A driving fatigue recognition method based on wrist acceleration signal

Abstract

The application provides a driving fatigue recognition method based on wrist acceleration signals, comprising: collecting three axial acceleration signals of a driver's wrist to obtain three axial acceleration values; calculating a resultant acceleration value according to the three axial acceleration values; calculating power spectrum densities of the Y-axis acceleration value and the resultant acceleration value respectively; calculating powers of the Y-axis acceleration and the resultant acceleration in the 0-4hz frequency band range of the segmented time length according to the power spectrum densities respectively; and recognizing and warning driving fatigue according to the powers of the Y-axis acceleration and the resultant acceleration and a driving fatigue threshold value. The application can avoid the influence of the driver's personal driving habits and road conditions on the recognition of driving fatigue, and can realize the recognition of driving fatigue only by using the wrist acceleration signals, so that the analysis and calculation process is simple and convenient to implement.

Description

Technical Field

[0001] This invention relates to the field of motion signal processing technology, and more specifically to a method for driver fatigue recognition based on wrist acceleration signals. Background Technology

[0002] Driver fatigue is a significant factor contributing to road traffic accidents. Timely detection and early warning of driver fatigue in its early stages can prevent accidents and is of great practical importance to road traffic safety.

[0003] Currently, research on driver fatigue is extensive, including indirect methods such as lane departure and steering wheel angle measurement, as well as direct measurement of driver physiological signals and behavioral characteristics to determine driver fatigue. Indirect measurement methods are easily affected by road conditions and individual driver habits, limiting their practical application. Direct measurement methods, on the other hand, can directly monitor the driver's physiological or physical states, enabling timely identification and warning when driver fatigue occurs, thus offering better results.

[0004] Wrist acceleration signals can reflect behavioral changes in drivers, and driver fatigue affects wrist movement characteristics; analysis of wrist acceleration signals can identify driver fatigue levels. However, when using wrist acceleration signals to identify driver fatigue, individual driving habits (such as the amplitude and duration of steering wheel turns) and road conditions may influence the identification of driver fatigue. Summary of the Invention

[0005] To address the shortcomings of existing technologies and to avoid the influence of drivers' personal driving habits and road conditions on the identification of driver fatigue, this invention proposes a method for driver fatigue identification by performing frequency domain analysis on wrist acceleration signals.

[0006] The technical solution adopted in this invention is as follows:

[0007] Firstly, a method for driver fatigue recognition based on wrist acceleration signals is provided, including:

[0008] Three axial acceleration signals from the driver's wrist were collected to obtain the acceleration values ​​for the three axes.

[0009] Calculate the resultant acceleration value based on the acceleration values ​​in the three axes;

[0010] Calculate the power spectral density of the Y-axis acceleration and the resultant acceleration, respectively.

[0011] Calculate the power of the Y-axis acceleration and the resultant acceleration in the 0-4 Hz frequency band within the segmented time length based on the power spectral density.

[0012] Based on the power of the Y-axis acceleration and the power of the resultant acceleration, combined with the driver fatigue threshold, driver fatigue is identified and warned.

[0013] Furthermore, the driver fatigue threshold includes:

[0014] When the power value of the Y-axis acceleration is greater than 37.9, the driver is in a state of fatigue.

[0015] When the power value of the Y-axis acceleration is greater than or equal to 33.2 and less than or equal to 37.9, the driver is in the transition period;

[0016] When the power value of the Y-axis acceleration is less than 33.2, the driver is in a conscious state.

[0017] Furthermore, if the power value of the combined acceleration is less than 10 for two consecutive minutes, a warning will be activated to remind the driver to be aware of fatigue.

[0018] Furthermore, the three axes include the three directions in the overall human body coordinate system: the midline horizontal axis, the midline sagittal axis, and the midline vertical axis.

[0019] Furthermore, the power spectral density is calculated using the multi-window method based on the acceleration value.

[0020] Furthermore, the power in the 0-4 Hz frequency band with a segmented time length is calculated using the composite Simpson algorithm based on the power spectral density.

[0021] Furthermore, the segment duration is greater than or equal to 15 seconds and less than or equal to 60 seconds.

[0022] Furthermore, before calculating the acceleration values ​​in the three axes, the acceleration signals in the three axes are filtered and denoised.

[0023] Secondly, a driver fatigue recognition device based on wrist acceleration signals is provided, comprising:

[0024] An accelerometer is used to collect acceleration signals from the driver's wrist in three axes.

[0025] The calculator, wirelessly connected to the accelerometer, is used to receive acceleration signals from three axes and calculate the power of the Y-axis acceleration and the power of the resultant acceleration based on the acceleration signals; it is also used to identify and warn of driver fatigue based on the power combined with a driver fatigue threshold.

