Vehicle collision alarm method, device and equipment and computer program product
By integrating IMU and CAN bus data to construct a multi-dimensional collision feature system, the problems of insufficient rear perception, environmental interference, and unknown collision degree in existing vehicle collision alarm systems are solved. This achieves highly accurate collision detection and differentiated alarms, thereby improving the active safety performance of vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-12
- Publication Date
- 2026-03-13
AI Technical Summary
Existing vehicle collision warning systems suffer from insufficient rear and lateral perception capabilities, high false alarm rates due to environmental interference, and unknown collision severity, making it impossible to fully assess collision risks and severity.
By integrating IMU data and CAN bus data, a multi-dimensional collision feature system is constructed. A preset collision level classification model is used to determine the collision level, and differentiated alarm strategies are executed according to the level, including alarm strategies for minor, moderate and severe collisions.
It significantly improves the comprehensiveness and accuracy of collision detection, reduces excessive interference from minor collisions, ensures the efficiency of emergency rescue in severe collisions, and enhances the vehicle's active safety performance and user experience.
Smart Images
Figure CN121661864A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of vehicle safety technology, and in particular to a vehicle collision alarm method, device and equipment, and computer program product. Background Technology
[0002] With the rapid development of the automotive industry, active safety technology has become a core element in ensuring driving safety. Existing vehicle collision warning systems mainly rely on visual sensors (such as cameras) and millimeter-wave radar to detect obstacles ahead. Among them, visual sensors analyze the distance, speed, and trajectory of vehicles, pedestrians, or obstacles ahead through image recognition technology; millimeter-wave radar uses the principle of electromagnetic wave reflection to monitor the relative speed and distance of target objects in real time, and triggers alarms or automatic emergency braking (AEB) systems when there is a potential collision risk.
[0003] Although existing technologies have improved driving safety to some extent, their core shortcomings significantly limit their effectiveness, specifically in the following aspects:
[0004] (1) Blind Spot Issue: Insufficient rear and side collision detection capabilities. Existing systems primarily rely on front collision warnings, exhibiting significant blind spots in their ability to detect collisions behind and to the sides of the vehicle. For example:
[0005] 1) Rear-end collision scenario: When a rear-end collision occurs due to the following vehicle's failure to brake in time or distraction, the front vision sensors and radar cannot monitor the dynamics of the following vehicle, causing the system to fail to issue an early warning; 2) Side collision scenario: When changing lanes or turning at intersections, vehicles on the side may cause a collision due to obstructed vision or excessive speed, but the existing system lacks lateral perception capabilities and cannot identify risks in time.
[0006] (2) High false alarm rate: Environmental interference leads to a decrease in system reliability. Visual sensors and millimeter-wave radar are susceptible to external environmental influences, resulting in false alarms or missed alarms. For example:
[0007] Visual sensor defects: 1) Light interference: Under strong light (such as backlight, high beams at night) or low light conditions (such as tunnels, no streetlights at night), the camera's imaging quality deteriorates and it cannot accurately identify the target; 2) Rain and fog interference: In rainy, snowy, or foggy weather, the lens may be blurred or reflective, causing the target to be lost. The system may misjudge as "no obstacle" and miss the alarm, or the alarm may be falsely triggered due to raindrop / fog reflection.
[0008] Millimeter-wave radar defects: 1) Metal interference: Radar waves generate multipath reflections on metal objects (such as guardrails and billboards), resulting in distance measurement errors; 2) Velocity confusion: The reflected signals of stationary objects (such as roadside trees) and slow-moving objects (such as pedestrians) may be misjudged as high-speed approaching targets, causing false alarms.
[0009] (3) Problem of unknown collision severity: A single data source cannot fully assess the severity. Existing systems only estimate collision risk through visual or radar data, lacking a quantitative assessment of the severity of the collision. Summary of the Invention
[0010] This application provides a vehicle collision alarm method, device, equipment, and computer program product to improve the accuracy and reliability of vehicle collision alarms.
