Method and device for detecting whether assisted driving is being spoofed

Abstract

The application relates to the field of assisted driving and discloses a method and device for detecting whether assisted driving is cheated, which comprises the following steps: acquiring a steering wheel torque in real time; subtracting a reference torque from the steering wheel torque to obtain a difference torque, wherein the reference torque is a steering wheel torque required for overcoming a current road environment and keeping a lane in the middle in the case that no intervention of a driver is applied, and the reference torque is calculated in real time based on various sensor data; and judging whether there is fine adjustment of random fluctuation according to data distribution characteristics of the difference torque in a preset time period, so as to judge whether cheating of assisted driving exists. Through the technical scheme, the cheating mode of the assisted driving system on the steering wheel is effectively identified, and the driving safety is improved.

Description

Technical Field

[0001] This invention relates to the field of driver assistance systems, and more particularly to a method and device for detecting whether a driver assistance system has been deceived. Background Technology

[0002] Advanced driver assistance systems (ADAS) are designed to require drivers to keep their hands on the steering wheel, focus on road conditions, and be ready to take over from the system at any time to deal with emergencies. However, some drivers over-rely on ADAS and stop holding the steering wheel after activating it. In this case, the torque sensor in the steering wheel can detect that the driver has taken their hands off the wheel and will issue a hands-off warning or even disengage the ADAS function.

[0003] However, in an attempt to deceive the driver assistance system's hands-off detection, drivers might attach heavy objects to the steering wheel, such as clay or water bottles, using the weight to mislead the system into thinking the driver is holding the wheel. Currently, driver attention can be detected, such as by using a cockpit camera to determine if the driver's gaze is on the road ahead. However, drivers might wear dark sunglasses, making visual detection impossible. Therefore, it is currently impossible to accurately determine whether a driver is attempting to deceive the driver assistance system.

[0004] In view of this, the present invention is hereby proposed. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and device for detecting whether assisted driving is being deceived, effectively identifying ways in which the assisted driving system is being deceived on the steering wheel, thereby improving driving safety.

[0006] This invention provides a method for detecting whether assisted driving has been deceived, the method comprising:

[0007] Real-time acquisition of steering wheel torque;

[0008] The difference torque is obtained by subtracting the reference torque from the steering wheel torque. The reference torque is calculated in real time based on data from multiple sensors and is the steering wheel torque required to overcome the current road conditions and maintain lane centering without driver intervention.

[0009] Based on the data distribution characteristics of the difference torque over a preset time period, it is determined whether there is any random fluctuation in the fine-tuning, thereby determining whether there is any deception of the assisted driving.

[0010] This invention provides an electronic device, the electronic device comprising:

[0011] Processor and memory;

[0012] The processor executes the steps of the detection method for whether the assisted driving has been deceived, as described in any embodiment, by calling the program or instructions stored in the memory.

[0013] The embodiments of the present invention have the following technical effects:

[0014] By acquiring the steering wheel torque in real time and subtracting the reference torque from it, the differential torque is obtained. The reference torque is calculated in real time based on data from multiple sensors and is the steering wheel torque required to overcome the current road environment and maintain lane centering without driver intervention. This differential torque is then used for deception detection. Furthermore, the data distribution characteristics of the differential torque over a preset time period are used to determine whether there are random fluctuations or fine adjustments, thereby identifying whether there is deception of the driver assistance system. This allows the vehicle to respond promptly and avoid dangerous driving situations, effectively identifying methods of deceiving the driver assistance system using the steering wheel, thus improving driving safety. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for detecting whether assisted driving has been deceived, provided by an embodiment of the present invention;

[0017] Figure 2 This is a flowchart of another method for detecting whether assisted driving has been deceived, provided by an embodiment of the present invention;

[0018] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0020] The method for detecting whether an assisted driving system has been spoofed, provided in this invention, is primarily applicable to preventing drivers from evading the steering wheel-off-hands detection of the assisted driving system. This method can be executed by the assisted driving system, vehicle controller, or other electronic devices.

[0021] Figure 1 This is a flowchart of a method for detecting whether assisted driving has been deceived, provided by an embodiment of the present invention. See also... Figure 1 The specific methods for detecting whether the assisted driving system has been deceived include:

[0022] S110: Real-time acquisition of steering wheel torque.

[0023] Among them, steering wheel torque is the torque measured by a torque sensor installed on the steering wheel.

[0024] Specifically, the steering wheel torque can be obtained in real time based on the torque sensor installed on the steering wheel.

[0025] S120. Subtract the reference torque from the steering wheel torque to obtain the difference torque.

[0026] The reference torque is calculated in real-time based on data from multiple sensors. It represents the steering wheel torque required to overcome the current road conditions and maintain lane centering without driver intervention. The differential torque is the steering wheel torque minus the reference torque. The differential torque largely eliminates the influence of driving environment factors such as slope, speed, and curve curvature, primarily reflecting the additional or active force applied to the steering wheel by the driver or a deceptive object.

[0027] Specifically, the reference torque is obtained in real time, and the steering wheel torque is used as the minuend to subtract the reference torque to obtain the difference torque at the current moment.

