CN-122009215-A - Early warning method for fatigue driving of driver and vehicle

CN122009215ACN 122009215 ACN122009215 ACN 122009215ACN-122009215-A

Abstract

The application discloses a method for early warning fatigue driving of a driver and the vehicle, the method comprises the steps of obtaining calibration driving data obtained after the driver finishes multiple times of preset calibration driving, wherein the preset calibration driving covers different driving scenes and driving time periods, obtaining historical driving data of the driver in a historical driving process, constructing a digital twin body corresponding to the driver based on the calibration driving data and the historical driving data, wherein the digital twin body comprises baseline data of the driver between a non-fatigue driving state and a fatigue driving state, determining the current fatigue level of the driver based on real-time scene environment data of current vehicle driving, real-time physiological state data and real-time driving behavior data of the driver and combining the digital twin body, and executing a grading intervention strategy matched with the digital twin body according to the current fatigue level. The technical scheme provided by the application can improve the accuracy of early warning of fatigue driving of the driver.

Inventors

SUN JIANWEI
LIU ZIHAO
WANG LINHAO
Xiong Hongcan
GAO XINPENG

Assignees

东风汽车集团股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260319

Claims (10)

1. The early warning method for fatigue driving of the driver is characterized by comprising the following steps: Obtaining calibration driving data obtained after a driver finishes multiple times of preset calibration driving, wherein the preset calibration driving covers different driving scenes and driving periods; Acquiring historical driving data of the driver in a historical driving process, and constructing a digital twin corresponding to the driver based on the calibration driving data and the historical driving data, wherein the digital twin comprises baseline data of the driver between a non-fatigue driving state and a fatigue driving state; Determining the current fatigue level of the driver by combining the digital twin body based on the real-time scene environment data of the current vehicle driving, the real-time physiological state data and the real-time driving behavior data of the driver; and executing a hierarchical intervention strategy matched with the digital twin body according to the current fatigue level.
2. The method of claim 1, wherein the constructing a digital twin corresponding to the driver based on the calibrated driving data and the historical driving data comprises: based on the calibration driving data, extracting physiological reference features and driving behavior reference features of the driver in a non-fatigue state to construct a digital twin corresponding to the driver; And dynamically updating the digital twin body based on the physiological state change data, the driving habit change data and the intervention feedback preference data in the historical driving data.
3. The method of claim 2, wherein after dynamically updating the digital twin, the method further comprises: Carrying out validity check on the baseline data in the updated digital twin body, and judging whether the baseline data is in a preset individual reasonable fluctuation range; if the baseline data exceeds the reasonable fluctuation range of the individual, discarding the update, and reserving the digital twin body of the previous version.
4. The method of claim 1, wherein the determining the current fatigue level of the driver in conjunction with the digital twins based on the real-time scene environment data of the current vehicle driving, the real-time physiological state data of the driver, and the real-time driving behavior data comprises: Inputting the real-time scene environment data, the calibration driving data and the historical driving data in the digital twin body into a gate control circulation unit neural network so as to predict the fatigue risk value of the driver in a fatigue driving state in real time; If the fatigue risk value exceeds a preset fatigue risk value threshold, comparing the characteristic data in the real-time physiological state data and the real-time driving behavior data with the baseline data, and calculating the real-time fatigue index of the driver through a convolution long-short-term memory fusion neural network; And determining the current fatigue level of the driver based on the real-time fatigue index and a preset grading threshold, wherein the preset grading threshold is associated with the age, the driving age and the historical driving habit data of the driver in the digital twin.
5. The method of claim 4, wherein prior to calculating the real-time fatigue index of the driver via a convolutional long-term memory fusion neural network, the method further comprises: based on the real-time scene environment data, identifying the scene type of the current running of the vehicle, wherein the scene type comprises a high-speed straight running scene, a urban congestion scene, a night running scene and a special road condition scene; based on the scene type, dynamically adjusting the weight duty ratio corresponding to each characteristic data in the real-time physiological state data and the real-time driving behavior data; And fusing the characteristic data based on the adjusted weight ratio so as to input the characteristic data into the convolution long-term and short-term memory fused neural network.
6. The method of claim 1, wherein said performing a hierarchical intervention strategy matching said digital twins in accordance with said current fatigue level comprises: if the current fatigue level is mild fatigue, executing a first-level intervention strategy, wherein the first-level intervention strategy comprises at least one of seat support dynamic adjustment, personalized fragrance release and mild voice prompt; If the current fatigue level is moderate fatigue, executing a second-level intervention strategy, wherein the second-level intervention strategy comprises at least one of seat vibration reminding, vehicle-mounted terminal dynamic reminding, navigation pushing near rest places and wearable equipment synchronous reminding; And if the current fatigue level is heavy fatigue, executing a third-level intervention strategy, wherein the third-level intervention strategy comprises at least one of seat combination vibration and lateral support tightening, grading audible and visual alarm, vehicle and road collaborative inquiry of an emergency parking area and notification of a preset emergency contact person.
7. The method of claim 6, wherein in performing a hierarchical intervention strategy matching the digital twins, the method further comprises: If the real-time physiological state data and the real-time driving behavior data show that the intervention has no obvious effect, the intervention intensity and the intervention dimension are gradually improved; And if the real-time physiological state data and the real-time driving behavior data show that the stress response of the driver is excessive, reducing the intervention intensity, and switching to an alternative intervention mode of the driver preference recorded in the digital twin body.
8. The method of claim 7, wherein after performing a hierarchical intervention strategy matching the digital twins, the method further comprises: Collecting manual scoring feedback data, driving operation feedback data and physiological state recovery feedback data of the driver, and constructing a triple feedback data set; based on the triple feedback data set, the digital twin is updated by an edge computing node local to the vehicle.
9. An early warning device for driver fatigue driving, the device comprising: the system comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring calibration driving data obtained after a driver finishes multiple times of preset calibration driving, and the preset calibration driving covers different driving scenes and driving time periods; A construction unit, configured to acquire historical driving data of the driver during the historical driving, and construct a digital twin corresponding to the driver based on the calibration driving data and the historical driving data, where the digital twin includes baseline data of the driver between a non-fatigue driving state and a fatigue driving state; The determining unit is used for determining the current fatigue level of the driver by combining the digital twin body based on the real-time scene environment data of the current vehicle driving, the real-time physiological state data and the real-time driving behavior data of the driver; and the execution unit is used for executing a hierarchical intervention strategy matched with the digital twin body according to the current fatigue level.
10. A vehicle comprising one or more processors and one or more memories, the one or more memories having stored therein at least one program code loaded and executed by the one or more processors to implement the method of any of claims 1-8.