[0011] The embodiments of this application adopt the following technical solutions:
[0012] In a first aspect, embodiments of this application provide a vehicle collision alarm method, the vehicle collision alarm method comprising:
[0013] Acquire vehicle IMU data and CAN bus data;
[0014] Based on the IMU data and the CAN bus data, multi-dimensional collision features of the vehicle are extracted using a multi-dimensional collision feature extraction strategy.
[0015] Once the multi-dimensional collision features of the vehicle are extracted, the collision level of the vehicle is determined using a preset collision level classification model based on the multi-dimensional collision features of the vehicle.
[0016] The alarm strategy corresponding to the vehicle is determined based on the collision level of the vehicle, and the alarm strategy is executed.
[0017] Optionally, acquiring the vehicle's IMU data and CAN bus data includes:
[0018] Determine the current driving scenario of the vehicle;
[0019] When the preset scenario conditions are triggered in the current driving scenario, the vehicle's IMU data and CAN bus data are acquired.
[0020] Optionally, the multi-dimensional collision features include at least two of time-domain features, frequency-domain features, kinematic features, and energy features. The step of extracting the multi-dimensional collision features of the vehicle using a multi-dimensional collision feature extraction strategy based on the IMU data and the CAN bus data includes:
[0021] The time-domain features and the frequency-domain features are determined based on the IMU data;
[0022] The kinematic characteristics and energy characteristics are determined based on the IMU data and the CAN bus data.
[0023] Optionally, the IMU data includes acceleration data and angular velocity data, the CAN bus data includes vehicle speed data and steering angle data, the time-domain features include at least one of acceleration peak value and the duration of acceleration peak value, the frequency-domain features include the frequency components of the collision signal, the kinematic features include at least one of velocity change features, attitude angle change features, and trajectory change features, and the energy features include collision vibration energy.
[0024] Optionally, the collision level is divided into minor collision, moderate collision, and severe collision, and determining the corresponding alarm strategy for the vehicle based on the collision level includes:
[0025] If the collision level of the vehicle is minor collision, then the alarm strategy corresponding to the vehicle is determined to be the minor collision alarm strategy, which includes the vehicle generating a first collision warning.
[0026] If the collision level of the vehicle is medium collision, then the alarm strategy corresponding to the vehicle is determined to be a medium collision alarm strategy, which includes the vehicle generating a second collision alert and reporting the collision information to the cloud.
[0027] If the collision level of the vehicle is a severe collision, then the alarm strategy corresponding to the vehicle is determined to be a severe collision alarm strategy, which includes making an emergency call and simultaneously notifying emergency contacts.
[0028] Optionally, the preset collision level classification model is trained in the following manner;
[0029] Collect multi-dimensional collision feature data of vehicles;
[0030] Based on the multi-dimensional collision feature data, the multi-dimensional collision feature data is labeled with corresponding collision level tags using preset collision level classification conditions.
[0031] The collision level classification model is trained based on the multi-dimensional collision feature data and the corresponding collision level labels to obtain the preset collision level classification model.
[0032] Optionally, before acquiring the vehicle's IMU data and CAN bus data, the vehicle collision alarm method further includes:
[0033] Determine the vehicle's current driving status;
[0034] When the current driving state is stationary, acquire IMU data in the stationary state and establish a gravity direction reference.
[0035] Secondly, embodiments of this application also provide a vehicle collision alarm device, the vehicle collision alarm device comprising:
[0036] The acquisition unit is used to acquire vehicle IMU data and CAN bus data;
[0037] An extraction unit is used to extract multi-dimensional collision features of the vehicle based on the IMU data and the CAN bus data using a multi-dimensional collision feature extraction strategy.
[0038] The first determining unit is used to determine the collision level of the vehicle based on the multi-dimensional collision features of the vehicle and a preset collision level classification model, after extracting the multi-dimensional collision features of the vehicle.
[0039] An alarm unit is used to determine the alarm strategy corresponding to the vehicle based on the collision level of the vehicle, and to execute the alarm strategy.
