Intelligent hand-off detection method, device and vehicle

Abstract

This application discloses an intelligent hands-free detection method, device, and vehicle. The method is applied to an intelligent hands-free detection system, which includes a driver assistance system, an image acquisition device for acquiring images of the driver's hands, and an electric power steering system for acquiring steering wheel torque information. The method includes: when the driver assistance system is activated, determining the visual grip probability based on acquired spatial parameters and the driver's hand image; the visual grip probability characterizes the confidence level of the driver's hands visually gripping the steering wheel; and determining the torque grip probability based on the acquired steering wheel torque information; the torque grip probability characterizes the confidence level of the driver's hands physically gripping the steering wheel. The driver's hands-free state is determined based on the consistency result of the visual grip probability and the torque grip probability, as well as the visual grip probability and the torque grip probability. Therefore, by fusing visual grip information, torque grip information, and their consistency information for hands-free judgment, detection accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to an intelligent hands-free detection method, device, and vehicle. Background Technology

[0002] In current vehicles, Advanced Driver Assistance Systems (ADAS) are typically based on the activation of a single-lane centering control function. When ADAS is activated, the detection of whether the driver has taken their hands off the steering wheel is usually performed using the steering wheel torque output by the Electric Power Steering (EPS) system.

[0003] The method of detecting steering wheel torque relies on the deformation of the torsion bar caused by the driver turning the steering wheel to obtain the hand torque. However, during lane centering control, even if the driver holds the steering wheel continuously, the small circular motion of the steering wheel may not generate significant torque, which can easily lead to misjudgment of letting go.

[0004] Another capacitive detection method requires pre-embedding segmented capacitors inside the steering wheel, which places higher demands on the steering wheel structure and manufacturing process, and increases the overall vehicle manufacturing cost.

[0005] Therefore, how to accurately detect when a driver takes their hands off the steering wheel and improve vehicle driving safety is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, one aspect of this application provides an intelligent hands-free detection method, the method being applied to an intelligent hands-free detection system, the system comprising a driving assistance system, an image acquisition device for acquiring images of the driver's hands, and an electric power steering system for acquiring steering wheel torque information, comprising: When the driving assistance system is in an active state, the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information are acquired. Based on the spatial parameters and the driver's hand image, the visual grip probability is determined; the visual grip probability is used to characterize the confidence level that the driver's hands are visually gripping the steering wheel. Based on the steering wheel torque information, the torque grip probability is determined; the torque grip probability is used to characterize the confidence level of the driver's hand gripping the steering wheel under physical force. Based on the visual grip probability and the torque grip probability, a consistency result characterizing the driver's hand grip on the steering wheel is determined; The driver's hands-free state is determined based on the consistency results, the visual grip probability, and the torque grip probability.

[0007] Optionally, the spatial parameters include a preset vehicle reference point, the center point of the airbag cover, and the steering wheel size; determining the visual grip probability based on the spatial parameters and the driver's hand image includes: Construct a vehicle coordinate system based on the vehicle reference point and the center point of the airbag cover; Based on the steering wheel size, a spatial detection area corresponding to the edge of the steering wheel is constructed in the vehicle coordinate system; Extract key feature points of the driver's hand from the image of the driver's hand; The coordinates of the key hand feature points in the vehicle coordinate system are determined to obtain the driver's hand feature points; The visual grip probability is determined based on the spatial relationship between the driver's hand feature points and the spatial detection area.

[0008] Optionally, determining the visual grip probability based on the spatial relationship between the driver's hand feature points and the spatial detection area includes: Determine a first number of the driver's hand feature points, and a second number of the driver's hand feature points that fall within the spatial detection area; Based on the first quantity and the preset number of feature points, the completeness of the hand features is determined; the completeness of the hand features is used to characterize the degree of completeness of the driver's hand in the driver's hand image; Based on the first quantity and the second quantity, the hand rim contact degree is determined; the hand rim contact degree is used to characterize the degree of contact between the driver's hand and the steering wheel; The grip confidence level is determined based on the positional relationship between specified hand feature points and the spatial detection area; the grip confidence level is used to characterize the degree of confidence in the driver's hand gripping the steering wheel; The visual grip probability is determined based on the completeness of the hand features, the contact degree of the hand rim, and the confidence of the grip posture.

[0009] Optionally, determining the torque grip probability based on the steering wheel torque information includes: Obtain the steering wheel torque sequence within a preset time window; The torque amplitude characteristic, torque fluctuation characteristic, and effective torque duration characteristic are determined using the steering wheel torque sequence; wherein, the torque amplitude characteristic is used to characterize the intensity of the force applied by the driver's hands to the steering wheel; the torque fluctuation characteristic is used to characterize the stability of the applied force; and the effective torque duration characteristic is used to characterize the duration of the applied force. The torque holding probability is determined based on the torque amplitude characteristics, the torque fluctuation characteristics, and the effective torque duration characteristics.

[0010] Optionally, the torque amplitude feature includes the mean of the absolute torque values ​​in the steering wheel torque sequence; the torque fluctuation feature includes the standard deviation of the torque values ​​in the steering wheel torque sequence; the effective torque duration feature includes the number of times the ratio of the absolute torque value to the preset base torque value is greater than a preset ratio within the preset time window. Specifically, when the torque amplitude characteristic is greater than a first preset value, the effective torque duration characteristic is greater than a second preset value, and the torque fluctuation characteristic is less than a third preset value, the torque holding probability is reduced.

[0011] Optionally, determining the visual grip probability and the torque grip probability, characterizing the consistency of the driver's hand gripping the steering wheel, includes: If the visual grip probability is greater than the first probability threshold, then the visual grip result is determined to be gripping; otherwise, the visual grip result is determined to be releasing. If the torque grip probability is greater than the second probability threshold, then the torque grip result is determined to be a grip; otherwise, the torque grip result is determined to be a release. The consistency result is determined based on whether the visual grip result and the torque grip result are the same.

