driver and the same scene category, so that the selected target vehicle operation parameter set is more suitable for the current actual driving situation of the driver.

Optionally, in another embodiment of the present invention, the step of "S4000 determining an adaptive cruise parameter according to the driver identity information, the scene type, and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter" includes the steps of:

s4100, determining the corresponding calibrated sample cluster and the corresponding target vehicle operation parameter set according to the driver identity information and the scene category;

s4200 calculating a distance of the vehicle handling parameter information to a cluster center of each of the sample clusters;

s4300 selects parameters in the target vehicle control parameter set corresponding to the sample cluster with the minimum distance as the adaptive cruise parameters, and controls the vehicle to run according to the adaptive cruise parameters.

Specifically, in this embodiment, when it is detected that the vehicle is in the adaptive cruise control state, the corresponding calibrated sample cluster and the corresponding target vehicle operating parameter set are determined according to the driver identity information and the scene category, where the sample cluster is in one-to-one correspondence with the target vehicle operating parameter set, and the sample clusters are in the calibrated number, so that it is necessary to further determine the corresponding sample cluster to which the current vehicle state belongs.

And calculating the distance from the vehicle operation parameter information to the clustering center of each sample cluster, such as Manhattan distance, namely the sum of absolute values of differences between all vehicle state parameters in the vehicle operation parameter information and the vehicle state parameters of the clustering centers corresponding to the sample clusters. And attributing the vehicle operation parameter information to the nearest cluster center, so that the parameters in the target vehicle operation parameter set corresponding to the sample cluster with the minimum distance are selected as the adaptive cruise parameters, and the vehicle is controlled to run according to the adaptive cruise parameters.

The method is based on the sample cluster of deep learning, when the vehicle is in the self-adaptive cruise control state, the real-time vehicle operation parameter information of the current vehicle is further divided into the corresponding sample clusters, and the vehicle driving is controlled through the parameters in the corresponding target vehicle operation parameter set, so that the selected target vehicle operation parameter set is more suitable for the current actual driving situation of a driver.

Optionally, in another embodiment of the present invention, after the step of "S4000 determining an adaptive cruise parameter according to the driver identity information, the scene type, and the vehicle operation parameter information, and controlling the vehicle to run according to the adaptive cruise parameter", the method includes the following steps:

s5000, when the situation that the vehicle is switched from the self-adaptive cruise control state to the driver driving state is detected, acquiring new vehicle operation parameter information after switching;

and S6000, determining a new target vehicle manipulation parameter set again by training a new vehicle manipulation parameter set sample through deep learning, wherein the new vehicle manipulation parameter set sample comprises the vehicle manipulation parameter set sample of the adaptive cruise control state and the new vehicle manipulation parameter information.

Specifically, in this embodiment, when it is detected that the vehicle is switched from the adaptive cruise control state to the driver driving state, it indicates that the driver takes over the vehicle, that is, the driver does not adapt to the adaptive cruise parameters executed in the adaptive cruise control state, and needs to change the adaptive cruise parameters. Therefore, new vehicle operation parameter information adjusted by a driver after switching is obtained, and a new calibrated number of sample clusters are determined again through deep learning training of new vehicle operation parameter set samples to obtain a new target vehicle operation parameter set, wherein the deep learning mode is known as described in the above embodiment, and the new vehicle operation parameter set samples used for the deep learning include vehicle operation parameter set samples of an adaptive cruise control state, new vehicle operation parameter information, and the rest of history records.

In addition, for the reason for the driver to take over, when the vehicle is in the adaptive cruise control state, the distance between the vehicle operation parameter information and the cluster center of each sample cluster is calculated, and the sample cluster with the minimum distance is determined, and the vehicle state parameter corresponding to the parameter value with the maximum influence distance value is selected as the reason for the driver to take over the vehicle.

According to the method and the device, when the situation that the vehicle is switched from the self-adaptive cruise control state to the driver driving state is detected, deep learning is carried out again based on the new vehicle operation parameter information adjusted by the driver after switching, so that a target vehicle operation parameter set obtained through deep learning can be more fit with the real driving habit of the driver.