[0040] Thirdly, embodiments of this application also provide an apparatus, comprising:
[0041] A processor; and a memory arranged to store computer-executable instructions, which, when executed, cause the processor to perform any of the aforementioned vehicle collision alarm methods.
[0042] Fourthly, embodiments of this application also provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement any of the aforementioned vehicle collision alarm methods.
[0043] The vehicle collision alarm method of this application embodiment achieves the following beneficial effects: First, it acquires the vehicle's IMU data and CAN bus data; then, based on the IMU data and CAN bus data, it extracts the vehicle's multi-dimensional collision features using a multi-dimensional collision feature extraction strategy; subsequently, based on the extracted multi-dimensional collision features, it determines the vehicle's collision level using a preset collision level classification model; finally, it determines the corresponding alarm strategy for the vehicle based on the collision level and executes the alarm strategy. The vehicle collision alarm method of this application embodiment, by fusing IMU and CAN bus data, constructs a multi-dimensional collision feature system covering time, space, and kinematics, breaking through the perception limitations of traditional solutions that rely solely on a single sensor, significantly improving the comprehensiveness and accuracy of collision detection. Furthermore, the differentiated alarm strategy links in-vehicle alarms, remote notifications, and vehicle control according to the collision level, reducing excessive interference from minor collisions while ensuring the efficiency of emergency rescue in severe collisions, thus improving overall vehicle active safety performance and user experience. Attached Figure Description
[0044] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:
[0045] Figure 1 This is a flowchart illustrating a vehicle collision alarm method according to an embodiment of this application;
[0046] Figure 2 This is a schematic diagram of the structure of a vehicle collision alarm device according to an embodiment of this application;
[0047] Figure 3 This is a schematic diagram of the structure of a device according to an embodiment of this application. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0049] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.
[0050] This application provides a vehicle collision alarm method, such as... Figure 1 The diagram shows a flowchart of a vehicle collision alarm method according to an embodiment of this application. The vehicle collision alarm method includes the following steps S110 to S140:
[0051] Step S110: Obtain the vehicle's IMU data and CAN bus data.
[0052] With its high sampling rate (>100Hz) and six-degree-of-freedom (6-DOF) motion sensing capabilities, the IMU (Integrated Measurement Unit) can capture real-time acceleration and angular velocity changes at the moment of collision, making it an ideal choice for improving collision detection accuracy. Therefore, this embodiment first acquires vehicle IMU data in real time through an onboard sensor network. This requires the prior deployment of a three-axis accelerometer and gyroscope, and then the vehicle's three-axis acceleration (X / Y / Z axes) and three-axis angular velocity (pitch, roll, yaw) before and after the collision are acquired. The sampling frequency is set to, for example, above 100Hz to capture high-frequency vibration signals such as the shock wave at the moment of collision. The raw data is filtered, such as by Kalman filtering, to eliminate noise interference, and the transformation relationship between the sensor coordinate system and the vehicle coordinate system is calibrated to ensure data consistency.
[0053] For CAN bus data acquisition, key parameters such as vehicle speed (calculated via wheel speed sensors), steering angle (steering wheel angle sensor), and braking pressure (brake master cylinder pressure sensor) can be read from the CAN bus. The sampling frequency is, for example, 50Hz, reflecting the vehicle's dynamic control status. A sliding window smoothing process is applied to the vehicle speed and steering angle data to avoid misreadings caused by short-term sudden changes such as road bumps, which could affect subsequent analysis.
[0054] By aligning IMU data and CAN bus data using timestamps, we can ensure that the two types of data are strictly synchronized in the time dimension, providing a foundation for multi-dimensional feature extraction.
[0055] Step S120: Based on the IMU data and the CAN bus data, extract the multi-dimensional collision features of the vehicle using a multi-dimensional collision feature extraction strategy.
[0056] Based on the synchronized data, collision features need to be extracted from multiple dimensions, such as temporal and kinematic features. Temporal features, such as the duration of acceleration exceeding a threshold, can distinguish between short-lived impacts like scrapes and sustained collisions like frontal impacts. Kinematic features, such as the rate of change of velocity, can quantify impact intensity. These features are combined into a multi-dimensional vector, which serves as input to the subsequent classification model.