[0012] Optionally, the driver's hands-free state is determined based on the consistency result, the visual grip probability, and the torque grip probability, including: If the consistency results represent the same state, then the consistency results are taken as the released state; the consistency results are either all holding or all released. If the consistency results are inconsistent, then the following steps are performed: Determine the consistency probability between the visual grip probability and the torque grip probability; The target grip probability is obtained by weighted summing of the visual grip probability, the torque grip probability, and the consistency probability. Determine whether the target holding probability is greater than the target probability threshold; if it is greater, determine that the release state is holding; if it is not greater, determine that the release state is releasing.

[0013] Optionally, the target grip probability is obtained by weighted summation of the visual grip probability, the torque grip probability, and the consistency probability, including: Obtain the preset first weight corresponding to the visual grip probability, the second weight corresponding to the torque grip probability, the third weight corresponding to the consistency probability, and the visual clarity of the driver's hand image; If the visual grip result represents gripping, the torque grip result represents release, and the visual clarity is greater than the clarity threshold, then the first weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is not greater than the clarity threshold, then the second weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is greater than the clarity threshold, then the first weight is increased and / or the second weight is decreased. Based on the adjusted weights, the visual grip probability, the torque grip probability, and the consistency probability are weighted and summed to obtain the target grip probability.

[0014] Another aspect of this application provides an intelligent hands-off detection device, which is applied to an intelligent hands-off detection system. The system includes a driving assistance system, an image acquisition device for acquiring images of the driver's hands, and an electric power steering system for acquiring steering wheel torque information. The device includes: The information acquisition module is used to acquire the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information when the driving assistance system is in an active state. A visual grip probability determination module is used to determine the visual grip probability based on the spatial parameters and the driver's hand image; the visual grip probability is used to characterize the confidence level of the driver's hand visually gripping the steering wheel; The torque grip probability determination module is used to determine the torque grip probability based on the steering wheel torque information; the torque grip probability is used to characterize the confidence level of the driver's hand gripping the steering wheel under physical force. A consistency determination module is used to determine a consistency result characterizing the driver's hand gripping the steering wheel based on the visual grip probability and the torque grip probability; The hands-free state determination module is used to determine the driver's hands-free state based on the consistency result, the visual grip probability, and the torque grip probability.

[0015] Another aspect of this application provides a vehicle, the vehicle including a driving assistance system, an image acquisition device for acquiring images of the driver's hands, an electric power steering system for acquiring steering wheel torque information, a memory, and a processor; The memory stores a computer program that can run on the processor, and when the processor executes the computer program, it implements the steps of the intelligent hands-free detection method.

[0016] The intelligent hands-off detection method, device, and vehicle provided in this application offer the following advantages: By using the driver's hand image and steering wheel torque information, the visual grip probability and torque grip probability are determined respectively. The driver's hands-off state is then judged by combining the consistency results of these two grip probabilities, achieving a fusion of visual detection and physical force detection. Therefore, even when the steering wheel's circular motion is small and torque characteristics are not obvious during lane centering control, the spatial relationship between the driver's hands and the steering wheel can be used to assist in judging the grip state, reducing false alarms caused by relying solely on steering wheel torque. Simultaneously, there is no need to pre-embed capacitors in the steering wheel, reducing the requirements for steering wheel structure and manufacturing processes, and improving the accuracy of hands-off detection while controlling costs. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an intelligent hand-free detection method provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of the structure of an intelligent hand-free detection system provided in an embodiment of this application.

[0019] Figure 3 This is a schematic diagram illustrating the principle of an intelligent hand-free detection method provided in an embodiment of this application.

[0020] Figure 4 This is a schematic diagram illustrating the principle of determining visual grip probability, provided in an embodiment of this application.

[0021] Figure 5 This is a schematic diagram illustrating the principle of determining torque grip probability as provided in an embodiment of this application.

[0022] Figure 6 This is a schematic diagram of the structure of an intelligent hand-free detection device provided in an embodiment of this application.

[0023] Figure 7 This is a structural schematic diagram of a vehicle provided in an embodiment of this application.

[0024] The reference numerals in the attached figures are as follows: 20 is the driver assistance system, 21 is the image acquisition device, 22 is the electric power steering system, 23 is the hands-off detection controller, 60 is the information acquisition module, 61 is the visual grip probability determination module, 62 is the torque grip probability determination module, 63 is the consistency determination module, 64 is the hands-off state determination module, 70 is the memory, 71 is the processor, 72 is the display screen, 73 is the input / output interface, 74 is the communication interface, 75 is the power supply, 76 is the communication bus, 701 is the computer program, 702 is the operating system, and 703 is the data. Detailed Implementation

[0025] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0026] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0027] Figure 1 This is a flowchart illustrating an intelligent hand-free detection method provided in an embodiment of this application. Figure 2 This is a schematic diagram of an intelligent hand-drop detection system provided in an embodiment of this application. The method provided in this embodiment is applied to the intelligent hand-drop detection system, such as... Figure 2 As shown, the system includes a driver assistance system (ADAS) 20, an image acquisition device 21 for acquiring images of the driver's hands, an electric power steering system (EPS) 22 for acquiring steering wheel torque information, and a hands-off detection controller 23.

[0028] The image acquisition device 21 can be any one of an in-cabin camera, a driver monitoring camera, or other image sensors capable of capturing images of the driver's hands. The EPS 22 can output a steering wheel torque sequence, a hand torque signal, or a steering signal related to the force applied to the steering wheel.

[0029] It should be noted that the entity implementing the method provided in this application can be... Figure 2The hands-free detection controller 23 shown can also be any one of the vehicle controller, domain controller, cockpit controller, or body controller on the vehicle, and this application does not limit it.

[0030] like Figure 1 As shown, the method includes: S10: When the driving assistance system is in an active state, acquire the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information; In a specific embodiment, the functional state of ADAS 20 may include an inactive state and an active state. When ADAS 20 is inactive, the vehicle may not initiate the intelligent hands-free detection process, or may only perform low-frequency monitoring. When ADAS 20 is in an active state with single-lane centering control, lane keeping assist, or other lateral control functions, the vehicle initiates the intelligent hands-free detection method process provided in this application.

