CN107891866B

CN107891866B - Method for determining a road surface on the basis of vehicle data

Info

Publication number: CN107891866B
Application number: CN201611100584.5A
Authority: CN
Inventors: 徐海珍; 李东烈; 金时浚
Original assignee: Hyundai Motor Co
Current assignee: Hyundai Motor Co
Priority date: 2016-10-04
Filing date: 2016-12-02
Publication date: 2021-02-05
Anticipated expiration: 2036-12-02
Also published as: CN107891866A; KR101827152B1

Abstract

A method of determining a road surface based on vehicle data is disclosed. There is provided a method of determining a road surface based on vehicle data by a controller, the method comprising: preprocessing vehicle data; inputting the preprocessed data to a learned neural network to obtain a corresponding output; sequentially extracting a first reference number of samples from an output of the neural network and detecting a value of a sample located in the middle of the samples (hereinafter referred to as "middle value"); and determining whether the road surface is a high-friction road surface or a low-friction road surface based on the detected intermediate value.

Description

Method for determining a road surface on the basis of vehicle data

Cross Reference to Related Applications

This application is based on and claimed in the benefit of priority of korean patent application No. 10-2016-0127656, filed in korean intellectual property office at 10/4/2016, which is hereby incorporated by reference in its entirety.

Technical Field

The present disclosure relates to a method of determining a road surface based on vehicle data, and more particularly, to a technique of determining whether a road surface on which a vehicle is running is a high-friction road surface or a low-friction road surface based on vehicle data obtained from an on-vehicle network.

An on-board network according to an exemplary embodiment of the present disclosure includes a Controller Area Network (CAN), a Local Interconnect Network (LIN), a FlexRay, and a Media Oriented System Transmission (MOST).

Background

To ensure the safety of the driver, vehicles have been equipped with various user-friendly systems, such as an anti-lock brake system (ABS), an Electronic Stability Control (ESC) system, a Smart Cruise Control (SCC) system, and an Advanced Driver Assistance System (ADAS).

For optimum performance, these user-friendly systems can control the behavior of the vehicle by taking into account the road conditions. Here, the road surface condition refers to a high friction road surface such as a dry asphalt road and a dry cement road and a low friction road surface such as a rainy road, a snowy road and a dirt road.

Conventionally, there are the following methods: a method of determining whether a road surface is a high friction road surface or a low friction road surface based on dynamic data such as wheel speed, engine torque, and vehicle speed; and a method of determining whether the road surface is a high-friction road surface or a low-friction road surface based on data from various sensors such as a road surface directional ultrasonic sensor and a microphone.

First, the method of determining a road surface based on dynamic data determines whether the road surface is a high friction road surface or a low friction road surface based on a vehicle slip phenomenon. Therefore, when the vehicle is running on a particular pattern of road that does not accelerate or decelerate quickly, it will be difficult to determine whether the road surface on which the vehicle is running is a high friction road surface or a low friction road surface.

Second, the method of determining the road surface based on the data from the road surface directivity ultrasonic sensor requires additional installation of the sensor, resulting in an increase in the production cost of the vehicle.

Disclosure of Invention

The present disclosure is directed to solving the above-mentioned problems occurring in the prior art while maintaining the advantages achieved by the prior art unaffected.

Aspects of the present disclosure provide a method of determining a road surface based on vehicle data by: preprocessing the vehicle data in a manner corresponding to the characteristic of each vehicle data; inputting the preprocessed data to a learning neural network; post-processing the output of the neural network; and determining whether the road surface is a high-friction road surface or a low-friction road surface, thereby quickly and accurately determining the road surface condition regardless of the road type.

The object of the present disclosure is not limited to the foregoing object, and any other objects and advantages not mentioned herein will be clearly understood from the following description. The inventive concept will be more clearly understood from the exemplary embodiments of the present disclosure. Further, it is apparent that the objects and advantages of the present disclosure can be achieved by the elements and features claimed in the claims and combinations thereof.

According to one aspect of the disclosure, a method of determining a road surface based on vehicle data by a controller includes: preprocessing vehicle data; inputting the preprocessed data to a learned neural network to obtain a corresponding output; sequentially extracting a first reference number of samples from an output of the neural network, and detecting a value of a sample located in the middle of the samples (midle) (hereinafter referred to as an "intermediate value"); and determining whether the road surface is a high-friction road surface or a low-friction road surface based on the detected intermediate value.