The embodiment provides a self-learning parameter calibration method and system of a scene-based adaptive cruise system, and a functional block diagram of the method is shown in fig. 2, and a logical algorithm block diagram of the system is shown in fig. 3. The self-learning parameter calibration method and system function realization of the scene-based self-adaptive cruise system mainly comprise the following steps:

(1) obtaining information such as environment sensing information S0 (set) and vehicle operation parameter information detected by the sensing device 10, for example, the number n of surrounding vehicles, the number m of surrounding pedestrians, the lateral and longitudinal vehicle speeds Vy and Vx, the lateral and longitudinal distances dy and dx, the vehicle speed Vh, the longitudinal acceleration ax, the lateral acceleration ay, the steering wheel angle θ, the road condition information such as the road curvature radius R and the ramp i, the accelerator pedal depth of the vehicle

Depth of brake pedal

And the like, the information of surrounding vehicles can be obtained by a mode of V2V and the like, and the road condition information such as the curvature radius R of a road, the slope i and the like can be obtained by a mode of high-precision map, V2I and the like.

(2) Based on the environment perception information obtained in (1), the self-learning parameter module 20 determines the current scene category C, such as scenes of cruising, conventional follow-up driving, congested road conditions driving, follow-up stop, curve speed limit, crossroads, country roads, busy road sections, highways, and the like. Each scene type has a set of corresponding main characteristic parameters and characteristic threshold values C1 (set), and the scenes are enumerated to obtain a scene type C0 (set), wherein the characteristic threshold values are mainly derived from manual calibration. Therefore, in this step, the environmental awareness information is compared with the corresponding C1 of each scene type one by one, and the current driving scene type C is determined. Assuming that the number n of surrounding vehicles is greater than a certain threshold value n0 and the number m of surrounding pedestrians is greater than a certain threshold value m0, it may be preliminarily determined that the own vehicle is in the heavy road scene category, and further, if the sensors display more tall buildings around, it may be further assisted to determine that the own vehicle is in the heavy road scene category, where n and m are derived from the acquired environment perception information and n0 and m0 are derived from C1. And if the accelerator pedal and the brake pedal of the vehicle are frequently switched and the depth change rate is large, the vehicle is preliminarily judged to be in the traffic jam road condition for running. Similarly, the remaining scene categories may also be determined according to the main scene feature parameters and the feature threshold S0.

(3) According to the ACC function state, whether the self-learning parameter module 20 is in actual driver operation or not is judged, if yes, the self-learning parameter module 20 self-adaptively determines a vehicle operation parameter set sample S1 (set) representing the driver habit according to the current scene type C, such as the accelerator pedal depth of the self-vehicle

And rate of change, depth of brake pedal of bicycle

And its rate of change, vehicle-ahead distance d, vehicle speed Vh, vehicle longitudinal acceleration ax, vehicle lateral acceleration ay, vehicle response-ahead sensitivity S, etc., and statistically analyzing a personalized threshold of driver behavior, i.e., a target vehicle maneuver parameter set S2, according to a plurality of vehicle maneuver parameter set samples, wherein the personalized threshold is obtained by a deep learning method, which can be embodied in the K-Means algorithm, but is not limited thereto. Assuming that the current scene type C is in the congested road condition driving, the output parameters representing the driver habits in the scene type mainly include the accelerator pedal depth of the vehicle

And rate of change, depth of brake pedal of bicycle

The change rate, the distance d between the vehicles and the sensitivity S of the vehicles responding to the front vehicles are obtained, the parameters form an S1 (set), and the information of the S1 parameters of the vehicles running on the congested road condition for many times is obtainedPersonalized threshold for the parameters of driver behavior in this scene category. And assuming that the current scene type C is the curve speed limit, the parameter for outputting the behavior of the driver under the scene type mainly comprises the depth of the brake pedal of the self-vehicle

And the change rate, the speed Vh and the lateral acceleration ay of the vehicle form an S1 (set), and the personalized threshold of the habit parameter of the driver passing the curve under the scene category is obtained based on the road curvature radius R and the S1 parameter information under the multi-curve speed-limiting scene.

The implementation method for obtaining the personalized threshold value by the deep learning method of the K-Means algorithm is described in detail below. Based on the acquired main characteristic parameters S0 (set) corresponding to the scene type C and the parameters S1 (set) corresponding to the scene type C, firstly, k0 (which is a calibration quantity) sets of S1 data under the scene type C are randomly selected as initial sample clusters, meanwhile, the initial sample clusters are also cluster centers, the distances between the rest k1 sets of S1 data under the scene type C and the cluster centers of all the sample clusters are calculated, the distance in the method adopts Manhattan distance, namely the sum of absolute values of differences between all the parameters S1 and the cluster centers of the corresponding sample clusters, the data S1 under each set of scene type C is allocated to the sample cluster closest to the scene type C, and then the cluster centers of the sample clusters are recalculated until no data S1 under the rest scene type C exists. And aiming at the k0 sample clusters, calculating a target vehicle operation parameter set S2 which is the personalized threshold value of the corresponding driver habit by adopting a mean method.

