Adaptive emergency braking method and system based on multi-modal fusion intent and medium

Abstract

The application relates to a multi-modal fusion intention-based adaptive emergency braking method and system and a medium, and relates to the technical field of intelligent driving.The method comprises the following steps: acquiring traffic signal lamp states, self-vehicle states, front-vehicle states and front-vehicle brake light states; determining a self-vehicle passing intention based on the self-vehicle states, and determining a front-vehicle passing intention based on the front-vehicle states and the front-vehicle brake light states; selecting a corresponding target automatic emergency braking mode from a predefined decision matrix according to the combination of the self-vehicle passing intention and the front-vehicle passing intention; and when a real-time collision risk parameter reaches a trigger threshold corresponding to the target automatic emergency braking mode, performing braking intervention according to a deceleration strategy corresponding to the target automatic emergency braking mode.The application can significantly reduce the false triggering rate of an automatic emergency braking system in a city intersection traffic signal lamp switching scene, and improve the accuracy of collision risk identification and the timeliness of system response.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving technology, and in particular to an adaptive emergency braking method, system, and medium based on multimodal fusion intent. Background Technology

[0002] Automatic Emergency Braking (AEB), a core component of modern automotive active safety technology, uses sensors to monitor road conditions in real time and automatically activates the braking system when a potential collision risk is detected, thereby reducing or avoiding traffic accidents. Traditional AEB systems primarily rely on sensors such as radar, lidar, and cameras to obtain motion parameters like relative distance and speed between vehicles, and make braking decisions based on fixed thresholds such as time-to-collision (TTC). With the development of intelligent driving technology, AEB systems have demonstrated good performance in highway and simple road scenarios, providing strong protection for road traffic safety.

[0003] However, existing AEB systems still face numerous technical challenges in complex urban intersection scenarios, especially during traffic light changes. During signal transitions such as flashing green and yellow lights, drivers of the vehicle ahead face uncertainty in deciding whether to "go" or "stop," resulting in highly random and complex behavioral patterns. Traditional AEB systems, relying solely on fixed distance or time thresholds, cannot effectively understand and predict the driving intentions of the vehicle ahead, leading to frequent false triggers in such scenarios, severely impacting driving comfort and user experience. Furthermore, when the vehicle ahead suddenly changes its driving strategy, the limitations of fixed thresholds can cause system lag, failing to provide timely and effective collision avoidance measures. In addition, existing systems lack comprehensive utilization of key environmental information such as traffic light status and the braking signals of the vehicle ahead, resulting in relatively simplistic decision-making logic that struggles to adapt to dynamically changing traffic environments. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide an adaptive emergency braking method, system and medium based on multimodal fusion intent. By fusing multimodal information and predicting driving intent, the false trigger rate of the automatic emergency braking system in urban intersection traffic light switching scenarios can be significantly reduced, while improving the accuracy of collision risk identification and the timeliness of system response.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] In a first aspect, the present invention provides an adaptive emergency braking method based on multimodal fusion intent, which adopts the following technical solution: Get the status of traffic lights, your own vehicle, the vehicle in front, and the brake lights of the vehicle in front; The vehicle's intention to proceed is determined based on its own vehicle status, and the vehicle's intention to proceed is determined based on the preceding vehicle's status and the preceding vehicle's brake light status. Based on the combination of the vehicle's own travel intention and the preceding vehicle's travel intention, a corresponding target automatic emergency braking mode is selected from a predefined decision matrix; and In the target automatic emergency braking mode, when the real-time collision risk parameter reaches the trigger threshold corresponding to the mode, braking intervention is performed according to the deceleration strategy corresponding to the mode.

[0007] Furthermore, in the above adaptive emergency braking method, determining the vehicle's intention to proceed based on the vehicle's state includes: Get the vehicle's current speed and the distance between the vehicle and the stop line; Calculate the safe distance required for comfortable parking based on the vehicle's current speed; When the distance between the vehicle and the stop line is less than or equal to the safe distance, it is determined that the vehicle's intention to continue passing is to continue. When the distance between the vehicle and the stop line is greater than the safe distance, the vehicle's intention to proceed is determined to be to stop.

[0008] Furthermore, in the above adaptive emergency braking method, determining the preceding vehicle's intention to proceed based on the preceding vehicle's state and brake light state includes: Get the status and duration of the brake lights on the vehicle in front; Calculate the collision time between the vehicle in front and the stop line; When the continuous illumination time is greater than a preset time threshold and the collision time is greater than a preset collision time threshold, the intention of the vehicle in front is determined to be preparing to stop. If the brake lights of the preceding vehicle are not illuminated, the duration of illumination is less than or equal to the preset time threshold, or the collision time is less than or equal to the preset collision time threshold, it is determined that the preceding vehicle's intention to proceed is to continue.

[0009] Furthermore, in the above adaptive emergency braking method, the preset time threshold is 0.5 seconds, and the preset collision time threshold is 3 seconds.