[0057] Step S130: After extracting the multi-dimensional collision features of the vehicle, determine the collision level of the vehicle using a preset collision level classification model based on the multi-dimensional collision features of the vehicle.
[0058] The training of the pre-defined collision level classification model can be carried out using machine learning models such as XGBoost. Historical collision data (including real accident cases and simulated test data) is used to label the collision level (such as light / medium / heavy). The input features include the multi-dimensional features extracted in step S120, and the output is a discrete level label.
[0059] The feature vectors extracted in real time are input into the trained model, which outputs a collision level probability distribution, such as 80% for mild, 15% for moderate, and 5% for severe, and the highest probability is taken as the final level.
[0060] Furthermore, the classification threshold can be dynamically adjusted based on vehicle speed and driving scenario, such as high speed / low speed, straight driving / turning. For example, at high speeds (e.g., greater than 80 km / h), the energy threshold for severe collisions is reduced to increase sensitivity to high-speed impacts; when turning, the weight of side collisions is increased to optimize the recognition accuracy of lane change scenarios.
[0061] Step S140: Determine the alarm strategy corresponding to the vehicle based on the collision level of the vehicle, and execute the alarm strategy.
[0062] Based on the collision level determined by the aforementioned steps, differentiated alarms and responses are triggered, that is, different alarm strategies are set for different collision levels to ensure the effectiveness and timeliness of alarms.
[0063] The vehicle collision alarm method in this application integrates IMU and CAN bus data to construct a multi-dimensional collision feature system covering time, space, and kinematics. This overcomes the limitations of traditional solutions that rely on a single sensor, significantly improving the comprehensiveness and accuracy of collision detection. Furthermore, the differentiated alarm strategy links in-vehicle alarms, remote notifications, and vehicle control based on the collision level, reducing excessive interference in minor collisions while ensuring efficient emergency rescue in severe collisions, thus improving overall vehicle active safety performance and user experience.
[0064] In some embodiments of this application, acquiring the vehicle's IMU data and CAN bus data includes: determining the current driving scenario of the vehicle; and acquiring the vehicle's IMU data and CAN bus data when the current driving scenario triggers preset scenario conditions.
[0065] This application embodiment can determine in real time whether the vehicle's current scene meets the preset "visual / radar failure conditions" by combining multimodal environment perception with a rule engine. The specific process is as follows:
[0066] On one hand, the vision sensor captures real-time images through a forward-facing camera and uses image processing algorithms (such as OpenCV) to calculate image sharpness (e.g., edge detection gradient values) and visibility (e.g., contrast threshold). If the sharpness is below 30% or the visibility is below 50 meters, it is marked as a "visually limited scene." On the other hand, the radar sensor collects target reflection intensity through millimeter-wave radar. If multipath interference caused by rain or fog is detected (e.g., an abnormal increase of more than 30% in target point cloud density) or the effective detection range is shortened to less than 50% of the original range, it is marked as a "radar-limited scene." In addition, real-time weather information (such as rain sensor data and visibility level) can be obtained through onboard weather stations or network APIs. If the weather is "heavy rain," "foggy," or "snowy," it is directly marked as a "severe weather scene."
[0067] If the above three types of data are input into the rule engine, and any of the following conditions are met, the current scenario is determined to be a "visual / radar failure scenario":
[0068] 1) Visually limited scenarios and radar-limited scenarios;
[0069] 2) Scenarios with severe weather and limited vision (such as rain causing the camera to be blurry);
[0070] 3) The radar detects strong interference (such as water spray from a vehicle ahead) and the visual system cannot identify the outline of the obstacle.
[0071] If the above conditions are not met, it is determined to be a "normal scenario" and the original vision / radar solution will continue to be used.