According to another aspect of the present disclosure, a method of determining a road surface by a controller includes: preprocessing vehicle data; inputting the preprocessed data to a learned neural network to obtain a corresponding output; sequentially extracting a first reference number of samples from an output of the neural network and calculating an average value of the extracted samples; and determining whether the road surface is a high-friction road surface or a low-friction road surface based on the calculated average value.

Drawings

The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which:

FIG. 1 illustrates a system for determining a road surface based on vehicle data, applying the concepts of the present invention;

FIG. 2 illustrates an output of a neural network according to an exemplary embodiment of the present disclosure;

FIG. 3 shows a block diagram of state transitions of a road surface in accordance with an exemplary embodiment of the present disclosure;

FIG. 4 shows a flowchart of a method of determining a road surface based on vehicle data according to an example embodiment of the present disclosure; and

FIG. 5 shows a flowchart of a method of determining a road surface based on vehicle data according to another exemplary embodiment of the present disclosure.

Detailed Description

The above and other objects, features and advantages of the present disclosure will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, so that those skilled in the art to which the present disclosure pertains can easily carry out the technical ideas described herein. In addition, detailed descriptions of well-known technologies associated with the present disclosure will be excluded so as not to unnecessarily obscure the gist of the present disclosure. Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

FIG. 1 illustrates logic for determining a road surface based on vehicle data according to an exemplary embodiment of the present disclosure.

As shown in fig. 1, logic for determining a road surface based on vehicle data according to an exemplary embodiment of the present disclosure is a process executed by a processor (controller) interconnected with a memory having program instructions recorded therein, and includes a preprocessing operation 10, a neural network 20, and a post-processing operation 30.

First, in the preprocessing operation 10, the vehicle data collected from the on-vehicle network may be preprocessed in a manner corresponding to the characteristic of each vehicle data.

For example, Longitudinal Acceleration Sensor (LAS) data, Wheel Speed Sensor (WSS) data, accelerator pedal position sensor (APS) data, and steering wheel angle sensor (SAS) data may be collected at 10ms intervals from an on-board network, such as a Controller Area Network (CAN) bus. Here, in the preprocessing operation 10, inputs of LAS data, WSS data, APS data, and SAS data may also be received from separate data collectors (not shown).

Thereafter, with respect to the APS data of the motive power (as a main output generated by the operation of the vehicle), an average value and a difference value of APS data values may be calculated to check the magnitude and variation thereof.

In addition, with respect to the vehicle wheel rotation speed and the vehicle longitudinal acceleration that represent the vehicle behavior results, the standard deviation may be calculated to check the variance of the window. By calculating the standard deviation with respect to the difference in rotation speed between the front and rear wheels and between the left and right wheels, the behavior characteristics caused by the unevenness of the road surface (unevenness of the slippery road surface) can be emphasized.

In addition, with respect to the steering wheel angle, a mean value and a difference (difference) value may be calculated so as to logically stably respond to a change in the behavior characteristic caused by the turning of the vehicle.

Hereinafter, the preprocessing operation 10 will be described in detail. Here, the number of data constituting a single window (processing unit) may be determined to take into account physical characteristics (transmission time difference) of each data. For example, the number of data may be 50, and may be arbitrarily changed.

1) The standard deviation LAS _ Std of 50 LAS data (values) may be calculated.

2) The standard deviation FR _ Diff _ Std may be calculated with respect to 50 values obtained by subtracting the average speed of the rear wheels from the average speed of the front wheels. In other words, after the calculation of subtracting the average speed of the rear wheels from the average speed of the front wheels is performed 50 times, the standard deviation of 50 result values may be calculated. Here, the front wheels include left and right front wheels, and the rear wheels include left and right rear wheels.

3) The standard deviation LR _ Diff _ Std can be calculated with respect to 50 values obtained by subtracting the average speed of the left wheel from the average speed of the right wheel. Here, the right wheel includes a right front wheel and a right rear wheel, and the left wheel includes a left front wheel and a left rear wheel.

4) An average APS _ Avg of 50 APS data (values) may be calculated.

5) A sum APS _ Diff of 50 difference values (50 result values obtained by subtracting the previous APS value from the current APS value) of the APS data may be calculated.

6) A sum SAS _ Diff of 50 differences (50 result values obtained by subtracting the previous SAS value from the current SAS value) of the SAS data may be calculated.

7) An average value SAS _ Avg of 50 SAS data (values) may be calculated.

These preprocessed data may be input to the neural network 20 that has completed the learning process.