[0031] Figure 3 This is a schematic diagram illustrating the principle of an intelligent hand-free detection method provided in an embodiment of this application, as shown below. Figure 3 As shown, when ADAS 20 is activated, it acquires images of the driver's hands through image acquisition device 21 and steering wheel torque information through EPS 22. In addition, it also needs to acquire the spatial parameters of the steering wheel, including parameters that describe the position and size of the steering wheel in the vehicle space.

[0032] The spatial parameters of the steering wheel may include, but are not limited to, calibrated values ​​such as the fixed reference point of the entire vehicle, the center point of the airbag cover, the steering wheel radius, the steering wheel diameter, the steering wheel rim diameter, the rim cross-sectional radius, and the rim width. The driver's hand image can be acquired in real time by the image acquisition device 21 to visually determine whether the hand is located in the steering wheel rim area. Steering wheel torque information can be acquired in real time by the EPS 22 to determine whether the driver's hand is applying force to the steering wheel from a physical force perspective.

[0033] Therefore, by only initiating the fusion detection of whether the driver's hands are off the steering wheel when the ADAS 20 function is active, the hands-off detection process can be focused on scenarios that require driver supervision and assisted driving, reducing the waste of computing resources and false triggers in irrelevant conditions.

[0034] S11: Determine the visual grip probability based on spatial parameters and driver's hand image; the visual grip probability is used to characterize the confidence level of the driver's hands visually gripping the steering wheel; Further, see Figure 3The hands-off detection controller 23 performs visual calculations based on the received spatial parameters and driver's hand image to obtain the visual grip probability, which can characterize the confidence level of the driver's hands visually gripping the steering wheel. In other words, the visual grip probability is used to reflect how much the driver is currently visually gripping the steering wheel.

[0035] Specifically, a vehicle coordinate system can be established based on spatial parameters, and a spatial detection area corresponding to the steering wheel rim can be established based on the steering wheel size. Key hand feature points can be extracted from the driver's hand image. By judging the distance, overlap, or encirclement relationship between the key hand feature points and the spatial detection area, the visual grip probability can be obtained.

[0036] It should be noted that the process of determining the visual grip probability is not just about identifying whether a hand appears in the image, but rather about judging the spatial relationship between the key feature points of the hand and the spatial detection area of ​​the steering wheel rim in the whole vehicle coordinate system, in order to distinguish different states such as the hand being close to the steering wheel, obscuring the steering wheel, and actually gripping the steering wheel.

[0037] S12: Determine the torque grip probability based on the steering wheel torque information; the torque grip probability is used to characterize the confidence level of the driver's hands gripping the steering wheel under physical force. While performing visual computing, see Figure 3 Based on the steering wheel torque information, torque calculation is performed to obtain the torque grip probability, which reflects the probability that the driver grips the steering wheel under physical force. Specifically, continuous steering wheel torque within a specified window can be collected, and the torque grip probability can be determined based on the ratio of the average steering wheel torque within that window to a preset value.

[0038] It should be noted that, in one optional embodiment, steering wheel torque information can be acquired by an EPS 22 torque sensor. The EPS 22 torque sensor may include an input shaft, an output shaft, an upper rotor, a lower rotor, and sensor circuitry. When the steering wheel is subjected to force from the driver's hand, the torsion bar deforms, causing the sensor circuitry to output a corresponding torque signal. This torque signal does not always represent the actual grip state when the ADAS 20 continuously controls the steering wheel; therefore, it needs to be combined with visual grip probability for judgment.

[0039] S13: Determine the consistency results characterizing the driver's hand grip on the steering wheel based on the visual grip probability and torque grip probability; S14: Determine the driver's hands-free state based on the consistency results, visual grip probability, and torque grip probability.

[0040] Furthermore, to improve the accuracy of determining whether the driver has taken their hands off the steering wheel, a consistency analysis was conducted on the visual calculation results and the torque calculation results. The consistency results include three scenarios: both the visual and torque sides indicate holding the steering wheel; both the visual and torque sides indicate holding the steering wheel; and one of the visual and torque sides indicates holding the steering wheel while the other indicates holding the steering wheel.

[0041] In one optional embodiment, if both the visual and torque sides represent a grip or both represent a release, the corresponding release state can be directly output. If they are inconsistent, the driver's release state is determined by combining the visual grip probability and the torque grip probability based on the consistency result. Specifically, in one optional embodiment, the probability of the consistency result can be calculated, which can be used to characterize the confidence level of the consistency structure. Further, the final target grip probability is calculated by combining the three probability values, and the driver's release of the steering wheel is determined based on whether the target grip probability is greater than a threshold.

[0042] In specific embodiments, during straight-line centering control scenarios, the driver lightly grips the steering wheel, but the EPS 22 has low hand torque, making the visual results more reflective of the actual situation. When the camera is obstructed but the EPS 22 has normal hand torque, the torque-side results are more valuable. When the vision is clear and the torque side indicates grip but the visual side indicates release, it is necessary to be wary of evasive behavior where an object substitutes for the hand to apply torque.

[0043] Therefore, the intelligent hands-off detection method provided in this application determines the visual grip probability and torque grip probability respectively by using the driver's hand image and steering wheel torque information, and judges the driver's hands-off state by combining the consistency results of the two grip probabilities, thus achieving the fusion of visual detection and physical force detection. This allows for auxiliary judgment of the grip state by utilizing the spatial relationship between the driver's hands and the steering wheel during lane centering control when the steering wheel's circular motion is small and torque characteristics are not obvious, reducing false alarms caused by relying solely on steering wheel torque. Simultaneously, it eliminates the need for pre-embedded capacitors in the steering wheel, reducing the requirements for steering wheel structure and manufacturing process, and improving the accuracy of hands-off detection while controlling costs.

[0044] In one optional embodiment, the spatial parameters include a preset vehicle reference point, the center point of the airbag cover, and the steering wheel size; based on the spatial parameters and the driver's hand image, the visual grip probability is determined, including: Construct a vehicle coordinate system based on the vehicle reference point and the center point of the airbag cover; Based on the steering wheel size, a spatial detection area corresponding to the edge of the steering wheel is constructed in the vehicle coordinate system; Extract key feature points of the driver's hand from the image of the driver's hand; Determine the coordinates of key hand feature points in the vehicle coordinate system to obtain the driver's hand feature points; The probability of visual grip is determined based on the spatial relationship between the driver's hand feature points and the spatial detection area.

