Automobile automatic emergency braking system and control method based on fusion data

Abstract

The present invention discloses an automobile automatic emergency braking system based on fused data and a control method thereof, which belongs to the technical field of automobile braking control, and specifically includes: obtaining driving environment information and vehicle internal data through sensors and Internet of Things methods, and fusing these data to obtain accurate automobile status; using a hierarchical warning model and a deep learning algorithm to predict dangerous scenarios, and giving different levels of warning information according to the prediction results; when receiving a collision or emergency braking warning, the system adopts a target recognition strategy to assess the degree of danger and sends an emergency braking instruction; through vehicle-to-vehicle communication, multi-vehicle coordinated braking is achieved, and automatic emergency braking is performed using an adaptive control strategy; the system receives sensor feedback, optimizes the control strategy in real time, and feeds back to the traffic management system through vehicle-to-infrastructure communication, thereby improving driving safety and traffic efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of automobile braking control, and in particular to an automobile automatic emergency braking system based on fusion data and a control method thereof. Background Art

[0002] With economic development, rising national incomes, and improved road conditions, cars are becoming the preferred mode of transportation for daily commutes and are becoming a common consumer good for millions of households. As the number of cars continues to grow, people are demanding higher levels of operational convenience, driving comfort, and safety. Many electric vehicles are equipped with automatic emergency braking to prevent collisions.

[0003] For example, patent publication number CN111845668A discloses an automatic emergency braking method for electric vehicles, including: the vehicle controller determines whether the braking torque T is less than or equal to the motor braking torque value calibrated during vehicle initialization; if so, the vehicle controller sends a braking torque request value T to the motor to implement motor braking when the remaining power battery is less than 80% of the calibrated power, there is no fault prohibiting charging, and the brake pedal opening is zero; if not, the vehicle controller sends a torque request signal to the electronic stability module to directly implement hydraulic braking. The above technical solution is simple, practical, safe, reliable, and low-cost. During automatic emergency braking, energy is recovered by motor braking to charge the power battery, which increases battery range to a certain extent, reduces wear on the ESC reversing valve, and extends its service life.

[0004] The above existing technologies all have the following problems: 1) lack of a comprehensive early warning model; 2) difficulty in system maintenance; 3) lack of flexibility to adapt to different driving conditions and emergency situations, resulting in unsatisfactory or insufficient braking effect; 4) lack of ability to coordinate braking with other vehicles. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention proposes an automobile automatic emergency braking system and a control method based on fused data. The system obtains driving environment information and vehicle internal data through sensors and Internet of Things methods, and fuses these data to obtain the accurate vehicle status; uses a hierarchical warning model and a deep learning algorithm to predict dangerous scenarios, and gives different levels of warning information based on the prediction results; when receiving a collision or emergency braking warning, the system uses a target recognition strategy to assess the degree of danger and sends an emergency braking command; through vehicle-to-vehicle communication, multi-vehicle coordinated braking is achieved, and automatic emergency braking is performed using an adaptive control strategy; the system receives sensor feedback, optimizes the control strategy in real time, and feeds back to the traffic management system through vehicle-to-infrastructure communication, thereby improving driving safety and traffic efficiency.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The automobile automatic emergency braking system based on fusion data includes: a decision module and a control module;

[0008] The decision module includes a target recognition unit and a braking decision unit; the target recognition unit is configured with a target recognition strategy based on contours and segmentation, which is used to identify and classify objects in the captured visual information and distinguish between pedestrians, vehicles and obstacles; the braking decision unit is configured with a priority strategy, which is used to set different priorities according to the urgency and importance of different factors and issue emergency braking instructions;

[0009] The control module includes a braking control unit and a power control unit; the braking control unit is used to implement the emergency braking command issued by the decision module; the power control unit is configured with a fast coordination control strategy, which includes responding within a short time when receiving the emergency braking command and coordinating the adjustment of engine and transmission operations.