The neural network 20 may be a supervised learning neural network. The neural network 20 has completed a learning process to input results of the preprocessing LAS data, the WSS data, the APS data, and the SAS data, and obtain corresponding outputs (degree of friction).

The output of the learned neural network 20 may be as shown in fig. 2.

In fig. 2, the y-axis represents the output of the neural network 20 and the x-axis represents time. Here, a value of 0.1 on the x-axis corresponds to 10 ms.

In fig. 2, "210" represents an output with respect to a high friction road surface such as a dry asphalt road and a dry cement road, and "220" represents an output with respect to a low friction road surface such as a snow road and a rain road.

Since the frictional force is uniform on a high-friction road surface, a stable output of less than 0.5 can be obtained. However, since the frictional force is not uniform on a low-friction road surface, an unstable output of 0.5 or more may be obtained.

In the post-processing operation 30, the output of the neural network 20 may be post-processed to determine whether the road surface on which the vehicle is running is a high friction road surface or a low friction road surface using the output of the neural network 20.

Hereinafter, the post-processing operation 30 will be described in detail. Here, the post-processing operation may be periodically performed, and the number of samples may be arbitrarily changed.

1) After 300 samples are taken from the output of the neural network 20, the maximum value can be detected therefrom. Here, the maximum value may represent an instantaneous change in the output of the neural network 20.

2) The standard deviation of the 300 samples taken can be calculated.

3) The sum of differences between neighboring samples with respect to the extracted 300 samples may be calculated as a relative value. Here, the sum of the differences between the samples may represent an absolute change in the output of the neural network 20.

For example, when the first, second, and third sample values are 1, 2, and 4, respectively, the difference between the first and second sample values is 1, and the difference between the second and third sample values is 2, and thus the sum of the differences is 3.

4) After sequentially extracting 1000 samples from the output of the neural network 20, the value of the 500 th sample may be detected as an intermediate value. Here, the intermediate value may represent the overall condition of the output of the neural network 20. Meanwhile, the average value of 1000 samples may also be detected as a median value.

Hereinafter, referring to fig. 3, a process of determining the road surface condition by the controller will be described.

In fig. 3, "310" represents an initial road condition. When the median value of the initial road condition 310 exceeds a first threshold, the road condition may be changed to a low friction road condition 320, and when it does not exceed the first threshold, the road condition may be changed to a high friction road condition 330.

When the median value of the high friction road condition 330 exceeds the second threshold, the road condition may be changed to the low friction road condition 320, and when it does not exceed the second threshold, the current road condition may be maintained.

When the median value of the low friction road condition 320 exceeds the third threshold, the current road condition may be maintained, and when it does not exceed the third threshold, the road condition may be changed to a high friction road condition 330.

Here, when the user puts priority on the stability of the system to which the inventive concept is applied, the order of the thresholds may satisfy a third threshold < a first threshold < a second threshold in the stable mode.

In addition, when the user places the priority on the low-friction road surface, the order of the thresholds may satisfy a first threshold < a third threshold < a second threshold in the low-friction priority mode.

Further, when the user places the priority on the high friction road surface, the order of the thresholds may satisfy a third threshold < a second threshold < a first threshold in the high friction priority mode.

In addition, when the sum of the maximum value, the standard deviation, and the relative value of the high friction road condition 330 exceeds a fourth threshold (e.g., 4), the road condition may be changed to the low friction road condition 320, and when it does not exceed the fourth threshold, the current road condition may be maintained.

When the maximum value of the low friction road condition 320 exceeds the fifth threshold, the current road condition may be maintained, and when it does not exceed the fifth threshold, the road condition may be changed to a high friction road condition 330.

Meanwhile, in order to improve the reliability of the road surface determination result, the determination may be suspended under the following conditions based on the slip estimation result output through the neural network 20:

1) the brakes being activated while driving

-determining that the brake is activated while driving when at least one of the average speed of the front wheels, the average speed of the rear wheels, the average speed of the right wheels and the average speed of the left wheels does not exceed a preset value.

2) Low speed section

When the average speed of the front wheel, the average speed of the rear wheel, the average speed of the right wheel, and the average speed of the left wheel do not exceed 10kph, it may be determined as a low speed section.

3) Steering

When the time (period) when the SAS value exceeds the reference value is maintained, steering may be determined.

4) Rough road

When the brake is activated more than or equal to a predetermined number of times within a predetermined period of time, it may be determined as a rough road.

