An intelligent driving human active exit detection method and system based on multi-modal perception

Abstract

This invention relates to the field of intelligent vehicle control technology, and discloses a method and system for detecting voluntary exit of intelligent driving based on multimodal perception. The invention includes acquiring identity authentication data, human contact data, steering wheel grip data, and active operation data; performing temporal correlation within a preset voluntary exit determination time window to form a hard-gated evidence chain; calculating a voluntary exit intent assessment value when each evaluation parameter meets basic threshold conditions, and generating exit safety parameters based on vehicle operating state parameters; and determining a control strategy based on the voluntary exit intent assessment value and the exit safety parameters, including direct switching, controlled handover, or maintaining the intelligent driving mode. This invention does not use steering wheel wobbling information, steering wheel displacement information, or steering wheel torque change information as factors in calculating the voluntary exit intent assessment value, which helps improve the reliability of voluntary exit recognition and the safety of driving handover.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle control technology, and in particular to an intelligent driver's active exit detection method and system based on multimodal perception. Background Technology

[0002] Controlling the exit from intelligent driving mode or the transfer of driving control is a crucial aspect of intelligent driving control. In scenarios where the driver actively requests to exit intelligent driving mode, it is necessary to simultaneously recognize the driver's true intention and ensure the safety of the control transfer process. The steering wheel is the most direct human-machine interface between the driver and the vehicle, and it is also the primary signal source for existing related detection solutions.

[0003] Existing steering wheel status detection solutions related to autonomous driving disengagement or takeover often use signals such as steering wheel torque, capacitance, pressure, or visual hands-off detection as the basis for determining whether the driver is holding the steering wheel, whether intervention is required, or whether a warning is needed. For example: Publication number CN114852092A discloses a method, device, readable storage medium, and vehicle for detecting steering wheel hands-off. This solution periodically acquires the steering wheel capacitance and torque values, using the change in steering wheel torque as a key factor in hands-off detection. The detection result is determined by combining this with the operating state of the capacitance sensor, belonging to a detection logic based on physical disturbances or gripping states of the steering wheel. The drawback of this solution is that in scenarios where the driver actively requests to exit intelligent driving mode, changes in steering wheel torque may originate from driver input, road bumps, vehicle vibrations, or hand posture adjustments. Relying solely on torque and capacitance changes for judgment carries the risk of misidentifying these non-active exit factors as exit-related signals.

[0004] Publication No. CN113247078B discloses a method and system for detecting steering wheel hands-off. This scheme calculates a standard hand-grip threshold based on a steering wheel capacitive sensor, steering wheel heating state, heating duration, and initial temperature, and then determines whether the steering wheel is in a hands-off state. In this scheme, temperature-related information is mainly used to correct or calculate the capacitive hands-off detection threshold. The shortcomings of this scheme are: the temperature signal is only used as an auxiliary correction parameter for capacitive detection; it does not treat temperature as an independent sensing dimension to determine whether there is real, continuous human contact between the driver's hands and the steering wheel, nor does it address the collaborative relationship between the temperature signal and other sensing signals in the exit determination process.

[0005] Regarding driver identification, there are already publicly available methods for identification using the steering wheel area, for example: Publication number CN108263338B discloses a method for identifying a car, steering wheel, and driver. This solution installs sensors on locations such as the steering wheel that come into contact with the driver's hands to acquire the driver's hand features and identify the driver's identity. The identification function is mainly used for driver authentication or vehicle-related permission / configuration management. The shortcomings of this solution are: the identification function is mainly used for driver authentication or vehicle-related permission / configuration management, and it does not involve continuous identity verification of the operator initiating the exit request during intelligent driving operation, nor does it correlate the authentication result with the hand contact state, grip state, and active operation commands in a timely manner.

[0006] Publication number CN109703487A discloses a fingerprint recognition system and method based on a car steering wheel. This solution incorporates a fingerprint recognition detection module within the steering wheel to identify the driver and access their personalized settings. However, this solution is limited to the use of fingerprint recognition for comfort settings and does not address driver control transfer confirmation in intelligent driving modes. Furthermore, it does not disclose a technical solution for making fingerprint recognition a necessary condition in exit procedures.

[0007] Regarding steering wheel temperature and grip strength detection, there are already solutions that combine both for driving condition monitoring, for example: Publication number CN107198526B discloses a method and system for assisted driving. This scheme collects the driver's fingertip temperature and steering wheel grip force in real time, and obtains driving safety adaptability values ​​related to hand temperature and grip force through a metrological model to reflect the driving state in real time. This scheme focuses on monitoring the driver's safety adaptability or dangerous driving conditions, and its temperature and grip force signals are used for driving state assessment. The shortcomings of this scheme are: while the temperature and grip force signals are used for continuous monitoring and trend analysis of the driving state, they do not address using temperature and grip force as evidence of whether the driver's hands maintain genuine, continuous, and stable contact before and after an active exit operation event, nor do they perform time-series correlation determination with authentication signals and active operation signals.

[0008] Regarding the use of grip force for driving intention recognition, relevant technologies have been disclosed, such as: Publication number CN113460059A discloses a driver engagement identification device and method based on a smart steering wheel. This solution installs pressure sensors on the left and right hand grip areas of the steering wheel to collect grip force data and uses an identification model to identify the driver's engagement. The shortcomings of this solution are: while using grip force signals for classifying and identifying engagement, it does not consider grip force signals as a grip stability parameter in assessing intention to actively disengage, nor does it perform joint determination with identity authentication signals, human contact signals, and active operation signals within the same time window.

[0009] Regarding the activation and deactivation of intelligent driving functions, there are already solutions that trigger driving function switching through user interface components, for example: Publication number CN112041192B discloses an activation and deactivation method for an automated driving function with longitudinal and lateral control. This solution involves a user interface and input components for activating or deactivating the driving function. The drawback of this solution is that, although it involves a user interface input component and driving function activation or deactivation control, it does not disclose the temporal correlation verification of the input trigger signal with continuous evidence such as driver authentication, hand contact status, and grip status within the same active exit determination time window. This makes it difficult to distinguish between authorized driver's intentional operation and accidental or unauthorized operation by the passenger.

[0010] Regarding the transfer of driving rights and the safety control of vehicle status, relevant solutions have already been made public, for example: Publication number CN106873596B discloses a vehicle control method and device. This scheme acquires human-operated information about the vehicle, determines the driver's intervention intention based on this information, and when the vehicle is in autonomous driving mode and the intervention intention is a slow intervention, controls the vehicle to reach a predetermined safe state before transferring control to the driver. This scheme focuses on the safety state control and control transfer after the driver's intervention intention, with the intervention intention primarily derived from human-operated information. The shortcomings of this scheme are: its emphasis is on detecting the safety state control and control transfer after the driver's intervention intention, but it does not disclose the formation of a hard-gating evidence chain through the temporal correlation of identity authentication, human contact, steering wheel grip, and active operation data during the active exit request formation stage, nor does it disclose using the active exit intention evaluation value and exit safety parameters together as decision inputs for direct switching, controlled transfer, or maintaining intelligent driving.

[0011] In summary, the existing technologies described above have the following shortcomings: First, existing steering wheel torque, capacitance, pressure, or hands-off detection schemes are mainly used to determine whether the driver is holding the steering wheel or whether there are signs of intervention. In scenarios where the driver actively requests to exit the intelligent driving mode, steering wheel wobbling, displacement, or torque changes are easily affected by road bumps, vehicle vibrations, and hand posture adjustments, making it difficult to independently and stably represent the driver's clear intention to actively exit.

[0012] Second, existing identity recognition, temperature detection, grip strength detection, and button triggering technologies are typically used for identity verification, health monitoring, grip status judgment, or driving function switching, respectively. They do not use the moment of active operation data triggering as a benchmark to temporally correlate continuous evidence such as identity authentication, human contact, and steering wheel grip with the instantaneous event of active operation within the preceding and following time windows, nor do they form a hard-gating evidence chain for active exit scenarios.

[0013] Third, while existing driving handover or safety status control schemes can perform a safe handover after detecting the driver's intention to intervene, they typically do not introduce a multimodal hard-gating evidence chain consisting of identity authentication, human contact, steering wheel grip, and active operation during the active exit request formation stage. Nor do they combine the active exit intention assessment value generated by this evidence chain with the exit safety parameters formed by the vehicle's operating status to determine differentiated control strategies for direct switching, controlled handover, or maintaining intelligent driving. Summary of the Invention

[0014] Given that existing technologies suffer from issues such as the susceptibility of steering wheel torque, displacement, wobbling, capacitance, or hands-off detection signals to road bumps, vehicle vibrations, and driver hand posture adjustments, making it difficult to stably represent the driver's true intention to actively exit the intelligent driving mode; and that existing signals such as identity recognition, human contact detection, grip detection, and button triggering are mostly used for vehicle identity verification, driving status monitoring, grip status judgment, or driving function switching, lacking a temporal correlation, hard-gated evidence chain, and joint decision-making mechanism for safe exit in intelligent driving scenarios, resulting in insufficient reliability of active exit recognition and safety of driving handover, a multimodal perception-based method and system for detecting active exit of intelligent driving is proposed.