[0010] The vehicle automatic emergency braking control method based on fusion data includes:

[0011] Step S1: Acquire driving environment information and vehicle internal data through sensors and Internet of Things methods, and fuse the driving environment information and vehicle internal data from different sensors to obtain fused vehicle data;

[0012] Step S2: Construct a hierarchical warning model and a dangerous scene prediction model based on a deep learning algorithm. Use the fused vehicle data to predict existing dangerous scenes. Based on the prediction results and combined with real-time traffic information, the warning levels are divided into safe driving level, collision warning level, and emergency braking level, and corresponding warning information is given.

[0013] Step S3: If a collision warning level or emergency braking level warning information is received, the object recognition strategy based on contour and segmentation is used to identify and classify the objects in the captured visual information, distinguishing between pedestrians, vehicles and obstacles, and assessing the degree of danger based on the recognition results and the preset priority strategy. If the degree of danger reaches the preset threshold, an emergency braking command is issued;

[0014] Step S4: Based on the warning information and the vehicle dynamics model, an adaptive control strategy is formulated to perform multi-vehicle coordinated braking through vehicle-to-vehicle communication. The adaptive control strategy is sent to the brake in the braking control unit to perform automatic emergency braking. At the same time, when an emergency braking command is received, a fast coordinated control strategy is used to reduce the vehicle speed;

[0015] Step S5: Receive feedback signals from sensors, monitor the braking effect and driving status in real time, and optimize and adjust the adaptive control strategy based on the braking effect and driving status using a machine learning algorithm. Feedback the optimized adaptive control strategy to the traffic management system through vehicle-to-infrastructure communication.

[0016] Specifically, the specific steps of step S2 include:

[0017] S2.1: Set the warning level to safe driving level, collision warning level, and emergency braking level. Based on the fused car data, the warning thresholds for each level are set according to the vehicle driving status, relative distance, and relative speed data. 、 、 ;

[0018] S2.2: Based on the preset warning threshold 、 、 , determine the warning level of the current driving state, and obtain a hierarchical warning model based on the judgment result;

[0019] S2.3: Obtain a pre-trained deep learning prediction model. Simultaneously, perform feature extraction on the fused vehicle data to obtain dangerous scene data. Pre-process the dangerous scene data and input the pre-processed dangerous scene data into the pre-trained deep learning model to obtain a dangerous scene prediction model.

[0020] S2.4: Based on the prediction results output by the dangerous scenario prediction model and combined with the hierarchical warning model, the current warning level is determined, and corresponding warning information is output through sound and light alarms according to the warning level.

[0021] Specifically, the specific steps of step S3 include:

[0022] S3.1: Receive warning information and extract the warning level, determine whether the warning level is a collision warning level or an emergency braking level, and if so, capture real-time image frames through the camera on the vehicle and pre-process the captured image frames;

[0023] S3.2: Use the improved contour detection algorithm to extract the object contour in the image frame and obtain the contour image. The formula is:

[0024] ;

[0025] in, Represents an image frame f The second derivative of Represented in the image frame three-dimensional coordinate system x ,y , z The direction of i The coordinates of the dimensions, n Indicates the number of dimensions, Indicates the i The weight of the contour image in dimensions, Indicates the i The scale characteristics of the dimensions, represents partial derivative.

[0026] Specifically, the specific steps of step S3 also include:

[0027] S3.3: Set the second-order derivative threshold of the image frame to , obtain the threshold comparison result;

[0028] like , then the coordinates The corresponding pixel belongs to the contour area;

[0029] like , then the coordinates The corresponding pixel belongs to the background;

[0030] S3.4: According to the threshold comparison result, the contour image is converted into a binary contour image, and the binary contour image is divided into m Class object, obtain the segmentation result of binary contour image;

[0031] S3.5: Extract features from the segmented objects and match the extracted features with the pre-trained random forest classifier to identify pedestrians, vehicles, and obstacles in the image.