By applying the inventive concept to ABS, ESC, ADAS, 4WD, etc., the performance of the corresponding system can be improved.

FIG. 4 shows a flowchart of a method of determining a road surface based on vehicle data according to an example embodiment of the present disclosure. Which shows a process performed by a processor (controller).

First, vehicle data may be preprocessed in 401.

Next, at 402, the preprocessed data may be input to a learned neural network, thereby obtaining a corresponding output.

At 403, after a first reference number of samples are sequentially extracted from the output of the neural network, a value of a sample located in the middle of the samples (hereinafter referred to as "middle value") may be detected.

Then, at 404, it may be determined whether the road surface is a high friction road surface or a low friction road surface based on the detected intermediate value.

FIG. 5 shows a flowchart of a method of determining a road surface based on vehicle data according to another exemplary embodiment of the present disclosure. Which shows a process performed by a processor (controller).

First, vehicle data may be preprocessed 501.

Next, at 502, the preprocessed data can be input to a learned neural network, thereby obtaining a corresponding output.

In 503, after the first reference number of samples are sequentially extracted from the output of the neural network, an average of the extracted samples may be calculated.

Then, in 504, it may be determined whether the road surface is a high friction road surface or a low friction road surface based on the calculated average value.

On the other hand, the above-described method according to the exemplary embodiment of the present disclosure may be written as a computer program. Codes and code segments constituting the program can be easily inferred by computer programmers in the art. The written program may be stored in a computer-readable recording medium (information storage medium) and read and executed by a computer, thereby implementing a method according to an exemplary embodiment of the present disclosure. The recording medium includes all types of computer-readable recording media.

As described above, the method of determining a road surface based on vehicle data may be characterized in that: the method includes preprocessing vehicle data in a manner corresponding to a feature of each vehicle data, inputting the preprocessed data to a learned neural network, post-processing an output of the neural network, and determining whether a road surface on which the vehicle is running is a high-friction road surface or a low-friction road surface, thereby quickly and accurately determining a road surface condition regardless of a road type.

In the foregoing, although the present disclosure has been described with reference to the exemplary embodiments and the accompanying drawings, the present disclosure is not limited thereto, but various modifications and changes can be made by those skilled in the art to which the present disclosure pertains without departing from the spirit and scope of the present disclosure claimed in the appended claims.

Symbols of the various elements of the drawings

310 initial road condition

320 low friction road conditions

330 high friction road conditions

401 preprocessing vehicle data

402 inputting the preprocessed data into a learned neural network to obtain corresponding outputs

403 sequentially extracts a first reference number of samples from an output of the neural network, and detects a value of a sample located in the middle of the samples (hereinafter, middle value)

404 determines whether the road surface is a high-friction road surface or a low-friction road surface based on the detected intermediate value

501 preprocessing vehicle data

502 inputs the preprocessed data into a learned neural network to obtain corresponding outputs

503 sequentially extracting a first reference number of samples from the output of the neural network, and calculating an average of the extracted samples

504 determines whether the road surface is a high-friction road surface or a low-friction road surface based on the calculated average value

Claims

1. A method of determining a road surface by a controller based on vehicle data, the method comprising the steps of:

preprocessing vehicle data;

inputting the preprocessed data to a learned neural network to obtain a corresponding output;

sequentially extracting a first reference number of samples from the output of the neural network and detecting a value of a sample located in the middle of the first reference number of samples as an intermediate value; and

determining whether the road surface is a high friction road surface or a low friction road surface based on the detected intermediate value, wherein the determining step includes: changing an initial road condition to a low friction road condition when the intermediate value in the initial road condition exceeds a first threshold value, and changing the initial road condition to a high friction road condition when the intermediate value in the initial road condition does not exceed the first threshold value.

2. The method of claim 1, wherein the determining step further comprises:

changing the high friction road condition to the low friction road condition when the median value under the high friction road condition exceeds a second threshold value, and maintaining a current road condition when the median value under the high friction road condition does not exceed the second threshold value; and

maintaining a current road condition when the intermediate value in the low friction road condition exceeds a third threshold, and changing the low friction road condition to the high friction road condition when the intermediate value in the low friction road condition does not exceed the third threshold.

3. The method of claim 2, wherein the third threshold < the first threshold < the second threshold are satisfied in a threshold order in a stable priority mode.

4. The method of claim 2, wherein the first threshold < the third threshold < the second threshold are satisfied in a threshold order in a low friction priority mode.