[0015] The purpose is to address scenarios where drivers actively request to exit intelligent driving mode during operation. By using the onboard intelligent driving controller to sequentially correlate identity authentication data, human contact data, steering wheel grip data, and active operation data, a hard-gated evidence chain is formed between continuous evidence and instantaneous active operation events. Combined with the vehicle's operating status, exit safety parameters are generated. This allows the system to determine the exit control strategy—direct switching, controlled handover, or maintaining intelligent driving—based on the vehicle's current operating safety status while confirming the driver's genuine intention to exit. This improves the reliability of intelligent driving active exit detection and the safety of the driving handover process.

[0016] In this application, the driver's active exit from intelligent driving mode refers to the technical process in which, while the vehicle is operating in intelligent driving mode, the driver requests to end the intelligent driving mode through a preset active operation method, and the on-board intelligent driving controller determines the exit control strategy based on the collected multimodal perception data and vehicle operating status parameters; the exit control strategy includes three types of strategies: directly switching to manual driving mode, entering a controlled handover process, and maintaining intelligent driving mode.

[0017] The active exit intention assessment in this application refers to the technical process by which the on-board intelligent driving controller performs parameterized processing, temporal correlation, hard gate judgment, and weighted fusion calculation on identity authentication data, human contact data, steering wheel grip data, and active operation data to generate an active exit intention assessment result. The active exit intention assessment generates an assessment result based on sensor-collected data and vehicle control parameters. It is not a direct judgment of the driver's subjective psychological state, nor is it a rule-based judgment process executed manually.

[0018] The exit safety assessment in this application refers to the technical process by which the onboard intelligent driving controller generates vehicle operating state parameters based on vehicle speed, road type, autonomous driving level, lateral control state, longitudinal control state, distance to surrounding vehicles, and collision risk level, and generates exit safety parameters based on the vehicle operating state parameters; the exit safety parameters are used to characterize the degree to which the current vehicle operating state meets the safe exit conditions, and serve as input parameters for determining the exit control strategy.

[0019] The technical solution of this invention is a method for detecting active driver exit in intelligent driving based on multimodal perception, executed by an onboard intelligent driving controller, characterized by comprising: S1. Acquire driver multimodal perception data, which includes identity authentication data, human body contact data, steering wheel grip data, and active operation data.

[0020] S2. Within the preset active exit judgment time window, perform time-series correlation on identity authentication data, human body contact data, steering wheel grip data and active operation data to generate identity credibility parameters, contact validity parameters, grip stability parameters and operation initiative parameters respectively.

[0021] Among them, the temporal correlation takes the trigger time of the active operation data as the benchmark, and associates the identity authentication data, human body contact data, and steering wheel grip data as continuous evidence with the active operation data within a preset time window before and after the trigger time of the active operation data.

[0022] S3. Generate an assessment result of the voluntary exit intention based on the identity credibility parameter, contact validity parameter, grip stability parameter, and operation initiative parameter. The generation of the voluntary exit intention assessment result includes: Based on whether the identity credibility parameter, contact validity parameter, grip stability parameter, and operation initiative parameter respectively meet their respective basic threshold conditions, hard gating parameters are generated. .

[0023] Among them, hard gating parameters This is a binary decision parameter used to control whether the evaluation value of the intention to actively withdraw is included in the weighted calculation.

[0024] When the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter all meet their respective basic threshold conditions, the hard-gating parameter... The value of the intention to actively withdraw is set to 1, and the weighted fusion result based on the identity credibility parameter, contact validity parameter, grip stability parameter, and operation initiative parameter is used to calculate the evaluation value. .

[0025] When the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter do not all meet their respective basic threshold conditions, the hard gating parameter... Set it to 0, and set the value of the intention to actively exit as the evaluation value. Set to zero.

[0026] S4. Obtain vehicle operating status parameters and generate exit safety parameters based on the vehicle operating status parameters. ; S5. Based on the assessment results of the intention to exit and the exit security parameters. Determine the exit control strategy, including: When the intention to actively withdraw is evaluated ≥Preset intent threshold and exit security parameter When the speed reaches or exceeds the preset safety threshold, the vehicle will switch from intelligent driving mode to manual driving mode.

[0027] When the intention to actively withdraw is evaluated ≥Preset intent threshold and exit security parameter When a preset safety threshold is reached, the vehicle is controlled to enter a controlled handover process, which includes: A secondary confirmation prompt is output, progressive deceleration control is executed, and the safety parameter is exited. Once the preset safety threshold is met, the vehicle will switch from intelligent driving mode to manual driving mode.

[0028] When exiting the security parameters If the temperature remains below the preset safety threshold for an extended period of time, maintain the intelligent driving mode.

[0029] When the intention to actively withdraw is evaluated When the preset intent threshold is reached, the in-vehicle intelligent driving controller maintains the intelligent driving mode.

[0030] During the generation of the active exit intention assessment result, steering wheel wobbling information, steering wheel displacement information, and steering wheel torque change information are not included in the active exit intention assessment value. The calculation is not sufficient to trigger an active exit condition on its own.

[0031] Furthermore, the identity authentication data is fingerprint recognition data. The fingerprint template is stored in the vehicle's local secure storage unit, and the identity comparison is completed locally on the vehicle's intelligent driving controller. The vehicle's intelligent driving controller generates identity credibility parameters based on the fingerprint comparison and matching results.

[0032] Furthermore, the human contact data is temperature detection data, and the contact effectiveness parameter is generated based on whether the temperature detection data is within the preset human contact temperature range, the duration of temperature change, the rate of temperature rise, and the result of ambient temperature compensation.

[0033] Furthermore, the steering wheel grip data is grip force detection data, and the grip stability parameter is generated based on whether the grip force detection data is within the preset grip force range, grip force duration, grip force fluctuation amplitude, and the consistency of grip on both sides of the steering wheel.

[0034] Furthermore, the active operation data is physical button trigger data. The physical button trigger modes include long press mode and double-click mode. The long press mode is determined based on a preset long press duration threshold, and the double-click mode is determined based on a preset double-click interval threshold. The preset long press duration threshold and the preset double-click interval threshold are set according to the parameter configuration corresponding to the driver's identity.

[0035] Furthermore, in S2, the identity authentication data, human body contact data, steering wheel grip data, and active operation data are correlated in a time sequence, including: within a preset time window before and after the active operation data triggering time, determining whether the identity credibility parameter, contact validity parameter, and grip stability parameter all meet their respective basic threshold conditions; only when the identity credibility parameter, contact validity parameter, and grip stability parameter continuously meet the basic threshold conditions within the time window, and the active operation data conforms to the preset active triggering mode, the vehicle intelligent driving controller recognizes the active operation data as a valid active exit event.

[0036] Furthermore, in S3, the evaluation value of the intention to actively withdraw. It is generated by combining hard gating judgment with linear weighted fusion.

[0037] When the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter each meet their respective basic threshold conditions, the hard-gating parameter... Set to 1, the evaluation value for the intention to actively exit. Calculate using the following formula: When not all of the basic threshold conditions are met among the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter, the hard gating parameter... Set to 0, indicating an intention to actively exit the evaluation value. Set to 0; in, For identity credibility parameters, For contact effectiveness parameters, For grip stability parameters, For the operational initiative parameter, , , , All parameters are normalized and have a pre-set basic threshold for identity credibility. Basic threshold for contact effectiveness Basic threshold for grip stability and the basic threshold of operational initiative ,when , , , At that time, hard-gating parameters Take 1, when , , , When any parameter in the parameter does not meet its corresponding basic threshold, the hard-gated parameter Take 0, , , , For preset weighting coefficients, ,and .

[0038] Preset active exit conditions include hard-gating parameters =1 and the assessment value of the intention to actively withdraw. ≥Preset Intent Threshold , .

[0039] Furthermore, in S4, vehicle operating status parameters include vehicle speed, road type, level of autonomous driving, lateral control status, longitudinal control status, distance to surrounding vehicles, and collision risk level; exit safety parameters... Calculate using the following formula: in, For vehicle speed safety parameters, For road type safety parameters, For lateral control stability parameters, For longitudinal control stability parameters, For safety parameters regarding the distance to surrounding vehicles, For collision risk safety parameters, , , , , , All are normalized parameters. , , , , , For preset weighting coefficients, ,and Preset safe exit conditions include ≥Preset safety threshold.

[0040] Furthermore, the present invention also provides a multimodal perception-based intelligent driver proactive exit system for implementing the above-mentioned multimodal perception-based intelligent driver proactive exit detection method, the system comprising: The multimodal sensing module includes a fingerprint recognition unit, a temperature sensing unit, and a grip force detection unit located in the steering wheel grip area, as well as a physical button control unit located in the multi-function operation area of ​​the steering wheel. The multimodal sensing module is used to collect identity authentication data, human body contact data, steering wheel grip data, and active operation data.

