Control method, device and equipment for automatic emergency braking of vehicle and medium

Abstract

The embodiment of the invention provides a control method, device, equipment and medium for automatic emergency braking of a vehicle, and the method comprises the steps: collecting first sensing data of the surrounding environment of the vehicle through a low-cost laser radar, collecting second sensing data of the surrounding environment of the vehicle through a low-resolution millimeter-wave radar, performing space-time synchronization on the first sensing data and the second sensing data; dynamically determining a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter wave radar according to different conditions of the target; performing weighted fusion on the first sensing data and the second sensing data according to the first weight and the second weight to obtain multi-modal data; and calculating relative driving state information of the vehicle and the target by adopting a lightweight algorithm model based on the multi-modal data, and triggering graded braking according to different relative driving state information. According to the scheme, the hardware cost and the system complexity are reduced, the environmental adaptability and the sensing precision are improved, and the real-time performance of the AEB system is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to a control method, device, equipment and medium for automatic emergency braking of a vehicle. Background Art

[0002] Automatic emergency braking (AEB) is a core feature of active safety in intelligent driving. Its technology has evolved from a single sensor to multi-modal fusion. The core challenges of existing AEB systems include:

[0003] Cost and complexity: LiDAR is used in high-end vehicles (such as the Ideal AD Max system) due to its high precision, but its high cost (single-line LiDAR costs about 500, while multi-line LiDAR (such as 128-line) costs over 2000) makes it difficult to popularize in the mid- and low-end markets.

[0004] Insufficient environmental adaptability: Pure vision solutions (such as Tesla's early Autopilot) are significantly affected by lighting, rain, and fog. Although millimeter-wave radar has strong anti-interference capabilities, low-resolution versions (such as 24GHz radar) have poor static target recognition capabilities.

[0005] Real-time bottleneck: Traditional fusion algorithms (such as target-level fusion) need to process heterogeneous data from multiple sensors, resulting in response delays (usually >150ms), which makes it difficult to meet emergency braking requirements.

[0006] The existing automatic emergency braking implementation solutions are as follows:

[0007] Solution 1: Camera + millimeter-wave radar fusion system

[0008] Structural principle: A 77GHz millimeter-wave radar (detection range 150-250m) is combined with a monocular / binocular camera (resolution 1280×720), and the radar distance / speed data and camera semantic information are associated through target-level fusion (such as the Hungarian algorithm).

[0009] Representative products:

[0010] Bosch's third-generation AEB system integrates forward-facing radar and cameras to support pedestrian detection in urban conditions. However, it relies on high-resolution radar (costing approximately $300) and has a high false trigger rate for static targets.

[0011] Volvo City Safety: This system uses radar and camera fusion and a TTC (time to collision) model for decision-making, but its missed detection rate for low obstacles (such as wire mesh) exceeds 15%.

[0012] limitation:

[0013] High-resolution radar and cameras require complex calibration, with a system cost of approximately $500. Low-resolution millimeter-wave radar (e.g., 24GHz) has insufficient accuracy (±0.5m) at close range (<30m), leading to false braking or missed detections. Data fusion also takes a long time (averaging 120ms).

[0014] Option 2: Low-cost LiDAR + millimeter-wave radar fusion

[0015] Structural principle: A multi-line laser radar (angular resolution 0.1°) is combined with a low-resolution millimeter-wave radar (such as 79GHz), and target tracking accuracy is improved through spatiotemporal synchronization of point cloud and radar signals (such as Kalman filtering).

[0016] Representative products:

[0017] Ideal Lidar Pilot: Integrates a 128-line laser radar and millimeter-wave radar, enabling braking from 80 km / h in rain and fog. However, the system costs over $1,500 and consumes high power (>30W).

[0018] Mobileye EyeQ5: Supports lidar raw data fusion, but relies on a dedicated ASIC chip and has poor hardware compatibility.

[0019] limitation:

[0020] The cost and volume of multi-line lidar limit lightweight design; the point cloud of single-line lidar is sparse (thousands of points per second), making it difficult to classify targets at long distances (>80m).

[0021] Solution 3: Lightweight multimodal fusion solution

[0022] Structural principle: Use low-resolution millimeter-wave radar (such as 4D imaging radar) and vision fusion, and compress model parameters (<10MB) through lightweight neural networks (such as MobileNet) to reduce computing power requirements.

[0023] Representative products:

[0024] Huawei MDC platform: supports millimeter-wave radar + camera fusion, with a response time of 110ms, but relies on a high-computing domain controller (power consumption > 20W);

[0025] JIMU Intelligent Commercial Vehicle AEB: Based on vision + radar fusion, the false trigger rate in backlit scenes is less than 5%, but without integrated lidar, the missed detection rate of small targets at night is still high.