[0072] To avoid misjudgments, the scene status can be reassessed periodically, such as every 10 seconds. For example, when a vehicle enters a tunnel from clear weather, the camera may experience a brief period of darkness due to the sudden change in light, initially marking it as "visually limited." However, if the radar can still reliably detect vehicles ahead, the rule engine will exclude the misjudgment and maintain the normal approach. Similarly, when a vehicle is driving in heavy rain, if the radar generates dense false targets due to water droplet reflections, but the vision system maintains an image clarity greater than a certain threshold (e.g., 40%) through the windshield wipers, then visual data will be prioritized, avoiding switching to the IMU / CAN solution.
[0073] Upon determining a "visual / radar failure scenario," the system automatically switches to IMU and CAN bus data acquisition mode. In this mode, the sampling frequency of the accelerometer and gyroscope can be increased from the conventional 50Hz to 200Hz to capture subtle tire slippage or vehicle sway signals (such as sudden changes in lateral acceleration) caused by slippery road surfaces in rainy or foggy weather. Furthermore, a bandpass filter can be used to retain low-frequency impact signals related to collisions, addressing the high-frequency vibration noise generated by raindrops hitting the vehicle body. For CAN bus data, in addition to the usual vehicle speed and steering angle, additional data such as brake master cylinder pressure (reflecting the driver's emergency braking intention), ESP operating status (e.g., whether traction control is activated), and tire pressure (abnormal tire pressure in rainy or foggy weather may increase the risk of collision) can be read.
[0074] By synchronizing the collected IMU and CAN data using hardware timestamps, it is ensured that the acceleration at the moment of collision is strictly aligned with the vehicle speed and braking pressure, thus avoiding feature extraction errors due to time errors.
[0075] Furthermore, after switching to IMU / CAN mode, it can automatically cache sensor data from the most recent period, such as 3 seconds, to cover critical warning stages before a collision (such as a vehicle swerving due to sudden braking in the rain). If the subsequent scenario is determined to be a false trigger (such as when vision becomes clear after the rain stops), the cached data is automatically cleared and the system switches back to the normal solution to avoid redundant calculations.
[0076] This application's embodiments achieve intelligent switching of collision detection schemes through dynamic scene recognition and adaptive data acquisition strategies, significantly improving system robustness in complex environments. On one hand, it enhances environmental adaptability: in scenarios where vision / radar failure occurs, such as rain, fog, or strong interference, it automatically switches to IMU and CAN bus data, directly reflecting the collision impact using physical quantities such as acceleration and angular velocity, avoiding missed or false detections caused by sensor failure in traditional schemes. On the other hand, it optimizes resource utilization efficiency: in normal scenarios, the original vision / radar scheme is retained, reducing the computational load caused by high-frequency sampling of the IMU and CAN bus, while extending sensor lifespan. Furthermore, it improves the comprehensiveness of collision assessment: by fusing dynamic impact data from the IMU with vehicle control status from the CAN bus, the severity of the collision can be assessed more accurately, providing a reliable basis for subsequent graded alarms.
[0077] In some embodiments of this application, the multi-dimensional collision features include at least two of time-domain features, frequency-domain features, kinematic features, and energy features. The step of extracting the multi-dimensional collision features of the vehicle using a multi-dimensional collision feature extraction strategy based on the IMU data and the CAN bus data includes: determining the time-domain features and the frequency-domain features based on the IMU data; and determining the kinematic features and the energy features based on the IMU data and the CAN bus data.
[0078] The multi-dimensional collision features in this application embodiment may include time-domain features, frequency-domain features, kinematic features, and energy features. Time-domain features can capture the dynamic impact at the moment of collision by directly analyzing the time-series changes of the IMU raw signal. Frequency-domain features can convert the time-domain signal into a spectrum using Fourier transform to analyze the frequency components of the collision impact. Kinematic features combine the dynamic data from the IMU with the vehicle control status of the CAN bus to reflect the vehicle's motion patterns before and after the collision. Energy features calculate the energy changes during the collision process using a physical model to quantify the severity of the collision.