[0041] The vehicle operating status acquisition module is used to acquire vehicle operating status parameters.

[0042] The vehicle-mounted intelligent driving controller is electrically connected to a multimodal perception module and a vehicle operating status acquisition module. The vehicle-mounted intelligent driving controller includes an active exit intention assessment unit, an exit safety assessment unit, and an exit control unit.

[0043] The active exit intent assessment unit is configured to generate active exit intent assessment results based on identity authentication data, human body contact data, steering wheel grip data, and active operation data.

[0044] The exit safety assessment unit is configured to generate exit safety parameters based on vehicle operating status parameters.

[0045] The exit control unit is configured to determine the exit control strategy based on the results of the active exit intent assessment and the exit safety parameters.

[0046] When abnormal situations occur, such as missing or abnormal identity authentication data, missing or abnormal human contact data, missing or abnormal steering wheel grip data, or missing or abnormal active operation data, the vehicle intelligent driving controller restricts the execution of the direct exit strategy and outputs risk warnings and secondary confirmation prompts. After secondary confirmation, if the preset downgrade judgment conditions are met, the vehicle intelligent driving controller determines whether to output a controlled handover process instruction based on the exit safety parameters. If the preset downgrade judgment conditions are not met, the intelligent driving mode is maintained.

[0047] Furthermore, the present invention also provides a vehicle, including a steering wheel, and the aforementioned intelligent driver active exit system based on multimodal perception.

[0048] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a temporal correlation mechanism within a preset active exit determination time window. Within the time window before and after the active operation event trigger, it correlates continuous evidence such as identity authentication, human contact, and steering wheel grip with active operation data, forming a hard-gated evidence chain composed of identity authentication data, human contact data, steering wheel grip data, and active operation data. When the evaluation parameters corresponding to the above four types of data all meet their respective basic threshold conditions, the hard-gated evidence chain closes, and the active exit intention evaluation value enters the weighted calculation process; when the above evaluation parameters do not all meet their respective basic threshold conditions, the active exit intention evaluation result is limited. This hard-gated evidence chain mechanism helps distinguish between driver active exit behavior and interference situations such as road bumps, brief hand displacement, unauthorized touch, or accidental operation, improving the reliability of active exit intention recognition and helping to reduce the probability of false triggering.

[0049] 2. This invention introduces an exit safety parameter, quantifying vehicle operating states such as vehicle speed, road type, autonomous driving level, lateral control stability, longitudinal control stability, distance to surrounding vehicles, and collision risk level into safety indicators. These indicators are then combined with the active exit intention assessment value to determine the exit control strategy. When both the active exit intention assessment value and the exit safety parameter meet threshold conditions, the vehicle can be directly switched to manual driving mode. When the active exit intention assessment value meets the threshold conditions but the exit safety parameter does not, a controlled handover process is initiated, outputting a secondary confirmation prompt, executing progressive deceleration control, and switching driving modes only after safety conditions are met. By selecting different exit methods based on the vehicle's operating state, the safety of both the driver's active exit request and the handover of control is balanced.

[0050] 3. This invention uses fingerprint recognition data as identity authentication data and takes the identity credibility parameter as one of the necessary conditions for the hard-gated evidence chain. When an unauthorized person touches the steering wheel and triggers the physical button, if the identity credibility parameter does not meet the basic threshold condition, the hard-gated evidence chain cannot be closed. The system can maintain the intelligent driving mode and output a prompt, thereby helping to reduce the risk of unauthorized operation causing abnormal exit of the intelligent driving mode and improving the reliability of exit permission control.

[0051] 4. This invention uses temperature detection data to generate contact effectiveness parameters and grip force detection data to generate grip stability parameters. This allows temperature and grip force signals to serve as evidence of genuine, continuous hand contact and stable grip on the steering wheel in active exit scenarios, rather than solely for fatigue monitoring or driving condition assessment. By including contact effectiveness and grip stability parameters as necessary conditions in the hard-gated evidence chain, it helps confirm that the active exit request is initiated while the driver's hands are continuously in contact with and stably gripping the steering wheel, thus improving the reliability of the exit determination.

[0052] 5. In generating the active exit intention assessment value, this invention does not use steering wheel wobbling information, steering wheel displacement information, and steering wheel torque change information as calculation factors for the active exit intention assessment value, and the aforementioned information does not constitute sufficient conditions for triggering active exit on its own. Physical disturbances to the steering wheel caused by road bumps, vehicle vibrations, or driver hand adjustments are unlikely to constitute a valid active exit event on its own, thereby helping to reduce the impact of road disturbances on active exit determination and improving the continuity of intelligent driving mode operation in non-exit scenarios.

[0053] 6. When identity authentication data, human body contact data, steering wheel grip data, or active operation data are missing or abnormal, this invention can restrict the execution of the direct exit strategy and output risk warnings and secondary confirmation prompts. When the preset degradation judgment conditions are met, it can determine whether to enter the controlled handover process based on the exit security parameters. By setting a degradation processing mechanism in abnormal situations, it helps to reduce the adverse impact of abnormal data on exit decisions and improve the robustness and functional security of the system. Attached Figure Description

[0054] Figure 1 This is a block diagram of a multimodal perception-based intelligent driving active exit detection system according to the present invention. Figure 2 This is a flowchart of a method for detecting the active exit of intelligent driving based on multimodal perception, according to the present invention. Figure 3 This is a schematic diagram of the timing correlation of the present invention; Figure 4 This is a schematic diagram illustrating the calculation logic of the active exit intention evaluation value of the present invention; Figure 5 This is a schematic diagram of the logic for calculating the exit safety parameter of the present invention; Figure 6 This is a schematic diagram of the exit control strategy decision logic of the present invention; Figure 7 This is a schematic diagram of the controlled transfer process of the present invention. Detailed Implementation

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0056] This invention can be applied to vehicles with intelligent driving mode and manual driving mode switching functions. The intelligent driving mode includes L2, L3, or L4 level intelligent driving modes. The following embodiments use L3 level intelligent driving mode as an example for illustration and do not constitute a limitation on the level of intelligent driving.

[0057] Example 1, referring to Figure 1 - Figure 7 This is the first embodiment of the present invention, which provides a method for detecting active exit of intelligent driving based on multimodal perception. This embodiment is illustrated by a scenario in which the vehicle is in L3 level intelligent driving mode, driving on urban roads with a speed lower than a preset threshold, and the driver actively requests to exit by pressing and holding a physical button.

[0058] This method operates in an onboard intelligent driving controller, which is electrically connected to a multimodal perception module and a vehicle operating status acquisition module.

[0059] The multimodal sensing module includes a fingerprint recognition unit, a temperature sensing unit, and a grip force detection unit located in the steering wheel grip area, as well as a physical button control unit located in the steering wheel multi-function operation area.

[0060] The vehicle operating status acquisition module is used to acquire vehicle operating status parameters from the vehicle CAN bus or in-vehicle Ethernet.

[0061] Data is transmitted between the modules via the vehicle communication bus.

[0062] The multimodal perception module sends the collected identity authentication data, human contact data, steering wheel grip data, and active operation data to the vehicle intelligent driving controller.

[0063] The vehicle operation status acquisition module sends the acquired vehicle operation status parameters, such as vehicle speed, road type, autonomous driving level, lateral control status, longitudinal control status, distance to surrounding vehicles, and collision risk level, to the on-board intelligent driving controller.

[0064] The active exit intention assessment unit, exit safety assessment unit, and exit control unit inside the vehicle intelligent driving controller process the above data in sequence, and finally output the exit control strategy to the vehicle actuator.

[0065] Specifically, the method of this embodiment includes the following steps: Step S1: Acquire driver multimodal perception data.

[0066] The onboard intelligent driving controller acquires driver multimodal perception data collected by the multimodal perception module. This driver multimodal perception data includes identity authentication data, human contact data, steering wheel grip data, and active operation data.

[0067] The fingerprint recognition unit, located in the steering wheel grip area, collects the driver's fingerprint information and generates fingerprint recognition data.

[0068] A temperature sensing unit located in the steering wheel grip area collects temperature detection data of the area where the driver's palm contacts the steering wheel.

[0069] A grip force detection unit, located in the steering wheel grip area or inside the steering wheel, collects data on the grip force exerted by the driver's hands on the steering wheel.

[0070] The physical button control unit located in the multi-function control area of ​​the steering wheel monitors button trigger signals and generates physical button trigger data.

[0071] Step S2: Generate identity credibility parameters Contact effectiveness parameters Grip stability parameters and operational initiative parameters .

[0072] Within a preset active exit determination time window, the vehicle intelligent driving controller performs time-series correlation on identity authentication data, human body contact data, steering wheel grip data, and active operation data to generate identity credibility parameters. Contact effectiveness parameters Grip stability parameters and operational initiative parameters .

[0073] To facilitate unified processing of sensor data from different sources and with different dimensions by the onboard intelligent driving controller, identity credibility parameters are used. Contact effectiveness parameters Grip stability parameters and operational initiative parameters All parameters are represented by normalized parameters, with values ​​ranging from 0 to 1.