(4) The ACC system control module 30 performs self-learning customization to adjust ACC function key control parameters such as driving time interval Tg, acceleration amplitude axm of the vehicle, acceleration change rate of the vehicle, lateral acceleration ay of the vehicle and the like based on the personalized threshold value representing the driver habit obtained in the step (3) and the scene category C. When the self-vehicle is in ACC system operation, calculating the distance D between the S0 and k0 cluster centers under the current scene category C, obtaining the cluster corresponding to the minimum distance, taking the personalized threshold value corresponding to the parameter S1 of the driver habit corresponding to the cluster as the input of the ACC system, and customizing and adjusting the key control parameter of the ACC function.

D＝|S0i(1)-S2(1)|+|S0i(2)-S2(2)|+……+|S0i(p)-S2(p)|

Wherein S0i represents parameter values corresponding to vehicle state parameters in the S0 set under the current scene category C, S2 represents personalized thresholds corresponding to vehicle state parameters in the S1 set of the cluster center of a certain sample cluster, and p represents the sizes of the S0 and S2 sets.

Assuming that the current scene type C is running under congested road conditions, performing statistical analysis based on the multiple data results by combining the parameter S1 of the behavior of the driver under the scene type obtained in step (3) and the personalized threshold thereof, and if the following-stop following distance expected by the driver under the scene type is obtained from the following-stop following distance under the running under the congested road conditions, the value can also be used as the driving time distance Tg. And (3) combining the obtained parameter S1 of the driver habit in the scene category and the personalized threshold thereof, and performing statistical analysis based on multiple data results, wherein the value can be used as the expected lateral acceleration of the vehicle when the vehicle lateral acceleration ay is used for ACC control, so as to obtain the speed limit value of the curve running for the curve speed limit function. It should be noted that, as the actual driver's operation mileage (scene category and number thereof) increases, the key control parameter for adjusting the ACC function obtained in this step is roll optimization and makes the ACC vehicle control effect closer to the actual driver's operation effect.

(5) In the step (3), if the self-vehicle is in ACC system operation, the ACC system control module 30 performs customized calibration based on the personalized threshold corresponding to the driver habit parameter S1 obtained in the step (4), and at the same time, monitors whether the driver has takeover behavior in the process, and if so, performs online evaluation and analysis of the takeover reason, and re-optimizes the relevant key parameters to perform customized calibration. The reason for taking over the online evaluation and analysis is mainly based on each subentry in the distance calculation expression in the step (4), and finding out the subentry with a larger value as a key consideration object; and taking the S0 and S1 parameters of the driver taking over process as new data to participate in the calculation of the clustering center in the step (3) to obtain a new personalized threshold value corresponding to the parameter S1 of the driver habit, so that the ACC system performance is basically consistent with the driving process of the actual driver under different scene categories, the driving styles of different drivers are customized, and the trust of the driver on the ACC function and the utilization rate of the ACC function are improved.

As shown in fig. 4, an adaptive cruise control system 100 includes:

a parameter obtaining module 110, configured to obtain environment sensing information and vehicle operation parameter information, where the vehicle operation parameter information includes a plurality of vehicle state parameters and corresponding parameter values;

the scene analysis module 120 is in communication connection with the parameter acquisition module 110, and is configured to determine a current scene category of the vehicle according to the environment sensing information;

the information acquisition module 130 is used for acquiring the identity information of a driver when the vehicle is detected to be in the self-adaptive cruise control state;

and a parameter analysis module 140, communicatively connected to the parameter obtaining module 110, the scene analysis module 120, and the information obtaining module 130, configured to determine an adaptive cruise parameter according to the driver identity information, the scene type, and the vehicle operation parameter information, control a vehicle to run according to the adaptive cruise parameter, and control a vehicle to run according to the adaptive cruise parameter.