[0032] Specifically, the specific steps of step S3 also include:

[0033] S3.6: Set the priority strategy and obtain the Every The number of class objects , and assign a priority weight to each type of object according to the preset priority strategy;

[0034] S3.7: For each object in each class, measure the distance from the object to the vehicle, normalize the distance, and calculate a hazard score based on its type, priority weight, and normalized distance, using the formula:

[0035] ;

[0036] in, represents the total risk score, M represents the k the number of class objects, Indicates thek Class object j The weight of the object, D represents the normalized distance, m represents the object category, k≤m, j≤M;

[0037] S3.8: Set the risk score threshold H, if , emergency braking is triggered and an emergency braking command is sent to the vehicle braking system, and the vehicle braking system performs emergency braking according to the command.

[0038] Specifically, the priority strategy in S3.6 includes:

[0039] ;

[0040] in, Indicates the level of high-speed vehicles, Indicates the level of slow-moving pedestrians, Indicates the level of large vehicles and buildings, Indicates the level of small vehicles and pedestrians, Indicates pedestrians and vehicles moving in the same or opposite direction as the vehicle, and the distance between vehicles Meter level, Indicates the distance between vehicles Meters and in the direction of vehicle movement Angle of pedestrian and vehicle levels, Indicates the distance between pedestrians and vehicles Meter level, Indicates the level of accelerating pedestrians and vehicles, Indicates the level of pedestrians or vehicles that are decelerating and moving at a constant speed. Indicates the level of vehicles on the highway, Indicates the level of pedestrians on the highway, Indicates the level of pedestrians on the sidewalk, Indicates the level of vehicles on the sidewalk.

[0041] Specifically, an electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of a method for controlling automatic emergency braking of an automobile based on fused data are implemented.

[0042] Specifically, a computer-readable storage medium stores computer instructions, which, when executed, execute the steps of a method for controlling an automatic emergency braking of an automobile based on fused data.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention proposes an automobile automatic emergency braking system based on fused data, and optimizes and improves the architecture, operating steps and processes. The system has the advantages of simple processes, low investment and operating costs, and low production costs.

[0045] 2. The present invention proposes a method for controlling automatic emergency braking of automobiles based on fused data. It uses sensors and Internet of Things methods to obtain real-time driving environment information and vehicle internal data, and combines deep learning algorithms to predict dangerous scenarios. It can give early warnings when potential collision risks occur, reducing the driver's reaction time, thereby effectively avoiding or mitigating collision accidents. Upon receiving collision warning-level or emergency braking-level warning information, the system can quickly identify the target and assess the degree of danger. When the degree of danger reaches a preset threshold, it automatically sends an emergency braking command to ensure that the vehicle takes braking measures in the shortest possible time, reducing the possibility of accidents.

[0046] 3. The present invention proposes a method for controlling automatic emergency braking of vehicles based on fused data. This method uses an adaptive control strategy, combined with a vehicle dynamics model and real-time traffic information, to develop a more accurate braking plan, avoiding sudden vehicle jitter or excessive deceleration during emergency braking, and improving the comfort and smoothness of the braking process. It also reduces vehicle speed through a rapid coordinated control strategy, allowing for more rapid speed control in emergency situations, reducing collisions caused by untimely or insufficient braking. Furthermore, it enables multi-vehicle coordinated braking through vehicle-to-vehicle communication, improving the safety and efficiency of road traffic. In emergency situations, multi-vehicle coordinated braking can avoid or mitigate chain collisions caused by inconsistent braking between vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is an architecture diagram of the automobile automatic emergency braking system based on fusion data of the present invention;

[0048] Figure 2 A schematic diagram of an automobile automatic emergency braking control method based on fused data according to the present invention;

[0049] Figure 3 This is a flowchart of the automobile automatic emergency braking control method based on fusion data of the present invention;