[0028] For example, a baseline torque curve is constructed based on the current navigation data, vehicle motion state data, and a baseline mathematical model corresponding to the current navigation information. The relevant physical information of the current road segment shown in the current navigation information may include the curvature angle, slope, and speed limit. The vehicle motion state data is the state data during vehicle movement and may include vehicle speed, steering wheel angle, and vehicle attitude (pitch angle, roll angle, steering angle, etc.). The current navigation data includes curvature data and slope data. The baseline mathematical model is trained using various training navigation data and the corresponding uninterrupted steering wheel torque, and is used to simulate the torque situation of the steering wheel without any intervention. The uninterrupted steering wheel torque is the steering wheel torque required to overcome the current road environment (such as centrifugal force during turning, gravity component during uphill driving, road friction, etc.) and maintain lane centering when no external force is applied to the steering wheel. Specifically, by inputting the current navigation data and vehicle motion state data corresponding to the current navigation information into the baseline mathematical model, the baseline torque at the current moment can be obtained.

[0029] S130. Based on the data distribution characteristics of the differential torque over a preset time period, determine whether there is any random fluctuation in the fine-tuning, thereby determining whether there is any deception of the assisted driving.

[0030] The preset time period is a pre-defined duration for using differential torque to detect deception. Data distribution characteristics can include time-domain and frequency-domain features. Assisted driving can be understood as a driver assistance system that can detect when the driver takes their hands off the steering wheel.

[0031] Specifically, the differential torque is collected over a preset time period, and the data distribution characteristics of each differential torque within that preset time period are analyzed. The data distribution characteristics are compared with a preset judgment threshold to determine whether there is any random fluctuation in fine-tuning, that is, whether the difference is too large. Based on the judgment result, it can be further determined whether there is any deception of assisted driving.

[0032] Based on the above example, the data distribution characteristic is the standard deviation of the difference torque data distribution. If the standard deviation is less than or equal to the first threshold, it is determined that there is deception of assisted driving.

[0033] The first threshold is a pre-set judgment threshold related to the standard deviation of the difference torque data distribution.

[0034] Specifically, the standard deviation of the differential torque within a preset time period is analyzed, and this standard deviation is compared with a first threshold to determine whether the standard deviation is less than or equal to the first threshold. If so, it is determined that there is deception of the assisted driving; otherwise, there is no deception of the assisted driving.

[0035] Based on the above example, a judgment can be made through time-domain analysis. Specifically, the difference torque forms a time-domain curve that fluctuates around a reference value within a preset time period. The reference value is a value determined by combining driving conditions and road conditions. The data distribution characteristic is the frequency at which the time-domain curve crosses the reference value. If the frequency is lower than or equal to a second threshold, it is judged that there is deception of the assisted driving.

[0036] The second threshold is a pre-set judgment threshold related to the frequency at which the time-domain curve crosses the reference value.

[0037] Specifically, the difference torques within a preset time period are arranged in chronological order to form a time-domain curve, which should fluctuate around a reference value. The frequency at which this time-domain curve crosses the reference value is used as a data distribution feature. This frequency is compared with a second threshold to determine whether the frequency is lower than or equal to the second threshold. If it is, it is determined that there is deception of the assisted driving system; otherwise, it is determined that there is no deception of the assisted driving system.

[0038] Based on the above example, a judgment can be made by frequency domain transformation on the basis of time domain analysis. Specifically, the difference torque forms a time domain curve in a preset time period. The difference torque spectrum is obtained by performing a Fourier transform on the time domain curve. The data distribution characteristics are the difference torque spectrum. If the distribution of the energy characteristics of the difference torque spectrum in the range of 0~aHz is greater than or equal to the third threshold, it is judged that there is deception of assisted driving.

[0039] The differential torque spectrum is the result of frequency domain transformation of the time-domain curve of the differential torque within a preset time period. The third threshold is a pre-set judgment threshold related to the energy distribution of the differential torque spectrum in the range of 0~aHz.

[0040] Specifically, the difference torque within a preset time period is plotted as a time-domain curve, i.e., the time-domain curve of the difference torque. A Fourier transform is performed on this time-domain curve to obtain the spectrum, which is the difference torque spectrum. In this case, the data distribution characteristic is the difference torque spectrum. Analyzing the energy characteristics of the difference torque spectrum, the energy distributed in the range of 0~aHz is determined. It is then judged whether this portion of energy is greater than or equal to a third threshold. If so, it is determined that there is deception of the assisted driving system; otherwise, it is determined that there is no deception of the assisted driving system.

[0041] Based on the example above, the range of a is [0.05, 0.15].

[0042] Based on the above example, the data distribution characteristics are at least one of the following:

[0043] Standard deviation σ, zero crossover rate ZCR, mean μ, , , ;

[0044] Wherein: standard deviation σ is the standard deviation of the data distribution of the difference moment; the difference moment forms a time-domain curve that fluctuates around a reference value within a preset time period, the reference value being a value determined by considering driving conditions and road conditions; zero crossover rate ZCR is the frequency at which the time-domain curve crosses the reference value; mean μ is the mean of the difference moment data; The first derivative of the standard deviation; Low-frequency energy ratio; Mid-frequency ratio; Low-frequency energy ratio This represents the ratio of low-frequency energy to mid-frequency energy in the spectrum obtained after Fourier transforming the time-domain curve, where mid-frequency accounts for a certain percentage. This represents the proportion of mid-frequency energy in the spectrum obtained after Fourier transforming the time-domain curve.

