Man-machine co-driving detection method and device based on steering wheel torque and medium

Abstract

The invention discloses a man-machine co-driving detection method and device based on steering wheel torque and a medium. The method comprises the steps that a steering wheel torque accumulated value of a target vehicle is obtained; performing nonlinear normalization processing on the steering wheel torque accumulated value to obtain a torque confidence coefficient, and if the torque confidence coefficient is greater than or equal to a driver take-over intervention threshold value, calculating to obtain a comprehensive confidence coefficient by combining the torque confidence coefficient and the correction confidence coefficient; and if the comprehensive confidence is greater than or equal to the driver take-over final judgment threshold, judging that the driver of the target vehicle intervenes in driving. According to the man-machine co-driving detection method and device based on the torque of the steering wheel and the medium, the extensive judgment mode of a single torque threshold value is abandoned through a progressive technical means in accordance with the real operation characteristics of a driver, layer-by-layer accurate screening and verification logic is formed from signal collection to comprehensive judgment, and the detection accuracy is improved. The problem that whether a driver actually intervenes in vehicle driving operation is difficult to accurately recognize and judge when auxiliary driving is started can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of automobiles, and in particular to a method and device for detecting human-machine co-driving based on steering wheel torque, and a medium. BACKGROUND

[0002] The current vigorous development of the new energy industry promotes the automobile automatic driving technology to become the focus of attention in the industry, but due to the limitations of technology and regulations, it is difficult to achieve complete automatic driving, and human-machine co-driving has become the current mainstream trend. The switching of the driving subject between the vehicle and the driver is the core issue of human-machine co-driving, and the torque sensing steering wheel becomes the mainstream component due to its low cost and mature technology, and its torque also becomes an important basis for determining the intervention and withdrawal of the driver co-driving. The current human-machine co-driving intervention and withdrawal detection methods mainly have two types: one is to use a fixed steering wheel torque value as the determination time point, and to complete the intervention or withdrawal determination after maintaining the corresponding time; the other is to set corresponding torque values according to different vehicle speeds and curvatures by combining vehicle speed and road curvature, and to realize more scene-adaptive intervention and withdrawal determination by establishing a three-dimensional MAP table of vehicle speed-curvature-torque.

[0003] However, when the auxiliary driving is started, the method of determining by a fixed steering wheel torque value cannot adapt to different driving styles, and the fixed maintenance time can easily cause the auxiliary driving to intervene in advance, which has the risk of steering wheel confrontation; although the method combining vehicle speed and road curvature is scene-adaptive, it needs to establish a three-dimensional MAP table, the dimension of the calibration parameters is high, and it also needs to be recalibrated with the change of vehicle type, EPS hardware or driving style, the adaptation cost is high, and it is difficult to accurately and reliably determine whether the driver truly intervenes in the driving operation. SUMMARY

[0004] The present application provides a human-machine co-driving detection method and device based on steering wheel torque to solve the problem of difficulty in accurately identifying and determining whether the driver truly intervenes in the driving operation of the vehicle when the auxiliary driving is started.

[0005] To achieve the above-mentioned purpose, the present application provides a human-machine co-driving detection method based on steering wheel torque, comprising: obtaining a steering wheel torque cumulative value of a target vehicle within a discrete sampling period when an auxiliary driving function is started; performing nonlinear normalization processing on the steering wheel torque cumulative value to obtain a torque confidence, and calculating a comprehensive confidence by combining the torque confidence and a correction confidence if the torque confidence is greater than or equal to a driver takeover intervention threshold value; determining that the driver of the target vehicle intervenes in driving if the comprehensive confidence is greater than or equal to a driver takeover final determination threshold value.

[0006] The application first acquires a steering wheel torque cumulative value in a discrete sampling period instead of a single point torque value, effectively filters out transient invalid torque signals caused by road bumps, steering mechanism shaking and hand touch, only captures the effective action of the driver continuously exerting torque, matches the behavior characteristics of real intervention driving, and guarantees the reliability of the determination from the collection source. Meanwhile, the torque cumulative value is subjected to nonlinear normalization to obtain a torque confidence, which can accurately convert the differentiated torque physical quantity into a quantified intervention intention index; when the torque confidence is greater than or equal to a driver takeover intervention threshold, it indicates that the torque continuously exerted by the driver has met the basic conditions of intervention determination, completes the first layer of effective screening, excludes invalid torque signal interference, confirms that the intervention intention has preliminary rationality, and enters the subsequent multi-dimensional verification link. Thereafter, the comprehensive confidence is calculated in combination with the correction confidence, which makes up for the determination deviation of a single torque dimension, realizes multi-dimensional complementary verification of the environment, the driver state and the torque intention; when the comprehensive confidence is greater than or equal to a driver takeover final determination threshold, it indicates that the intervention intention of the driver has been completely confirmed after multi-dimensional verification, reaches the final takeover determination standard, and completes the second layer of accurate determination. This double-layer threshold determination logic not only sensitively identifies the operation of the driver's active takeover, but also avoids misjudgment, accurately distinguishes real intervention from invalid signal interference, and realizes accurate and reliable determination of the real intervention behavior of the driver.