[0010] Furthermore, in the above adaptive emergency braking method, the predefined decision matrix includes: When both the vehicle in front and the vehicle in front intend to continue driving, select the relaxed automatic emergency braking mode. When the vehicle in front intends to continue passing and the vehicle in front intends to stop, select the standard automatic emergency braking mode. When the vehicle's intention is to stop, select the standard automatic emergency braking mode.

[0011] Furthermore, in the above adaptive emergency braking method, the trigger threshold corresponding to the relaxed automatic emergency braking mode is greater than the trigger threshold corresponding to the standard automatic emergency braking mode, and the trigger threshold corresponding to the standard automatic emergency braking mode is greater than or equal to the trigger threshold corresponding to the conventional automatic emergency braking mode.

[0012] Furthermore, in the above adaptive emergency braking method, the deceleration strategy includes: The relaxed automatic emergency braking mode employs a first-level gentle deceleration. The standard automatic emergency braking mode uses a second-level progressive deceleration, which is greater than the first-level gentle deceleration. The conventional automatic emergency braking mode uses the maximum emergency deceleration, which is greater than the second-stage progressive deceleration.

[0013] Furthermore, in the above adaptive emergency braking method, the real-time collision risk parameter is the collision time.

[0014] Furthermore, the aforementioned adaptive emergency braking method also includes: Real-time monitoring of changes in collision risk; When a sharp increase in collision risk is detected in either the relaxed automatic emergency braking mode or the standard automatic emergency braking mode, immediately switch to the regular automatic emergency braking mode.

[0015] Furthermore, in the above adaptive emergency braking method, the traffic light state includes a flashing green state and a yellow light state.

[0016] Secondly, the present invention provides an adaptive emergency braking system based on multimodal fusion intent, which adopts the following technical solution: The information acquisition module is used to acquire the status of traffic lights, the status of the vehicle itself, the status of the vehicle in front, and the status of the brake lights of the vehicle in front. The intent determination module is used to determine the driving intent of the vehicle based on the vehicle's status, and to determine the driving intent of the vehicle in front based on the status of the vehicle in front and the status of the vehicle in front's brake lights. The mode selection module is used to select the corresponding target automatic emergency braking mode from a predefined decision matrix based on the combination of the vehicle's travel intention and the preceding vehicle's travel intention; and The braking control module is used to intervene in braking according to the deceleration strategy corresponding to the mode when the real-time collision risk parameter reaches the trigger threshold corresponding to the mode in the target automatic emergency braking mode.

[0017] Thirdly, the present invention provides a readable storage medium, which adopts the following technical solution: A readable storage medium storing computer instructions that, when executed by a processor, implement the adaptive emergency braking method as described in any one of the first aspects above.

[0018] In summary, compared with the prior art, the present invention has at least one of the following beneficial technical effects: The adaptive emergency braking method of this invention integrates multimodal information such as traffic light status, the vehicle's status, the status of the vehicle in front, and the brake light status of the vehicle in front. This allows for accurate determination of the driving intentions of both the vehicle and the vehicle in front, and adaptively selects the appropriate braking mode and strategy based on the combination of intentions. Compared to traditional AEB systems based on fixed thresholds, this invention significantly reduces the false trigger rate in urban intersection traffic light switching scenarios, avoiding unnecessary emergency braking that could impact driving comfort. Furthermore, through real-time collision risk parameter monitoring and a dynamic mode switching mechanism, this invention provides more intelligent and user-friendly braking intervention while ensuring safety, effectively improving the adaptability and reliability of the automatic emergency braking system in complex traffic environments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of an adaptive emergency braking method based on multimodal fusion intent is shown.

[0021] Figure 2 A flowchart of the vehicle intent determination method is shown.

[0022] Figure 3 A flowchart illustrating the method for determining the driving intention of the vehicle in front is shown.

[0023] Figure 4 A flowchart of the adaptive emergency braking mode selection method is shown.

[0024] Figure 5 A flowchart illustrating the comprehensive application of the adaptive emergency braking method is shown.

[0025] Figure 6 The modular architecture diagram of an adaptive emergency braking system based on multimodal fusion intent is shown. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Furthermore, it should be understood that the specific embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application.

[0027] It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments of this application. Furthermore, the descriptions of each embodiment in the following embodiments have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0028] Reference Figure 1 An adaptive emergency braking method 100 based on multimodal fusion intent includes multiple sequentially executed steps for intelligent braking intervention in urban intersection traffic light switching scenarios. This method 100 is integrated into the vehicle's intelligent driving computing unit and addresses the problems of false triggering and inappropriate intervention timing during green and yellow light transitions in traditional automatic emergency braking methods through multimodal information fusion and intent prediction technology.

[0029] In some implementations, method 100 begins at step 102, acquiring the traffic light status, the vehicle's status, the status of the vehicle in front, and the status of the vehicle in front's brake lights. The environmental perception module identifies the traffic light status via a forward-facing camera, including flashing green and yellow lights, and simultaneously detects the illumination status of the vehicle in front's brake lights. The vehicle status perception module acquires the vehicle's real-time speed and acceleration data, as well as the distance information between the vehicle and the stop line. The vehicle in front's status information includes the vehicle's motion status, relative distance to the vehicle, and relative speed.