[0045] In this context, intermediate frequency energy refers to the energy within a preset intermediate frequency range, while low-frequency energy refers to the energy within a preset low-frequency range. The preset low-frequency range is the frequency range below the preset intermediate frequency range. For example, the preset intermediate frequency range is 0.1-2Hz, and the preset low-frequency range is 0-0.1Hz. The intermediate frequency ratio describes the proportion of energy used for manual fine-tuning. The low-frequency energy ratio measures the concentration of DC / ultra-low frequency energy.

[0046] Specifically, the energy distribution of the spectrum obtained after Fourier transforming the time-domain curve is analyzed to determine the mid-frequency energy and low-frequency energy. The larger of the mid-frequency energy and a preset minimum value is used as the denominator to avoid a denominator of zero. The low-frequency energy is then divided by the denominator to obtain the low-frequency energy ratio. Finally, the sum of the mid-frequency and low-frequency energies is used as the denominator, and the mid-frequency energy is used as the numerator to calculate the mid-frequency proportion.

[0047] For example, the preset mid-frequency range is 0.1-2Hz, and the preset low-frequency range is 0-0.1Hz. The low-frequency energy ratio and mid-frequency proportion are determined by the following formula:

[0048]

[0049]

[0050] in, For low-frequency energy ratio, For the proportion of mid-frequency frequencies, Low-frequency energy, It is medium-frequency energy. This is the preset minimum value.

[0051] For example, the standard deviation in the time-domain features can indicate that the standard deviation of the time-domain curve of the differential torque obtained by normal driving is usually significantly higher than that of the time-domain curve of the differential torque obtained by driving with a heavy suspension. The zero-crossing rate in the time-domain features is used because the torque frequently switches between positive and negative (left and right) during fine-tuning by human drivers, resulting in a high zero-crossing rate. Conversely, the zero-crossing rate of a heavy suspension is extremely low. Regarding the frequency-domain features, the spectrum corresponding to the time-domain curve during active driving distributes energy across multiple frequencies. The energy of the target spectrum used by a heavy suspension to deceive the driver assistance system is highly concentrated near 0Hz (DC component), with high-frequency components almost zero.

[0052] Based on the above example, if driving in a straight line, the following method can be used to determine whether there are random fluctuations in fine-tuning based on the data distribution characteristics of the differential torque over a preset time period, thereby determining whether there is deception of the assisted driving system:

[0053] Under straight-line driving conditions, the standard deviation σ is less than or equal to the fourth threshold, and the zero crossover rate (ZCR) is less than or equal to the fifth threshold. Less than or equal to the sixth threshold, and Greater than or equal to the seventh threshold or Less than or equal to the eighth threshold;

[0054] Alternatively, the standard deviation σ is less than or equal to the ninth threshold, and the zero crossover rate ZCR is less than or equal to the tenth threshold;

[0055] Alternatively, the mean μ is greater than or equal to the eleventh threshold, and the standard deviation σ is less than or equal to the twelfth threshold, and Greater than or equal to the thirteenth threshold;

[0056] Then it is determined that there is deception of the assisted driving system.

[0057] The fourth, fifth, sixth, seventh, eighth, ninth, tenth, eleventh, twelfth, and thirteenth thresholds are all determined based on experience or experimentation to determine the existence of deception in assisted driving. For example, the fourth threshold is 0.1, the fifth threshold is 0.2, the sixth threshold is 0.06, the seventh threshold is 2.0, the eighth threshold is 0.4, the ninth threshold is 0.05, the tenth threshold is 0.15, the eleventh threshold is 0.2, the twelfth threshold is 0.1, and the thirteenth threshold is 1.5.

[0058] Optionally, if driving in a straight line, the following method can be used to determine whether there are random fluctuations in fine-tuning based on the data distribution characteristics of the differential torque over a preset time period, thereby determining whether there is any deception of the assisted driving system:

[0059] Under straight-line driving conditions, the standard deviation σ is greater than the nineteenth threshold, and the zero crossover rate (ZCR) is greater than the twentieth threshold. Greater than or equal to the twenty-first threshold, and It is greater than or equal to the twenty-second threshold, and the mean μ is less than or equal to the twenty-third threshold;

[0060] Then it is determined that there is deception of the assisted driving system.

[0061] Among them, the nineteenth, twentieth, twenty-first, twenty-second, and twenty-third thresholds are all thresholds determined based on experience or experiments to determine whether there is any deception of the assisted driving system. For example, the nineteenth threshold is 0.2, the twentieth threshold is 1, the twenty-first threshold is 0.1, the twenty-second threshold is 0.6, and the twenty-third threshold is 0.3.