Description

Early warning method for fatigue driving of driver and vehicle Technical Field The application belongs to the technical field of vehicles and driving monitoring, and particularly relates to a driver fatigue driving early warning method and a vehicle. Background With the rapid development of intelligent and networking technologies of automobiles, a vehicle-mounted active safety protection system is gradually perfected, and a fatigue driving early warning technology is used as a core technical means for preventing traffic accidents of a heavy extra-large road, is widely applied to various passenger and commercial intelligent vehicles, can trigger early warning intervention by identifying the fatigue state of a driver, and provides important technical support for driving safety. However, the current mainstream fatigue driving early warning technology still has a technical bottleneck which is difficult to break through. Most of the existing schemes set a fixed and uniform judgment threshold value based on fatigue characteristics of a general group, and fatigue state identification is completed only through facial characteristics or driving behavior data of a single dimension. The solidifying judging and early warning mode is extremely easy to cause the problems of high misjudgment rate and high missed judgment rate, can not accurately capture the real fatigue state of a driver, is difficult to realize effective fatigue prevention and control through an adaptive intervention strategy, and forms a remarkable potential threat to the safety of the driver driving the vehicle. Based on the above, how to improve the accuracy of early warning for fatigue driving of a driver has become a technical problem to be solved in the art. Disclosure of Invention The embodiment of the application provides a method and a device for early warning driver fatigue driving, a computer program product, a computer readable storage medium and a vehicle, and further can improve the early warning accuracy of the driver fatigue driving at least to a certain extent. Other features and advantages of the application will be apparent from the following detailed description, or may be learned by the practice of the application. According to a first aspect of the embodiment of the application, a method for early warning fatigue driving of a driver is provided, and the method comprises the steps of obtaining calibration driving data obtained after the driver finishes multiple times of preset calibration driving, wherein the preset calibration driving covers different driving scenes and driving periods, obtaining historical driving data of the driver in a historical driving process, constructing a digital twin corresponding to the driver based on the calibration driving data and the historical driving data, wherein the digital twin comprises baseline data of the driver between a non-fatigue driving state and a fatigue driving state, determining the current fatigue level of the driver based on real-time scene environment data of current vehicle driving, real-time physiological state data and real-time driving behavior data of the driver, and executing a grading intervention strategy matched with the digital twin according to the current fatigue level. In some embodiments of the application, based on the foregoing scheme, the constructing a digital twin corresponding to the driver based on the calibration driving data and the historical driving data includes extracting a physiological reference feature and a driving behavior reference feature of the driver in a non-fatigue state based on the calibration driving data to construct a digital twin corresponding to the driver, and dynamically updating the digital twin based on physiological state change data, driving habit change data and intervention feedback preference data in the historical driving data. In some embodiments of the present application, after dynamically updating the digital twin based on the foregoing scheme, the method further includes performing validity check on the baseline data in the updated digital twin, determining whether the baseline data is in a preset individual reasonable fluctuation range, and discarding the update and retaining the digital twin of the previous version if the baseline data exceeds the individual reasonable fluctuation range. In some embodiments of the present application, based on the foregoing scheme, the determining the current fatigue level of the driver by combining the real-time scene environment data based on the current vehicle driving, the real-time physiological state data of the driver and the real-time driving behavior data includes inputting the real-time scene environment data, the calibration driving data and the historical driving data in the digital twin to a gated loop unit neural network to predict the fatigue risk value of the driver entering the fatigue driving state in real time, comparing each feature data in the real-time physiologica