[0030] Method 100 continues with step 104, determining the vehicle's intention to proceed based on its status. This step analyzes the vehicle's current speed and distance from the stop line to calculate the safe distance required for comfortable parking, and compares the actual distance with the safe distance to determine whether the vehicle chooses to continue or prepare to stop. In some implementations, when the distance between the vehicle and the stop line is less than or equal to the safe distance, the vehicle's intention to proceed is determined to be to continue.

[0031] like Figure 1As shown, method 100 then proceeds to step 106, determining the preceding vehicle's intention to proceed based on the preceding vehicle's status and brake light status. This step integrates multi-dimensional information such as the preceding vehicle's brake light illumination status, duration of illumination, and collision time with the stop line. By analyzing whether the preceding vehicle's brake lights remain illuminated for an extended period and the decision-making pressure faced by the preceding vehicle, it determines whether the preceding vehicle intends to continue proceeding or is preparing to stop.

[0032] Method 100 then proceeds to step 108, whereby, based on the combination of the vehicle's travel intention and the preceding vehicle's travel intention, a corresponding target automatic emergency braking mode is selected from a predefined decision matrix. Referring to Table 1, the decision matrix includes a relaxed automatic emergency braking mode, a standard automatic emergency braking mode, and a regular automatic emergency braking mode, each corresponding to a different trigger threshold and deceleration strategy.

[0033] Table 1

[0034] Continue to refer to Figure 1 Method 100 executes decision step 110 to determine whether the real-time collision risk parameters have reached the trigger threshold corresponding to the target automatic emergency braking mode. The real-time collision risk parameters include safety assessment indicators such as collision time. When the risk parameters reach the trigger threshold, Method 100 proceeds to step 112 to perform braking intervention; when the risk parameters do not reach the trigger threshold, Method 100 proceeds to step 114 to continue monitoring.

[0035] In step 112, method 100 performs braking intervention according to the deceleration strategy corresponding to the target automatic emergency braking mode. Different modes employ different deceleration strategies: the lenient mode uses a first-level gentle deceleration, the standard mode uses a second-level progressive deceleration, and the normal mode uses the maximum emergency deceleration. In some embodiments, method 100 also includes real-time monitoring of changes in collision risk, and immediately switching from the lenient or standard mode to the normal mode when a sharp increase in risk is detected.

[0036] Step 114 implements the continuous monitoring function, and method 100 returns to the risk parameter detection stage, forming a closed-loop control. This step ensures that method 100 can dynamically respond to changes in the traffic environment and provide continuous safety protection throughout the entire driving process.

[0037] Reference Figure 2 A vehicle intent determination method 200 is used to determine the driving intent of a vehicle in a traffic light switching scenario. This method 200 achieves accurate determination of the vehicle's driving intent by analyzing the spatial relationship between the vehicle's current state and the stop line.

[0038] Method 200 begins with step 201, acquiring the vehicle's current speed and the distance between the vehicle and the stop line. The vehicle state perception device monitors the vehicle's speed v_ego in real time, and simultaneously determines the actual distance between the vehicle and the stop line at the intersection ahead using visual perception technology. This basic data provides input parameters for subsequent intent determination.

[0039] like Figure 2 As shown, method 200 proceeds to step 202, calculating the safe distance required for comfortable parking based on the vehicle's current speed. The safe distance D_safe is calculated based on the vehicle's speed v_ego, taking into account driver reaction time and vehicle braking performance to ensure the vehicle can stop smoothly at a comfortable deceleration. In some implementations, the safe distance D_safe is calculated using a formula that comprehensively considers vehicle speed, road conditions, and braking system response characteristics.

[0040] Method 200 then proceeds to decision step 203, determining whether the distance between the vehicle and the stop line is less than or equal to the safe distance. This step compares the actual distance with the calculated safe distance D_safe, forming the basis for the intent determination. When the distance relationship meets specific conditions, method 200 proceeds to different intent output branches.

[0041] Continue to refer to Figure 2 When the distance between the vehicle and the stop line is less than or equal to the safe distance, method 200 executes step 204, determining that the vehicle's intention to proceed is to continue. This determination indicates that the vehicle has entered the passage decision area, and stopping would result in uncomfortable emergency braking; therefore, the vehicle chooses to continue through the intersection. When the distance between the vehicle and the stop line is greater than the safe distance, method 200 executes step 205, determining that the vehicle's intention to proceed is to prepare to stop, indicating that the vehicle has sufficient distance for comfortable braking.

[0042] Reference Figure 3 A method 300 for determining the driving intention of a preceding vehicle is proposed, which determines the preceding vehicle's intention by fusing analysis of the preceding vehicle's brake light status and collision time. This method 300 solves the problem that traditional methods, which rely solely on relative motion relationships, cannot accurately predict the behavior of the preceding vehicle.

[0043] Method 300 begins at step 302, acquiring the status and duration of the brake lights on the preceding vehicle. The visual recognition device detects whether the brake lights on the preceding vehicle are illuminated and records the duration of the brake lights' illumination, T_light. This parameter is used to distinguish between slight deceleration and a firm braking intention from the preceding vehicle.