[0062] Understandably, drivers are most prone to distraction and deceptive behavior on long, straight sections of highways or urban expressways. In this scenario, the vehicle primarily maintains its lane, with minimal changes in steering wheel torque. During straight-line driving, the core decision-making logic remains based on the curve characteristics analysis of the time-domain curve of the differential torque. On an ideal, straight, level road surface, the baseline torque curve is theoretically zero. In reality, environmental factors such as road camber and crosswinds need to be compensated for. The ECU calculates a small, gently changing baseline torque based on the vehicle's lateral drift trend and the IMU attitude sensor to maintain the vehicle's precise trajectory. This torque is typically very small (e.g., 0.1-0.3 Nm). During normal driving (hands on the steering wheel), unconscious fine-tuning by the driver generates continuous, random, and high-frequency torque inputs. Therefore, the differential torque curve appears as a "noise" signal fluctuating around zero (or a very small mean), with a large standard deviation and a high zero-crossing rate. If a weight deception is used, the force applied by the weight is constant or changes slowly with slight vehicle swaying. The time-domain curve of the differential torque will be very smooth, close to a straight line (possibly non-zero, depending on whether the weight generates additional deflection torque), with a very small standard deviation and highly concentrated frequency domain energy at 0Hz. If the hands are completely released (no weight), the torque sensor reading will be very close to the system's calculated baseline torque curve. The amplitude of the time-domain curve of the differential torque is close to zero, and the standard deviation is even smaller than that of the "weight deception." This is a condition that can be identified by traditional hands-free detection and is not within the detection range of this example.

[0063] Based on the above example, if the vehicle is turning, the following method can be used to determine whether there are random fluctuations in the fine-tuning of the differential torque over a preset time period, thereby identifying any deception of the assisted driving system:

[0064] When turning

[0065] The standard deviation σ is less than or equal to the fourteenth threshold;

[0066] The zero crossover rate (ZCR) is less than or equal to the fifteenth threshold.

[0067] First derivative standard deviation Less than or equal to the sixteenth threshold; and

[0068] Less than or equal to the seventeenth threshold or Greater than or equal to the eighteenth threshold;

[0069] Then it is determined that there is deception of the assisted driving system.

[0070] Among them, the fourteenth, fifteenth, sixteenth, seventeenth, and eighteenth thresholds are all thresholds determined based on experience or experiments to determine whether there is deception of the assisted driving system. For example, the fourteenth threshold is 0.15, the fifteenth threshold is 0.5, the sixteenth threshold is 0.12, the seventeenth threshold is 0.45, and the eighteenth threshold is 1.5.

[0071] Understandably, when navigating curves, ramps, or making U-turns, the steering wheel requires a significant and continuously varying torque input. This is a critical scenario testing the driver's attention. In this scenario, accurate calculation of the baseline torque is crucial; it's a prerequisite for isolating environmental influences and extracting driver behavior characteristics. The ECU calculates the theoretical torque required to overcome steering resistance and centrifugal force in real time based on vehicle speed, steering wheel angle, navigation information (predicted curve curvature), and IMU data (vehicle roll angle). The baseline torque smoothly increases upon entering a curve, remains at a high level during the curve, and smoothly decreases upon exiting the curve, forming a predictable and smooth macroscopic torque curve. During normal driving (with hands on the steering wheel), the driver, while following the general trend of the baseline torque, adds fine, high-frequency adjustments to correct the line. The steering wheel torque is the superposition of the baseline torque and these fine adjustments. Therefore, the time-domain curve of the differential torque still presents a high-frequency, high-variance "noise" shape, reflecting the driver's active control. If a heavy object is used as a deception, the heavy object cannot provide intelligent, active torque. Steering wheel torque is mainly composed of the reference torque applied by the system and simple physical forces generated by the weight (such as the change of the gravity component with the steering angle). Therefore, the time-domain curve of the differential torque will be a very smooth, low-frequency curve, the shape of which is only related to the position of the weight and the steering wheel angle, and is not related to road condition fine-tuning.

[0072] The present invention has the following technical effects:

[0073] By acquiring the steering wheel torque in real time and subtracting the reference torque from it, the differential torque is obtained. The reference torque is calculated in real time based on data from multiple sensors and is the steering wheel torque required to overcome the current road environment and maintain lane centering without driver intervention. This differential torque is then used for deception detection. Furthermore, the data distribution characteristics of the differential torque over a preset time period are used to determine whether there are random fluctuations or fine adjustments, thereby identifying whether there is deception of the driver assistance system. This allows the vehicle to respond promptly and avoid dangerous driving situations, effectively identifying methods of deceiving the driver assistance system using the steering wheel, thus improving driving safety.

[0074] Figure 2 This is a flowchart of another method for detecting whether assisted driving has been deceived, provided by an embodiment of the present invention. See also... Figure 2 The specific methods for detecting whether the assisted driving system has been deceived include:

[0075] S210: Real-time acquisition of steering wheel torque.

[0076] S220. Subtract the reference torque from the steering wheel torque to obtain the difference torque.

[0077] S230. Based on the data distribution characteristics of the differential torque over a preset time period, determine whether there is any random fluctuation in the fine-tuning, thereby determining whether there is any deception of the assisted driving.

[0078] S240. If it is impossible to determine whether there is deception of the assisted driving based on the data distribution characteristics, then the steering wheel torque is input into a pre-trained large model, and the large model determines whether there is deception of the assisted driving.

[0079] Among them, the pre-trained large model can analyze the rationality of the differential torque to determine whether there is any deception of assisted driving.

[0080] Specifically, if the data distribution characteristics cannot determine whether there is deception of the assisted driving system, it indicates that the system is in a fuzzy range. The above methods cannot directly determine whether there is deception of the assisted driving system. Therefore, further analysis is required. Thus, the steering wheel torque can be input into a pre-trained large model, which can then determine whether there is deception of the assisted driving system.