[0050] Figure 4 This is a vehicle emergency braking flow chart of the vehicle automatic emergency braking control method based on fusion data of the present invention. DETAILED DESCRIPTION

[0051] Example 1

[0052] See also Figure 1, an embodiment provided by the present invention: an automobile automatic emergency braking system based on fusion data, comprising: a perception module, a decision module, a control module, a communication module, and an interaction module;

[0053] The perception module is used to collect information about the vehicle's surroundings, including vehicles ahead, pedestrians, obstacles, and road conditions;

[0054] The decision module receives data from the perception module and uses advanced algorithms to analyze whether there is a collision risk and decide whether to initiate emergency braking.

[0055] The control module is used to execute the instructions issued by the decision module and control the operation of the braking system;

[0056] The communication module is responsible for the communication between the modules within the system and the communication with the external vehicle intelligent system;

[0057] The interaction module is used for interaction between the system and the driver, such as displaying system status and receiving driver instructions.

[0058] The perception module includes: sensor unit, ranging unit, and data fusion unit;

[0059] A sensor unit for collecting environmental information around the vehicle;

[0060] The ranging unit is used to calculate the distance between the vehicle and the obstacle in front through radar;

[0061] The data fusion unit is used to integrate data from different sensors to improve the accuracy and robustness of the data.

[0062] The decision-making module includes a target recognition unit and a braking decision unit. The target recognition unit is equipped with a target recognition strategy based on contours and segmentation. This strategy is used to identify and classify objects in the captured visual information and distinguish between pedestrians, vehicles and obstacles. The braking decision unit is equipped with a priority strategy. The priority strategy is used to set different priorities based on the urgency and importance of different factors and issue emergency braking commands. This decision takes into account multiple factors such as vehicle speed, distance to the obstacle, and the nature of the obstacle. Once a collision risk is determined, a braking command is issued to the control module.

[0063] The decision module also includes a data processing unit; the data processing unit is used to receive data from the perception module, conduct in-depth analysis and processing of the data, and use machine learning algorithms to identify existing collision risks and evaluate the consequences of different response strategies;

[0064] The control module includes a brake control unit and a power control unit; the brake control unit is used to implement the emergency braking command issued by the decision module, control the vehicle's hydraulic or electronic braking system, and perform emergency braking at the fastest speed and maximum braking force to avoid or mitigate collisions; the power control unit is equipped with a rapid coordination control strategy, which includes responding within a short time when receiving an emergency braking command and coordinating the adjustment of engine and transmission operations.

[0065] The communication module includes: on-board communication unit and roadside communication unit;

[0066] The vehicle communication unit is responsible for communicating with nearby vehicles and infrastructure to obtain additional environmental information and transmit the vehicle's status information;

[0067] The roadside communication unit is responsible for communicating with the traffic management center, receiving traffic status information, and also sending emergency braking events.

[0068] Example 2

[0069] See also Figure 2-Figure 4 Another embodiment of the present invention provides a method for controlling automatic emergency braking of an automobile based on fusion data, comprising:

[0070] Step S1: Acquire driving environment information and vehicle internal data through sensors and Internet of Things methods, and fuse the driving environment information and vehicle internal data from different sensors to obtain fused vehicle data;

[0071] Among them, driving environment information includes: road conditions ahead, obstacle locations, and vehicle speed; vehicle internal data includes: vehicle speed, engine status, and brake system status.

[0072] Step S2: Construct a hierarchical warning model and a dangerous scene prediction model based on a deep learning algorithm. Use the fused vehicle data to predict existing dangerous scenes. Based on the prediction results and combined with real-time traffic information, the warning levels are divided into safe driving level, collision warning level, and emergency braking level, and corresponding warning information is given.

[0073] Dangerous scenarios include: pedestrians crossing the road, sudden deceleration of the vehicle ahead, and road obstacles;

[0074] Real-time traffic information includes: traffic flow, signal light status, road construction conditions;

[0075] Warning information includes: road conditions, status of other vehicles, and obstacle information.