[0044] like Figure 3As shown, method 300 simultaneously executes step 304, calculating the collision time between the preceding vehicle and the stop line. This step calculates the collision time TTC_front_stopline using the preceding vehicle's current position, speed, and stop line position. This parameter quantifies the decision-making pressure faced by the preceding vehicle, reflecting whether it has sufficient time to stop comfortably.

[0045] Method 300 proceeds to decision step 306, determining whether the duration of the brake lights on the preceding vehicle exceeds a preset time threshold. The preset time threshold T1 is set to 0.5 seconds, used to distinguish between the preceding vehicle's intermittent braking warning behavior and its firm braking intention. When the duration exceeds 0.5 seconds, it indicates that the preceding vehicle has a clear intention to decelerate; when the duration does not exceed this threshold, Method 300 directly proceeds to step 312.

[0046] Continue to refer to Figure 3 When the brake lights remain illuminated for a duration exceeding a preset time threshold, method 300 executes step 308 to determine whether the collision time between the preceding vehicle and the stop line exceeds a preset collision time threshold. The preset collision time threshold T2 is set to 3 seconds to determine whether the preceding vehicle has sufficient time to stop comfortably. When TTC_front_stopline is greater than 3 seconds, it indicates that the preceding vehicle has sufficient time and space to choose to stop.

[0047] In some implementations, when the brake lights of the preceding vehicle remain illuminated for more than 0.5 seconds and the collision time between the preceding vehicle and the stop line is more than 3 seconds, method 300 executes step 310 to determine that the preceding vehicle's intention to proceed is to prepare to stop. This determination is based on the preceding vehicle having both a clear braking signal and sufficient stopping conditions. When the brake lights of the preceding vehicle are not illuminated, remain illuminated for less than or equal to 0.5 seconds, or the collision time is less than or equal to 3 seconds, method 300 executes step 312 to determine that the preceding vehicle's intention to proceed is to continue.

[0048] Method 300 also considers an additional collision time threshold T3, set to 2 seconds, to determine whether the vehicle ahead faces an emergency decision. When TTC_front_stopline is less than or equal to 2 seconds, the vehicle ahead can only choose to accelerate through the intersection or brake suddenly, increasing the likelihood that the vehicle ahead will continue to proceed. This multi-layered threshold design improves the accuracy and reliability of determining the vehicle ahead's intentions.

[0049] Reference Figure 4 An adaptive emergency braking mode selection method 400 achieves intelligent mode selection based on intent combination through a predefined decision matrix. This method 400 solves the problem that traditional braking methods cannot provide differentiated intervention according to different traffic scenarios. By combining and analyzing the driving intent of the vehicle itself and the driving intent of the vehicle in front, it achieves adaptive adjustment of the braking strategy.

[0050] Method 400 begins at step 401, where the user's travel intention and the preceding vehicle's travel intention are input. This step receives analysis results from the intention determination phase, including whether the user chooses to continue or prepare to stop, and whether the preceding vehicle chooses to continue or prepare to stop. These two intention parameters constitute the input variables of the decision matrix.

[0051] like Figure 4 As shown, method 400 continues to execute decision step 402 to determine whether the vehicle's intention to proceed is to continue. This step uses the vehicle's intention as the first-level branch condition of the decision matrix. When the vehicle's intention is to continue, method 400 proceeds to further determine the intention of the vehicle in front; when the vehicle's intention is to prepare to stop, method 400 directly jumps to step 407 to select the normal automatic emergency braking mode.

[0052] When the vehicle's intention to proceed is to continue, method 400 executes decision step 403 to determine whether the preceding vehicle's intention to proceed is also to continue. This step constitutes the second-level branch condition of the decision matrix, determining the final braking mode based on the different choices of the preceding vehicle's intention. This two-level decision structure achieves complete coverage of the four possible combinations of intentions.

[0053] Continue to refer to Figure 4 When both the vehicle in front and the vehicle in front intend to continue through the intersection, method 400 executes step 404, selecting the relaxed automatic emergency braking mode and applying the first-level gentle deceleration strategy. This mode is suitable for scenarios where both vehicles choose to pass through the intersection, employing a relatively relaxed intervention strategy to avoid unnecessary braking interference.

[0054] When the vehicle intends to continue proceeding and the vehicle in front intends to stop, method 400 executes step 405, selecting the standard automatic emergency braking mode and applying the second-level progressive deceleration strategy. This mode is suitable for conflict scenarios where the vehicle plans to proceed but the vehicle in front is preparing to stop, requiring timely intervention to avoid a rear-end collision.

[0055] In some implementations, when the vehicle's intention is to stop, regardless of the intention of the vehicle in front, method 400 executes step 407, selecting a conventional automatic emergency braking mode and applying a maximum emergency deceleration strategy. This mode ensures that the vehicle can stop safely, avoiding collisions with vehicles in front or the stop line.