[0026] Furthermore, the accelerometer is a smart bracelet, and the calculator is a smartphone.

[0027] As can be seen from the above technical solution, the beneficial technical effects of the present invention are as follows:

[0028] 1. Driver fatigue can be identified using only wrist acceleration signals, and the analysis and calculation process is simple and easy to implement;

[0029] 2. It does not require the collection of multiple physiological and driving behavior-related signals, and has strong anti-interference capabilities;

[0030] 3. The recognition interval can be set as needed, and the recognition is real-time. Attached Figure Description

[0031] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0032] Figure 1 This is a flowchart of the driver fatigue recognition method according to Embodiment 1 of the present invention;

[0033] Figure 2 This is a schematic diagram of the driver fatigue recognition device architecture according to Embodiment 3 of the present invention. Detailed Implementation

[0034] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0035] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0036] Example 1

[0037] This embodiment provides a method for driver fatigue recognition based on wrist three-axis acceleration signals. It includes the following steps:

[0038] 1. Collect acceleration signals from the driver's wrist in three axes and calculate the acceleration values ​​in each of the three axes.

[0039] In a specific implementation, considering that the driver's right hand will operate the gear shift lever to change gears from time to time, it is preferable to collect the three-axis acceleration signal of the driver's left wrist; the sampling frequency is 64Hz, which can obtain a good quality acceleration signal.

[0040] The acceleration signals of the driver's wrist in three axes (X, Y, and Z axes) were measured using a MEMS accelerometer (-2g, +2g) and converted into acceleration values ​​using the following formula:

[0041]

[0042] Where x is the converted acceleration value, X is the original X-axis data measured by the accelerometer, and Fs is the range of the accelerometer (2g); the acceleration values ​​y and z along the Y and Z axes are obtained using the same method.

[0043] 2. Calculate the resultant acceleration value based on the acceleration values ​​along the three axes.

[0044] In a specific implementation, the resultant acceleration value is calculated using the following formula:

[0045]

[0046] Among them, A T The resultant acceleration value is represented by x, y, and z, which represent the three directions of the overall human body coordinate system: the horizontal axis, the sagittal axis, and the vertical axis, respectively, i.e., the acceleration values ​​in the (X, Y, Z directions) mentioned above.

[0047] 3. Calculate the acceleration value along the Y-axis and the power spectral density of the resultant acceleration value.

[0048] The power spectral density is calculated using the multi-window method. This method generates a series of windowed data using a discrete long-sphere sequence composed of a series of orthogonal conical windows, and calculates the average of the periodograms of these windowed data as a spectral estimate of the signal. In addition to orthogonality, these conical windows also possess optimal time-frequency convergence characteristics. Therefore, the spectral estimate obtained by the multi-window method has small variance and high frequency resolution.

[0049] 4. Calculate the Y-axis acceleration and the power of the resultant acceleration in the 0-4 Hz frequency band within the segmented time length based on the power spectral density.

[0050] The power of acceleration along each of the three axes is calculated based on the power spectral density of the three axes; and the power of the resultant acceleration is calculated based on the power spectral density of the resultant acceleration. The composite Simpson's rule is used to calculate the power in segments with time lengths of 15-60 seconds and a specific frequency range of 0-4 Hz.

[0051] The segment time length is set to 15-60 seconds to take into account the need to make a judgment on whether the driver is fatigued within a short time interval. The preferred segment time length is 30 seconds. If necessary, the segment time length can be further shortened to match the sampling time of the accelerometer to meet higher real-time requirements.

[0052] The power in the 0-4Hz frequency band is calculated because the power in this band varies significantly depending on whether the driver is driving normally or experiencing fatigue.

[0053] In a specific implementation, steps 3 and 4 can be implemented in MATLAB.

[0054] 5. Based on the power of the Y-axis acceleration and the power of the resultant acceleration, combined with the driver fatigue threshold, driver fatigue can be identified and warned.

[0055] The driver fatigue threshold was determined using statistical methods and experimental data. The specific experimental process is as follows:

[0056] Multiple healthy volunteers were recruited as drivers, covering all eligible age groups. Volunteers participated in two driving simulation experiments: one under normal rest conditions, and the other under conditions of insufficient sleep (4-6 hours of sleep) followed by fatigued driving. MEMS accelerometers were used to measure the acceleration data of the driver's wrist along three axes, constructing datasets of wrist movement behavior under both awake and fatigued states. Analysis using existing statistical methods revealed significant differences in the power of Y-axis acceleration and the power of the resultant acceleration between fatigued and awake states. The driving fatigue threshold was set as follows:

[0057] 1) When the power value of the Y-axis acceleration is greater than 37.9, the driver's wrist movement speed slows down, the low-frequency power of the signal increases, and the driver is in a state of fatigue.