[0026] limitation:

[0027] In dynamic scenarios, fixed sensor weights (such as relying on radar for long distances) lead to errors in speed estimation of close-range targets; data preprocessing (such as point cloud filtering) occupies more than 30% of computing resources, and the delay bottleneck is significant.

[0028] Therefore, the existing technical solutions have the following disadvantages:

[0029] (1) High hardware cost and complex system

[0030] Dependence on multi-line lidar: Existing solutions (such as Ideal Lidar Pilot) require a 128-line lidar (costing over $2,000) to achieve high-precision point clouds, resulting in high system costs.

[0031] High-resolution sensor redundancy: Solutions such as Bosch and Volvo rely on a combination of high-resolution millimeter-wave radar (77GHz, costing $300+) and cameras. This requires high hardware redundancy and complex calibration (the error must be <0.1°).

[0032] (2) Insufficient environmental adaptability and perception accuracy

[0033] Shortcomings of a single sensor: Pure vision solutions (Tesla's early Autopilot) fail in backlight, rain, and fog; low-resolution millimeter-wave radar (24GHz) has a large ranging error (±0.5m) for static targets;

[0034] Rigid fusion strategy: Existing fusion systems (such as Huawei's MDC) have fixed sensor weights, relying on radar for long-range detection and cameras for close-range detection. This results in target velocity estimation errors exceeding 15% in dynamic scenarios (such as vehicle cutting in).

[0035] (3) Poor real-time performance and high computing power requirements

[0036] Data fusion latency: Traditional target-level fusion (such as Mobileye EyeQ5) requires multi-level data alignment and matching, taking an average of 120-150ms.

[0037] Preprocessing resource usage: Point cloud filtering and feature extraction occupy more than 30% of computing power (such as the JIMU intelligent solution), resulting in decision delays exceeding 100ms.

[0038] (4) Lightweight and insufficient energy efficiency

[0039] High power consumption of multi-line radars: Systems such as Ideal Lidar Pilot consume over 30W of power, requiring independent heat dissipation and making integration into small vehicles difficult.

[0040] Dependence on dedicated hardware: The Mobileye solution requires ASIC chip support, resulting in poor hardware compatibility and high upgrade costs. Summary of the Invention

[0041] In view of this, an embodiment of the present invention provides a method for controlling vehicle automatic emergency braking to address the technical problems of existing automatic emergency braking solutions, such as high cost, complex system, insufficient environmental adaptability and perception accuracy, and poor real-time performance. The method includes:

[0042] collecting first perception data of the vehicle's surrounding environment through a low-cost laser radar, collecting second perception data of the vehicle's surrounding environment through a low-resolution millimeter-wave radar, and performing spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar;

[0043] Dynamically determine a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target;

[0044] performing weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data;

[0045] A lightweight algorithm model is used to calculate relative driving state information between the vehicle and the target based on the multimodal data, and graded braking is triggered according to differences in the relative driving state information.

[0046] The present invention also provides a vehicle automatic emergency braking control device to address the technical problems of existing automatic emergency braking solutions, such as high cost, complex systems, insufficient environmental adaptability and perception accuracy, and poor real-time performance. The device includes:

[0047] a first data acquisition module, configured to collect first perception data of the vehicle's surrounding environment through a low-cost laser radar, collect second perception data of the vehicle's surrounding environment through a low-resolution millimeter-wave radar, and perform spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar;

[0048] a weight determination module, configured to dynamically determine a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target;

[0049] a data fusion module, configured to perform weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data;

[0050] A braking trigger module is used to calculate relative driving state information between the vehicle and the target using a lightweight algorithm model based on the multimodal data, and trigger graded braking according to differences in the relative driving state information.

[0051] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any of the above-mentioned control methods for automatic emergency braking of a vehicle is implemented to solve the technical problems of the existing automatic emergency braking solutions, such as high cost, complex system, insufficient environmental adaptability and perception accuracy, and poor real-time performance.

[0052] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program for executing any of the above-mentioned vehicle automatic emergency braking control methods, so as to solve the technical problems of the existing automatic emergency braking solutions, such as high cost, complex system, insufficient environmental adaptability and perception accuracy, and poor real-time performance.