[0074] A value of 0 indicates that the condition is not met at all, a value of 1 indicates that the condition is met at all, and intermediate values ​​indicate the degree of partial satisfaction.

[0075] When the normalized calculation result is greater than 1, the on-board intelligent driving controller will truncate the corresponding parameter to 1; when the normalized calculation result is less than 0, the on-board intelligent driving controller will truncate the corresponding parameter to 0.

[0076] In this embodiment, each normalization parameter can be generated using interval matching, duration ratio, deviation deduction, or minimum safety margin. For parameters that need to be generated based on the degree of deviation, the on-board intelligent driving controller compares the measured value with a preset effective interval, a preset threshold, or a preset margin. When the measured value meets the corresponding condition, the normalization parameter is set to 1; when the measured value does not meet the corresponding condition, a normalization value between 0 and 1 is generated based on the deviation of the measured value from the corresponding threshold. Normalization values ​​greater than 1 are truncated to 1, and values ​​less than 0 are truncated to 0.

[0077] Identity credibility parameters Generation process: Identity credibility parameters This is used to characterize the degree of matching between the currently collected driver identity authentication data and the pre-stored identity template in the vehicle's local secure storage unit. The vehicle intelligent driving controller compares the driver's fingerprint information collected by the fingerprint recognition unit with the pre-stored fingerprint template in the vehicle's local secure storage unit. The comparison algorithm outputs a matching score. and maximum matching score ,and Greater than 0. Identity credibility parameter Calculate using the following formula: The higher the degree of match between the driver's fingerprint and the pre-stored fingerprint template, the higher the identity credibility parameter. The closer the value is to 1, the lower the degree of matching; the higher the identity credibility parameter. The closer it is to 0.

[0078] In this embodiment, the driver's fingerprint is matched with the authorized fingerprint template in the vehicle's local secure storage unit, and the matching score is equal to the maximum matching score. .

[0079] Contact effectiveness parameters The generation process: contact effectiveness parameters This parameter is used to characterize whether a real, continuous, and effective human contact is formed between the driver's hands and the steering wheel grip area. The in-vehicle intelligent driving controller uses a multi-sub-index linear weighted fusion method to generate the contact effectiveness parameter. .

[0080] Among them, temperature range matching parameters Used to characterize whether the temperature detection data is within a preset human contact temperature range; temperature duration parameter Used to characterize whether the time during which temperature detection data remains within the valid range meets the requirements; temperature rise rate parameter It is used to characterize whether the temperature change process conforms to the characteristics of heat conduction changes when in contact with the human body. , and All are normalized parameters.

[0081] The vehicle-mounted intelligent driving controller first determines whether the temperature detection data falls within the preset human contact temperature range, or whether it falls within the preset human contact temperature range after ambient temperature compensation. When the temperature detection data falls within the preset human contact temperature range, Set to 1; when the temperature detection data deviates from the preset human contact temperature range, The normalized value is taken as a number between 0 and 1, depending on the degree of deviation.

[0082] Then, the onboard intelligent driving controller determines whether the time the temperature remains within the effective range has reached the preset contact duration. When the temperature duration reaches the preset contact duration, Set to 1; when the temperature duration is less than the preset contact duration, Generates based on the ratio of the actual duration to the preset contact duration.

[0083] Next, determine whether the rate of temperature change conforms to the characteristics of heat conduction changes upon human contact. When the rate of temperature change is within the preset range for human contact temperature change, Set to 1; when the rate of temperature change deviates from the preset range, The normalized value is taken as a number between 0 and 1, depending on the degree of deviation.

[0084] Contact effectiveness parameters Calculate using the following formula: in, , , For preset weighting coefficients, ,and .

[0085] In this embodiment, the temperature detection value of the contact area is 36.5℃, which is within the exemplary preset human body contact temperature range of 33℃ to 42℃. Take 1.

[0086] The temperature duration is 10 seconds, which exceeds the exemplary preset contact duration of 5 seconds. Take 1.

[0087] The rate of temperature change conforms to the characteristics of heat conduction changes upon human contact. Set to 1. Example preset weight coefficient: , =0.3, The calculation yields: Grip stability parameters The generation process of the holding stability parameter: It is used to indicate whether the driver holds the steering wheel in a stable manner.

[0088] The vehicle-mounted intelligent driving controller uses a linear weighted fusion method based on multi-dimensional grip features to generate grip stability parameters. .

[0089] Among them, grip strength range matching parameters Used to characterize whether the grip strength detection data is within a preset grip strength range; grip duration parameter Used to characterize whether the grip strength duration meets the requirements; grip strength fluctuation amplitude parameter Used to characterize whether changes in grip strength are stable; bilateral grip consistency parameter It is used to indicate whether the gripping state on the left and right sides of the steering wheel is consistent. , , and All are normalized parameters.

[0090] The vehicle's intelligent driving controller first determines whether the grip strength detection data falls within a preset grip strength range. When the grip strength detection data is within the preset grip strength range... Set to 1; when the grip strength detection data deviates from the preset grip strength range, Cg takes a normalized value between 0 and 1 according to the degree of deviation.

[0091] Then it checks whether the grip strength duration has reached the preset grip duration. When the grip strength duration reaches the preset grip duration... Set to 1; when the grip strength duration is less than the preset grip duration, It is generated based on the ratio of the actual holding duration to the preset holding duration.

[0092] Next, determine whether the grip strength fluctuation amplitude is less than or equal to a preset fluctuation threshold. When the grip strength fluctuation amplitude is less than or equal to the preset fluctuation threshold, ... Set to 1; when the grip strength fluctuation amplitude is greater than the preset fluctuation threshold, The normalized value is taken between 0 and 1 based on the degree to which the fluctuation exceeds the preset threshold.

[0093] Finally, it is determined whether the difference in grip strength on both sides of the steering wheel is less than or equal to a preset threshold. If the difference in grip strength on both sides of the steering wheel is less than or equal to the preset threshold, then... Set to 1; when the difference in grip strength on both sides of the steering wheel exceeds the preset difference threshold, The normalized value is taken as a value between 0 and 1 based on the degree to which the difference exceeds the preset threshold.

[0094] Grip stability parameters Calculate using the following formula: in, , , , For preset weighting coefficients, ,and .

[0095] In this embodiment, the grip strength detection data is 35N, which falls within the exemplary preset grip strength range of 20N to 50N. Take 1.

[0096] The grip duration is 10 seconds, which exceeds the exemplary preset grip duration of 5 seconds. Set to 1. The grip strength fluctuation amplitude is 2N, which is less than the exemplary preset fluctuation threshold of 5N. Take 1.

[0097] The grip strength on the left and right sides of the steering wheel is 35N and 33N respectively, with a difference of 2N, which is less than the example preset difference threshold of 5N. Take 1.

[0098] Example of preset weighting coefficient , , , The calculation yields: Operational initiative parameters The generation process: Operational initiative parameters This parameter is used to characterize whether the current active operation data reflects a clearly issued active exit request from the driver. The onboard intelligent driving controller generates the operation initiative parameter using a linear weighted fusion method of operational behavior features and temporal matching features. .

[0099] Among them, the operation duration parameter Used to indicate whether the button's duration meets the preset long-press mode requirements; trigger count parameter. Used to indicate whether the number of key presses meets the preset trigger mode requirements; trigger interval parameter Used to characterize whether the interval between two consecutive triggers in double-click mode meets the requirements; operation timing matching parameters Used to characterize active action events and identity credibility parameters Contact effectiveness parameters and grip stability parameters The degree of matching within a preset time window , , and All are normalized parameters.

[0100] When the physical button triggers a long press, the vehicle's intelligent driving controller determines whether the button's duration is greater than or equal to a preset long press duration threshold. If the button's duration is greater than or equal to the preset long press duration threshold... Set to 1; when the duration of the key press is less than the preset long press duration threshold, It is generated based on the ratio of the actual button press duration to the preset long press duration threshold.

[0101] When the physical button trigger data meets the preset trigger count requirement Take 1; otherwise, Take 0 or a normalized value between 0 and 1 based on the matching degree of the number of triggers.

[0102] In long press mode, trigger interval parameter The vehicle-mounted intelligent driving controller will not be used as an invalidating factor in the determination. Set to the default valid value that matches the current trigger mode. In double-click mode, It is generated based on the relationship between the time interval between two consecutive triggers and the preset double-click interval threshold.

[0103] Operation timing matching parameters Used to characterize physical button-triggered events and identity credibility parameters Contact effectiveness parameters and grip stability parameters The consistency of time between them. Within the preset time window before and after the physical button is triggered, , , When all conditions continue to meet their respective basic threshold conditions, Take 1; otherwise, according to , , The degree to which the basic threshold condition is consistently met within the time window is a normalized value between 0 and 1.

[0104] The operational initiative parameter D is calculated using the following formula: in, , , , For preset weighting coefficients, ,and .