[0007] Compared with the prior art, the application effectively solves the problem of accurate determination of the real intervention driving operation of the driver in assisted driving by using progressive technical means that fit the real operation characteristics of the driver, and the core is to abandon the extensive determination mode of a single torque threshold, and form layer-by-layer accurate screening and verification logic from signal collection to comprehensive determination, so that the problem that it is difficult to accurately identify and determine whether the driver truly intervenes in the driving operation of the vehicle when the assisted driving is started can be solved.

[0008] As a preferred scheme, the correction confidence is obtained in the following manner: The relative distance and relative speed between the target vehicle and the front obstacle are collected in real time, the collision time is calculated according to the relative distance and the relative speed, the collision time is substituted into an exponential function designed based on a forward collision warning system test standard for calculation and amplitude limiting processing to obtain an environmental collision risk confidence; The eyelid closure ratio of the driver in a preset time window is linearly mapped to obtain a driver fatigue confidence; The smaller one of the environmental collision risk confidence and the driver fatigue confidence is selected as the correction confidence.

[0009] The preferred embodiment obtains a correction confidence by fusing the environmental collision risk and the driver fatigue state, and realizes accurate multi-dimensional consideration of the driving scene. The exponential function designed based on the forward collision warning system test standard calculates the environmental collision risk confidence, ensuring the authority and accuracy of the environmental risk assessment, and accurately quantifying the safety hazards brought by the front obstacles; the driver fatigue confidence is obtained by linear mapping of the eyelid closure ratio, which can capture the influence of the driver's physiological state on the driving intervention ability in real time. Selecting the smaller value of the two as the correction confidence can effectively avoid the one-sidedness of single-dimensional evaluation and avoid correction deviation caused by environmental risk misjudgment or driver state neglect, providing a reliable correction basis for subsequent comprehensive confidence calculation and improving the safety and rationality of human-machine co-driving detection.

[0010] As a preferred embodiment, the driver takeover intervention threshold is obtained in the following manner: Nonlinear vehicle speed adaptive modeling is performed according to the real-time vehicle speed of the target vehicle and a speed gain coefficient, the influence of vehicle speed on driver steering torque demand is quantified, and the vehicle speed nonlinear correction term is obtained by dynamically adjusting the takeover intervention threshold; Linear curvature adaptive modeling is performed according to the road curvature of the current driving road of the target vehicle and a curvature gain coefficient, the influence of road curvature on steering torque demand is quantified, and the road curvature linear correction term is obtained; The driver takeover intervention threshold is calculated according to the vehicle speed nonlinear correction term and the road curvature linear correction term in combination with a preset basic torque parameter.

[0011] The preferred embodiment quantifies the influence of vehicle speed on steering torque demand through nonlinear vehicle speed adaptive modeling, solving the problem that a fixed threshold value cannot match the actual operation demand of the driver at different speeds, such as dynamically adjusting the threshold value to improve the judgment sensitivity when driving at high speed; the influence of road curvature is quantified through linear curvature adaptive modeling, which can accurately adapt to the differences in steering torque demand in different road scenes such as curves and straight roads. The threshold value is calculated in combination with the preset basic torque parameter and the two correction terms, which not only retains the stability of the basic parameter, but also compensates for the scene limitations of the fixed threshold value through dynamic correction, making the takeover intervention threshold more suitable for actual driving conditions, reducing misjudgment and omission, and improving the coordination of human-machine interaction.

[0012] As a preferred embodiment, the driver takeover final judgment threshold is obtained in the following manner: Linear vehicle speed adaptive modeling is performed according to the real-time vehicle speed of the target vehicle and a speed gain compensation coefficient, the influence of vehicle speed on driver takeover judgment sensitivity is quantified, and the vehicle speed linear compensation term is obtained; The driver takeover final judgment threshold is calculated according to the vehicle speed linear compensation term in combination with a preset basic threshold parameter and an environmental penalty correction term.