[0081] For example, a pre-trained large model can analyze the similarity between the difference torque curve (the time-domain curve of the difference torque) and the target torque curve over a certain period of time to determine whether there is any deception of assisted driving.

[0082] The target torque curve includes a true torque curve and / or a dummy torque curve. The true torque curve simulates the torque experienced by a driver holding the steering wheel while driving on the road segment corresponding to the current navigation information. The dummy torque curve simulates the torque experienced by a vehicle traveling on the road segment corresponding to the current navigation information when a heavy object is suspended on the steering wheel.

[0083] Specifically, based on a pre-built torque curve simulation model, the current navigation information can be input into the model to obtain the real torque curve simulating the driver holding the steering wheel and / or the false torque curve simulating a heavy object suspended on the steering wheel.

[0084] Understandably, the torque curve simulation model can be trained based on the steering wheel torque curve obtained from the driver's actual driving under various collected navigation information and / or the steering wheel torque curve obtained from the suspension of heavy objects.

[0085] In the example above, before determining the corresponding target moment curve based on the current navigation information, the road condition scenario can be considered to trigger the subsequent detection method in this example. Specifically, it could be:

[0086] If the current navigation information meets the criteria for a high-risk scenario or the vehicle is on a shaking road surface, then the step of determining the corresponding target torque curve based on the current navigation information is triggered.

[0087] High-risk scenarios are those that require strict driver control, such as rapid deceleration, ramp merging, and continuous sharp turns.

[0088] Specifically, based on the current navigation information, it's determined whether the area is in a high-risk scenario. For example, navigation warnings such as "Construction area 2 km ahead, speed limit reduced from 120 km / h to 60 km / h," "Continuous sharp curves ahead," or "Approaching a merging ramp" all fall under the category of high-risk scenarios. Understandably, corresponding conditions can be set for high-risk scenarios, and these conditions can be configured as needed. If the current navigation information meets the criteria for a high-risk scenario, it indicates a complex road section requiring rigorous review, which necessitates triggering the step of determining the corresponding target moment curve based on the current navigation information.

[0089] Based on the above example, the corresponding target moment curve can be determined according to the current navigation information in the following way:

[0090] Input the current navigation information into a pre-trained real curve model to obtain the real torque curve corresponding to the current navigation information; and / or,

[0091] The current navigation information is input into a pre-trained spurious curve model to obtain the spurious torque curve corresponding to the current navigation information.

[0092] The true curve model is trained based on various training navigation information and the corresponding human intervention steering wheel torque curve, while the spurious curve model is trained based on various training navigation information and the corresponding weight-based steering wheel torque curve. The training navigation information includes navigation information under various road conditions and road segments. The human intervention steering wheel torque curve is the steering wheel torque curve obtained by the driver holding the steering wheel and controlling the vehicle on the road segment corresponding to the training navigation information. The weight-based steering wheel torque curve is the steering wheel torque curve obtained by a weight suspended on the steering wheel and controlling the vehicle on the road segment corresponding to the training navigation information.

[0093] Specifically, the current navigation information is input into a pre-trained real curve model. After the model analyzes the current navigation information, it can output a torque curve simulating the force exerted by a driver's hands on the steering wheel, which is the real torque curve corresponding to the current navigation information. And / or, the current navigation information is input into a pre-trained spoof curve model. After the model analyzes the current navigation information, it can output a torque curve simulating the force exerted by a heavy object suspended on the steering wheel, which is the spoof torque curve corresponding to the current navigation information.

[0094] Based on the above example, it is possible to determine whether there is any deception of the driver assistance system by analyzing the time-domain curve corresponding to the steering wheel torque and the target torque curve.

[0095] Specifically, the time-domain curve corresponding to the steering wheel torque is compared with the target torque curve to determine its similarity to the real torque curve and / or the dummy torque curve. Based on the similarity, it is determined whether there is deception of the driver assistance system. For example: if the target torque curve includes both the real torque curve and the dummy torque curve, if it is more similar to the real torque curve, then there is no deception of the driver assistance system; if it is more similar to the dummy torque curve, then there is deception of the driver assistance system. If the target torque curve includes the real torque curve, and the similarity is greater than the true similarity threshold, then there is no deception of the driver assistance system; otherwise, there is deception of the driver assistance system.

[0096] Based on the above example, the following methods can be used to more accurately determine whether there is deception of the driver assistance system by analyzing the time-domain curve corresponding to the steering wheel torque and the target torque curve:

[0097] The time-domain curve corresponding to the steering wheel torque and the target torque curve are input into a pre-trained curve matching large model to obtain the target similarity between the time-domain curve corresponding to the steering wheel torque and the target torque curve.

[0098] Based on target similarity, determine whether there is deception of the driver assistance system.

[0099] The curve matching large model is a comprehensive model used for matching curve similarity. It can analyze curves from multiple dimensions, such as curvature and inflection points. Target similarity is the similarity between the time-domain curve corresponding to the steering wheel torque and the target torque curve. If the target torque curve is either a true torque curve or a spurious torque curve, the target similarity is one; if the target torque curve is both a true torque curve and a spurious torque curve, the target similarity is two.