[0053] Compared with the existing technology, the beneficial effects achieved by at least one of the above technical solutions adopted in the embodiments of this specification include at least the following: a combination of low-cost laser radar + low-resolution millimeter-wave radar is proposed. The low-cost laser radar can achieve low cost and lightweight. The low-resolution millimeter-wave radar has strong anti-interference characteristics, so that the combination can reduce hardware costs and system complexity under the premise of ensuring basic perception capabilities, which is conducive to lightweight design and solves the problems of complexity and high cost of existing multi-sensor fusion AEB systems; at the same time, the first weight of the low-cost laser radar and the second weight of the low-resolution millimeter-wave radar are dynamically determined according to different conditions of the target to achieve the first sense. The known data and the second perception data are weightedly fused based on dynamic weights to achieve full complementarity of multimodal data, so that the weather sensitivity of the lidar can be compensated by utilizing the all-weather characteristics of the low-resolution millimeter-wave radar. At the same time, the accuracy of the millimeter-wave radar in close-range target recognition is improved with the help of the high-precision point cloud data of the lidar, which is conducive to improving environmental adaptability and perception accuracy, and solving the problem of insufficient perception reliability in severe weather and complex environments. In addition, based on the multimodal data, a lightweight algorithm model is used to calculate the relative driving state information between the vehicle and the target to trigger graded braking, which is conducive to shortening the response time, achieving rapid response, and improving the real-time performance of the AEB system. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0055] Figure 1 This is a flow chart of a method for controlling automatic emergency braking of a vehicle provided by an embodiment of the present invention;

[0056] Figure 2 This is a principle framework diagram of a control method for implementing vehicle automatic emergency braking provided by an embodiment of the present invention;

[0057] Figure 3 is a schematic diagram of a lightweight algorithm model provided by an embodiment of the present invention;

[0058] Figure 4 This is a structural block diagram of a computer device provided by an embodiment of the present invention;

[0059] Figure 5 This is a structural block diagram of a vehicle automatic emergency braking control device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0061] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the features in the following embodiments and embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative work are within the scope of protection of this application.

[0062] In an embodiment of the present invention, a method for controlling automatic emergency braking of a vehicle is provided. Figure 1 As shown, the method includes:

[0063] S101: collecting first perception data of the vehicle's surrounding environment using a low-cost laser radar, collecting second perception data of the vehicle's surrounding environment using a low-resolution millimeter-wave radar, and performing spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar;

[0064] S102: Dynamically determining a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target;

[0065] S103: Performing weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data;

[0066] S104: Calculating relative driving state information between the vehicle and the target using a lightweight algorithm model based on the multimodal data, and triggering graded braking according to differences in the relative driving state information.

[0067] Depend on Figure 1 As can be seen from the illustrated process, in an embodiment of the present invention, a combination of a low-cost lidar and a low-resolution millimeter-wave radar is proposed. The low-cost lidar can be low-cost and lightweight, and the low-resolution millimeter-wave radar has strong anti-interference characteristics. This combination can reduce hardware cost and system complexity while ensuring basic perception capabilities, which is conducive to lightweight design and solves the problem of high complexity and cost of existing multi-sensor fusion AEB systems. At the same time, by dynamically determining the first weight of the low-cost lidar and the second weight of the low-resolution millimeter-wave radar according to different target conditions, the first perception data and the second perception data are weightedly fused based on dynamic weights, achieving full complementarity of multimodal data. This allows the all-weather characteristics of the low-resolution millimeter-wave radar to compensate for the weather sensitivity of the lidar. At the same time, the accuracy of the millimeter-wave radar in close-range target recognition is improved by leveraging the high-precision point cloud data of the lidar, thereby facilitating improved environmental adaptability and perception accuracy, and solving the problem of insufficient perception reliability in severe weather and complex environments. In addition, a lightweight algorithm model is used to calculate the relative driving state information between the vehicle and the target based on the multimodal data to trigger graded braking, which is conducive to shortening response time, achieving rapid response, and improving the real-time performance of the AEB system.

[0068] In specific implementation, by combining a single-line laser radar (whose cost is less than $500) with a low-resolution millimeter-wave radar (24GHz / 79GHz, whose cost is less than $100), the hardware cost can be reduced by more than 60%. At the same time, the calibration process is simplified, and the first perception data and the second perception data are synchronized in time and space, which can control the multi-sensor calibration error within ±0.2°, which helps to reduce the complexity of deployment. The combination of a single-line laser radar (volume <0.5L) and a low-resolution millimeter-wave radar (power consumption <5W) has a total system power consumption of <15W, which is suitable for compact vehicles.