[0013] The driver takeover final decision threshold acquisition system constructed by the preferred scheme realizes the double adaptation of the threshold to the vehicle speed and the environment. The vehicle speed linear compensation term is obtained by linear vehicle speed adaptation modeling, and the decision sensitivity can be adjusted according to the real-time vehicle speed, such as reducing the sensitivity to avoid unnecessary intervention decision when driving at low speed, and increasing the sensitivity to ensure safety when driving at high speed. The final threshold is calculated by combining the preset basic threshold and the two correction terms, which takes into account the uniformity of the basic decision standard and the flexibility of the scene adaptation, ensuring that the takeover decision under different vehicle speeds and different environments can meet the safety requirements, and improving the scene adaptability and reliability of the detection method.

[0014] As a preferred scheme, the steering wheel torque cumulative value is subjected to nonlinear normalization processing to obtain a torque confidence, specifically: The steering wheel torque cumulative value and the vehicle calibration constant are subjected to ratio operation, and the calculated ratio value is subjected to nonlinear mapping to obtain a mapping result. The mapping result is subjected to scaling and offset calibration to obtain the torque confidence.

[0015] The preferred scheme can eliminate the evaluation deviation caused by the torque parameter difference of different vehicle models through the ratio operation of the steering wheel torque cumulative value and the vehicle calibration constant. The nonlinear mapping processing can accurately fit the nonlinear relationship between the torque cumulative value and the driver intervention intention, solving the problem that linear processing cannot accurately reflect the correlation between the torque signal and the intervention intention. The system error and the sampling deviation can be further offset, so that the final torque confidence can truly and accurately quantify the intervention intention of the driver through the steering wheel torque, providing high-quality core data support for subsequent comprehensive judgment and improving the core precision of human-machine co-driving detection.

[0016] As a preferred scheme, the comprehensive confidence is calculated by combining the torque confidence and the correction confidence, specifically: The torque confidence and the correction confidence are subjected to weighted calculation according to the first weight coefficient corresponding to the torque confidence and the second weight coefficient corresponding to the correction confidence, to obtain the comprehensive confidence; wherein the sum of the first weight coefficient and the second weight coefficient is one.

[0017] The preferred scheme can allocate weights according to the importance of the torque signal and the correction signal by setting the first and second weight coefficients and ensuring that the sum of the two is one, so that the comprehensive confidence highlights the torque signal as the core intervention basis, and also takes into account the auxiliary correction information such as the environment and the driver state. This weighted fusion method avoids the limitations of a single data dimension. For example, when the torque signal is weak but the environmental risk is high, the weight adjustment can highlight the influence of the correction confidence; when the torque signal is clear, its core decision-making position is retained. The comprehensive confidence obtained ultimately more comprehensively and objectively reflects the driver's intervention willingness and the driving scene risk, and improves the scientificity and accuracy of the takeover decision.

[0018] As a preferred scheme, if the camera for detecting the driving environment of the vehicle, the radar or the driver monitoring system for detecting the driver state fails, the second weight coefficient is zero, and it is determined that the comprehensive confidence is equal to the torque confidence.

[0019] The preferred scheme designs a weight self-adaptive adjustment mechanism for the equipment failure scene, which significantly improves the robustness and fault tolerance of the detection method. When the environmental detection equipment or the driver monitoring system fails, the correction confidence data will lose reliability. At this time, the second weight coefficient is set to zero and the torque confidence is used to replace the comprehensive confidence, which can effectively avoid the interference of faulty data on the decision result and avoid misjudgment caused by unreliable correction information. At the same time, this design ensures the continuity of the detection function in the case of equipment failure, and does not need to suspend the man-machine co-driving detection due to equipment failure, ensuring continuous monitoring of the driver's intervention state during driving, balancing system stability and driving safety, and making the detection method more suitable for complex equipment working conditions in actual driving.

[0020] The application also provides a man-machine co-driving detection device based on steering wheel torque, which comprises an acquisition module, a calculation module and a determination module. The acquisition module is configured to obtain the steering wheel torque cumulative value of the target vehicle within a discrete sampling period when the auxiliary driving function is turned on. The calculation module is configured to perform nonlinear normalization processing on the steering wheel torque cumulative value to obtain a torque confidence, and calculate a comprehensive confidence by combining the torque confidence and a correction confidence if the torque confidence is greater than or equal to a driver takeover intervention threshold. The determination module is configured to determine that the driver of the target vehicle intervenes in driving if the comprehensive confidence is greater than or equal to a driver takeover final determination threshold.