[0045] Figure 4 This is a schematic diagram illustrating the principle of determining visual grip probability provided in an embodiment of this application, as shown below. Figure 4 As shown in the specific embodiment, when constructing the vehicle coordinate system, the vehicle reference point can be a fixed point in the vehicle body structure, and the center point of the airbag cover is located near the center of the steering wheel. The two can be used together to determine the position reference of the steering wheel in the vehicle coordinate system.

[0046] It should be noted that since the relationship between the steering wheel, driver's seat, camera mounting position and hand space position all need to be determined within a unified vehicle space, the coordinate system constructed in this application is the vehicle coordinate system, rather than an isolated coordinate system established only for the steering wheel itself.

[0047] In a specific embodiment, the steering wheel dimensions may include, but are not limited to, the steering wheel diameter, steering wheel radius, steering wheel rim diameter, rim cross-sectional radius, or rim width. Based on the steering wheel dimensions, the vehicle determines a spatial detection area corresponding to the edge of the steering wheel in the vehicle coordinate system. In a specific embodiment, the spatial detection area may be a three-dimensional annular region formed around the steering wheel rim. This three-dimensional annular region includes at least the steering wheel center, the plane on which the steering wheel is located, the steering wheel outer radius, the rim cross-sectional radius or rim width, and the spatial tolerance along the normal direction of the steering wheel.

[0048] like Figure 4 As shown, while determining the spatial detection region, key hand feature points of the driver are extracted from the driver's hand image. These key hand feature points may include one or more of the following: wrist point, palm center point, finger root joint point, fingertip point, thumb joint point, and fist outline point.

[0049] In one optional embodiment, the driver's hand image output by the image acquisition device 21 can first enter the hand detection model or key point recognition model to obtain the position of the key feature points of the hand in the image coordinate system or camera coordinate system. Then, combined with the intrinsic and extrinsic parameters of the image acquisition device 21 and the vehicle installation calibration relationship, the key feature points of the hand are transformed into the vehicle coordinate system to obtain the driver's hand feature points.

[0050] After obtaining the driver's hand feature points, the visual grip probability is determined based on the spatial relationship between the driver's hand feature points and the spatial detection area. The spatial relationship may include whether the driver's hand feature points fall within the spatial detection area, the distance between the driver's hand feature points and the outer surface of the steering wheel rim, the overlap or enclosing relationship between the driver's hand feature points and the spatial detection area, etc., which are not limited in this application.

[0051] Therefore, the method provided in this application embodiment determines the spatial detection area of ​​the steering wheel rim by using the whole vehicle reference point, the center point of the airbag cover and the steering wheel size. The driver's hand image provides key hand feature points, thereby transforming visual judgment from planar image recognition to judgment of the whole vehicle spatial relationship, improving the accuracy of judgment of hand gripping the steering wheel.

[0052] Based on the above embodiments, as an optional embodiment, the visual grip probability is determined according to the spatial relationship between the driver's hand feature points and the spatial detection area, including: Determine the first number of driver's hand feature points, and the second number of driver's hand feature points that fall within the spatial detection area; The completeness of hand features is determined based on the first quantity and the preset number of feature points; the completeness of hand features is used to characterize the degree of completeness of the driver's hand in the driver's hand image; Based on the first and second quantities, the hand-wheel contact degree is determined; the hand-wheel contact degree is used to characterize the degree of contact between the driver's hands and the steering wheel; The confidence level of the grip posture is determined based on the positional relationship between specified hand feature points and the spatial detection area; the confidence level of the grip posture is used to characterize the degree of confidence in the driver's grip on the steering wheel; The probability of visual gripping is determined based on the integrity of hand features, the contact degree of the hand rim, and the reliability of the grip posture.

[0053] In a specific embodiment, the first quantity is the actual number of valid feature points detected, and the second quantity is the number of points in the first quantity that fall within the three-dimensional annular space detection area of ​​the steering wheel rim. If the number of detected feature points is too small, it indicates that the hand information in the driver's hand image is incomplete, and the visual confidence level is low.

[0054] See Figure 4 The visual grip probability is determined from three dimensions: hand feature completeness, hand rim contact degree, and grip posture reliability. Specifically, hand feature completeness can be determined based on a first quantity and a preset number of feature points. Hand feature completeness characterizes the degree of completeness of the driver's hand in the driver's hand image. In an optional embodiment, the result of dividing the first quantity by the preset number of feature points can be used as the hand feature completeness, where a larger first quantity indicates a higher hand feature completeness.

[0055] Furthermore, based on the first quantity and the second quantity, the vehicle determines the hand rim contact degree, which characterizes the degree of contact between the driver's hands and the steering wheel. In an optional embodiment, the ratio of the second quantity to the first quantity can be used as the hand rim contact degree; the larger the second quantity, the higher the hand rim contact degree.

[0056] Simultaneously, the grip confidence level is determined based on the positional relationship between designated hand feature points and the spatial detection area. Grip confidence level characterizes the degree of confidence in the driver's hand gripping the steering wheel. In one optional embodiment, designated hand feature points may include the center point of the palm, the thumb joint, the index finger joint, the wrist point, or the outline of the fist. In a specific embodiment, if the center of the palm is located near the steering wheel rim, and multiple finger joints are distributed on both sides of the rim or near the outer surface of the rim, the grip confidence level is high. If only the wrist or a few fingertips are close to the rim, the grip confidence level is low.

[0057] It should be noted that this application does not limit the calculation method for grip posture reliability. As an optional embodiment, it can be determined by a preset lookup table rule. For example, multiple posture conditions can be set, specifically including the center of the palm being near the steering wheel rim, at least two key finger points being near the rim, the wrist being on the side of the palm away from the steering wheel center, and the fingers and palm forming a bent posture towards the rim. In a specific embodiment, the more posture conditions that are met, the higher the grip posture reliability. Specifically, a mapping relationship can be established between the number of posture conditions met and the grip posture reliability, and the specific grip posture reliability can be determined based on the mapping relationship.