[0069] In specific implementations, a single-line lidar can use a mechanically rotating single-line lidar (such as the Hesai PandarXT, costing less than $500). This system acquires a two-dimensional point cloud (maximum detection range 150m) through horizontal scanning (e.g., 120° horizontal field of view, 3° vertical field of view, and 0.1° angular resolution). Vertically, only 3° coverage simplifies data processing. Single-line lidars are less than 0.5L in size and consume less than 8W, making them suitable for installation in compact vehicles. Low-resolution millimeter-wave radars can use 24GHz radars (such as the Continental ARS540, costing less than $80), with a detection range of 80m and a resolution of 5°. This radar can also improve velocity measurement accuracy (±0.1m / s) through Doppler shift compensation. Millimeter-wave radars can also be upgraded to 79GHz radars, cost-effectively improving rain and fog penetration.

[0070] In specific implementation, a combination of solid-state MEMS lidar (such as InnovizOne, which has a similar cost to a single-line lidar) and a low-resolution millimeter-wave radar (24GHz / 79GHz, whose cost is less than $100) can be used to replace multi-line lidar (such as a 128-line radar for $2000+) and high-resolution radar, thereby reducing hardware costs and deployment complexity.

[0071] When implementing it specifically, Figure 2 As shown, taking a single-line laser radar as an example, after the low-cost laser radar collects the first perception data (such as two-dimensional point cloud data), the first perception data is preprocessed. After the low-resolution millimeter-wave radar collects the second perception data, the second perception data (such as the velocity matrix) is preprocessed. In order to further improve the response speed, parallel preprocessing acceleration is proposed. For example, the first perception data is subjected to distance-based adaptive voxel filtering to achieve point cloud filtering optimization. The voxel size of the close-range is 0.1m and the voxel size of the long-range is 0.5m, which can reduce the amount of calculation by 60%; at the same time, Doppler compensation and static target filtering are performed in parallel on the second perception data to achieve millimeter-wave radar data correction, which can be accelerated by GPU and take less than 10ms.

[0072] In specific implementation, during data preprocessing and the spatiotemporal synchronization of the preprocessed first and second sensory data, in order to further simplify the calibration process, it is proposed that a synchronization pulse signal can be generated through an FPGA to align the timestamps of the lidar and millimeter-wave radar data (i.e., the first and second sensory data) (error <1ms). Software interpolation can also be used to achieve the timestamp alignment of the first and second sensory data. The spatial alignment of the first and second sensory data can be achieved by converting the radar polar coordinate data into the lidar coordinate system based on an extrinsic calibration matrix (calibration error <±0.2°).

[0073] When implementing it specifically, Figure 2 As shown, after preprocessing the first and second perception data, the first and second perception data can be fused. To improve data accuracy and environmental adaptability, it is proposed to perform weighted fusion of the first and second perception data based on dynamic weights. The first weight of the low-cost laser radar and the second weight of the low-resolution millimeter-wave radar are dynamically determined according to different target conditions, including:

[0074] If the target is a vehicle, different first weights and second weights are determined according to different distances between the vehicle and the target;

[0075] If the target is a target of another type other than a vehicle, the first weight is increased and the first weight is greater than the second weight.

[0076] Specifically, if the target is a vehicle, different first weights and second weights are determined according to different distances between the vehicle and the target, and the sum of the first weight and the second weight is 100%, including:

[0077] If the distance between the ego vehicle and the target is less than a first distance threshold (e.g., 30 meters), the first weight is determined to be greater than or equal to 70%, and the second weight is determined to be less than 30%. In this case, the high-precision point cloud (i.e., the first perception data) is relied upon, and the second perception data of the radar is only compensated for the speed information.

[0078] If the distance between the vehicle and the target is greater than or equal to the first distance threshold and less than or equal to a second distance threshold (e.g., 80 m), determining that the first weight and the second weight are close in value and the second weight is greater than the first weight (e.g., the first weight is 40% and the second weight is 60%), wherein the second distance threshold is greater than the first distance threshold;

[0079] If the distance between the vehicle and the target is greater than the second distance threshold, the second weight is determined to be greater than or equal to 90%, and if the second weight is determined to be less than 10%, the laser radar is only used for target verification.

[0080] If the target is a target other than a vehicle, such as a pedestrian, an obstacle, a cyclist, etc., the first weight is increased and the first weight is greater than the second weight. For example, the first weight may be 85%.

[0081] In a specific implementation, in order to reduce data fusion delay, improve response speed, and increase system lightweight during the data fusion process, it is proposed to use a lightweight algorithm model based on the multimodal data to calculate the relative driving state information between the ego vehicle and the target, including:

[0082] The lightweight neural network model is subjected to model compression processing to obtain a compressed neural network model, the multimodal data is input into the compressed neural network model, and the relative driving state information is calculated by the compressed neural network model, wherein the lightweight neural network model is a backbone network or YOLO-Fastest.