[0076] Step S3: If a collision warning level or emergency braking level warning information is received, the object recognition strategy based on contour and segmentation is used to identify and classify the objects in the captured visual information, distinguishing between pedestrians, vehicles and obstacles, and assessing the degree of danger based on the recognition results and the preset priority strategy. If the degree of danger reaches the preset threshold, an emergency braking command is issued;

[0077] Step S4: Based on the warning information and the vehicle dynamics model, an adaptive control strategy is formulated to perform multi-vehicle coordinated braking through vehicle-to-vehicle communication. The adaptive control strategy is sent to the brake in the braking control unit to perform automatic emergency braking. At the same time, when an emergency braking command is received, a fast coordinated control strategy is used to reduce the vehicle speed;

[0078] Among them, the adaptive control strategy includes: braking timing, braking force distribution, and braking time control; the rapid coordination control strategy includes: adjusting engine torque and activating the electronic stability program.

[0079] Furthermore, the specific steps of step S4 include:

[0080] (1) Receive and input warning information, vehicle status information, and vehicle dynamics model parameters;

[0081] (2) Analyze, classify, and calculate the input data to determine whether emergency braking measures are necessary;

[0082] (3) Formulate adaptive control strategies and rapid coordinated control strategies based on the analysis results;

[0083] (4) Converting the adaptive control strategy into specific braking instructions and sending them to the brake via communication. The control strategy includes detailed instructions on how to adjust the braking force based on the current vehicle state, road conditions, and predicted dangerous scenarios. These instructions are sent to the vehicle MCU via the vehicle controller, which further controls the output torque of the main motor and auxiliary motor. At the same time, the vehicle controller also directly controls the electronic hydraulic brake to output the corresponding braking torque.

[0084] (5) The brake executes the command to generate braking deceleration by increasing the friction between the brake disc and the brake pad, slowing down or stopping the vehicle, while monitoring the system status and braking effect;

[0085] (6) Optimize and adjust the adaptive control strategy based on real-time data during the braking process.

[0086] Step S5: Receive feedback signals from sensors, monitor the braking effect and driving status in real time, and optimize and adjust the adaptive control strategy based on the braking effect and driving status using a machine learning algorithm. Feedback the optimized adaptive control strategy to the traffic management system through vehicle-to-infrastructure communication.

[0087] Among them, the driving status includes: vehicle speed change, braking distance, and vehicle body posture.

[0088] The specific steps of step S2 include:

[0089] S2.1: Set the warning level to safe driving level, collision warning level, and emergency braking level. Based on the fused car data, the warning thresholds for each level are set according to the vehicle driving status, relative distance, and relative speed data. 、 、 ;

[0090] S2.2: Based on the preset warning threshold 、 、 , determine the warning level of the current driving state, and obtain a hierarchical warning model based on the judgment result;

[0091] S2.3: Obtain a pre-trained deep learning prediction model. Simultaneously, perform feature extraction on the fused vehicle data to obtain dangerous scene data. Pre-process the dangerous scene data, and input the pre-processed dangerous scene data into the pre-trained deep learning model to obtain a dangerous scene prediction model. The method for obtaining the pre-trained deep learning prediction model is prior art in this field and does not constitute an inventive solution of the present application, and is not further described herein.

[0092] S2.4: Based on the prediction results output by the dangerous scenario prediction model and combined with the hierarchical warning model, the current warning level is determined, and corresponding warning information is output through sound and light alarms according to the warning level.

[0093] The specific steps of step S3 include:

[0094] S3.1: Receive warning information and extract the warning level, determine whether the warning level is a collision warning level or an emergency braking level, and if so, capture real-time image frames through the camera on the vehicle and pre-process the captured image frames;

[0095] S3.2: Use the improved contour detection algorithm to extract the object contour in the image frame and obtain the contour image. The formula is:

[0096] ;

[0097] in, Represents an image frame f The second derivative of Represented in the image frame three-dimensional coordinate system x , y , z The direction of i The coordinates of the dimensions, n Indicates the number of dimensions, Indicates the i The weight of the contour image in dimensions, Indicates the i The scale characteristics of the dimensions, represents partial derivative.