[0100] Specifically, the time-domain curve corresponding to the steering wheel torque and the target torque curve are input into a pre-trained curve matching model to obtain the target similarity between the time-domain curve corresponding to the steering wheel torque and the target torque curve. If the target similarity is one, a pre-set similarity threshold can be used to determine whether it is sufficiently similar to the target torque curve. If it is greater than or equal to the similarity threshold, it is sufficiently similar to the target torque curve. Based on the method of obtaining the target torque curve, it can be determined whether there is deception of the driver assistance system. That is, if the target torque curve is a true torque curve, it is determined that there is no deception of the driver assistance system; if the target torque curve is a false torque curve, it is determined that there is deception of the driver assistance system. If it is less than the similarity threshold, it is basically dissimilar to the target torque curve. Based on the method of obtaining the target torque curve, it can be determined whether there is deception of the driver assistance system. That is, if the target torque curve is a true torque curve, it is determined that there is deception of the driver assistance system; if the target torque curve is a false torque curve, it is determined that there is no deception of the driver assistance system. If there are two targets with similarity scores, the magnitude of the two similarity scores can be compared. Based on the acquisition method corresponding to the target with the larger similarity score, it can be determined whether there is any deception of the driver assistance system.

[0101] Based on the above example, if the target torque curve consists of both a true torque curve and a spoof torque curve, then the target similarity includes the first similarity between the time-domain curve corresponding to the steering wheel torque and the true torque curve, and the second similarity between the time-domain curve corresponding to the steering wheel torque and the spoof torque curve. The following method can be used to determine whether there is deception of the driver assistance system based on the target similarity:

[0102] Based on the first and second similarity scores, determine whether there is any deception of the driver assistance system.

[0103] The first similarity metric is the similarity between the time-domain curve of the steering wheel torque analyzed by the large model and the true torque curve. The second similarity metric is the similarity between the time-domain torque curve of the steering wheel analyzed by the large model and the spurious torque curve.

[0104] Specifically, the time-domain curve, true torque curve, and spurious torque curve corresponding to the steering wheel torque are input into a pre-trained curve matching model. The model analyzes the similarity between the time-domain curve and the true torque curve, using this as the first similarity score. It then analyzes the similarity between the time-domain curve and the spurious torque curve, using this as the second similarity score. By comparing the first and second similarity scores, it is determined whether the time-domain curve corresponding to the steering wheel torque is more similar to the true torque curve or the spurious torque curve, thereby determining whether deception of the driver assistance system exists. For example: if the first similarity score is less than a first threshold, deception of the driver assistance system is confirmed; if the second similarity score is greater than the second threshold, deception of the driver assistance system is confirmed. If the first similarity score is greater than or equal to the first threshold and the second similarity score is less than or equal to the second threshold, the first and second similarity scores are compared. If the first similarity score is greater than the second similarity score, no deception of the driver assistance system is confirmed; if the first similarity score is less than or equal to the second similarity score, deception of the driver assistance system is confirmed.

[0105] Based on the above example, in response to deception of assisted driving, the level of deception can be determined, and the current vehicle can be controlled according to the level of deception.

[0106] The deception level is used to distinguish different vehicle alert and takeover control strategies, such as low level, medium level and high level; it can also be level one, level two, level three and level four, and the specific number and division of levels can be set according to actual needs.

[0107] Specifically, if deception of the driver assistance system is confirmed, the level of deception needs further analysis. This can be determined by factors such as the frequency of deception, the similarity between the time-domain curve of steering wheel torque and the spurious torque curve, and other parameters. Specific rules can be set according to requirements. Furthermore, different handling strategies can be configured for different levels of deception, such as voice vibration alerts, alarm sounds, restrictions on driver assistance functions, and automatic vehicle takeover. Therefore, the vehicle can be controlled according to the handling strategy corresponding to the deception level to avoid driving hazards caused by prolonged hands-free driving.

[0108] Based on the above example, the level of deception can be determined in the following way, and the current vehicle can be controlled according to the level of deception:

[0109] Obtain the number of deceptions and the current deception confidence level within the first preset time period;

[0110] Based on the number of deceptions and the current confidence level of the deception, the deception level and the corresponding vehicle control operation are determined.

[0111] The first preset time period is a time interval used to measure the frequency of deceiving the driver assistance system, and it is a period of time adjacent to the current moment. The number of deceptions is the number of times deception of the driver assistance system is confirmed within the first preset time period. The current deception confidence level is the confidence level of the currently confirmed existence of deception of the driver assistance system; it can be a second similarity score. If deception can be confirmed without calculating the second similarity score, then the deception confidence level can be set to 1. Vehicle control operations include vehicle alert operations, function control operations, and driving control operations. Vehicle alert operations include alerts to the driver through sound and light. Function control operations include limiting comfort functions, such as stopping multimedia playback to prevent driver distraction, and also include downgrading driver assistance functions to return driving functions to the driver. Driving control operations include disengaging driver assistance functions and controlling the vehicle to detect surrounding conditions and stop.

[0112] Specifically, the number of deceptions within a first preset time period prior to the current moment is obtained, and the deception confidence level at the current moment is determined. Based on the number of deceptions and the current deception confidence level, and compared with pre-defined deception level conditions, the deception level can be obtained, and the corresponding pre-configured vehicle control operation can be determined according to the deception level.