[0056] The trigger threshold for the relaxed automatic emergency braking mode is higher than that for the standard automatic emergency braking mode, and the trigger threshold for the standard automatic emergency braking mode is greater than or equal to that for the conventional automatic emergency braking mode. In some implementations, the relaxed automatic emergency braking mode is triggered by a 3.0-second collision time threshold, the standard automatic emergency braking mode is triggered by a 2.2-second collision time threshold, and the conventional automatic emergency braking mode is triggered by a 2.0-second collision time threshold.

[0057] For example, Table 2 lists the specific parameter configurations and application scenarios for various automatic emergency braking modes.

[0058] Table 2

[0059] Specifically, Table 2 shows four braking mode configuration schemes based on the combination of the vehicle's and the preceding vehicle's intentions to proceed. When both the vehicle and the preceding vehicle intend to proceed, Mode A (Relaxed AEB) is used, with a larger trigger threshold (e.g., TTC = 3.0 seconds) and a gentle, small deceleration (e.g., -3 m / s²), aiming to provide preparatory and comfortable braking intervention. When the vehicle intends to proceed but the preceding vehicle intends to stop, Mode B (Standard AEB) is selected, using a standard or slightly relaxed trigger threshold (e.g., TTC = 2.2 seconds) and a moderately progressive deceleration strategy, aiming for proactive intervention to avoid conflict. When the vehicle intends to stop, regardless of the preceding vehicle's intention, Mode C (Regular AEB) is used, employing a standard trigger threshold (e.g., TTC = 2.0 seconds) and a maximum emergency deceleration strategy to ensure collision avoidance. This decision matrix, through differentiated threshold settings and deceleration strategies, achieves graded intervention from preparatory braking to emergency braking, optimizing driving comfort while ensuring safety.

[0060] The deceleration strategy includes three escalating levels of braking intensity. The relaxed automatic emergency braking mode employs a first-level gentle deceleration, set at -3 m / s² in some implementations, to achieve a smooth and continuous braking effect. The standard automatic emergency braking mode employs a second-level progressive deceleration, which is greater than the first-level gentle deceleration, providing moderate braking intervention. The conventional automatic emergency braking mode employs the maximum emergency deceleration, which is greater than the second-level progressive deceleration, ensuring effective braking in emergency situations.

[0061] This tiered deceleration strategy achieves a smooth transition from preparatory braking to emergency braking. The first stage of gentle deceleration mimics the driver's preparatory braking behavior, the second stage of progressive deceleration provides moderate intervention that balances safety and comfort, and the maximum emergency deceleration ensures a safety baseline in dangerous situations. This differentiated braking strategy avoids the single maximum braking mode of traditional methods and improves the humanization of braking intervention.

[0062] In some implementations, the real-time collision risk parameter is the Time To Collision (TTC). TTC is calculated using the relative distance and relative speed between the vehicle and the vehicle in front, representing the time required for a collision to occur under the current motion conditions. The formula for TTC is relative distance divided by relative speed; when the relative speed is zero or negative, it indicates no collision risk. This parameter provides a quantitative risk assessment basis for braking decisions.

[0063] The application of Time-to-Collapse (TTC) in braking decisions is based on the trigger threshold settings for different automatic emergency braking modes. The lenient automatic emergency braking mode uses a larger TTC threshold, allowing for a longer reaction time; the standard automatic emergency braking mode uses a medium TTC threshold, providing a balanced intervention opportunity; and the regular automatic emergency braking mode uses a smaller TTC threshold to ensure timely and safe intervention. The dynamic changes in the TTC value reflect the real-time status of the collision risk.

[0064] The adaptive emergency braking method also includes the ability to monitor changes in collision risk in real time. This monitoring function continuously calculates and updates the time-to-collision (TTC), tracking changes in the relative motion between the vehicle and the vehicle in front. The monitoring process analyzes the trend of TTC changes, identifies patterns of increasing or decreasing collision risk, and provides triggering conditions for dynamic mode switching.

[0065] When a sharp increase in collision risk is detected in either the relaxed automatic emergency braking mode or the standard automatic emergency braking mode, the system immediately switches to the regular automatic emergency braking mode. The determination of a sharp increase in collision risk is based on a rapid decrease in the TTC (Traffic Trace Control) value. When the TTC value drops below a preset emergency threshold, the mode switching mechanism is triggered. This emergency threshold is typically set to 1.5 seconds or less to ensure a rapid response in hazardous situations.

[0066] The mode switching mechanism enables a seamless and rapid transition. When a sharp increase in risk is detected, the method immediately overrides the current lenient or standard braking strategy, activating the maximum emergency deceleration of the conventional automatic emergency braking mode. This switching process is completed within milliseconds, avoiding safety risks caused by mode transition delays. The switched conventional mode remains active until the collision risk is eliminated or the vehicle comes to a complete stop.

[0067] Traffic light states include flashing green and yellow, which constitute the triggering conditions for adaptive emergency braking decisions. A flashing green indicates the green light is about to end, forcing the driver to choose between proceeding or stopping; a yellow light indicates the light is about to turn red, increasing the urgency of the driving decision. These two traffic light states create the traffic scenarios with the highest uncertainty regarding the behavior of the vehicle ahead.