[0098] The specific steps of step S3 also include:

[0099] S3.3: Set the second-order derivative threshold of the image frame to , obtain the threshold comparison result;

[0100] like , then the coordinates The corresponding pixel belongs to the contour area;

[0101] like , then the coordinates The corresponding pixel belongs to the background;

[0102] Among them, the image frame second-order derivative threshold It can be adjusted according to the specific image and the contour details that need to be extracted. If the threshold is set too high, important contour information will be ignored; if the threshold is set too low, too much noise or unnecessary details will be introduced.

[0103] S3.4: According to the threshold comparison result, the contour image is converted into a binary contour image, and the binary contour image is divided into m Classify the object and obtain the segmentation result of the binary contour image, wherein the clustering method is the existing technical content in this field and is not the inventive solution of this application, and will not be described in detail here;

[0104] When converting a contour image to a binary contour image, the values ​​of pixels belonging to the contour area are set to 1 (or 255, depending on the image data type and representation), and the values ​​of pixels belonging to the background area are set to 0.

[0105] S3.5: Extract features from the segmented objects and match the extracted features with the pre-trained random forest classifier to identify pedestrians, vehicles, and obstacles in the image.

[0106] The specific steps of step S3 also include:

[0107] S3.6: Set the priority strategy and obtain the Every The number of class objects , and assign a priority weight to each type of object according to the preset priority strategy;

[0108] S3.7: For each object in each class, measure the distance from the object to the vehicle, normalize the distance, and calculate a hazard score based on its type, priority weight, and normalized distance, using the formula:

[0109] ;

[0110] in, represents the total risk score, M represents the k the number of class objects, Indicates the k Class object j The weight of the object, D represents the normalized distance, m Indicates the object category, k≤m; j≤M;

[0111] S3.8: Set the risk score threshold H, if , emergency braking is triggered and an emergency braking command is sent to the vehicle braking system, and the vehicle braking system performs emergency braking according to the command.

[0112] The priority strategies in S3.6 include:

[0113] ;

[0114] in, Indicates the level of high-speed vehicles, Indicates the level of slow-moving pedestrians, Indicates the level of large vehicles and buildings, Indicates the level of small vehicles and pedestrians, Indicates pedestrians and vehicles moving in the same or opposite direction as the vehicle, and the distance between vehicles Meter level, Indicates the distance between vehicles Meters and in the direction of vehicle movement Angle of pedestrian and vehicle levels, Indicates the distance between pedestrians and vehicles Meter level, Indicates the level of accelerating pedestrians and vehicles, Indicates the level of pedestrians or vehicles that are decelerating and moving at a constant speed. Indicates the level of vehicles on the highway, Indicates the level of pedestrians on the highway, Indicates the level of pedestrians on the sidewalk, Indicates the level of vehicles on the sidewalk.

[0115] It can be further summarized as follows: fast-moving objects have higher priority than slow-moving objects because they may cause danger in a shorter time; large obstacles have higher priority than small obstacles because they may take up more road space or cause greater damage; if the objects are moving in the same or opposite direction as the vehicle and are closer, they have higher priority than objects at a certain angle to the vehicle's direction of movement; objects that are accelerating have higher priority than objects that are decelerating or moving at a constant speed because they may approach the vehicle faster; on highways, vehicles may have higher priority, while on sidewalks, pedestrians may have higher priority.

[0116] Example 3

[0117] An electronic device includes a memory and a processor, wherein the memory stores a computer program and the processor implements the steps of a method for controlling automatic emergency braking of an automobile based on fused data when executing the computer program.