[0113] Building upon the above example, after controlling the current vehicle based on the level of deception, further processing can be performed for security reasons, specifically:

[0114] In response to the current vehicle performing a driving control operation and stopping, the driver response status of the current vehicle is obtained;

[0115] If the driver's response status is "no response", the emergency call system will be triggered, and the vehicle data for the second preset time period will be encrypted and locked.

[0116] The driver response status refers to the driver's ability to respond, which can be used to determine if the driver is conscious and alert. The emergency call system can automatically connect to rescue organizations and provide information such as vehicle location and status to assist in rapid rescue. The second preset time period is used to retain and protect data for subsequent analysis and use as evidence. Vehicle data includes data on the surrounding environment while the vehicle is driving, as well as the vehicle's driving data. If the vehicle has health monitoring functions, it may also include the driver's physiological data.

[0117] Specifically, if the vehicle executes driving control operations and stops, it indicates that the driver assistance function has been severely compromised, possibly due to driver unconsciousness. Therefore, it is necessary to obtain the driver's response status, for example, by analyzing images captured by the in-vehicle camera or interacting with the driver through the voice system. If the driver is unresponsive, the emergency call system is triggered to call for assistance, and vehicle data for a second preset time period is encrypted and locked for data preservation, analysis, and use as evidence.

[0118] For example, if the number of deceptions within a first preset time period is 1 and the current deception confidence level is less than the first confidence level, then the deception level is determined to be 1. The triggered vehicle control operations include: 1. Multimedia reminder: reminding the driver through instrument panel text ("Please hold the steering wheel"), a gentle prompt tone, and a high-frequency weak vibration of the steering wheel; 2. Closed-loop interaction: the system issues a voice command, such as "To confirm that you are focused on driving, please turn the steering wheel slightly" or "Please click to confirm on the screen." The system will wait for the driver's valid operation as a response. If a response is received, the warning is lifted, but the system will increase the monitoring frequency in the short term. If the number of deceptions within the first preset time period is greater than 1 but less than the preset number, or the current deception confidence level is greater than or equal to the first confidence level but less than the second confidence level, then the deception level is determined to be 2. The triggered vehicle control operations include: 1. Warning escalation: the prompt tone becomes more urgent, and a conspicuous red warning indicator appears on the instrument panel; 2. Comfort function restriction: suspending the multimedia entertainment system and reducing the air conditioning fan speed to prompt the driver to focus on the driving task; 3. Driving assistance function downgrade: disabling automatic lane changing, disabling active overtaking, and disabling the speed increase set by the driver. The vehicle will still maintain its lane and follow the car in front, but its "degree of freedom" is limited, partially returning control to the driver. If the number of deceptions within a first preset time period is greater than or equal to the preset number, or the current deception confidence level is greater than or equal to the second confidence level, the deception level is determined to be 3. The triggered vehicle control actions include: activating hazard lights to warn vehicles behind; gradually and smoothly reducing speed; using blind spot monitoring and rear radar to determine if the right lane or emergency lane is safe; if absolutely safe, slowly changing lanes to the far right lane or emergency lane; if changing lanes is unsafe (e.g., heavy traffic, complex road conditions), continuing to decelerate within the current lane until stopping. On highways, this is an action that requires extremely careful handling, but it is a lower-risk option compared to an unmonitored vehicle traveling at high speed. When the vehicle comes to a complete stop or its speed is extremely low, the driver assistance system is officially disengaged, and the system may be locked, requiring a visit to a service center for reset. The second confidence level is greater than the first confidence level. If the vehicle is brought to a stop under Level 3 deception, but the driver remains unresponsive (indicated by an unresponsive driver status, possibly due to a medical emergency such as fainting), the following measures can be taken: Automatically trigger an ECALL (Emergency Call) to connect to the emergency services center and report the vehicle's location and the "driver unresponsive" status; lock the Event Data Recorder (EDR), which encrypts and locks all relevant vehicle data (steering wheel torque curve, video, navigation information, vehicle dynamic data, system response measures, etc.) before and after the deception assessment for subsequent accident analysis and liability determination.

[0119] Based on the above example, road surface conditions can be detected in advance to avoid the impact of differential torque caused by road vibration on the detection triggering effect. Specifically, this can be done as follows:

[0120] If the vehicle is on a shaking road surface, the process of inputting steering wheel torque into a pre-trained large model is triggered, and the large model determines whether there is any deception of the assisted driving.

[0121] Among them, a shaking road surface is a continuously bumpy road surface, such as a washboard road. The shaking road surface can be identified by a pre-collected set of sensors for various types of bumpy and shaking road surfaces, which are analyzed and judged based on a trained model.

[0122] Specifically, if the vehicle is currently on a shaking road surface, for example, if the vehicle's IMU or suspension sensors detect continuous strong bumps, since such bumps may contaminate the differential torque, and such bumps are similar to what is considered shaking, the judgment of data distribution characteristics may fail. Therefore, the process of directly triggering the execution of inputting the steering wheel torque into a pre-trained large model, and having the large model determine whether there is any deception of the assisted driving, is initiated.