[0105] In this embodiment, the physical button trigger data is a long press mode, with a button duration of 4 seconds. An exemplary preset long press duration threshold is 3 seconds. (Operation duration parameter) It is generated based on the ratio of the actual duration to the preset long-press duration threshold, i.e. Because the normalization result is greater than 1, it is truncated to 1. The button trigger conforms to the preset long-press trigger pattern. Take 1. Long press mode Set as the default valid value. Take 1. Within the 3-second pre-trigger window and the 1-second post-trigger window of the physical button trigger moment, , , All of them continuously meet their respective basic threshold conditions. Set to 1. Example preset weight coefficient: , , , The calculation yields: After identity verification, the vehicle-mounted intelligent driving controller retrieves the active operation parameter configuration corresponding to the current driver's identity. The active operation parameter configuration includes preset long press duration threshold, preset double-click interval threshold, and operation initiative parameter weighting coefficient.

[0106] Different driver identities correspond to different parameter configurations. These parameter configurations can be preset and stored by the driver through the in-vehicle human-machine interface, or updated by the in-vehicle intelligent driving controller based on the statistical results of the driver's historical valid operation data.

[0107] For example, update the preset long press duration threshold based on the mean, quantile, or sliding average of historical long press durations.

[0108] The specific execution method of timing correlation is as follows: the on-board intelligent driving controller maintains a circular buffer in the memory and continuously records data at a preset sampling frequency. , , Time series data.

[0109] When a physical button trigger event is detected, the vehicle intelligent driving controller uses the trigger time as a reference to capture the parameter records in the preceding time window and the parameter records in the following time window.

[0110] The vehicle's intelligent driving controller determines that within the aforementioned time window, , , Whether each button continuously meets its respective basic threshold conditions, and whether the duration or trigger mode of the physical button trigger signal conforms to the preset active trigger mode.

[0111] If all the above conditions are met, the active operation event is recognized as a valid active exit event.

[0112] In another embodiment, the active operation data is double-click trigger data. The vehicle intelligent driving controller detects that the physical button is triggered twice consecutively within a preset double-click interval threshold, and within a preset time window before and after the trigger moment, , , If the basic threshold conditions are met continuously, the double-click operation is recognized as a valid active exit event.

[0113] at this time Take the valid value that matches the double-click mode. Take 1, Generate based on the actual time interval between two triggers. Take 1.

[0114] Step S3: Generate the assessment results of the intention to actively exit.

[0115] The vehicle intelligent driving controller is based on identity trust parameters Contact effectiveness parameters Grip stability parameters and operational initiative parameters Generate assessment results of the intention to actively withdraw.

[0116] The onboard intelligent driving controller uses a two-stage calculation method that combines hard gate control judgment with linear weighted fusion to generate an assessment value of the intention to actively exit the vehicle. .

[0117] The first stage involves hard gate control judgment. This is determined by the onboard intelligent driving controller. , , , Whether each condition meets its respective basic threshold. Basic threshold values ​​are pre-set. , , , ,in This serves as the basic threshold for identity credibility. This is the basic threshold for contact effectiveness. The basic threshold for grip stability. This is the basic threshold for operational initiative.

[0118] when , , , When all parameters meet their respective basic threshold conditions, the hard-gating parameters Take 1; when , , , When not all basic threshold conditions are met, the hard gating parameters Set it to 0, and set the value of the intention to actively exit as the evaluation value. Set to 0.

[0119] The second stage is linear weighted fusion. (Using hard-gated parameters...) When 1 is selected, the on-board intelligent driving controller... , , , Perform linear weighted fusion to generate an assessment value of the intention to actively withdraw. .

[0120] Active Exit Intent Assessment Value Calculate using the following formula: in, For identity credibility parameters, For contact effectiveness parameters, For grip stability parameters, For the operational initiative parameter, , , , All are normalized parameters. , , , For preset weighting coefficients, ,and .

[0121] In this embodiment, the basic threshold value , , , For example, all values ​​are 0.7. ≥0.7, ≥0.7, ≥0.7, ≥0.7, , , , All meet their respective basic threshold conditions, therefore .

[0122] Example preset weight coefficient , , , All are taken as 0.25. Because... , , , , ,therefore: Preset Intent Threshold For example, 0.8. ≥0.8, therefore the on-board intelligent driving controller determines that the assessment result of the intention to actively exit meets the preset active exit conditions.

[0123] Step S4: Generate exit safety parameters .

[0124] The onboard intelligent driving controller acquires vehicle operating status parameters through the vehicle operating status acquisition module and generates exit safety parameters. .

[0125] Exit security parameters This is a normalized parameter used to characterize the degree to which the current vehicle operating state meets the conditions for safe exit.

[0126] In this embodiment, the vehicle operating parameters are as follows: vehicle speed is 35km / h, road type is urban ordinary road, autonomous driving level is L3, lateral control is stable, longitudinal control is smooth, distance to the vehicle in front is 50 meters, distance to vehicles in adjacent lanes is 40 meters, and collision risk level is low.

[0127] The onboard intelligent driving controller generates vehicle speed safety parameters based on vehicle operating status parameters. Road type safety parameters Lateral control stability parameters Longitudinal control stability parameters Safety parameters for distance from surrounding vehicles Collision risk safety parameters .

[0128] Vehicle speed safety parameters This is used to characterize the impact of current vehicle speed on the safety of driver handover. The onboard intelligent driving controller is generated using a piecewise linear normalized mapping method. . Calculate using the following formula: in, Current vehicle speed To preset the low-speed threshold, This is a preset high-speed threshold.

[0129] In this embodiment, =35km / h, =40km / h, =100km / h, ≤ ,therefore =1.

[0130] Road type safety parameters This is generated based on a preset correspondence between the current road type and safety level. In this embodiment, the road type is a typical urban road. Take 1.

[0131] Lateral control stability parameters It is generated based on lane centering deviation, lateral acceleration, or lateral control error.

[0132] In this embodiment, the lane centering deviation is 0.1 meters, which is less than the exemplary preset offset threshold of 0.3 meters. Take 1.

[0133] Longitudinal control stability parameters It is generated based on longitudinal acceleration, rate of change of longitudinal acceleration, or vehicle following stability.

[0134] In this embodiment, the absolute value of the longitudinal acceleration is 0.5 m / s². 2 Less than the upper limit of the example preset stable range of 1m / s 2 , Take 1.

[0135] Safety parameters for distance from surrounding vehicles The vehicle-mounted intelligent driving controller uses a minimum safety margin normalization method to generate a parameter that indicates whether the traffic distance around the vehicle meets the safety requirements for the handover of driving rights. And the calculation results are truncated to the upper limit. Calculate using the following formula: in, Distance to the vehicle in front. This represents the distance between vehicles in adjacent lanes. This is a preset safe distance threshold.

[0136] In this embodiment, =50 meters, =40 meters =30 meters, calculated as follows .

[0137] Collision risk safety parameters This is generated based on a preset correspondence between collision risk levels and safety parameters. In this embodiment, the collision risk level is low. Take 1.

[0138] Exit security parameters This is used to characterize whether the current vehicle operating state is suitable for a handover of driving control. The onboard intelligent driving controller is generated using a multi-factor linear weighted fusion method based on vehicle operating state. . Calculate using the following formula: in, For vehicle speed safety parameters, For road type safety parameters, For lateral control stability parameters, For longitudinal control stability parameters, For safety parameters regarding the distance to surrounding vehicles, For collision risk safety parameters, , , , , , All are normalized parameters. , , , , , For preset weighting coefficients, ,and Preset safe exit conditions include ≥Preset safety threshold .

[0139] In this embodiment, , , , , , All are set to 1. Example preset weight coefficients are set to... =0.25, =0.15, =0.15, =0.15, =0.15, =0.15, calculated as follows: .

[0140] Therefore, exit the security parameter Set to 1. Preset security threshold. For example, 0.7 is used. ≥0.7, therefore the onboard intelligent driving controller determines the exit safety level parameter. The preset safe exit conditions are met.

[0141] Step S5: Determine the exit control strategy.

[0142] The onboard intelligent driving controller assesses the intention to exit and the exit safety parameters. Determine the exit control strategy.

[0143] In this embodiment, the evaluation value of the intention to actively exit =1, greater than or equal to the preset intent threshold of 0.8; Exit security parameter =1, greater than or equal to the preset safety threshold of 0.7.

[0144] When both conditions are met, the on-board intelligent driving controller outputs a direct switching strategy to control the vehicle to switch from intelligent driving mode to manual driving mode, and at the same time provides feedback to the driver through instrument text display, sound prompts or steering wheel prompts to indicate that the exit is complete.

[0145] During the generation of the active exit intention assessment result, steering wheel wobbling information, steering wheel displacement information, and steering wheel torque change information are not included in the active exit intention assessment value. The calculation is not sufficient to trigger an active exit condition on its own.

[0146] Example 2, refer to Figure 1 - Figure 3 This is the second embodiment of the present invention. The difference between this embodiment and the first embodiment is that this embodiment is used to illustrate that when the steering wheel shakes, displaces, and changes torque in a scenario where the vehicle is driving on a bumpy road, the on-board intelligent driving controller maintains the intelligent driving mode.