[0083] When implementing it specifically, Figure 3 As shown, the lightweight neural network model takes the backbone network (such as MobileNetV3) as an example. YOLO-Fastest can also be used to compress the lightweight neural network model through model compression technology (channel pruning, parameter reduction by 40%, at the same time, 8-bit quantization, model size <5MB) to reduce model parameters and obtain a compressed neural network model. Then, multimodal data is input into the compressed neural network model for reasoning and calculation to obtain relative driving state information between the vehicle and the target, thereby improving the reasoning speed. For example, the above-mentioned compressed neural network model can achieve an inference speed of 25FPS (delay <40ms) running on the JetsonNano platform, which can compress the total system response time from 150ms of the traditional solution to 90ms; at the same time, resource usage is optimized, the computing load of the fusion algorithm is reduced by 50%, and it can be run on low-cost embedded platforms (such as Jetson Nano), is compatible with general hardware, and the algorithm supports mainstream vehicle computing platforms (such as TITDA4) without the need for dedicated ASIC chips.

[0084] When implementing it specifically, Figure 2 As shown, after obtaining relative driving state information through a multimodal data fusion process, a braking strategy is executed. To improve the accuracy and smoothness of brake triggering, a graded braking triggering method based on the relative driving state information is proposed. For example, if the combined confidence level of the low-cost lidar and the low-resolution millimeter-wave radar is greater than or equal to a preset confidence threshold, the corresponding braking mechanism is triggered according to the relative driving state information belonging to different risk levels.

[0085] If the combined confidence of the low-cost lidar and the low-resolution millimeter-wave radar is less than a preset confidence threshold (e.g., a combined confidence of 60-85%), the target is verified to exist in a preset number of consecutive frames based on auxiliary data. If so, the corresponding braking mechanism is triggered based on the different risk levels of the relative driving status information, wherein the auxiliary data includes the historical predicted trajectory of the vehicle (e.g., Kalman filter) or the camera data of the vehicle.

[0086] In specific implementation, the relative driving state information can be TTC (Time to Collision between the host vehicle and the target) or PDR (Remaining Distance between the host vehicle and the target). The risk level can be divided according to the numerical values of TTC or PDR respectively (e.g., high risk, medium risk, low risk), and then different braking mechanisms can be triggered based on different risk levels. For example, taking TTC as an example,

[0087] When the comprehensive confidence level of the sensor (lidar + millimeter wave radar)>85% (i.e., the preset confidence threshold), the first TTC threshold is 1.5s, and the second TTC threshold is 3s. Different braking mechanisms are triggered based on different risk levels as follows:

[0088] High risk (TTC < 1.5s): Trigger full braking (deceleration > 6m / s

[0089] );

[0089] Medium risk (1.5s < TTC < 3s): Trigger pre-braking (deceleration is 2 - 4m / s 2 ) and issue an audible and visual alarm;

[0090] Low risk (TTC > 3s): Only give a warning without braking.

[0091] In specific implementation, the above comprehensive confidence level can be the weighted sum or the mean of the confidence levels of a low-cost lidar and a low-resolution millimeter wave radar.

[0092] ​​​​​​​​​​​​​​​​​By using single-line lidar (such as Hesai PandarXT, cost <500) and low-resolution millimeter-wave radar (such as 24GHz radar <500) and low-resolution millimeter-wave radar (such as 24GHz radar <100), hardware costs can be reduced by more than 60%;

[0098] The volume of a single-line laser radar is <0.5L, the power consumption is <8W, the power consumption of the millimeter-wave radar is <5W, and the total system power consumption is <15W, making it suitable for compact vehicles.

[0099] Technical support: By simplifying sensor configuration and optimizing space occupancy, we achieve breakthroughs in both low cost and lightweight.

[0100] 2. Improved environmental adaptability and perception accuracy

[0101] Compared with existing technologies: Pure vision solutions (Tesla's early Autopilot) fail in backlight / rain and fog, and low-resolution millimeter-wave radar (24GHz) has a ranging error of ±0.5m.

[0102] Advantages of the present invention:

[0103] Multimodal complementarity: millimeter-wave radar compensates for the failure of lidar in rain and fog, and lidar corrects the close-range error of millimeter-wave radar (ranging accuracy within 30m ±0.1m);

[0104] Dynamic weight allocation: Adjust sensor weights based on target distance (e.g., the weight of millimeter-wave radar is increased to 80% in rainy and foggy weather) to improve robustness in complex scenarios.

[0105] Technical support: The fusion strategy breaks through the limitations of a single sensor and achieves all-weather high-precision perception.

[0106] 3. Optimization of response speed and computing efficiency

[0107] Compared with existing technologies: Traditional fusion algorithms (such as Mobileye EyeQ5) have a response delay of 120-150ms and pre-processing computing power occupies >30%.