[0068] The flashing green and yellow light states play a role in scene recognition during the triggering of adaptive emergency braking decisions. When these specific traffic light states are detected, the method activates decision logic based on multimodal fusion intent, distinct from standard braking logic in other traffic scenarios. This scene-specific activation mechanism ensures that the adaptive braking strategy only takes effect in the relevant traffic environment, avoiding false triggering in inapplicable scenarios.

[0069] When the light is flashing green, the driver of the vehicle in front often hesitates between accelerating through and decelerating to a stop, leading to a high degree of uncertainty in the vehicle's behavior. The yellow light further exacerbates this uncertainty, as the vehicle in front may choose to accelerate through at the beginning of the yellow light or choose to brake suddenly at the end. Adaptive emergency braking methods identify these specific traffic light states, anticipate potential complex traffic interaction scenarios, and prepare appropriate braking strategies in advance.

[0070] In summary, referring to Figure 5 A specific embodiment of the adaptive emergency braking method based on multimodal fusion intent of the present invention includes method 500, which realizes complete closed-loop control from multimodal information acquisition to intelligent braking execution through five sequentially executed steps, reflecting the collaborative working mechanism of various technical features.

[0071] Method 500 begins at step 502, acquiring the traffic light status, the vehicle's status, the status of the vehicle in front, and the status of the vehicle in front's brake lights. This step integrates multi-dimensional information acquisition functions of environmental perception and vehicle status perception. It uses a forward-facing camera to identify the flashing green and yellow lights, obtains the distance relationship between the vehicle and the stop line, and monitors the motion parameters and brake light status of the vehicle in front. This step provides a complete data foundation for subsequent intent analysis and decision-making.

[0072] like Figure 5 As shown, method 500 continues to execute step 504 to determine the driving intentions of both the vehicle and the vehicle in front. This step integrates the dual analysis functions of determining the driving intention of the vehicle and the vehicle in front. It determines the driving intention of the vehicle by calculating the relationship between the vehicle's safe parking distance and the actual distance, and simultaneously determines the driving intention of the vehicle in front by combining multi-parameter fusion analysis of the duration of the vehicle in front's brake lights and the collision time between the vehicle in front and the stop line. This step achieves an intelligent conversion from raw perception data to understanding driving intentions.

[0073] Method 500 then proceeds to step 506, selecting the target automatic emergency braking mode. This step, based on a predefined decision matrix, combines and analyzes the driving intentions of the vehicular and preceding vehicles, selecting the most suitable target mode from among the relaxed, standard, and regular automatic emergency braking modes for the current traffic scenario. This step embodies an intention-based intelligent decision-making mechanism, enabling scenario-based adaptive adjustment of the braking strategy.

[0074] Continue to refer to Figure 5 Method 500 executes step 508, calculating real-time collision risk parameters. This step calculates risk assessment indicators such as collision time based on the relative distance and relative speed between the vehicle and the vehicle in front, and compares the calculation results with the trigger threshold corresponding to the target automatic emergency braking mode. This step realizes the functions of risk quantification assessment and trigger condition determination, providing a quantitative basis for braking intervention decisions.

[0075] Method 500 ultimately executes step 510, which involves braking intervention. When the real-time collision risk parameters reach the trigger threshold corresponding to the target automatic emergency braking mode, this step performs braking control according to the deceleration strategy of the corresponding mode. The relaxed mode uses a first-level gentle deceleration to achieve preparatory braking, the standard mode uses a second-level progressive deceleration to provide balanced intervention, and the normal mode uses the maximum emergency deceleration to ensure a safety baseline. This step also includes real-time risk monitoring and emergency mode switching functions; when a sharp increase in risk is detected, it immediately switches to the normal braking mode.

[0076] Method 500 demonstrates the collaborative working mechanism of various technical features, forming a complete intelligent braking technology solution. Multimodal information acquisition provides data support for intent determination, intent determination provides decision-making basis for mode selection, mode selection provides strategic guidance for braking execution, and braking execution is dynamically adjusted through risk monitoring. This complete process achieves a technological leap from passive response to proactive prediction, solving the limitations of traditional automatic emergency braking methods in complex traffic scenarios.

[0077] Method 500's technological advantages are reflected in its multi-layered synergistic optimization. At the information acquisition level, multimodal fusion improves the accuracy of environmental understanding; at the intent determination level, multi-parameter analysis enhances the reliability of behavior prediction; at the decision-making level, matrix-based logic improves the precision of strategy matching; and at the execution control level, tiered braking enhances the humanization of intervention effects. This multi-layered technological integration achieves a unity of safety, comfort, and intelligence.

[0078] This invention also discloses an adaptive emergency braking system based on multimodal fusion intent.

[0079] Reference Figure 6An adaptive emergency braking system 600 based on multimodal fusion intent adopts a modular architecture design, achieving intelligent braking intervention through the collaborative work of four functional modules. This system 600 is integrated into the vehicle's intelligent driving computing unit, and through multimodal information fusion and intent prediction technology, it overcomes the technical limitations of traditional automatic emergency braking systems in urban intersection traffic light switching scenarios.