[0147] Specifically, the vehicle is in intelligent driving mode and driving on a bumpy road. The driver holds the steering wheel normally with both hands, but does not press any physical buttons. Due to the bumpy road surface, the steering wheel shakes, shifts, and changes in torque, and the driver's hands experience slight displacement.

[0148] At this time, the multimodal perception module collected fingerprint recognition data, temperature detection data, and grip force detection data, but did not collect physical button trigger data that matched the preset active trigger mode.

[0149] After acquiring the aforementioned multimodal perception data, the onboard intelligent driving controller performs temporal correlation within a preset active exit determination time window. Identity authentication data is normal, and the identity credibility parameter... Take 1; due to short-term fluctuations in temperature detection data caused by hand displacement, the temperature duration parameter decreased, and the contact effectiveness parameter... The grip strength test data is calculated using a multi-sub-index linear weighted fusion method; the grip strength fluctuation data is affected by road surface bumps, which reduces the grip strength fluctuation amplitude parameter and the grip stability parameter. The operation initiative parameter is calculated using a linear weighted fusion method based on multi-dimensional grip features; since no physical button trigger data matching the preset active trigger mode was detected, the operation initiative parameter is... Take 0.

[0150] The vehicle-mounted intelligent driving controller generates identity credibility parameters according to the parameter generation rules in Example 1. Contact effectiveness parameters Grip stability parameters and operational initiative parameters .

[0151] Among them, the identity authentication data is normal. Take the higher value; temperature and grip strength data may fluctuate briefly due to road bumps and slight hand movements. and Corresponding changes occur; since no physical button trigger data matching the preset active trigger mode is detected, The basic threshold conditions are not met.

[0152] Specifically, due to , , and Not all basic threshold conditions are met; hard-gating parameters Set to 0, indicating an intention to actively exit the evaluation value. It was set to 0.

[0153] The onboard intelligent driving controller determines that the assessment result of the intention to actively exit does not meet the preset conditions for active exit, and maintains the intelligent driving mode.

[0154] In this embodiment, although the steering wheel may wobble, shift, and change torque, the above information is not included in the calculation of the active exit intention evaluation value I, and does not constitute a sufficient condition for triggering active exit. Therefore, the vehicle will not exit the intelligent driving mode due to physical disturbance of the steering wheel caused by road bumps.

[0155] The remaining structure is the same as the steps in Example 1.

[0156] Example 3, referring to Figure 1 , Figure 2 , Figure 4 and Figure 6 The difference between this embodiment and the second embodiment is that this embodiment is used to illustrate that when an unauthorized person touches the steering wheel and presses a physical button, the vehicle intelligent driving controller prevents the unauthorized exit request through identity trust parameters and hard gating mechanism.

[0157] Specifically, the vehicle is in intelligent driving mode. Unauthorized personnel touch the steering wheel grip area and press physical buttons.

[0158] The fingerprint recognition unit collects fingerprint information from unauthorized personnel, the physical button control unit detects button trigger signals, and the temperature sensing unit and grip force detection unit detect human contact data and steering wheel grip data, respectively.

[0159] The vehicle-mounted intelligent driving controller generates identity credibility parameters according to the parameter generation rules in Example 1. Contact effectiveness parameters Grip stability parameters and operational initiative parameters .

[0160] Because the unauthorized person's fingerprint does not match the pre-stored fingerprint template in the vehicle's local secure storage unit, the identity trust parameter... Below the basic threshold.

[0161] even though , and They may meet their respective basic threshold conditions. , , and The basic threshold conditions are not yet fully met, therefore the hard-gating parameters... Set to 0, indicating an intention to actively exit the evaluation value. It was set to 0.

[0162] Specifically, if the onboard intelligent driving controller determines that the assessment result of the intention to actively exit does not meet the preset conditions for active exit, it will maintain the intelligent driving mode and output an unauthorized operation prompt or risk warning.

[0163] The remaining steps are the same as those in Example 2.

[0164] Example 4, refer to Figure 1 , Figure 2 , Figure 3 , Figure 4 and Figure 6 This is the fourth embodiment of the present invention. The difference between this embodiment and the third embodiment is that: when an authorized driver accidentally touches a physical button briefly, the vehicle intelligent driving controller distinguishes between accidental touch operation and effective active exit operation through the operation initiative parameter.

[0165] Specifically, the vehicle is in intelligent driving mode, authorizing the driver to hold the steering wheel normally. Fingerprint recognition data, temperature detection data, and grip strength detection data are all within normal range.

[0166] The driver accidentally touched the physical button briefly, but the duration of the touch did not reach the preset long press duration threshold and did not meet the double-click trigger mode.

[0167] The vehicle-mounted intelligent driving controller generates parameters according to the parameter generation rules in Example 1. , , and .in, , and Each meets its respective basic threshold conditions; due to the short duration of button triggering and the triggering mode not conforming to the preset active triggering mode, the operation initiative parameter... Below the basic threshold.

[0168] Specifically, due to , , and Not all basic threshold conditions are met; hard-gating parameters Set to 0, indicating an intention to actively exit the evaluation value. It was set to 0.

[0169] The onboard intelligent driving controller determines that the assessment result of the intention to actively exit does not meet the preset conditions for active exit, and maintains the intelligent driving mode.

[0170] In another embodiment, the active operation data is double-click trigger data. The vehicle intelligent driving controller detects that the physical button is triggered twice consecutively within a preset double-click interval threshold, and within a preset time window before and after the trigger moment, the identity credibility parameter... Contact effectiveness parameters and grip stability parameters If the basic threshold conditions are consistently met, the double-click operation is recognized as a valid active exit event, and the process proceeds to the subsequent hard-gating judgment and active exit intent evaluation process. The remaining steps are the same as those in Example 3.

[0171] Example 5, refer to Figure 1 , Figure 2 , Figure 5 , Figure 6 and Figure 7 This is the fifth embodiment of the present invention, which differs from the fourth embodiment in that: this embodiment illustrates how, when an authorized driver initiates an active exit request while the vehicle is traveling at high speed, the onboard intelligent driving controller controls the vehicle to enter a controlled handover process based on the exit safety parameter S; it also illustrates the process by which the onboard intelligent driving controller restricts the direct exit strategy and performs a downgrade process when multimodal perception data is missing or abnormal. The remaining structure is the same as that of embodiment 4.

[0172] Specifically, this embodiment includes the following two scenarios: The vehicle is in Level 3 autonomous driving mode and traveling at high speed on a highway. The authorized driver presses and holds a physical button to request to exit autonomous driving mode. The onboard autonomous driving controller generates parameters according to the parameter generation rules in Example 1. , , and .

[0173] In this embodiment, the authorized driver's fingerprint verification is successful, the palm contact temperature is within a preset human body contact temperature range, the grip strength detection data is within a preset grip strength range and the grip strength fluctuation amplitude is less than a preset fluctuation threshold, and the physical button trigger data conforms to the long press mode. Therefore, the identity credibility parameter... Contact effectiveness parameters Grip stability parameters and operational initiative parameters All satisfy their respective basic threshold conditions, hard-gating parameters The value is set to 1. The onboard intelligent driving controller calculates the active exit intention evaluation value according to the active exit intention evaluation value calculation method disclosed in Example 1. The calculation results meet the preset intent threshold, and the evaluation results of the intention to actively exit meet the preset active exit conditions.

[0174] In step S4, the on-board intelligent driving controller obtains vehicle operating status parameters through the vehicle operating status acquisition module.

[0175] In this embodiment, the vehicle operating status parameters include: current vehicle speed is higher than a preset high speed threshold, road type is highway, distance to the vehicle in front and distance to vehicles in adjacent lanes are lower than preset safe distance thresholds, and collision risk level is medium. The onboard intelligent driving controller generates exit safety parameters according to the exit safety parameter calculation method disclosed in Embodiment 1. Because the vehicle is currently traveling at high speed on the highway and the distance to surrounding vehicles is insufficient, the safety parameters are being deactivated. The value is below the preset safety threshold, and the preset safe exit condition is not met.

[0176] In step S5, the onboard intelligent driving controller evaluates the intention to actively exit the vehicle based on the assessment value. and exit security parameters Determine the exit control strategy. In this embodiment, the active exit intent evaluation value... The preset intent threshold is met, but the exit security parameter is not. The preset safety threshold was not met. At this point, the onboard intelligent driving controller does not perform a direct instantaneous switch, but instead controls the vehicle to enter a controlled handover process.

[0177] The controlled handover process includes: the onboard intelligent driving controller outputs a secondary confirmation prompt via instrument panel text display, audible prompts, or steering wheel indicator lights, alerting the driver that the vehicle is currently in a high-speed condition or a high-risk surrounding traffic situation, at which point the system will execute a controlled handover; after the driver completes the secondary confirmation, the onboard intelligent driving controller executes step-by-step deceleration control via the longitudinal controller, continuously acquiring vehicle operating status parameters during deceleration and updating the exit safety parameters in real time. .