[0108] Advantages of the present invention:

[0109] Fast response algorithm: lightweight neural network (MobileNetV3 compressed to <5MB) and parallel processing (GPU-accelerated point cloud filtering), system response time ≤ 90ms;

[0110] Decreased computing power requirements: Peak computing power <2TOPS, can run on low-cost embedded platforms such as Jetson Nano.

[0111] Technical support: Algorithm optimization breaks through the real-time bottleneck and meets the millisecond-level decision-making requirements for emergency braking.

[0112] 4. Dynamic decision-making mechanism and reduced false trigger rate

[0113] Compared with existing technologies: Bosch's AEB system has a false trigger rate of about 5%, and cannot trigger braking in high-speed scenarios (>85km / h).

[0114] Advantages of the present invention:

[0115] Graded TTC thresholds: high risk (TTC < 1.5s) means full braking, medium risk means pre-braking, and low risk means only warning;

[0116] Confidence fusion decision: Direct triggering when dual sensor confidence is >85%, secondary verification combined with trajectory prediction when confidence is 60-85%, and false triggering rate <1%.

[0117] Technical support: Multi-dimensional verification mechanism improves braking safety and passenger comfort.

[0118] 5. Technical compatibility and scalability

[0119] Compared with existing technologies: Bosch's solution relies on dedicated ASIC chips and has poor hardware compatibility.

[0120] Advantages of the present invention:

[0121] Supports mainstream in-vehicle computing platforms (such as TI TDA4) without the need for dedicated chips;

[0122] It can be expanded to include an integrated camera as a third verification source, or upgraded to a 79GHz radar for enhanced rain and fog penetration capabilities.

[0123] Technical support: The open architecture design reserves space for functional upgrades.

[0124] Table 1

[0125]

[0126] In specific implementation, the above-mentioned vehicle automatic emergency braking control method can also collect data through the following hardware combination to meet the braking requirements of different application scenarios, for example,

[0127] In high-speed scenarios (such as commercial vehicles or mid-range passenger vehicles suitable for high-speed scenarios (>100 km / h)), the first perception data of the vehicle's surrounding environment can be collected, and the second perception data of the vehicle's surrounding environment can be collected using a low-resolution camera (2 million pixels). The low-resolution camera performs target classification, and the Doppler compensation algorithm is used to improve the speed measurement accuracy (±0.05 m / s); or

[0128] In scenarios where the roads are dark or animals are frequently seen, the first perception data can be collected by a far-infrared thermal imaging camera, and the second perception data of the vehicle's surrounding environment can be collected by a low-resolution millimeter-wave radar.

[0129] Specifically, the Continental ARS540 high-resolution 4D imaging radar can be used. It boasts a vertical resolution of 2°, a detection range of 300m, and can generate point cloud-level data. The unit price of a high-resolution 4D imaging radar is less than 200 yuan, a 60% reduction in cost compared to lidar (less than 500 yuan per line). Millimeter-wave radar is unaffected by rain and fog, and can replace lidar in inclement weather, ensuring environmental adaptability.

[0130] Specifically, a far-infrared (LWIR) thermal imaging camera (such as Teledyne FLIR), with a detection range of 150m, identifies biological targets (pedestrians and animals) by temperature differences. Combined with a 24GHz low-resolution millimeter-wave radar, the two data are fused using a feature-level strategy (thermal imaging provides target classification, and radar provides motion parameters). This combination increases pedestrian detection success rate to 92% in 0.2 lux illumination (compared to 65% for traditional cameras), improving nighttime performance. Furthermore, it improves false trigger control, as thermal imaging can distinguish between living and non-living objects, reducing false positives such as metal signs.

[0131] In this embodiment, a computer device is provided, such as Figure 4 As shown, it includes a memory 401, a processor 402 and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any of the above-mentioned vehicle automatic emergency braking control methods is implemented.

[0132] Specifically, the computer device may be a computer terminal, a server or a similar computing device.

[0133] In this embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program for executing any of the above-mentioned vehicle automatic emergency braking control methods.

[0134] Specifically, computer-readable storage media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable storage media does not include transitory media such as modulated data signals and carrier waves.

[0135] Based on the same inventive concept, an embodiment of the present invention further provides a control device for automatic emergency braking of a vehicle, as described in the following embodiments. Since the principle of solving the problem by the control device for automatic emergency braking of a vehicle is similar to that of the control method for automatic emergency braking of a vehicle, the implementation of the control device for automatic emergency braking of a vehicle can refer to the implementation of the control method for automatic emergency braking of a vehicle, and the repeated parts will not be repeated. As used below, the terms "unit" or "module" can be a combination of software and / or hardware that implements predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceivable.