5. The method of claim 2, wherein the third threshold < the second threshold < the first threshold are satisfied in a threshold order in a high friction priority mode.

6. The method of claim 1, further comprising:

extracting a second reference number of samples from the output of the neural network and detecting a maximum value from the second reference number of samples;

calculating a standard deviation of the second reference number of samples; and

calculating a sum of differences between neighboring samples with respect to the second reference number of samples as a relative value.

7. The method of claim 6, wherein the determining step comprises: changing the high friction road condition to a low friction road condition when a sum of the maximum value, the standard deviation, and the relative value in the high friction road condition exceeds a fourth threshold, and maintaining a current road condition when the sum of the maximum value, the standard deviation, and the relative value in the high friction road condition does not exceed the fourth threshold.

8. The method of claim 6, wherein the determining step comprises: maintaining a current road condition when the maximum value under a low friction road condition exceeds a fifth threshold value, and changing the low friction road condition to a high friction road condition when the maximum value under the low friction road condition does not exceed the fifth threshold value.

9. The method of claim 1, wherein the vehicle data includes at least one of Longitudinal Acceleration Sensor (LAS) data, Wheel Speed Sensor (WSS) data, accelerator pedal position sensor (APS) data, and steering wheel angle sensor (SAS) data.

10. The method of claim 9, wherein the preprocessing step comprises:

calculating a standard deviation of the third reference number of longitudinal acceleration sensor data;

obtaining a value of a third reference number by subtracting the average speed of the rear wheels from the average speed of the front wheels and calculating a standard deviation thereof;

obtaining a value of a third reference number by subtracting the average speed of the left wheel from the average speed of the right wheel and calculating a standard deviation thereof;

calculating an average of a third reference number of accelerator pedal position sensor data;

obtaining a third reference number of accelerator pedal position sensor data differences and calculating a sum thereof;

obtaining a third reference number of steering wheel angle sensor data difference values and calculating a sum thereof; and

an average of the third reference number of steering wheel angle sensor data is calculated.

11. A method of determining a road surface by a controller based on vehicle data, the method comprising the steps of:

preprocessing vehicle data;

sequentially extracting a first reference number of samples from the output of the neural network and calculating an average of the extracted samples; and

determining whether the road surface is a high friction road surface or a low friction road surface based on the calculated average value, wherein the determining step includes: changing the initial road condition to a low friction road condition when the average value under the initial road condition exceeds a first threshold value, and changing the initial road condition to a high friction road condition when the average value under the initial road condition does not exceed the first threshold value.

12. The method of claim 11, wherein the determining step further comprises:

changing the high friction road condition to the low friction road condition when the average value in the high friction road condition exceeds a second threshold value, and maintaining a current road condition when the average value in the high friction road condition does not exceed the second threshold value; and

maintaining a current road condition when the average value under the low friction road condition exceeds a third threshold value, and changing the low friction road condition to the high friction road condition when the average value under the low friction road condition does not exceed the third threshold value.

13. The method of claim 12, wherein the third threshold < the first threshold < the second threshold are satisfied in a threshold order in a stable priority mode.

14. The method of claim 12, wherein the first threshold < the third threshold < the second threshold are satisfied in a threshold order in a low friction priority mode.

15. The method of claim 12, wherein the third threshold < the second threshold < the first threshold are satisfied in a threshold order in a high friction priority mode.

16. The method of claim 11, further comprising:

calculating a standard deviation of the second reference number of samples; and

17. The method of claim 16, wherein the determining step comprises: changing the high friction road condition to a low friction road condition when a sum of the maximum value, the standard deviation, and the relative value in the high friction road condition exceeds a fourth threshold, and maintaining a current road condition when a sum of the maximum value, the standard deviation, and the relative value in the high friction road condition does not exceed the fourth threshold.

18. The method of claim 16, wherein the determining step comprises: maintaining a current road condition when the maximum value under a low friction road condition exceeds a fifth threshold value, and changing the low friction road condition to a high friction road condition when the maximum value under the low friction road condition does not exceed the fifth threshold value.

19. The method of claim 11, wherein the vehicle data includes at least one of Longitudinal Acceleration Sensor (LAS) data, Wheel Speed Sensor (WSS) data, accelerator pedal position sensor (APS) data, and steering wheel angle sensor (SAS) data.

20. The method of claim 19, wherein the preprocessing step comprises:

obtaining a third reference number of steering wheel angle sensor data differences and calculating a sum thereof; and