[0079] This application embodiment significantly improves the accuracy and comprehensiveness of collision detection by fusing IMU and CAN bus data and extracting collision features from multiple dimensions such as time domain, frequency domain, kinematics, and energy. Time domain features directly reflect the impact intensity of the collision, frequency domain features reveal the collision type, kinematic features supplement the causes of the collision, and energy features quantify the consequences of the collision. The complementary features of multiple dimensions form a complete chain of evidence, enabling refined processing of collision assessment, avoiding misjudgment based on a single feature, and enhancing anti-interference capabilities.
[0080] In some embodiments of this application, the IMU data includes acceleration data and angular velocity data, the CAN bus data includes vehicle speed data and steering angle data, the time domain features include at least one of acceleration peak value and the duration of acceleration peak value, the frequency domain features include the frequency components of the collision signal, the kinematic features include at least one of velocity change features, attitude angle change features, and trajectory change features, and the energy features include collision vibration energy.
[0081] (1) Temporal feature extraction
[0082] 1) Acceleration amplitude characteristics:
[0083] Within a sliding time window (e.g., 100ms), calculate the maximum absolute value of the three-axis acceleration. If the acceleration of any axis is greater than or equal to a threshold (e.g., 2.5g), a collision warning is triggered. If the peak acceleration of the Z-axis (vertical direction) reaches 3.2g, it is determined to be a vertical collision impact.
[0084] 2) Duration characteristics of acceleration abrupt change:
[0085] The duration (Δt) for which the statistical acceleration exceeds the threshold is calculated. If Δt ≤ 50ms, it is considered a brief impact; if 50ms < Δt ≤ 200ms, it is considered a sustained collision, such as a rear-end collision. If the X-axis (longitudinal) acceleration is ≥ 2.5g for 30ms, it is considered a minor collision.
[0086] (2) Frequency domain feature extraction
[0087] 1) Frequency components of the collision signal:
[0088] A Fast Fourier Transform (FFT) is performed on the filtered acceleration time-domain signal to obtain a spectrum of 0-50Hz (covering the main energy distribution frequency band of a vehicle collision). The frequency component with the highest energy in the spectrum (dominant frequency) is extracted. If the dominant frequency is in the range of 10-20Hz, it may correspond to vehicle body structural resonance (such as a frontal collision); if the dominant frequency is >30Hz, it may correspond to a flexible collision (such as a bumper impact). If the dominant frequency of the spectrum is 15Hz and the energy accounts for 60%, it is determined to be a rigid collision.
[0089] (3) Extraction of kinematic features
[0090] 1) Characteristics of abrupt rate changes:
[0091] Calculate the rate of change of vehicle speed (Δv / Δt) over a period of time before and after the collision, such as within 1 second. If Δv / Δt ≥ 5 m / s 2If the speed deceleration threshold is reached, it is determined to be a sudden speed change caused by a collision. For example, if the vehicle speed before the collision is 60 km / h (16.7 m / s), and it drops to 10 km / h (2.8 m / s) within 0.5 seconds after the collision, then Δv / Δt = (16.7 - 2.8) / 0.5 = 27.8 m / s², triggering a collision determination.
[0092] 2) Characteristics of abrupt changes in attitude angle:
[0093] The vehicle's pitch, roll, and yaw angles are calculated by integrating the gyroscope. If the pitch angle changes abruptly by Δθ ≥ 15° at the moment of impact (such as the rear of the vehicle dropping during a rear-end collision), it is determined to be a sudden change in attitude angle. For example, if the pitch angle changes abruptly from 0° to -20° at the moment of impact, it is determined to be a vertical collision.
[0094] 3) Trajectory mutation characteristics:
[0095] By combining vehicle speed and steering angle data, the vehicle's trajectory before and after the collision is reconstructed through dead reckoning. If the lateral deviation of the trajectory is greater than a threshold such as 0.5m (e.g., a side collision when avoiding an obstacle), it is determined to be a trajectory abrupt change.