[0178] As vehicle speed decreases, distances to surrounding vehicles change, or the collision risk level decreases, the vehicle speed safety parameters, surrounding vehicle distance safety parameters, or collision risk safety parameters change accordingly, thus exiting the safety parameter category. It is continuously updated.

[0179] When exiting the security parameters When the preset safety threshold is reached or exceeded, the on-board intelligent driving controller controls the vehicle to switch from intelligent driving mode to manual driving mode and outputs a driving mode switching prompt.

[0180] When exiting the security parameters If the vehicle remains below the preset safety threshold for an extended period of time, the onboard intelligent driving controller will maintain the intelligent driving mode and continuously output risk warnings or request the driver to handle the situation according to the vehicle's preset takeover procedure.

[0181] Scenario 2: Anomaly Handling in Multimodal Sensing Data When the vehicle is operating in intelligent driving mode, if identity authentication data, human contact data, steering wheel grip data, or active operation data are missing, exceed the effective collection range, sampling is interrupted, sensor fault is marked, or data consistency verification fails, the on-board intelligent driving controller will mark the corresponding modal data as abnormal and restrict the execution of the direct exit strategy.

[0182] When identity authentication data is missing or abnormal, the on-board intelligent driving controller is unable to generate identity credibility parameters that meet the basic threshold conditions based on the identity authentication data. When human contact data is missing or abnormal, the onboard intelligent driving controller cannot generate contact effectiveness parameters that meet the basic threshold conditions based on the human contact data. When steering wheel grip data is missing or abnormal, the onboard intelligent driving controller cannot generate grip stability parameters that meet the basic threshold conditions based on the steering wheel grip data. When active operation data is missing or abnormal, the onboard intelligent driving controller cannot generate operation initiative parameters that meet the basic threshold conditions based on the active operation data. .

[0183] In the above-mentioned abnormal situations, the identity credibility parameter Contact effectiveness parameters Grip stability parameters and operational initiative parameters Not all basic threshold conditions were met, the hard-gated chain of evidence was not closed, and the hard-gated parameters... Set to 0, indicating an intention to actively exit the evaluation value. It can be 0 or is restricted.

[0184] Therefore, the onboard intelligent driving controller does not execute the control strategy of directly switching to manual driving mode.

[0185] Upon detecting the aforementioned abnormal situation, the vehicle intelligent driving controller outputs risk warnings and secondary confirmation prompts through instrument panel text displays, sound prompts, or steering wheel indicator lights, indicating to the driver that there is currently an abnormality in multimodal perception data and that the direct exit strategy is restricted.

[0186] After the driver completes the secondary confirmation, the on-board intelligent driving controller determines whether to allow the driver to enter the downgrade process based on the preset downgrade judgment conditions.

[0187] The preset degradation judgment conditions include: the vehicle is currently in a state that can maintain intelligent driving control, the vehicle operation status acquisition module normally outputs vehicle operation status parameters, the active operation data or backup confirmation signal meets the preset confirmation conditions, and the vehicle actuator is able to perform step-by-step deceleration control.

[0188] When the preset downgrade judgment conditions are met, the on-board intelligent driving controller further determines the processing method based on the exit safety level parameter S.

[0189] If exiting the security parameters Once the preset safety threshold is met, the on-board intelligent driving controller can output a controlled handover prompt after secondary confirmation and control the vehicle to switch from intelligent driving mode to manual driving mode.

[0190] If exiting the security parameters If the preset safety threshold is not met, the onboard intelligent driving controller will initiate a controlled handover process, output a secondary confirmation prompt, and execute progressive deceleration control before exiting the safety parameter. Once the preset safety threshold is met, the vehicle will be switched from intelligent driving mode to manual driving mode.

[0191] When the preset degradation judgment condition is not met, or when the security parameters are exited If the vehicle remains below the preset safety threshold for an extended period of time, the onboard intelligent driving controller will maintain the intelligent driving mode and continuously output risk warnings or request the driver to handle the situation according to the vehicle's preset takeover procedure.

[0192] Through the above processing method, when multimodal perception data is missing or abnormal, the vehicle intelligent driving controller will not directly execute the driving mode switch due to a single modal anomaly. Instead, it will restrict the direct exit strategy and combine secondary confirmation, downgrade judgment conditions, and exit safety parameters. Controlled processing helps reduce the adverse impact of anomalous data on proactive exit decisions.

[0193] Example 6, refer to Figure 1 , Figure 2and Figure 7 This is the sixth embodiment of the present invention. The difference between this embodiment and the fifth embodiment is that this embodiment is used to illustrate the processing method of restricting the direct exit strategy and outputting a prompt when multimodal perception data is missing or abnormal.

[0194] Specifically, when abnormal situations occur, such as missing or abnormal identity authentication data, missing or abnormal human body contact data, missing or abnormal steering wheel grip data, or missing or abnormal active operation data, the vehicle intelligent driving controller will mark the corresponding modal data as abnormal and restrict the execution of the direct exit strategy.

[0195] When identity authentication data is missing or abnormal, the onboard intelligent driving controller cannot generate identity credibility parameters based on normal identity comparison results. The hard-gate evidence chain does not meet the integrity requirements. At this point, the onboard intelligent driving controller outputs a risk warning and a secondary confirmation prompt. If, after secondary confirmation, the system confirms through preset downgrade judgment conditions that the driver is still ready to take over, then it will proceed according to the exit safety parameters. Determine whether to output a controlled handover process instruction; if the preset downgrade judgment conditions are not met, maintain the intelligent driving mode.

[0196] Before the preset degradation judgment conditions are met, the basic threshold conditions corresponding to the abnormal mode are considered not met, and the vehicle intelligent driving controller does not execute the direct exit strategy.

[0197] Specifically, the preset downgrade judgment conditions include: the human contact data and steering wheel grip data in the available modal data continuously meet the corresponding basic threshold conditions, the active operation data conforms to the preset active triggering mode, and the driver completes secondary confirmation through the in-vehicle human-machine interface.

[0198] When human contact data is missing or abnormal, the onboard intelligent driving controller cannot confirm the actual continuous contact state between the driver's hands and the steering wheel, thus affecting the contact effectiveness parameter. If the basic threshold conditions are not met, the on-board intelligent driving controller will restrict the execution of the direct exit strategy and output a risk warning and a secondary confirmation prompt.

[0199] When steering wheel grip data is missing or abnormal, the onboard intelligent driving controller cannot confirm whether the driver is holding the steering wheel stably, and the grip stability parameter... If the basic threshold conditions are not met, the on-board intelligent driving controller will restrict the execution of the direct exit strategy and output a risk warning and a secondary confirmation prompt.

[0200] When active operation data is missing or abnormal, the onboard intelligent driving controller cannot confirm whether the driver has issued an exit request through a preset active operation method, thus affecting the operation initiative parameters. If the basic threshold conditions are not met, the on-board intelligent driving controller will maintain the intelligent driving mode and output a risk warning or operation failure warning.

[0201] Specifically, in the aforementioned abnormal situations, when the onboard intelligent driving controller determines, through a secondary confirmation process and vehicle operating status assessment, that a transfer of driving control still needs to be performed, the onboard intelligent driving controller will determine the exit safety level parameters. Determine whether to proceed with the controlled handover process. When exiting the security parameters... If the preset security threshold is not met, a controlled handover process instruction is output and the direct exit strategy is restricted; when the exit security parameter is not met, the control handover process instruction is output and the direct exit strategy is restricted. When the preset safety threshold is met and the secondary confirmation process meets the preset confirmation conditions, the on-board intelligent driving controller performs controlled exit according to the preset downgrade judgment process. The remaining steps are the same as those in Example 5.

[0202] Combining Examples 1 to 6, this invention constructs the intelligent driver's active exit detection process into a three-layer progressive architecture of hard gating, soft scoring, and joint decision-making on exit safety.

[0203] The first layer is the hard gate layer. Within the preset active exit judgment time window, the vehicle intelligent driving controller uses the active operation data trigger time as a benchmark to sequentially associate continuous evidence such as identity authentication, human contact and steering wheel grip with the instantaneous event of active operation.

[0204] When the identity credibility parameter Contact effectiveness parameters Grip stability parameters and operational initiative parameters Hard gating parameters when not all basic threshold conditions are met Set to 0, indicating an intention to actively exit the evaluation value. Setting it to 0 will either maintain the intelligent driving mode or restrict the direct exit strategy.

[0205] The second layer is the soft scoring layer, when... , , and When all parameters meet their respective basic threshold conditions, the hard-gating parameters If the value is set to 1, the onboard intelligent driving controller uses a linear weighted fusion method to calculate the evaluation value of the intention to actively exit the vehicle. .

[0206] The soft scoring layer is used to further quantify the assessment results of the voluntary exit intention after the hard-gated evidence chain is closed.