[0136] Figure 5 FIG. 1 is a structural block diagram of a control device for automatic emergency braking of a vehicle according to an embodiment of the present invention. Figure 5 As shown, including:

[0137] A first data acquisition module 501 is configured to collect first perception data of the vehicle's surrounding environment using a low-cost laser radar, collect second perception data of the vehicle's surrounding environment using a low-resolution millimeter-wave radar, and perform spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar;

[0138] A weight determination module 502 is configured to dynamically determine a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target;

[0139] a data fusion module 503, configured to perform weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data;

[0140] The braking triggering module 504 is configured to calculate relative driving state information between the vehicle and the target using a lightweight algorithm model based on the multimodal data, and trigger graded braking according to differences in the relative driving state information.

[0141] In one embodiment, a weight determination module is used to determine different first weights and second weights based on different distances between the vehicle and the target if the target is a vehicle; and to increase the first weight and make the first weight greater than the second weight if the target is another type of target other than a vehicle.

[0142] In one embodiment, a weight determination module is configured to determine that the first weight is greater than or equal to 70% and the second weight is less than 30% if the distance between the vehicle and the target is less than a first distance threshold, wherein the sum of the first weight and the second weight is 100%; if the distance between the vehicle and the target is greater than or equal to the first distance threshold and less than or equal to a second distance threshold, determine that the values ​​of the first weight and the second weight are close and the second weight is greater than the first weight, wherein the second distance threshold is greater than the first distance threshold; if the distance between the vehicle and the target is greater than the second distance threshold, determine that the second weight is greater than or equal to 90% and the second weight is less than 10%.

[0143] In one embodiment, a braking trigger module is used to perform model compression processing on a lightweight neural network model to obtain a compressed neural network model, input the multimodal data into the compressed neural network model, and calculate the relative driving state information through the compressed neural network model, wherein the lightweight neural network model is a backbone network or YOLO-Fastest.

[0144] In one embodiment, the apparatus further comprises:

[0145] a preprocessing module for performing distance-based adaptive voxel filtering on the first perception data before weighted fusion of the first perception data and the second perception data according to the first weight and the second weight, and performing Doppler compensation and static target filtering on the second perception data in parallel.

[0146] In one embodiment, a braking trigger module is configured to trigger a corresponding braking mechanism according to whether the relative driving status information belongs to different risk levels if the combined confidence of the low-cost lidar and the low-resolution millimeter-wave radar is greater than or equal to a preset confidence threshold; and to verify whether the target exists in a preset number of consecutive frames based on auxiliary data if the combined confidence of the low-cost lidar and the low-resolution millimeter-wave radar is less than the preset confidence threshold, and if so, trigger a corresponding braking mechanism according to whether the relative driving status information belongs to different risk levels, wherein the auxiliary data includes the historical predicted trajectory of the ego vehicle or the camera data of the ego vehicle.

[0147] In one embodiment, the apparatus further comprises:

[0148] The second data acquisition module is used to collect the first perception data of the environment around the vehicle through a high-resolution 4D imaging radar in high-speed scenarios, and to collect the second perception data of the environment around the vehicle through a low-resolution camera; or; to collect the first perception data through a far-infrared thermal imaging camera in scenarios where roads are at night or animals are frequently present.

[0149] The embodiments of the present invention achieve the following technical effects: a combination of a low-cost laser radar and a low-resolution millimeter-wave radar is proposed. The low-cost laser radar can be low-cost and lightweight, and the low-resolution millimeter-wave radar has strong anti-interference characteristics. This combination can reduce hardware costs and system complexity while ensuring basic perception capabilities, facilitates lightweight design, and solves the problems of high complexity and cost in existing multi-sensor fusion AEB systems. At the same time, by dynamically determining a first weight for the low-cost laser radar and a second weight for the low-resolution millimeter-wave radar according to different target conditions, first and second perception data are weightedly fused based on dynamic weights, achieving full complementarity of multimodal data. This allows the all-weather characteristics of the low-resolution millimeter-wave radar to compensate for the weather sensitivity of the laser radar, while leveraging the high-precision point cloud data of the laser radar to improve the accuracy of the millimeter-wave radar in close-range target recognition, thereby facilitating improved environmental adaptability and perception accuracy, and solving the problem of insufficient perception reliability in severe weather and complex environments. Furthermore, a lightweight algorithm model is used to calculate the relative driving state information between the vehicle and the target based on the multimodal data to trigger graded braking, which helps shorten response time, achieve rapid response, and improve the real-time performance of the AEB system.