[0058] In one optional embodiment, determining the torque grip probability based on steering wheel torque information includes: Obtain the steering wheel torque sequence within a preset time window; The torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics are determined by the steering wheel torque sequence. The torque hold probability is determined based on the characteristics of torque amplitude, torque fluctuation, and effective torque duration.

[0059] Figure 5 This is a schematic diagram illustrating the principle of determining torque grip probability provided in an embodiment of this application, as shown below. Figure 5 As shown, a steering wheel torque sequence within a preset time window is obtained. This steering wheel torque sequence can be output by EPS 22 at a fixed sampling period, denoted as {T1, T2, ..., Tn}, where Ti is the steering wheel torque value at the i-th sampling moment, and n is the number of sampling points within the preset time window. In a specific embodiment, the preset time window is used to smooth instantaneous noise and avoid misjudgment of the hands-free state caused by abrupt changes in a single sampling point.

[0060] See Figure 5 By using the steering wheel torque sequence, the torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics are determined. Specifically, the torque amplitude characteristic characterizes the intensity of the force applied by the driver's hands to the steering wheel; the torque fluctuation characteristic characterizes the stability of the applied force; and the effective torque duration characteristic characterizes the duration of the applied force.

[0061] In one optional embodiment, the torque amplitude characteristic includes the mean of the absolute torque values ​​in the steering wheel torque sequence; the torque fluctuation characteristic includes the standard deviation of the torque values ​​in the steering wheel torque sequence; and the effective torque duration characteristic includes the number of times the ratio of the absolute torque value to the preset base torque value is greater than a preset ratio within a preset time window.

[0062] When the torque amplitude characteristic is greater than a first preset value, the effective torque duration characteristic is greater than a second preset value, and the torque fluctuation characteristic is less than a third preset value, the torque holding probability is reduced. It is understood that the first preset value is used to determine whether the torque amplitude has reached a significant force level, the second preset value is used to determine whether the force is persistent, and the third preset value is used to determine whether the torque is stable. It is important to note that when determining the effective torque duration characteristic, a lower base torque value can be set, and the ratio of the absolute value of the torque to the preset base torque value within a preset time window can be calculated. When this ratio is greater than a preset ratio, the currently calculated torque is considered effective torque. Therefore, the more effective torques there are within the preset time window, the better the torque persistence. In other words, in this embodiment, the holding persistence is not determined by a single threshold, but by the number of times the effective torque occurs over a period of time.

[0063] Therefore, if the torque amplitude is high, the continuity is high, but the fluctuation is low, it indicates that there is a continuous and stable external force on the steering wheel. This state may not be a natural hand grip, but may come from a suspension, fixed object, or other object that avoids the hand-off alarm. For example, it may be the force from a mineral water bottle. In this case, it is necessary to reduce the torque grip probability to avoid misjudgment of grip.

[0064] Furthermore, the torque holding probability is determined based on the torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics. In an optional embodiment, corresponding weights can be pre-set for the torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics, and the sum of each weight is 1. Further, the torque holding probability is obtained by weighted summation of the torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics based on each weight.

[0065] It should be noted that, in specific embodiments, the larger the torque amplitude, the higher the probability that the driver is applying force to the steering wheel, the more the torque fluctuation reflects natural hand input, and the higher the torque continuity characteristic, indicating that the force is continuously present.

[0066] Therefore, the method provided in this application determines three types of characteristics—amplitude, volatility, and duration—by using the steering wheel torque sequence within a preset time window, and determines the torque grip probability based on these three characteristics. This makes the judgment of the steering wheel grip state not only dependent on a certain instantaneous torque threshold, but also considers the force intensity, stability, and duration simultaneously, further improving the accuracy of the steering wheel grip state judgment.

[0067] In one alternative embodiment, determining the visual grip probability and torque grip probability to characterize the consistency of the driver's hand grip on the steering wheel includes: If the visual grip probability is greater than the first probability threshold, the visual grip result is determined to be gripping; otherwise, the visual grip result is determined to be releasing. If the torque grip probability is greater than the second probability threshold, the torque grip result is determined to be a grip; otherwise, the torque grip result is determined to be a release. Consistency is determined by whether the visual grip results and the torque grip results are the same.

[0068] In a specific embodiment, the first probability threshold and the second probability threshold can correspond to the judgment boundaries on the visual side and the torque side, respectively. If both the visual grip result and the torque grip result are gripping, or both are release, the consistency result indicates consistency. If the visual grip result is gripping and the torque grip result is release, or the visual grip result is release and the torque grip result is gripping, the consistency result indicates inconsistency.

[0069] Based on the above embodiments, when determining the driver's hands-free state according to the consistency result, visual grip probability, and torque grip probability, if the consistency result is consistent, then the consistency result is taken as the hands-free state. Specifically, if both the visual grip result and the torque grip result are grip, the hands-free state is grip. If both the visual grip result and the torque grip result are hands-free, the hands-free state is hands-free.

[0070] If the consistency results are inconsistent, determine the consistency probability between the visual grip probability and the torque grip probability; perform a weighted sum of the visual grip probability, torque grip probability and consistency probability to obtain the target grip probability; determine whether the target grip probability is greater than the target probability threshold; if it is greater, determine the release state as grip; if it is not greater, determine the release state as release.

[0071] In a specific embodiment, the consistency probability is used to characterize the degree to which two probability results jointly support the gripping state. In an optional embodiment, the product of the visual gripping probability and the torque gripping probability can be used as the consistency probability. The higher both the visual gripping probability and the torque gripping probability are, the higher the consistency probability is. The lower the probability on either side, the lower the consistency probability is.

[0072] Furthermore, the visual grip probability, torque grip probability, and consistency probability are weighted and summed to obtain the target grip probability, and the driver's hands-free state is determined based on the target grip probability.