Further comprising:

the information obtaining module 130 is further configured to obtain driver identity information when it is detected that the vehicle is in a driver driving state;

a deep learning module 150, communicatively connected to the parameter obtaining module 110, the scene analyzing module 120, and the information obtaining module 130, configured to obtain a plurality of vehicle operation parameter set samples according to the driver identity information and the scene category, where the vehicle operation parameter set samples include the vehicle operation parameter information; deep learning training is carried out on a plurality of vehicle operation parameter set samples, and a target vehicle operation parameter set is determined; the method specifically comprises the following steps: selecting calibrated vehicle operation parameter set samples from a plurality of vehicle operation parameter set samples to obtain a calibrated number of sample clusters; calculating the clustering center of each sample cluster, selecting any one vehicle operation parameter set sample from the rest vehicle operation parameter set samples, and calculating the distance from the selected vehicle operation parameter set sample to the clustering center of each sample cluster; distributing the selected vehicle control parameter set samples to clusters with the nearest distance, and recalculating the clustering centers of new sample clusters; selecting another vehicle operation parameter set sample from the rest vehicle operation parameter set samples to distribute until the rest vehicle operation parameter set samples are distributed into the sample clusters with the calibrated quantity; calculating the set of target vehicle handling parameters for each of the sample clusters.

In addition, the parameter analysis module 140 is further configured to: determining the corresponding calibrated sample cluster and the corresponding target vehicle operation parameter set according to the driver identity information and the scene category;

The deep learning module 150 is further configured to: when the situation that the vehicle is switched from the self-adaptive cruise control state to the driver driving state is detected, acquiring new vehicle operation parameter information after switching;

Specifically, the functions of each module in this embodiment have been described in detail in the corresponding method embodiment, and thus are not described in detail again.

Based on the same inventive concept, the embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements all or part of the method steps of the above method.

The present invention can implement all or part of the processes of the above methods, and can also be implemented by using a computer program to instruct related hardware, where the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above method embodiments can be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution media, and the like. It should be noted that the computer readable medium may contain other components which may be suitably increased or decreased as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, in accordance with legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunications signals.

Based on the same inventive concept, an embodiment of the present application further provides an electronic device, which includes a memory and a processor, where the memory stores a computer program running on the processor, and the processor executes the computer program to implement all or part of the method steps in the method.

The processor may be a Central Processing Unit (CP U), or may be other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, or the like. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, the processor being the control center of the computer device and the various interfaces and lines connecting the various parts of the overall computer device.

The memory may be used to store computer programs and/or modules, and the processor may implement various functions of the computer device by executing or executing the computer programs and/or modules stored in the memory, as well as by invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (e.g., a sound playing function, an image playing function, etc.); the storage data area may store data (e.g., audio data, video data, etc.) created according to the use of the cellular phone. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, server, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.

The present invention has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), servers and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.

Claims

1. A control method for adaptive cruise, comprising the steps of:

determining an adaptive cruise parameter according to the driver identity information, the scene category and the vehicle operation parameter information, controlling the vehicle to run according to the adaptive cruise parameter, and determining the current scene category of the vehicle according to the environment perception information, wherein the method comprises the following steps:

the deep learning training is carried out on a plurality of vehicle manipulation parameter set samples to determine a target vehicle manipulation parameter set, and the deep learning training is carried out on the plurality of vehicle manipulation parameter set samples to determine the target vehicle manipulation parameter set, wherein the deep learning training comprises the following steps:

distributing the rest vehicle operation parameter set samples to the sample clusters with the calibrated number one by one, and calculating the target vehicle operation parameter set of each sample cluster;

the step of allocating the rest vehicle operation parameter set samples to a calibrated number of the sample clusters one by one and calculating the target vehicle operation parameter set of each sample cluster comprises the following steps:

2. The adaptive cruise control method according to claim 1, wherein said determining adaptive cruise parameters based on said driver status information, said scene type, and said vehicle handling parameter information, and controlling vehicle driving based on said adaptive cruise parameters comprises the steps of:

3. The adaptive cruise control method according to claim 1, wherein said determining adaptive cruise parameters based on said driver status information, said scene type, and said vehicle handling parameter information, and controlling the vehicle according to said adaptive cruise parameters, after the step of controlling the vehicle to travel, comprises the steps of:

4. An adaptive cruise control system, comprising:

the parameter analysis module is in communication connection with the parameter acquisition module, the scene analysis module and the information acquisition module, and is used for determining an adaptive cruise parameter according to the driver identity information, the scene type and the vehicle operation parameter information and controlling the vehicle to run according to the adaptive cruise parameter;

5. A storage medium having stored thereon a computer program, characterized in that the computer program, when being executed by a processor, implements the adaptive cruise control method according to claim 1.

6. An electronic device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, wherein the processor implements the adaptive cruise control method according to claim 1 when executing the computer program.