[0080] like Figure 6 As shown, system 600 includes an information acquisition module 602, used to acquire traffic light status, vehicle status, preceding vehicle status, and preceding vehicle brake light status. The information acquisition module 602 uses various sensing devices such as a forward-facing camera, radar sensors, and vehicle status sensors to collect environmental information and vehicle motion parameters in real time. This module 602 identifies traffic light statuses such as flashing green and yellow lights, monitors the vehicle's real-time speed, acceleration, and distance to the stop line, acquires the motion status, relative distance, and relative speed of the preceding vehicle, and detects the illumination status and duration of the preceding vehicle's brake lights.

[0081] System 600 also includes an intent determination module 604, connected to the information acquisition module 602, used to determine the vehicle's intention to proceed based on its own vehicle status, and to determine the preceding vehicle's intention to proceed based on the preceding vehicle's status and brake light status. The intent determination module 604 receives multi-dimensional data from the information acquisition module 602, calculates the safe distance required for the vehicle to park comfortably, and compares it with the actual distance to determine whether the vehicle should continue proceeding or prepare to stop. This intent determination module 604 also integrates parameters such as the duration of the preceding vehicle's brake lights and the collision time between the preceding vehicle and the stop line to determine the preceding vehicle's intention to proceed.

[0082] Continue to refer to Figure 6 The system 600 includes a mode selection module 606 connected to the intent determination module 604. This module selects the corresponding target automatic emergency braking mode from a predefined decision matrix based on a combination of the vehicle's and the preceding vehicle's travel intentions. The mode selection module 606, based on dual intent input, determines the most suitable target mode for the current traffic scenario from three options: a relaxed automatic emergency braking mode, a standard automatic emergency braking mode, and a regular automatic emergency braking mode, by searching the preset decision matrix. This mode selection module 606 implements an intelligent decision-making mechanism based on intent collaboration.

[0083] System 600 further includes a braking control module 608 connected to the mode selection module 606. In the target automatic emergency braking mode, when the real-time collision risk parameters reach the trigger threshold corresponding to that mode, the braking control module 608 performs braking intervention according to the deceleration strategy corresponding to that mode. The braking control module 608 receives a mode selection command from the mode selection module 606, calculates risk parameters such as collision time in real time, and compares the risk parameters with the trigger threshold of the selected mode. When the trigger condition is met, the braking control module 608 executes the braking operation according to the corresponding deceleration strategy.

[0084] In some implementations, the braking control module 608 is connected to the decision control module as an execution module to receive and execute braking commands. This braking control module 608 implements differentiated braking control according to the requirements of different automatic emergency braking modes: a first-level gentle deceleration is used in the relaxed mode, a second-level progressive deceleration is used in the standard mode, and the maximum emergency deceleration is used in the normal mode. The braking control module 608 also has a real-time risk monitoring function; when a sharp increase in collision risk is detected, it immediately switches from the current mode to the normal automatic emergency braking mode.

[0085] The inter-module connections of system 600 form a linear information flow architecture. The information acquisition module 602 transmits perceived data to the intent determination module 604, the intent determination module 604 transmits intent analysis results to the mode selection module 606, and the mode selection module 606 transmits mode selection instructions to the braking control module 608. This sequential processing architecture ensures the effective fusion of multimodal information and the accurate execution of intelligent decisions, realizing a complete closed-loop control from environmental perception to braking intervention.

[0086] System 600 achieves a balance between functional decoupling and technological integration through modular design. The information acquisition module 602 focuses on multimodal data collection, the intent determination module 604 focuses on behavior prediction and analysis, the pattern selection module 606 focuses on strategy decision-making, and the braking control module 608 focuses on execution control implementation. This clearly defined architecture improves the system's maintainability and scalability while ensuring the coordinated optimization of various technical functions.

[0087] The adaptive emergency braking system based on multimodal fusion intent described in this invention improves the accuracy of environmental understanding through multimodal information fusion, solves the problem of traditional methods being unable to understand the behavior of the vehicle in front through intent prediction technology, and realizes humanized intervention from preparatory braking to emergency braking through a graded braking strategy. It significantly reduces the false trigger rate in urban intersection traffic light switching scenarios, optimizes driving comfort while ensuring safety, and improves the adaptability and reliability of the automatic emergency braking system in complex traffic environments.

[0088] This invention also discloses a readable storage medium.

[0089] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the adaptive emergency braking method described in any of the above embodiments. The computer-readable storage medium may include any entity or device capable of carrying a computer program, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), and a software distribution medium, etc. The computer program includes computer program code. The computer program code may be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable storage medium may include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), and a software distribution medium, etc.