[0073] In a specific embodiment, when the visual grip result is "grip" but the torque grip result is "release," it may be a scenario where the vehicle is being controlled for straight-line centering, the driver is lightly gripping the steering wheel, and the EPS 22 hand torque is low. When the visual grip result is "release" but the torque grip result is "grip," it may be a scenario where the camera is obstructed, the driver is actually gripping the steering wheel but the visual system does not recognize it, or it may be a scenario where a non-hand object generates torque.

[0074] Therefore, in scenarios where the visual grip probability and the torque grip probability are inconsistent, the calculation of consistency probability and target grip probability is introduced to further improve the accuracy of determining the steering wheel grip state.

[0075] In one optional embodiment, the target grip probability is obtained by weighted summation of the visual grip probability, torque grip probability, and consistency probability, including: The system obtains the first weight corresponding to the preset visual grip probability, the second weight corresponding to the torque grip probability, the third weight corresponding to the consistency probability, and the visual clarity of the driver's hand image. If the visual grip result represents grip, the torque grip result represents release, and the visual clarity is greater than the clarity threshold, then the first weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is not greater than the clarity threshold, then the second weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is greater than the clarity threshold, then increase the first weight and / or decrease the second weight. Based on the adjusted weights, the visual grip probability, torque grip probability, and consistency probability are weighted and summed to obtain the target grip probability.

[0076] In specific embodiments, visual clarity can be determined based on image brightness, contrast, the clarity of the hand detection box, occlusion ratio, etc., and this application does not limit this. Furthermore, it should be noted that this application does not limit the specific values ​​of the first weight, second weight, and third weight. In one optional embodiment, the first weight is greater than the second weight, the second weight is greater than the third weight, and the sum of the first weight, second weight, and third weight is 1. For example, in one optional embodiment, the first weight is 0.5, the second weight is 0.3, and the third weight is 0.2.

[0077] In one optional embodiment, if the visual grip result indicates grip, the torque grip result indicates release, and the visual clarity is greater than the clarity threshold, it indicates that ADAS 20 single-lane centering control continuously assists steering on a relatively straight road, with the driver lightly gripping or loosely touching the steering wheel, while the EPS 22 hand torque is not obvious, but the camera can clearly observe that the hand is on the steering wheel rim area. In this case, the visual side weight is increased, that is, the first weight is increased, to avoid misjudging a light grip as a release.

[0078] In another optional embodiment, if the visual grip result indicates slippage, the torque grip result indicates gripping, and the visual clarity is not greater than the clarity threshold, it indicates that the visual error may have occurred due to camera obstruction, the hand being blocked by the steering wheel, or image blurring, while the EPS 22 still detects normal hand torque. In this case, the torque side weight is increased, i.e., the second weight is increased, to avoid poor visual quality causing the actual grip to be misjudged as slippage.

[0079] In another alternative embodiment, if the visual grip result indicates that the hand is released, the torque grip result indicates that the hand is held, and the visual clarity is greater than the clarity threshold, indicating that the visual image is clear, but the driver's hand is not identified as being located in the steering wheel rim area, and the EPS 22 outputs a grip tendency, it is possible that there is a non-hand object such as a mineral water bottle applying continuous torque to the steering wheel. In this case, the first weight is increased and / or the second weight is decreased.

[0080] Finally, based on the adjusted weights, the visual grip probability, torque grip probability, and consistency probability are weighted and summed to obtain the target grip probability.

[0081] Therefore, by dynamically adjusting the first, second, and third weights, the driver's grip on the steering wheel can be accurately identified in any scenario.

[0082] The above embodiments have described the intelligent hand-drop detection method in detail. This application also provides an embodiment of an intelligent hand-drop detection device.

[0083] Figure 6This is a schematic diagram of the structure of an intelligent hand-removal detection device provided in an embodiment of this application. The device is applied to... Figure 2 The intelligent hands-free detection system shown includes a driver assistance system 20, an image acquisition device 21 for acquiring images of the driver's hands, and an electric power steering system 22 for acquiring steering wheel torque information. Figure 6 As shown, the device includes: The information acquisition module 60 is used to acquire the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information when the driving assistance system is in an active state. The visual grip probability determination module 61 is used to determine the visual grip probability based on spatial parameters and driver's hand image; the visual grip probability is used to characterize the confidence level of the driver's hands visually gripping the steering wheel; The torque grip probability determination module 62 is used to determine the torque grip probability based on the steering wheel torque information; the torque grip probability is used to characterize the confidence of the driver's hands gripping the steering wheel under physical force. The consistency determination module 63 is used to determine the consistency result characterizing the driver's hand grip on the steering wheel based on the visual grip probability and the torque grip probability; The hands-free state determination module 64 is used to determine the driver's hands-free state based on the consistency results, visual grip probability, and torque grip probability.

[0084] In some embodiments, the spatial parameters include a preset vehicle reference point, the center point of the airbag cover, and the steering wheel size; determining the visual grip probability based on the spatial parameters and the driver's hand image includes: Construct a vehicle coordinate system based on the vehicle reference point and the center point of the airbag cover; Based on the steering wheel size, a spatial detection area corresponding to the edge of the steering wheel is constructed in the vehicle coordinate system; Extract key feature points of the driver's hand from the image of the driver's hand; Determine the coordinates of key hand feature points in the vehicle coordinate system to obtain the driver's hand feature points; The probability of visual grip is determined based on the spatial relationship between the driver's hand feature points and the spatial detection area.

[0085] In some embodiments, determining the visual grip probability based on the spatial relationship between the driver's hand feature points and the spatial detection area includes: Determine the first number of driver's hand feature points, and the second number of driver's hand feature points that fall within the spatial detection area; The completeness of hand features is determined based on the first quantity and the preset number of feature points; the completeness of hand features is used to characterize the degree of completeness of the driver's hand in the driver's hand image; Based on the first and second quantities, the hand-wheel contact degree is determined; the hand-wheel contact degree is used to characterize the degree of contact between the driver's hands and the steering wheel; The confidence level of the grip posture is determined based on the positional relationship between specified hand feature points and the spatial detection area; the confidence level of the grip posture is used to characterize the degree of confidence in the driver's grip on the steering wheel; The probability of visual gripping is determined based on the integrity of hand features, the contact degree of the hand rim, and the reliability of the grip posture.