[0021] The application also provides a storage medium having a computer program stored thereon, wherein the computer program is invoked and executed by a computer to implement the man-machine co-driving detection method based on steering wheel torque.

[0022] The application further provides a computer program product comprising a computer program or instructions which, when executed by a communication device, implement a method for detecting human-machine co-driving based on steering wheel torque as described above. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of a method for detecting human-machine co-driving based on steering wheel torque provided by an embodiment of the application; Figure 2 is a logic flowchart provided by an embodiment of the application; Figure 3 is a structural schematic diagram of a device for detecting human-machine co-driving based on steering wheel torque provided by an embodiment of the application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the application.

[0025] In the description of the application, it should be understood that the terms "first" and "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" can explicitly or implicitly include one or more features. In the description of the application, unless otherwise specified, the meaning of "several" is two or more.

[0026] The method for detecting human-machine co-driving based on steering wheel torque provided by an embodiment of the application aims to solve the problems that the determination method of fixed steering wheel torque value cannot adapt to different driving styles, and the fixed maintenance time easily causes early intervention and steering wheel confrontation; and the method relying on a three-dimensional MAP table of vehicle speed-curvature-torque has high parameter dimension, high adaptation cost, and lacks multi-dimensional correction, which easily causes misjudgment in scenes such as bad weather and driver fatigue, and finally realizes accurate and reliable identification of the driver's intervention intention, and avoids invalid signal interference and safety hazards.

[0027] Embodiment one: Please refer to Figure 1 The embodiment of the application provides a method for detecting human-machine co-driving based on steering wheel torque, which comprises S1-S3, and the specific implementation steps are as follows: S1, obtaining the steering wheel torque cumulative value of the target vehicle within the discrete sampling period when the auxiliary driving function is turned on.

[0028] The step S1 of the embodiment of the present application is specifically: After the target vehicle auxiliary driving function is turned on, the steering wheel torque data is collected at a discrete sampling period, and the steering wheel torque value corresponding to each time point in the discrete sampling period n (i.e. the period from t=i-n to t=i) is summed up based on the current time point i. The sum operation is performed to obtain the steering wheel torque cumulative value of the time interval covered by the discrete sampling period n.

[0029] The "auxiliary driving function" refers to a function that the vehicle has a certain autonomous driving capability and can assist in performing steering, acceleration, braking and other driving operations in a specific scenario, which is not completely automatic driving and requires the driver to remain ready to take over. In the scenario of the present application, when the function is turned on, the system will judge whether the driver intervenes through monitoring the steering wheel torque and other signals, and then realize the switching of the driving subject between the vehicle and the driver.

[0030] S2, performing nonlinear normalization processing on the steering wheel torque cumulative value to obtain a torque confidence, and if the torque confidence is greater than or equal to a driver takeover intervention threshold, calculating a comprehensive confidence by combining the torque confidence and a correction confidence.

[0031] The step S2 of the embodiment of the present application includes S2.1-S2.3, which are specifically: S2.1, performing a ratio operation on the steering wheel torque cumulative value and a vehicle calibration constant to obtain a mapping result by substituting the ratio into an arctangent function (arctan) for nonlinear mapping; scaling the mapping result by 1 / π, and finally superimposing a value of 0.5 to realize offset calibration to obtain the torque confidence , which is specifically: wherein, is the steering wheel torque cumulative value; is the vehicle calibration constant, and the 95% maximum torque value of the torque distribution of the driver within a certain time is taken as a reference value.

[0032] The ratio operation of the steering wheel torque cumulative value and the vehicle calibration constant in the embodiment S2.1 can eliminate the evaluation deviation caused by the torque parameter difference of different vehicle models; the nonlinear mapping processing can accurately fit the nonlinear relationship between the torque cumulative value and the driver's intervention intention, solve the problem that linear processing cannot accurately reflect the correlation between the torque signal and the intervention intention, further offset the system error and the sampling deviation, so that the final torque confidence can truly and accurately quantify the intervention intention of the driver through the steering wheel torque, provide high-quality core data support for subsequent comprehensive judgment, and improve the core accuracy of human-machine co-driving detection.​​

[0033] S2.2, If the torque confidence level ≥ Driver takeover intervention threshold The initial assessment indicates the driver intended to intervene, initiating the subsequent judgment phase based on comprehensive confidence level calculation; if the "torque confidence level" is not met... ≥ Driver takeover intervention threshold If the condition is met, return to the previous step, re-collect and calculate the cumulative value of the steering wheel torque and the corresponding torque confidence level within the discrete sampling period, and ensure that further verification is only carried out for scenarios that have a preliminary reasonable basis for intervention.