[0207] The third layer is the exit safety assessment and joint decision-making layer, where the onboard intelligent driving controller generates exit safety parameters based on vehicle operating status parameters. And based on the assessment value of the intention to actively withdraw. and exit security parameters Jointly determine the exit control strategy.

[0208] When the intention to actively withdraw is evaluated The value is greater than or equal to the preset intent threshold, and the exit security parameter is also greater than or equal to the threshold value. When the value is greater than or equal to a preset safety threshold, the vehicle is directly switched to manual driving mode; when the intention assessment value is actively exited... The value is greater than or equal to the preset intent threshold, and the exit security parameter is also greater than or equal to the threshold value. When the value is below a preset safety threshold, the vehicle is controlled to enter a controlled handover process; when the value actively exits the intention assessment... When the threshold is less than the preset intention threshold, maintain intelligent driving mode.

[0209] Not participating in the assessment of intention to withdraw The calculated steering wheel wobbling, steering wheel displacement, and steering wheel torque change information can still be collected by other vehicle subsystems and used for lane keeping assist, electric power steering control, or road condition monitoring, but will not be used as the basis for triggering the active exit decision in intelligent driving active exit scenarios.

[0210] This invention does not simply combine identity authentication, temperature, grip strength, and button trigger signals in parallel. Instead, it uses the moment of active operation data triggering as the time reference, binds continuous evidence and instantaneous active operation events into the same active exit event, and uses hard-gated parameters. Controlled intention to exit assessment value Whether to proceed to the weighted scoring stage, and then assess the intention to voluntarily withdraw. With exit safety parameters Both are used together for selecting exit control strategies.

[0211] In the above embodiments, the preset human contact temperature range, preset grip strength range, preset long press duration threshold, preset double-click interval threshold, various basic thresholds, preset intent threshold, preset safety threshold, preset observation duration, various weighting coefficients, and various safety distance thresholds can all be configured according to the vehicle model, steering wheel structure, individual driver characteristics, ambient temperature, and intelligent driving level. The specific values ​​in the specification are for illustrative purposes only.

[0212] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting active driver exit in intelligent driving based on multimodal perception, executed by an onboard intelligent driving controller, characterized in that, include: Acquire driver multimodal perception data, which includes identity authentication data, human body contact data, steering wheel grip data, and active operation data; Within the preset active exit judgment time window, the identity authentication data, human body contact data, steering wheel grip data and active operation data are correlated in time to generate identity credibility parameters, contact validity parameters, grip stability parameters and operation initiative parameters respectively. The temporal correlation is based on the triggering time of the active operation data. Within a preset time window before and after the triggering time of the active operation data, the identity authentication data, human body contact data, and steering wheel grip data are used as continuous evidence to correlate with the active operation data. The voluntary exit intention assessment result is generated based on identity credibility parameters, contact validity parameters, grip stability parameters, and operational initiative parameters. The generation of the voluntary exit intention assessment result includes: Based on whether the identity credibility parameter, contact validity parameter, grip stability parameter, and operation initiative parameter respectively meet their respective basic threshold conditions, hard gating parameters are generated. ; Among them, hard gating parameters A binary decision parameter is used to control whether the evaluation value of the intention to actively withdraw is included in the weighted calculation; When the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter all meet their respective basic threshold conditions, the hard-gating parameter... The value of the intention to actively withdraw is set to 1, and the weighted fusion result based on the identity credibility parameter, contact validity parameter, grip stability parameter, and operation initiative parameter is used to calculate the evaluation value. ; When the identity credibility parameter, contact validity parameter, grip stability parameter, and operational initiative parameter do not all meet their respective basic threshold conditions, the hard gating parameter... Set it to 0, and set the value of the intention to actively exit as the evaluation value. Set to zero; Obtain vehicle operating status parameters and generate exit safety parameters based on these parameters. ; Based on the assessment results of the intention to actively exit and the exit safety parameters Determine the exit control strategy, including: When the intention to actively withdraw is evaluated ≥Preset intent threshold and exit security parameter When the speed reaches or exceeds the preset safety threshold, control the vehicle to switch from intelligent driving mode to manual driving mode; When the intention to actively withdraw is evaluated ≥Preset intent threshold and exit security parameter When a preset safety threshold is reached, the vehicle is controlled to enter a controlled handover process, which includes: A secondary confirmation prompt is output, progressive deceleration control is executed, and the safety parameter is exited. Once a preset safety threshold is met, the vehicle will switch from intelligent driving mode to manual driving mode. When exiting the security parameters If the temperature remains below the preset safety threshold for an extended period of time, maintain the intelligent driving mode. When the intention to actively withdraw is evaluated <When the preset intent threshold is reached, maintain intelligent driving mode; During the generation of the active exit intention assessment result, steering wheel wobbling information, steering wheel displacement information, and steering wheel torque change information are not included in the active exit intention assessment value. The calculation is not sufficient to trigger an active exit condition on its own.

2. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: The identity authentication data is fingerprint recognition data. The fingerprint template is stored in the vehicle's local secure storage unit, and the identity comparison is completed locally on the vehicle's intelligent driving controller.

3. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: Human contact data is temperature detection data. The contact effectiveness parameter is generated based on whether the temperature detection data is within the preset human contact temperature range, the duration of temperature change, the rate of temperature rise, and the result of ambient temperature compensation.

4. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: The steering wheel grip data is grip force detection data. The grip stability parameter is generated based on whether the grip force detection data is within the preset grip force range, grip force duration, grip force fluctuation amplitude, and the consistency of grip on both sides of the steering wheel.

5. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: Active operation data is physical button trigger data. The physical button trigger modes include long press mode and double press mode. The long press mode is determined according to the preset long press duration threshold, and the double press mode is determined according to the preset double press interval threshold. The preset long press duration threshold and preset double-click interval threshold are set according to the parameter configuration corresponding to the driver's identity.

6. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: Temporal correlations include: Within a preset time window before and after the moment of active data triggering, determine whether the identity credibility parameter, contact validity parameter, and grip stability parameter all meet their respective basic threshold conditions. Active operation data is considered a valid active exit event only when the identity credibility parameter, contact validity parameter, and grip stability parameter continuously meet the basic threshold conditions within the time window, and the active operation data conforms to the preset active triggering mode.

7. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 6, characterized in that: Active Exit Intent Assessment Value Calculated in the following way: When hard-gating parameters When the value is 1, the intention to actively exit the evaluation value is determined. Calculate using the following formula: When hard-gating parameters When the value is 0, the intention to actively exit the evaluation value is considered. Set to 0; in, For identity credibility parameters, For contact effectiveness parameters, For grip stability parameters, For the operational initiative parameter, , , , All are normalized parameters. , , , For preset weighting coefficients, ,and ; Preset active exit conditions include hard-gating parameters =1 and the assessment value of the intention to actively withdraw. ≥Preset intent threshold.

8. The intelligent driving driver's active exit detection method based on multimodal perception according to claim 1, characterized in that: Vehicle operating status parameters include vehicle speed, road type, autonomous driving level, lateral control status, longitudinal control status, distance to surrounding vehicles, and collision risk level; Exit security parameters Calculate using the following formula: in, For vehicle speed safety parameters, For road type safety parameters, For lateral control stability parameters, For longitudinal control stability parameters, For safety parameters regarding the distance to surrounding vehicles, For collision risk safety parameters, , , , , , All are normalized parameters. , , , , , For preset weighting coefficients, ,and Preset safe exit conditions include ≥Preset safety threshold.

9. A multimodal perception-based intelligent driver active exit system, used to implement the method of any one of claims 1 to 8, characterized in that, include: The multimodal sensing module includes a fingerprint recognition unit, a temperature sensing unit, and a grip force detection unit located in the steering wheel grip area, as well as a physical button control unit located in the multi-function operation area of ​​the steering wheel. The vehicle operating status acquisition module is used to acquire vehicle operating status parameters; The in-vehicle intelligent driving controller is electrically connected to the multimodal perception module and the vehicle operating status acquisition module. The in-vehicle intelligent driving controller includes an active exit intention assessment unit, an exit safety assessment unit, and an exit control unit. The active exit intention assessment unit is configured to generate an active exit intention assessment result based on identity authentication data, human contact data, steering wheel grip data, and active operation data. The exit safety assessment unit is configured to generate exit safety parameters based on vehicle operating status parameters. The exit control unit is configured to evaluate the results of the active exit intent and the exit safety parameters. Determine the exit control strategy; When abnormal situations occur, such as missing or abnormal identity authentication data, missing or abnormal human body contact data, missing or abnormal steering wheel grip data, or missing or abnormal active operation data, the vehicle intelligent driving controller restricts the execution of the direct exit strategy and outputs risk warnings and secondary confirmation prompts. After secondary confirmation, if the preset downgrade judgment conditions are met, the on-board intelligent driving controller determines whether to output the controlled handover process instruction based on the exit safety parameters. If the preset downgrade criteria are not met, the intelligent driving mode will be maintained.

10. A vehicle, characterized in that, Includes a steering wheel, and the intelligent driver active exit system based on multimodal perception as described in claim 9.

Images