[0090] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0091] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a system including a processing module or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0092] 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive emergency braking method based on multimodal fusion intent, characterized in that, include: Get the status of traffic lights, your own vehicle, the vehicle in front, and the brake lights of the vehicle in front; The vehicle's intention to proceed is determined based on its own vehicle status, and the intention of the vehicle in front is determined based on the status of the preceding vehicle and its brake light status. The determination of the preceding vehicle's intention based on its preceding vehicle status and brake light status includes: obtaining the preceding vehicle's brake light illumination status and duration; calculating the collision time between the preceding vehicle and the stop line; when the duration of illumination is greater than a preset time threshold and the collision time is greater than a preset collision time threshold, the preceding vehicle's intention to proceed is determined to be preparing to stop; when the preceding vehicle's brake light is not illuminated, the duration of illumination is less than or equal to the preset time threshold, or the collision time is less than or equal to the preset collision time threshold, the preceding vehicle's intention to proceed is determined to be continuing to proceed. Based on the combination of the vehicle's intention to proceed and the preceding vehicle's intention to proceed, a corresponding target automatic emergency braking mode is selected from a predefined decision matrix. The predefined decision matrix includes: when both the vehicle's intention to proceed and the preceding vehicle's intention to proceed are to continue, a lenient automatic emergency braking mode is selected; when both the vehicle's intention to proceed and the preceding vehicle's intention to proceed are to prepare to stop, a standard automatic emergency braking mode is selected; when the vehicle's intention to proceed is to prepare to stop, a regular automatic emergency braking mode is selected. The trigger threshold for the lenient automatic emergency braking mode is greater than the trigger threshold for the standard automatic emergency braking mode, and the trigger threshold for the standard automatic emergency braking mode is greater than or equal to the trigger threshold for the regular automatic emergency braking mode. In the target automatic emergency braking mode, when the real-time collision risk parameter reaches the trigger threshold corresponding to the mode, braking intervention is performed according to the deceleration strategy corresponding to the mode.

2. The adaptive emergency braking method according to claim 1, characterized in that, The determination of the vehicle's travel intention based on the vehicle's status includes: Get the vehicle's current speed and the distance between the vehicle and the stop line; Calculate the safe distance required for comfortable parking based on the vehicle's current speed; When the distance between the vehicle and the stop line is less than or equal to the safe distance, it is determined that the vehicle's intention to continue passing is to continue. When the distance between the vehicle and the stop line is greater than the safe distance, the vehicle's intention to proceed is determined to be to stop.

3. The adaptive emergency braking method according to claim 1, characterized in that, The preset time threshold is 0.5 seconds, and the preset collision time threshold is 3 seconds.

4. The adaptive emergency braking method according to claim 1, characterized in that, The deceleration strategy includes: The relaxed automatic emergency braking mode employs a first-level gentle deceleration. The standard automatic emergency braking mode uses a second-level progressive deceleration, which is greater than the first-level gentle deceleration. The conventional automatic emergency braking mode uses the maximum emergency deceleration, which is greater than the second-stage progressive deceleration.

5. The adaptive emergency braking method according to claim 1, characterized in that, The real-time collision risk parameter is the collision time.

6. The adaptive emergency braking method according to claim 1, characterized in that, Also includes: Real-time monitoring of changes in collision risk; When a sharp increase in collision risk is detected in either the relaxed automatic emergency braking mode or the standard automatic emergency braking mode, immediately switch to the regular automatic emergency braking mode.

7. The adaptive emergency braking method according to claim 1, characterized in that, The traffic light status includes flashing green and yellow.

8. An adaptive emergency braking system based on multimodal fusion intent, characterized in that, include: The information acquisition module is used to acquire the status of traffic lights, the status of the vehicle itself, the status of the vehicle in front, and the status of the brake lights of the vehicle in front. The intent determination module is used to determine the driving intent of the vehicle based on the vehicle's status and to determine the driving intent of the vehicle in front based on the status and brake light status of the vehicle in front. The determination of the driving intent based on the status and brake light status of the vehicle in front includes: acquiring the illumination status and duration of the brake lights of the vehicle in front; calculating the collision time between the vehicle in front and the stop line; determining the driving intent of the vehicle in front to be preparing to stop when the duration of illumination is greater than a preset time threshold and the collision time is greater than a preset collision time threshold; and determining the driving intent of the vehicle in front to continue driving when the brake lights of the vehicle in front are not illuminated, the duration of illumination is less than or equal to the preset time threshold, or the collision time is less than or equal to the preset collision time threshold. The mode selection module is used to select a corresponding target automatic emergency braking mode from a predefined decision matrix based on the combination of the vehicle's intention to proceed and the preceding vehicle's intention to proceed. The predefined decision matrix includes: a lenient automatic emergency braking mode when both the vehicle's intention to proceed and the preceding vehicle's intention to proceed are to continue; a standard automatic emergency braking mode when both the vehicle's intention to proceed and the preceding vehicle's intention to proceed are to stop; and a regular automatic emergency braking mode when both the vehicle's intention to proceed are to stop. The trigger threshold for the lenient automatic emergency braking mode is greater than the trigger threshold for the standard automatic emergency braking mode, and the trigger threshold for the standard automatic emergency braking mode is greater than or equal to the trigger threshold for the regular automatic emergency braking mode. The braking control module is used to intervene in braking according to the deceleration strategy corresponding to the mode when the real-time collision risk parameter reaches the trigger threshold corresponding to the mode in the target automatic emergency braking mode.

9. A readable storage medium, characterized in that, The readable storage medium stores computer instructions that, when executed by a processor, implement the adaptive emergency braking method as described in any one of claims 1-7.

Images