[0034] Among them, the driver takeover intervention threshold The method of obtaining it is: Based on the target vehicle's real-time speed and velocity gain coefficient Nonlinear speed adaptation modeling is performed. By quantifying the impact of vehicle speed on the driver's steering torque demand, the intervention threshold of the steering system is dynamically adjusted to obtain the nonlinear speed correction term. ; Based on the road curvature of the road the target vehicle is currently traveling on and curvature gain coefficient Linear curvature adaptation modeling is performed, and the influence of road curvature on steering torque demand is quantified to obtain the linear correction term for road curvature. ; Combined with preset basic torque parameters According to the nonlinear correction term of vehicle speed and the linear correction term for road curvature The driver takeover intervention threshold was calculated. Specifically: Where A is the average steering wheel torque applied by the driver on a straight road at 60km / h (16.67m / s), i.e., the preset base torque parameter; B is the speed gain coefficient, v is the real-time vehicle speed, and C is the curvature gain coefficient. The curvature of the road.

[0035] In this embodiment, S2.2, the impact of vehicle speed on steering torque demand is quantified through nonlinear speed adaptation modeling, solving the problem that fixed thresholds are difficult to match the actual operating needs of drivers at different speeds. For example, the threshold can be dynamically adjusted to improve judgment sensitivity when driving at high speeds. Linear curvature adaptation modeling quantifies the impact of road curvature, accurately adapting to the differences in steering torque demand in different road scenarios such as curves and straight roads. By combining preset basic torque parameters with two correction terms to calculate the threshold, the stability of the basic parameters is preserved, while dynamic correction compensates for the scenario limitations of fixed thresholds. This makes the takeover intervention threshold more closely match actual driving conditions, reducing misjudgments and omissions, and improving the coordination of human-machine interaction.

[0036] S2.3, Based on torque confidence level Corresponding first weight coefficient and corrected confidence level The corresponding second weighting coefficient Confidence level of torque and corrected confidence A weighted calculation is performed to obtain the overall confidence level. Among them, the first weighting coefficient Second weighting coefficient The sum is one. If the cameras, radars used to detect the vehicle's driving environment, or the driver monitoring system used to detect the driver's state malfunction or lack the corresponding hardware, then the second weighting coefficient... The overall confidence level is zero. Equal to torque confidence level .

[0037] The overall confidence level is: in, As the first weighting coefficient, ; This is the second weighting coefficient. , . For torque confidence level, To adjust the confidence level.

[0038] Among them, the corrected confidence level The method of obtaining it is: Real-time acquisition of the relative distance and relative speed between the target vehicle and the obstacle in front, and calculation of the collision time based on the relative distance and relative speed. Collision time Substituting the exponential function designed based on the forward collision warning (FCW) system test standard After calculation and amplitude limiting, the environmental collision risk confidence level is obtained. "Collision time" refers to the estimated time from the current moment when a collision is expected to occur, assuming that neither the target vehicle nor the obstacle in front takes any collision avoidance measures and their relative motion remains unchanged.

[0039] The driver monitoring system (DMS) obtains the driver's eyelid closure rate within a preset time window. The ratio of eyelid closure Perform a linear mapping to obtain the driver fatigue confidence level. ; From the confidence level of environmental collision risk and driver fatigue confidence The smaller value is selected as the corrected confidence level. .

[0040] The corrected confidence level is: The confidence level for environmental collision risk is: The driver fatigue confidence level is: in, This represents the collision time; 2.5 represents... The decay point is also the collision risk time commonly used in FCW testing. This represents the eyelid closure ratio.

[0041] In this embodiment, S2.3, by setting first and second weighting coefficients and ensuring their sum equals one, weights can be allocated according to the importance of the torque signal and the correction signal. This allows the overall confidence score to highlight the torque signal as the core intervention basis while also taking into account auxiliary correction information such as the environment and driver status. This weighted fusion method avoids the limitations of a single data dimension. For example, when the torque signal is weak but the environmental risk is high, the influence of the correction confidence score can be highlighted through weight adjustment; when the torque signal is clear, its core decision-making status is retained. The resulting overall confidence score more comprehensively and objectively reflects the driver's willingness to intervene and the risks of the driving scenario, improving the scientific nature and accuracy of takeover determination.