[0118] A computer-readable storage medium stores computer instructions, which, when executed, execute the steps of a method for controlling an automatic emergency braking of an automobile based on fused data.

[0119] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific embodiments. The above-mentioned specific embodiments are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention. These are all protected by the present invention.

Claims

1. An automatic emergency braking system for automobiles based on fusion data, characterized in that: include: Decision module and control module; The decision module includes a target recognition unit and a braking decision unit; The target recognition unit is configured with a target recognition strategy based on contours and segmentation, which is used to identify and classify objects in the captured visual information and distinguish between pedestrians, vehicles and obstacles; The braking decision unit is configured with a priority strategy, which is used to set different priorities according to the urgency and importance of different factors and send emergency braking instructions; The warning levels are divided into safe driving level, collision warning level, and emergency braking level, and corresponding warning information is given; If a collision warning or emergency braking warning is received, the system uses contour- and segmentation-based object recognition strategies to identify and classify objects in the captured visual information, distinguishing between pedestrians, vehicles, and obstacles. The system then assesses the degree of danger based on the recognition results and a preset priority strategy. If the degree of danger reaches a preset threshold, it issues an emergency braking command. The control module includes a brake control unit and a power control unit; the brake control unit is used to implement the emergency braking command issued by the decision module; the power control unit is configured with a fast coordinated control strategy, which includes responding to the emergency braking command within a short time and coordinating the operation of the engine and transmission; Based on the warning information and vehicle dynamics model, an adaptive control strategy is formulated to perform multi-vehicle coordinated braking through vehicle-to-vehicle communication, and the adaptive control strategy is sent to the brake in the braking control unit to perform automatic emergency braking. At the same time, when an emergency braking command is received, a fast coordinated control strategy is used to reduce the vehicle speed.

2. A method for controlling an automatic emergency braking system for an automobile based on fused data, implemented based on the automatic emergency braking system for an automobile based on fused data according to claim 1, characterized in that: include: Step S1: Acquire driving environment information and vehicle internal data through sensors and Internet of Things methods, and fuse the driving environment information and vehicle internal data from different sensors to obtain fused vehicle data; Step S2: Construct a hierarchical warning model and a dangerous scene prediction model based on a deep learning algorithm. Use the fused vehicle data to predict existing dangerous scenes. Based on the prediction results and combined with real-time traffic information, the warning levels are divided into safe driving level, collision warning level, and emergency braking level, and corresponding warning information is given. Step S3: If a collision warning level or emergency braking level warning information is received, the object recognition strategy based on contour and segmentation is used to identify and classify the objects in the captured visual information, distinguishing between pedestrians, vehicles and obstacles, and assessing the degree of danger based on the recognition results and the preset priority strategy. If the degree of danger reaches the preset threshold, an emergency braking command is issued; Step S4: Based on the warning information and the vehicle dynamics model, an adaptive control strategy is formulated to perform multi-vehicle coordinated braking through vehicle-to-vehicle communication. The adaptive control strategy is sent to the brake in the braking control unit to perform automatic emergency braking. At the same time, when an emergency braking command is received, a fast coordinated control strategy is used to reduce the vehicle speed; Step S5: Receive feedback signals from sensors, monitor braking effects and driving status in real time, and optimize and adjust the adaptive control strategy based on the braking effects and driving status using a machine learning algorithm. Feedback the optimized adaptive control strategy to the traffic management system via vehicle-to-infrastructure communication. The specific steps of step S3 include: S3.1: Receive warning information and extract the warning level, determine whether the warning level is a collision warning level or an emergency braking level, and if so, capture real-time image frames through the camera on the vehicle and pre-process the captured image frames; S3.2: Use the improved contour detection algorithm to extract the object contour in the image frame and obtain the contour image. The formula is: in, represents the second-order derivative of the image frame f, (x i ,y i ,z i ) represents the coordinate of the i-th dimension in the x, y, and z directions in the three-dimensional coordinate system of the image frame, n represents the number of dimensions, and w i represents the weight of the contour image in the i-th dimension, σ i represents the scale feature of the i-th dimension, represents partial derivative; The specific steps of step S3 also include: S3.3: Set the second-order derivative threshold of the image frame to h4 and obtain the threshold comparison result; like Then the coordinate (x i ,y i ,z i ) The corresponding pixel belongs to the contour area; like Then the coordinate (x i ,y i ,z i ) The corresponding pixel belongs to the background; S3.4: Convert the contour image into a binary contour image based on the threshold comparison result, and use a clustering method to divide the binary contour image into m types of objects to obtain a segmentation result of the binary contour image; S3.5: Extract features from the segmented objects and match the extracted features with a pre-trained random forest classifier to identify pedestrians, vehicles, and obstacles in the image. The specific steps of step S3 also include: S3.6: Set the priority strategy and obtain the number M of objects corresponding to each category based on the recognition results in S3.5 m , and assign a priority weight to each type of object according to the preset priority strategy; S3.7: For each object in each class, measure the distance from the object to the vehicle, normalize the distance, and calculate a hazard score based on its type, priority weight, and normalized distance, using the formula: Among them, S total represents the total risk score, M represents the number of objects in the kth category, and W k,j represents the weight of the jth object in the kth class of objects, D represents the normalized distance, m represents the object category, k≤m; j≤M; S3.8: Set the risk score threshold H, if S total ≥H, emergency braking is triggered and an emergency braking command is sent to the vehicle braking system, which performs emergency braking according to the command.