[0123] The present invention has the following technical effects: by inputting the steering wheel torque into a pre-trained large model, the large model can determine whether there is deception of the assisted driving. In cases where it is not possible to use the data distribution characteristics of the differential torque over a preset time period to determine whether there is deception of the assisted driving, a more precise method is used to determine the deception. This achieves coverage of all driving scenarios, greatly improves processing efficiency, and ensures vehicle driving safety.

[0124] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. For example... Figure 3 As shown, the electronic device 300 includes one or more processors 301 and memory 302.

[0125] The processor 301 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 300 to perform desired functions.

[0126] The memory 302 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 301 may execute the program instructions to implement the detection method for whether assisted driving has been deceived, as described above in any embodiment of the present invention, and / or other desired functions. Various contents such as initial extrinsic parameters and thresholds may also be stored in the computer-readable storage medium.

[0127] In one example, the electronic device 300 may further include an input device 303 and an output device 304, these components being interconnected via a bus system and / or other forms of connection mechanisms (not shown). The input device 303 may include, for example, a keyboard, a mouse, etc. The output device 304 may output various information to the outside, including warning messages, braking force, etc. The output device 304 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0128] Of course, for the sake of simplicity, Figure 3 Only some of the components of the electronic device 300 relevant to the present invention are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device 300 may include any other suitable components depending on the specific application.

[0129] In addition to the methods and devices described above, embodiments of the present invention may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps of the method for detecting whether assisted driving has been deceived, provided in any embodiment of the present invention.

[0130] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of the present invention. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0131] Furthermore, embodiments of the present invention may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps of the method for detecting whether assisted driving has been deceived, provided in any embodiment of the present invention.

[0132] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0133] It should be noted that the terminology used in this invention is for describing specific embodiments only and is not intended to limit the scope of this application. As shown in this specification, unless the context clearly indicates otherwise, words such as "a," "an," "an," and / or "the" do not specifically refer to the singular and may include the plural. The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, or apparatus. Without further limitations, an element defined by the phrase "comprising an..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes said element.

[0134] It should also be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Unless otherwise expressly specified and limited, the terms "installed," "connected," "linked," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components. For those skilled in the art, the specific meaning of the above terms in the present invention can be understood according to the specific circumstances.

[0135] 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 technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting whether assisted driving has been deceived, characterized in that, include: Real-time acquisition of steering wheel torque; The difference torque is obtained by subtracting the reference torque from the steering wheel torque. The reference torque is calculated in real time based on data from multiple sensors and is the steering wheel torque required to overcome the current road environment and maintain lane centering without driver intervention. as well as Based on the data distribution characteristics of the differential torque over a preset time period, it is determined whether there is any random fluctuation in the fine-tuning, thereby determining whether there is any deception of the assisted driving; The data distribution characteristic is at least one of the following: Standard deviation σ, zero crossover rate ZCR, mean μ, , , ; Wherein: the standard deviation σ is the data distribution standard deviation of the difference moment; The differential torque forms a time-domain curve that fluctuates around a reference value during the preset time period. The reference value is a value determined by combining driving conditions and road conditions. The zero crossover rate ZCR is the frequency at which the time-domain curve crosses the reference value. The mean μ is the mean of the difference torque data; The first derivative of the standard deviation is given. Low-frequency energy ratio; This represents the proportion of mid-frequency frequencies; The low-frequency energy ratio The ratio of low-frequency energy to mid-frequency energy in the spectrum obtained after Fourier transforming the time-domain curve is the mid-frequency proportion. The proportion of mid-frequency energy in the spectrum obtained after the Fourier transform of the time-domain curve; The step of determining whether there is random fluctuation in fine-tuning based on the data distribution characteristics of the difference torque over a preset time period, thereby determining whether there is deception of the assisted driving, includes: When driving in a straight line The standard deviation σ is less than or equal to the fourth threshold, and the zero crossover rate ZCR is less than or equal to the fifth threshold, and the Less than or equal to the sixth threshold, and the Greater than or equal to the seventh threshold or the stated Less than or equal to the eighth threshold; Alternatively, the standard deviation σ is less than or equal to the ninth threshold, and the zero crossover rate ZCR is less than or equal to the tenth threshold; Alternatively, the mean μ is greater than or equal to the eleventh threshold, and the standard deviation σ is less than or equal to the twelfth threshold, and the... Greater than or equal to the thirteenth threshold; Then it is determined that there is deception of the assisted driving system.

2. The method according to claim 1, characterized in that, Also includes: If the data distribution characteristics cannot determine whether there is deception of the assisted driving system, the steering wheel torque is input into a pre-trained large model, which then determines whether there is deception of the assisted driving system.

3. The method according to claim 1, characterized in that, The step of determining whether there is random fluctuation in fine-tuning based on the data distribution characteristics of the difference torque over a preset time period, thereby determining whether there is deception of the assisted driving, includes: When turning The standard deviation σ is less than or equal to the fourteenth threshold; The zero crossover rate (ZCR) is less than or equal to the fifteenth threshold. The first derivative standard deviation Less than or equal to the sixteenth threshold; and The Less than or equal to the seventeenth threshold or Greater than or equal to the eighteenth threshold; Then it is determined that there is deception of the assisted driving system.

4. An electronic device, characterized in that, The electronic device includes: Processor and memory; The processor executes the steps of the method for detecting whether assisted driving has been deceived as described in any one of claims 1 to 3 by calling the program or instructions stored in the memory.

Images