[0042] Furthermore, by integrating environmental collision risk and driver fatigue status to obtain the corrected confidence score, a multi-dimensional and accurate assessment of driving scenarios is achieved. The exponential function designed based on the forward collision warning system testing standards calculates the environmental collision risk confidence score, ensuring the authority and accuracy of the environmental risk assessment and accurately quantifying the safety hazards posed by obstacles ahead. The driver fatigue confidence score is obtained through a linear mapping of eyelid closure ratio, enabling real-time capture of the impact of the driver's physiological state on driving intervention capabilities. Selecting the smaller of the two values ​​as the corrected confidence score effectively avoids the one-sidedness of single-dimensional assessments and prevents correction deviations caused by misjudgments of environmental risks or neglect of driver status. This provides a reliable correction basis for subsequent comprehensive confidence score calculations, improving the safety and rationality of human-machine co-driving detection. Furthermore, an adaptive weight adjustment mechanism was designed for equipment failure scenarios, significantly improving the robustness and fault tolerance of the detection method. When the environmental detection equipment or driver monitoring system malfunctions, the calibration confidence data will lose its reliability. In this case, setting the second weight coefficient to zero and replacing the overall confidence with the torque confidence can effectively avoid interference from fault data on the judgment results and prevent misjudgments due to unreliable calibration information. At the same time, this design ensures the continuity of the detection function in the event of equipment failure, without interrupting the human-machine co-driving detection due to equipment failure. It ensures continuous monitoring of the driver's intervention status during driving, balancing system stability and driving safety, and making the detection method more adaptable to the complex equipment conditions in actual driving.

[0043] S3. If the overall confidence level is greater than or equal to the final threshold for driver takeover, then the driver of the target vehicle is determined to have intervened in driving.

[0044] Step S3 in this embodiment of the application is specifically as follows: If the overall confidence level Driver takeover final judgment threshold If the system detects that the driver of the target vehicle has intervened in driving, then a transfer of driving control from the assisted / automatic driving system to the driver will be executed: the assisted driving function of the target vehicle will then be disengaged, and the core driving control, such as steering, acceleration, and braking, will be completely handed over to the driver. The subsequent driving state of the vehicle will be determined by the driver's actual operation, such as steering wheel control and pedal operation, thus completing the switch from system-led to human-led driving in the human-machine co-driving scenario, ensuring that the driver can directly control the driving of the vehicle.

[0045] If the "overall confidence level" is not met Driver takeover final judgment threshold If the signal is "not clear and genuine", it indicates that the driver has not made a clear and genuine intention to intervene in driving. The system will maintain the current assisted driving state or switch back to the automatic driving mode, continue to collect the cumulative value of steering wheel torque through discrete sampling period and repeat the subsequent judgment process to avoid misjudging invalid signals as intervention operations and ensure the stability and safety of driving mode switching.

[0046] Among them, the final threshold for determining driver takeover The method of obtaining it is: Linear speed adaptation modeling is performed based on the real-time vehicle speed v and speed gain compensation coefficient D of the target vehicle. By quantifying the impact of vehicle speed on the sensitivity of driver takeover decision, the linear speed compensation term is obtained. ; Combining the preset base threshold parameter 0.5 and the environmental penalty correction term E, based on the vehicle speed linear compensation term... The final threshold for determining driver takeover was calculated. Specifically: Where D is the gain compensation coefficient for velocity v, with 60 km / h (16.67 m / s) as the reference value. E represents the penalty coefficient for environmental factors such as rain, snow, and nighttime. .

[0047] The driver takeover final determination threshold acquisition system constructed in embodiment S3 achieves dual adaptation of the threshold to vehicle speed and environment. A linear speed adaptation model is used to obtain a linear speed compensation term, which can adjust the determination sensitivity according to real-time vehicle speed. For example, sensitivity is reduced at low speeds to avoid unnecessary intervention determinations, while sensitivity is increased at high speeds to ensure safety. The final threshold is calculated by combining a preset basic threshold with two correction terms, balancing the uniformity of the basic determination standard with the flexibility of scenario adaptation. This ensures that takeover determinations at different vehicle speeds and in different environments meet safety requirements, improving the scenario adaptability and reliability of the detection method.