[0086] In some embodiments, determining the torque grip probability based on steering wheel torque information includes: Obtain the steering wheel torque sequence within a preset time window; The torque amplitude characteristics, torque fluctuation characteristics, and effective torque duration characteristics are determined by the steering wheel torque sequence. Among them, the torque amplitude characteristics are used to characterize the intensity of the force applied by the driver's hands to the steering wheel; the torque fluctuation characteristics are used to characterize the stability of the applied force; and the effective torque duration characteristics are used to characterize the duration of the applied force. The torque hold probability is determined based on the characteristics of torque amplitude, torque fluctuation, and effective torque duration.

[0087] In some embodiments, the torque amplitude characteristic includes the mean of the absolute torque values ​​in the steering wheel torque sequence; the torque fluctuation characteristic includes the standard deviation of the torque values ​​in the steering wheel torque sequence; and the effective torque duration characteristic includes the number of times the ratio of the absolute torque value to the preset base torque value is greater than a preset ratio within a preset time window. Specifically, when the torque amplitude characteristic is greater than the first preset value, the effective torque duration characteristic is greater than the second preset value, and the torque fluctuation characteristic is less than the third preset value, the torque holding probability is reduced.

[0088] In some embodiments, visual grip probability and torque grip probability are determined to characterize the consistency of driver hand grip on the steering wheel, including: If the visual grip probability is greater than the first probability threshold, the visual grip result is determined to be gripping; otherwise, the visual grip result is determined to be releasing. If the torque grip probability is greater than the second probability threshold, the torque grip result is determined to be a grip; otherwise, the torque grip result is determined to be a release. Consistency is determined by whether the visual grip results and the torque grip results are the same.

[0089] In some embodiments, determining the driver's hands-free state based on consistency results, visual grip probability, and torque grip probability includes: If the consistency results represent the same state, then the consistency results are taken as the unheld state; the consistency results are either all held or all unheld. If the consistency results are inconsistent, then perform the following steps: Determine the consistency probability between visual grip probability and torque grip probability; The target grip probability is obtained by weighted summing of the visual grip probability, torque grip probability, and consistency probability. Determine if the target holding probability is greater than the target probability threshold; if it is, determine the release state as holding; if it is not, determine the release state as releasing.

[0090] In some embodiments, the target grip probability is obtained by weighted summation of the visual grip probability, torque grip probability, and consistency probability, including: The system obtains the first weight corresponding to the preset visual grip probability, the second weight corresponding to the torque grip probability, the third weight corresponding to the consistency probability, and the visual clarity of the driver's hand image. If the visual grip result represents grip, the torque grip result represents release, and the visual clarity is greater than the clarity threshold, then the first weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is not greater than the clarity threshold, then the second weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is greater than the clarity threshold, then increase the first weight and / or decrease the second weight. Based on the adjusted weights, the visual grip probability, torque grip probability, and consistency probability are weighted and summed to obtain the target grip probability.

[0091] Figure 7 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application, such as... Figure 7 As shown, the vehicle includes: a driver assistance system 20, an image acquisition device 21 for acquiring images of the driver's hands, an electric power steering system 22 for acquiring steering wheel torque information, a memory 70, and a processor 71. Wherein: Memory 70 is used to store computer programs; The processor 71 is used to execute a computer program to implement the steps of the intelligent off-hand detection method as described in the above embodiments.

[0092] The vehicles provided in this embodiment may include, but are not limited to, sedans, sport utility vehicles (SUVs), multi-purpose vehicles (MPVs), off-road vehicles, pickup trucks, or other power-driven, non-rail-borne vehicles.

[0093] The processor 71 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 71 may be implemented using at least one of the following hardware forms: Digital Signal Processor (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 71 may also include a main processor and a coprocessor. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 71 may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 71 may also include an Artificial Intelligence (AI) processor, which is used to handle computational operations related to machine learning.

[0094] The memory 70 may include one or more computer-readable storage media, which may be non-transitory. The memory 70 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 70 is used to store at least the following computer program 701, which, after being loaded and executed by the processor 71, is capable of implementing the relevant steps of the intelligent hands-free detection method disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 70 may also include an operating system 702 and data 703, and the storage method may be temporary or permanent storage. The operating system 702 may include Windows, Unix, Linux, etc. The data 703 may include, but is not limited to, the relevant data involved in the intelligent hands-free detection method.

[0095] In some embodiments, the vehicle may also include a display screen 72, an input / output interface 73, a communication interface 74, a power supply 75, and a communication bus 76.

[0096] Those skilled in the art will understand that Figure 7 The structure shown does not constitute a limitation on the vehicle and may include more or fewer components than illustrated.

[0097] The vehicle provided in this application includes a memory and a processor. When the processor executes the program stored in the memory, it can implement the intelligent hands-free detection method described in the above embodiments.

[0098] It should be noted that although the operations are depicted in a specific order in the accompanying drawings, this should not be construed as requiring these operations to be performed in the specific order shown or sequentially, or requiring all illustrated operations to be performed to achieve the desired result. In some cases, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system modules and components in the above embodiments should not be construed as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

Claims

1. A smart hand-free detection method, characterized in that, The method is applied to an intelligent hands-free detection system, which includes a driving assistance system, an image acquisition device for acquiring images of the driver's hands, and an electric power steering system for acquiring steering wheel torque information, including: When the driving assistance system is in an active state, the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information are acquired. Based on the spatial parameters and the driver's hand image, the visual grip probability is determined; the visual grip probability is used to characterize the confidence level that the driver's hands are visually gripping the steering wheel. Based on the steering wheel torque information, the torque grip probability is determined; the torque grip probability is used to characterize the confidence level of the driver's hand gripping the steering wheel under physical force. Based on the visual grip probability and the torque grip probability, a consistency result characterizing the driver's hand grip on the steering wheel is determined; The driver's hands-free state is determined based on the consistency results, the visual grip probability, and the torque grip probability.