3. The method for controlling automatic emergency braking of an automobile based on fusion data according to claim 2, wherein: The specific steps of step S2 include: S2.1: Set the warning level to safe driving level, collision warning level, and emergency braking level. Based on the fused vehicle data, the warning thresholds for each level are set as h1, h2, and h3 according to the vehicle's driving status, relative distance, and relative speed data. S2.2: Determine the warning level to which the current driving state belongs based on the preset warning thresholds h1, h2, and h3, and obtain a graded warning model based on the judgment result; S2.3: Obtain a pre-trained deep learning prediction model. Simultaneously, perform feature extraction on the fused vehicle data to obtain dangerous scene data. Pre-process the dangerous scene data and input the pre-processed dangerous scene data into the pre-trained deep learning model to obtain a dangerous scene prediction model. S2.4: Based on the prediction results output by the dangerous scenario prediction model and combined with the hierarchical warning model, the current warning level is determined, and corresponding warning information is output through sound and light alarms according to the warning level.

4. The method for controlling automatic emergency braking of an automobile based on fusion data according to claim 3, wherein: The priority strategy in S3.6 includes: Among them, v vh Indicates the level of high-speed vehicles, p hu Indicates the level of slow-moving pedestrians, bar max Indicates the level of large vehicles and buildings, bar min Indicates the level of small vehicles and pedestrians, d min,θ′ Indicates pedestrians and vehicles moving in the same or opposite direction as the vehicle, and the distance between vehicles Meter level, d min,θ Indicates the distance between vehicles The level of pedestrians and vehicles that are at an angle θ to the direction of vehicle movement, d max Indicates the level of distance between pedestrians and vehicles d ≥ 500 meters, v a Indicates the level of accelerating pedestrians and vehicles, v u Indicates the level of pedestrians or vehicles that are slowing down and moving at a constant speed, fway vh Indicates the level of vehicles on the highway, fway hu Indicates the level of pedestrians on the highway, side hu Indicates the level of pedestrians on the sidewalk, side vh Indicates the level of vehicles on the sidewalk.

5. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the vehicle automatic emergency braking control method based on fusion data described in any one of claims 2 to 4 are implemented.

6. A computer-readable storage medium, characterized in that Computer instructions are stored thereon, and when the computer instructions are executed, the steps of the vehicle automatic emergency braking control method based on fusion data as described in any one of claims 2 to 4 are executed.

Images