[0058] 2) When the power value of the Y-axis acceleration is 33.2-37.9, the driver is in the transition period;

[0059] 3) When the power value of the Y-axis acceleration is less than 33.2, the driver's wrist movement changes rapidly, the low-frequency power decreases, and the driver is in a conscious state;

[0060] 4) When the power value of the combined acceleration is less than 10 for 2 consecutive minutes, the driver's wrist movement changes rapidly decrease, indicating that the driver may become fatigued in a short period of time. At this time, the warning is activated to remind the driver to pay attention to fatigue.

[0061] The technical solution of this embodiment can identify driver fatigue using only wrist acceleration signals. The analysis and calculation process is simple and easy to implement. It does not require the collection of multiple physiological and driving behavior-related signals, has strong anti-interference capabilities, and the identification interval can be set as needed, resulting in good real-time performance.

[0062] Example 2

[0063] To further improve the accuracy of identifying driver fatigue, based on Example 1, the three axial acceleration signals collected by the acceleration sensor are filtered and noise-reduced.

[0064] In a specific implementation, a filter is added to the output of the accelerometer to remove interference. The filter can be a Butterworth low-pass filter of order 8 with a cutoff frequency of 40Hz.

[0065] Example 3

[0066] This embodiment provides a driver fatigue recognition device based on wrist acceleration signals, which works on the principle of the technical solution provided in Embodiment 1, and includes: an acceleration sensor and a calculator.

[0067] An accelerometer is used to collect acceleration signals from the driver's wrist in three axes. In a specific implementation, a MEMS accelerometer with a range of (-2g, +2g) is selected as the accelerometer. The MEMS accelerometer is worn on the driver's left wrist, and its implementation is not limited, and can be implemented in any feasible manner in the prior art, such as a smart bracelet.

[0068] The calculator, wirelessly connected to an accelerometer, receives acceleration signals along three axes and calculates the power of the Y-axis acceleration and the power of the resultant acceleration based on these signals. It then uses this power, combined with a driver fatigue threshold, to identify and warn of driver fatigue. In specific implementations, its form is not limited and can be implemented in any feasible manner according to existing technology, such as a smartphone wirelessly connected to a smart bracelet.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for driver fatigue recognition based on wrist acceleration signals, characterized in that, include: Collect acceleration signals from the driver's wrist in three axes and calculate the acceleration values ​​in the three axes; Calculate the resultant acceleration value based on the acceleration values ​​in the three axes; Calculate the power spectral density of the Y-axis acceleration and the resultant acceleration, respectively. Calculate the power of the Y-axis acceleration and the resultant acceleration in the 0-4 Hz frequency band within the segmented time length based on the power spectral density. Based on the power of the Y-axis acceleration and the power of the resultant acceleration, combined with the driver fatigue threshold, driver fatigue is identified and warned. The driver fatigue threshold includes: when the power value of the Y-axis acceleration is greater than 37.9, the driver is in a state of fatigue; When the power value of the Y-axis acceleration is greater than or equal to 33.2 and less than or equal to 37.9, the driver is in a transition period; when the power value of the Y-axis acceleration is less than 33.2, the driver is in a conscious state; when the power value of the combined acceleration is less than 10 for 2 consecutive minutes, a warning is activated to remind the driver to pay attention to fatigue.

2. The driver fatigue recognition method according to claim 1, characterized in that, The three axes include the central horizontal axis, the central sagittal axis, and the central vertical axis in the overall human body coordinate system.

3. The driver fatigue recognition method according to claim 1, characterized in that, The power spectral density is calculated using the multi-window method based on the acceleration value.

4. The driver fatigue recognition method according to claim 1, characterized in that, The power in the 0-4 Hz frequency band with a segmented time length is calculated using the composite Simpson algorithm based on the power spectral density.

5. The driver fatigue recognition method according to claim 1, characterized in that, The segment time length is greater than or equal to 15 seconds and less than or equal to 60 seconds.

6. The driver fatigue recognition method according to claim 5, characterized in that, Before calculating the acceleration values ​​in the three axes, the acceleration signals in the three axes are filtered and denoised.

7. A driver fatigue recognition device based on wrist acceleration signals, characterized in that, The driver fatigue recognition method based on wrist acceleration signal as described in any one of claims 1-6 is used for driver fatigue recognition, wherein the driver fatigue recognition device comprises: An accelerometer is used to collect acceleration signals from the driver's wrist in three axes. The calculator, wirelessly connected to the accelerometer, is used to receive acceleration signals from three axes and calculate the power of the Y-axis acceleration and the power of the resultant acceleration based on the acceleration signals; it is also used to identify and warn of driver fatigue based on the power combined with a driver fatigue threshold.

8. The driver fatigue detection device according to claim 7, characterized in that, include: The accelerometer is a smart bracelet, and the calculator is a smartphone.

Images