2. The intelligent hand-free detection method as described in claim 1, characterized in that, The spatial parameters include a preset vehicle reference point, the center point of the airbag cover, and the steering wheel size; determining the visual grip probability based on the spatial parameters and the driver's hand image includes: Construct a vehicle coordinate system based on the vehicle reference point and the center point of the airbag cover; Based on the steering wheel size, a spatial detection area corresponding to the edge of the steering wheel is constructed in the vehicle coordinate system; Extract key feature points of the driver's hand from the image of the driver's hand; The coordinates of the key hand feature points in the vehicle coordinate system are determined to obtain the driver's hand feature points; The visual grip probability is determined based on the spatial relationship between the driver's hand feature points and the spatial detection area.

3. The intelligent hand-free detection method as described in claim 2, characterized in that, The step of determining the visual grip probability based on the spatial relationship between the driver's hand feature points and the spatial detection area includes: Determine a first number of the driver's hand feature points, and a second number of the driver's hand feature points that fall within the spatial detection area; Based on the first quantity and the preset number of feature points, the completeness of the hand features is determined; the completeness of the hand features is used to characterize the degree of completeness of the driver's hand in the driver's hand image; Based on the first quantity and the second quantity, the hand rim contact degree is determined; the hand rim contact degree is used to characterize the degree of contact between the driver's hand and the steering wheel; The grip confidence level is determined based on the positional relationship between specified hand feature points and the spatial detection area; the grip confidence level is used to characterize the degree of confidence in the driver's hand gripping the steering wheel; The visual grip probability is determined based on the completeness of the hand features, the contact degree of the hand rim, and the confidence of the grip posture.

4. The intelligent hand-free detection method as described in claim 1, characterized in that, The step of determining the torque grip probability based on the steering wheel torque information includes: Obtain the steering wheel torque sequence within a preset time window; The torque amplitude characteristic, torque fluctuation characteristic, and effective torque duration characteristic are determined using the steering wheel torque sequence; wherein, the torque amplitude characteristic is used to characterize the intensity of the force applied by the driver's hands to the steering wheel; the torque fluctuation characteristic is used to characterize the stability of the applied force; and the effective torque duration characteristic is used to characterize the duration of the applied force. The torque holding probability is determined based on the torque amplitude characteristics, the torque fluctuation characteristics, and the effective torque duration characteristics.

5. The intelligent hand-free detection method as described in claim 4, characterized in that, The torque amplitude feature includes the mean of the absolute torque values ​​in the steering wheel torque sequence; the torque fluctuation feature includes the standard deviation of the torque values ​​in the steering wheel torque sequence; the effective torque duration feature includes the number of times the ratio of the absolute torque value to the preset base torque value is greater than a preset ratio within the preset time window. Specifically, when the torque amplitude characteristic is greater than a first preset value, the effective torque duration characteristic is greater than a second preset value, and the torque fluctuation characteristic is less than a third preset value, the torque holding probability is reduced.

6. The intelligent hand-free detection method as described in any one of claims 1 to 5, characterized in that, Determining the visual grip probability and the torque grip probability, characterizing the consistency of the driver's hand grip on the steering wheel, includes: If the visual grip probability is greater than the first probability threshold, then the visual grip result is determined to be gripping; otherwise, the visual grip result is determined to be releasing. If the torque grip probability is greater than the second probability threshold, then the torque grip result is determined to be a grip; otherwise, the torque grip result is determined to be a release. The consistency result is determined based on whether the visual grip result and the torque grip result are the same.

7. The intelligent hand-free detection method as described in claim 6, characterized in that, Based on the consistency results, the visual grip probability, and the torque grip probability, the driver's hands-free state is determined, including: If the consistency results represent the same state, then the consistency results are taken as the released state; the consistency results are either all holding or all released. If the consistency results are inconsistent, then the following steps are performed: Determine the consistency probability between the visual grip probability and the torque grip probability; The target grip probability is obtained by weighted summing of the visual grip probability, the torque grip probability, and the consistency probability. Determine whether the target holding probability is greater than the target probability threshold; if it is greater, determine that the release state is holding; if it is not greater, determine that the release state is release.

8. The intelligent hand-free detection method as described in claim 7, characterized in that, The target grip probability is obtained by weighted summing of the visual grip probability, the torque grip probability, and the consistency probability, including: Obtain the preset first weight corresponding to the visual grip probability, the second weight corresponding to the torque grip probability, the third weight corresponding to the consistency probability, and the visual clarity of the driver's hand image; If the visual grip result represents gripping, the torque grip result represents release, and the visual clarity is greater than the clarity threshold, then the first weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is not greater than the clarity threshold, then the second weight is increased; If the visual grip result indicates release, the torque grip result indicates grip, and the visual clarity is greater than the clarity threshold, then the first weight is increased and / or the second weight is decreased. Based on the adjusted weights, the visual grip probability, the torque grip probability, and the consistency probability are weighted and summed to obtain the target grip probability.

9. An intelligent hand-removal detection device, characterized in that, The device is applied to an intelligent hands-off detection system, which includes a driving assistance system, an image acquisition device for acquiring images of the driver's hands, and an electric power steering system for acquiring steering wheel torque information. The device includes: The information acquisition module is used to acquire the spatial parameters of the steering wheel, the image of the driver's hands, and the steering wheel torque information when the driving assistance system is in an active state. A visual grip probability determination module is used to determine the visual grip probability based on the spatial parameters and the driver's hand image; the visual grip probability is used to characterize the confidence level of the driver's hand visually gripping the steering wheel; The torque grip probability determination module is used to determine the torque grip probability based on the steering wheel torque information; the torque grip probability is used to characterize the confidence level of the driver's hand gripping the steering wheel under physical force. A consistency determination module is used to determine a consistency result characterizing the driver's hand gripping the steering wheel based on the visual grip probability and the torque grip probability; The hands-free state determination module is used to determine the driver's hands-free state based on the consistency result, the visual grip probability, and the torque grip probability.

10. A vehicle, characterized in that, The vehicle includes a driver assistance system, an image acquisition device for acquiring images of the driver's hands, an electric power steering system for acquiring steering wheel torque information, a memory, and a processor; The memory stores a computer program that can run on the processor, and when the processor executes the computer program, it implements the steps of the intelligent off-hand detection method according to any one of claims 1 to 8.

Images