[0048] For application of the embodiments of the present invention, please refer to [reference needed]. Figure 2 , Figure 2 This is a logic flowchart provided in the embodiments of the present invention. Each determination step corresponds to a dedicated functional module, which fully demonstrates the entire process of human-machine co-driving intervention determination: First, the steering wheel torque value at each moment is obtained through the torque sampling module. The cumulative value of steering wheel torque within the discrete sampling period is obtained through the rolling accumulation module. The normalization module then performs nonlinear normalization to obtain the torque confidence level. ,judge Is it ≥ driver takeover intervention threshold? If the conditions are not met, resampling is performed; if the conditions are met, the process proceeds to the fusion decision module. The fusion decision module combines the environmental collision risk confidence level. and driver fatigue confidence Calculate the overall confidence level Final judgment whether Driver takeover final judgment threshold If the conditions are met, the driver is deemed to have intervened; otherwise, the system is maintained or switched back to assisted / automatic driving.

[0049] Overall, this embodiment has the following beneficial effects: This invention first acquires the cumulative steering wheel torque value within a discrete sampling period, rather than a single point-in-time torque value. This effectively filters out instantaneous invalid torque signals caused by road bumps, steering mechanism vibrations, and light hand touches, capturing only the valid action of the driver continuously applying torque. This aligns with the behavioral characteristics of real-world driving intervention, ensuring reliability of the judgment from the source of data acquisition. Simultaneously, nonlinear normalization is applied to the cumulative torque value to obtain a torque confidence score, accurately converting differentiated torque physical quantities into quantifiable intervention intention indicators. When the torque confidence score is greater than or equal to the driver takeover intervention threshold, it indicates that the torque continuously applied by the driver meets the basic conditions for intervention judgment, completing the first layer of effective screening, eliminating interference from invalid torque signals, confirming the initial rationality of the intervention intention, and proceeding to the subsequent multi-dimensional verification stage. Subsequently, a comprehensive confidence score is calculated by combining the corrected confidence score to compensate for the judgment bias of a single torque dimension, achieving multi-dimensional complementary verification of the environment, driver state, and torque intention. When the comprehensive confidence score is greater than or equal to the final driver takeover judgment threshold, it indicates that the driver's intervention intention has been fully confirmed after multi-dimensional verification, meeting the final takeover judgment standard, completing the second layer of accurate verification. This dual-threshold judgment logic not only sensitively identifies the driver's active takeover operation, but also avoids misjudgment, accurately distinguishes between real intervention and invalid signal interference, and achieves accurate and reliable judgment of the driver's real intervention behavior. In summary, the embodiments of this invention abandon the coarse judgment of a single torque threshold and the complex three-dimensional MAP calibration mode: by using the cumulative value of steering wheel torque over a continuous time as the core judgment basis, instead of the traditional single fixed torque value, it effectively improves the adaptability to different driving styles and avoids steering wheel resistance problems caused by premature intervention; at the same time, by replacing the three-dimensional MAP with an accurate formula, it significantly reduces the workload of parameter calibration, and by combining hardware such as cameras, millimeter-wave radar, and DMS to provide multi-dimensional correction, it can effectively avoid the risk of misoperation in scenarios such as bad weather and driver fatigue, and provide a solid guarantee for the driving safety of advanced ADAS systems.

[0050] Example 2: Please see Figure 3The embodiments of this application provide a human-machine co-driving detection device based on steering wheel torque, including a data acquisition module 10, a calculation module 20 and a judgment module 30; Among them, the acquisition module 10 is used to acquire the cumulative value of steering wheel torque within a discrete sampling period when the assisted driving function of the target vehicle is activated; The calculation module 20 is used to perform nonlinear normalization processing on the cumulative value of steering wheel torque to obtain torque confidence. If the torque confidence is greater than or equal to the driver takeover intervention threshold, the comprehensive confidence is calculated by combining the torque confidence and the correction confidence. The determination module 30 is used to determine that the driver of the target vehicle has intervened in driving if the overall confidence level is greater than or equal to the final determination threshold for driver takeover.

[0051] It should be noted that the technical concept of this second embodiment is completely consistent with that of the first embodiment. The two maintain a high degree of synergy at the technical logic level. The specific technical details can be referred to the relevant description of the first embodiment, which will not be repeated here.

[0052] Example 3: This application provides a computer-readable storage medium including a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the aforementioned human-machine co-driving detection method based on steering wheel torque; The aforementioned human-machine co-driving detection method based on steering wheel torque, when implemented as a software functional unit and used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0053] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video optical disc; or it can be a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or non-volatile storage medium, or may include both types of storage media.