[0150] Obviously, those skilled in the art should understand that the various modules or steps of the above-mentioned embodiments of the present invention can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices. Alternatively, they can be implemented using program code executable by the computing device, so that they can be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than herein, or they can be made into separate integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for implementation. Thus, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0151] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for controlling automatic emergency braking of a vehicle, characterized in that: include: collecting first perception data of the vehicle's surrounding environment through a low-cost laser radar, collecting second perception data of the vehicle's surrounding environment through a low-resolution millimeter-wave radar, and performing spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar; Dynamically determine a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target; performing weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data; A lightweight algorithm model is used to calculate relative driving state information between the vehicle and the target based on the multimodal data, and graded braking is triggered according to differences in the relative driving state information.

2. The method for controlling vehicle automatic emergency braking according to claim 1, wherein: Dynamically determining a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target, including: If the target is a vehicle, different first weights and second weights are determined according to different distances between the vehicle and the target; If the target is a target of another type other than a vehicle, the first weight is increased and the first weight is greater than the second weight.

3. The method for controlling vehicle automatic emergency braking according to claim 2, wherein: If the target is a vehicle, determining different first weights and second weights according to different distances between the vehicle and the target includes: If the distance between the vehicle and the target is less than a first distance threshold, determining that the first weight is greater than or equal to 70%, and determining that the second weight is less than 30%, wherein the sum of the first weight and the second weight is 100%; If the distance between the vehicle and the target is greater than or equal to the first distance threshold and less than or equal to a second distance threshold, determining that the first weight is close to the second weight and the second weight is greater than the first weight, wherein the second distance threshold is greater than the first distance threshold; If the distance between the vehicle and the target is greater than the second distance threshold, the second weight is determined to be greater than or equal to 90%, and the second weight is determined to be less than 10%.

4. The method for controlling automatic emergency braking of a vehicle according to claim 1, wherein: Calculating relative driving state information between the vehicle and the target using a lightweight algorithm model based on the multimodal data includes: The lightweight neural network model is subjected to model compression processing to obtain a compressed neural network model, the multimodal data is input into the compressed neural network model, and the relative driving state information is calculated by the compressed neural network model, wherein the lightweight neural network model is a backbone network or YOLO-Fastest.

5. The method for controlling automatic emergency braking of a vehicle according to any one of claims 1 to 4, wherein: Also includes: Before weighted fusion of the first perception data and the second perception data is performed according to the first weight and the second weight, distance-based adaptive voxel filtering is performed on the first perception data, and Doppler compensation and static target filtering are performed on the second perception data in parallel.

6. The method for controlling automatic emergency braking of a vehicle according to any one of claims 1 to 4, characterized in that: Triggering graded braking according to the relative driving state information includes: If the combined confidence level of the low-cost laser radar and the low-resolution millimeter-wave radar is greater than or equal to a preset confidence threshold, triggering a corresponding braking mechanism based on the risk level of the relative driving state information; If the combined confidence of the low-cost lidar and the low-resolution millimeter-wave radar is less than a preset confidence threshold, the target is verified to be present in a preset number of consecutive frames based on auxiliary data. If so, a corresponding braking mechanism is triggered based on whether the relative driving status information belongs to different risk levels, wherein the auxiliary data includes the historical predicted trajectory of the ego vehicle or the camera data of the ego vehicle.

7. The method for controlling automatic emergency braking of a vehicle according to any one of claims 1 to 4, characterized in that: Also includes: In a high-speed scenario, the first perception data of the vehicle's surrounding environment is collected by a high-resolution 4D imaging radar, and the second perception data of the vehicle's surrounding environment is collected by a low-resolution camera; or In nighttime roads or scenes where animals frequently appear, the first perception data is collected by a far-infrared thermal imaging camera.

8. A vehicle automatic emergency braking control device, characterized in that: include: a first data acquisition module, configured to collect first perception data of the vehicle's surrounding environment through a low-cost laser radar, collect second perception data of the vehicle's surrounding environment through a low-resolution millimeter-wave radar, and perform spatiotemporal synchronization on the first perception data and the second perception data, wherein the low-cost laser radar includes a single-line laser radar or a solid-state MEMS laser radar; a weight determination module, configured to dynamically determine a first weight of the low-cost laser radar and a second weight of the low-resolution millimeter-wave radar according to different conditions of the target; a data fusion module, configured to perform weighted fusion on the first perception data and the second perception data according to the first weight and the second weight to obtain multimodal data; A braking trigger module is used to calculate relative driving state information between the vehicle and the target using a lightweight algorithm model based on the multimodal data, and trigger graded braking according to differences in the relative driving state information.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the vehicle automatic emergency braking control method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program for executing the vehicle automatic emergency braking control method according to any one of claims 1 to 7 .

Images