[0054] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A human-machine co-driving detection method based on steering wheel torque, characterized in that, include: Obtain the cumulative value of steering wheel torque of the target vehicle within a discrete sampling period when the driver assistance function is activated; The cumulative steering wheel torque value is nonlinearly normalized to obtain the torque confidence level. If the torque confidence level is greater than or equal to the driver takeover intervention threshold, the comprehensive confidence level is calculated by combining the torque confidence level and the correction confidence level. If the overall confidence level is greater than or equal to the final threshold for driver takeover determination, then it is determined that the driver of the target vehicle has intervened in driving.

2. The human-machine co-driving detection method based on steering wheel torque as described in claim 1, characterized in that, The method for obtaining the correction confidence level is as follows: The relative distance and relative speed between the target vehicle and the obstacle in front are collected in real time. The collision time is calculated based on the relative distance and relative speed. The collision time is then substituted into an exponential function designed based on the test standard of the forward collision warning system for calculation and amplitude limiting to obtain the environmental collision risk confidence level. The driver's eyelid closure ratio within a preset time window is linearly mapped to obtain the driver's fatigue confidence level. The smaller value between the environmental collision risk confidence level and the driver fatigue confidence level is selected as the corrected confidence level.

3. The human-machine co-driving detection method based on steering wheel torque as described in claim 1, characterized in that, The method for obtaining the driver takeover intervention threshold is as follows: Based on the real-time vehicle speed and speed gain coefficient of the target vehicle, a nonlinear vehicle speed adaptation model is performed. By quantifying the impact of vehicle speed on the driver's steering torque demand, the intervention threshold is dynamically adjusted to obtain the vehicle speed nonlinear correction term. Linear curvature adaptation modeling is performed based on the road curvature and curvature gain coefficient of the current driving road of the target vehicle. By quantifying the influence of road curvature on steering torque demand, a linear road curvature correction term is obtained. The driver takeover threshold is calculated by combining the preset basic torque parameters with the vehicle speed nonlinear correction term and the road curvature linear correction term.

4. The human-machine co-driving detection method based on steering wheel torque as described in claim 1, characterized in that, The method for obtaining the final threshold for determining driver takeover is as follows: Linear speed adaptation modeling is performed based on the real-time vehicle speed and speed gain compensation coefficient of the target vehicle. By quantifying the influence of vehicle speed on the sensitivity of driver takeover judgment, the linear speed compensation term is obtained. The final threshold for driver takeover is calculated based on the vehicle speed linear compensation term, by combining the preset basic threshold parameters and environmental penalty correction terms.

5. The human-machine co-driving detection method based on steering wheel torque as described in claim 1, characterized in that, The cumulative steering wheel torque value is nonlinearly normalized to obtain the torque confidence level, specifically: The ratio of the cumulative steering wheel torque value to the vehicle calibration constant is calculated, and the calculated ratio is nonlinearly mapped to obtain the mapping result. The mapping result is scaled and offset calibrated to obtain the torque confidence level.

6. The human-machine co-driving detection method based on steering wheel torque as described in claim 1, characterized in that, The overall confidence level is calculated by combining the torque confidence level and the correction confidence level, specifically as follows: The torque confidence level and the corrected confidence level are weighted and calculated to obtain the comprehensive confidence level based on the first weighting coefficient corresponding to the torque confidence level and the second weighting coefficient corresponding to the corrected confidence level; wherein the sum of the first weighting coefficient and the second weighting coefficient is one.

7. The human-machine co-driving detection method based on steering wheel torque as described in claim 6, characterized in that, If the camera, radar, or driver monitoring system used to detect the vehicle's driving environment malfunctions, the second weighting coefficient is zero, and the overall confidence level is determined to be equal to the torque confidence level.

8. A human-machine co-driving detection device based on steering wheel torque, characterized in that, It includes a data acquisition module, a calculation module, and a decision-making module; The acquisition module is used to acquire the cumulative value of steering wheel torque within a discrete sampling period when the assisted driving function of the target vehicle is activated. The calculation module is used to perform nonlinear normalization processing on the cumulative value of the steering wheel torque to obtain the torque confidence score. If the torque confidence score is greater than or equal to the driver takeover intervention threshold, the comprehensive confidence score is calculated by combining the torque confidence score and the correction confidence score. The determination module is used to determine that the driver of the target vehicle has intervened in driving if the overall confidence level is greater than or equal to the final determination threshold for driver takeover.

9. A storage medium, characterized in that, The storage medium stores a computer program, which is called and executed by a computer to implement a human-machine co-driving detection method based on steering wheel torque as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the communication device, they implement the human-machine co-driving detection method based on steering wheel torque as described in any one of claims 1 to 7.

Images