[0096] (4) Energy feature extraction
[0097] 1) Collision vibration energy calculation:
[0098] The displacement (s) is obtained by performing a second integral on the acceleration data. The collision vibration energy is then calculated using a spring-mass model: E = 0.5 kS / s. 2 Where k is the vehicle body equivalent stiffness coefficient (calibrated experimentally, e.g., k = 3000 N / m in a frontal collision). If the Z-axis displacement is 0.2 m, then E = 0.5 × 3000 × 0.2 m. 2 =60J, reflecting the magnitude of the collision energy.
[0099] The embodiment significantly improves the accuracy and robustness of collision detection by fusing IMU and CAN bus data and extracting collision features from four dimensions: time domain, frequency domain, kinematics, and energy. Specifically:
[0100] (1) Multi-dimensional complementary features: Time domain features directly reflect the impact intensity of the collision, frequency domain features reveal the collision type, kinematic features supplement the cause of the collision, energy features quantify the consequences of the collision, and the four types of features form a complete chain of evidence to avoid misjudgment by a single feature.
[0101] (2) Enhanced anti-interference capability: IMU data is not affected by light, rain and fog, and CAN data provides vehicle control status. The fusion of the two can eliminate environmental noise and sensor error.
[0102] (3) Refined collision assessment: The collision direction can be distinguished by the characteristics of attitude angle change, and the collision offset can be assessed by the characteristics of trajectory change, providing a reliable basis for subsequent graded alarm and accident liability determination, and significantly improving the practical value of active safety system.
[0103] In some embodiments of this application, the collision level is divided into minor collision, moderate collision, and severe collision. Determining the alarm strategy corresponding to the vehicle based on its collision level includes: if the collision level is minor, then the alarm strategy corresponding to the vehicle is determined to be a minor collision alarm strategy, which includes the vehicle generating a first collision warning; if the collision level is moderate, then the alarm strategy corresponding to the vehicle is determined to be a moderate collision alarm strategy, which includes the vehicle generating a second collision warning and reporting the collision information to the cloud; if the collision level is severe, then the alarm strategy corresponding to the vehicle is determined to be a severe collision alarm strategy, which includes dialing an emergency number and simultaneously notifying emergency contacts.
[0104] The collision levels in this application embodiment can be divided into minor collisions, moderate collisions, and severe collisions. If the collision level output by the model is a minor collision (Level 1), the matching alarm strategy is the minor collision alarm strategy. The minor collision alarm strategy may include a first collision prompt, which may be, for example, triggering a buzzer to sound briefly (e.g., frequency 1Hz, lasting 3 seconds) and popping up a "minor collision" prompt (text + icon) on the central control screen. The prompt content includes the collision time, location (GPS positioning), and suggested inspection parts (e.g., "Please check the right front bumper").
[0105] If the model outputs a collision level of Level 2 (Medium Collision), then the alarm strategy will be the Level 2 collision alarm strategy. This strategy can include a second collision alert and cloud reporting. The second collision alert can trigger hazard lights to flash (e.g., at a frequency of 2Hz) and a continuous horn blare (e.g., for 10 seconds), while a "Medium Collision" warning pops up on the central control screen (e.g., red background + warning sound). Cloud reporting uses 4G / 5G communication modules to send collision information (e.g., time, location, speed, acceleration data, collision direction, etc.) to the cloud-based accident center. The cloud then generates an accident report and pushes it to the owner's app, displaying a message such as "Collision event recorded, please contact your insurance company."
[0106] If the model outputs a collision level of Level 3 (Severe Collision), then the matching alarm strategy will be the severe collision alarm strategy. The severe collision alarm strategy can include emergency calls and notification of emergency contacts. Emergency calls can include automatically dialing the emergency number and using the in-vehicle TTS module to announce the accident location via voice: "This is the intersection of XX Road and XX Road; a severe collision has occurred, and assistance is needed." Notifying emergency contacts can involve simultaneously sending SMS messages / APP push notifications to preset emergency contacts, containing the accident location, collision level, and suggested actions (such as "Your vehicle has been involved in a severe collision; please proceed to the scene immediately").
[0107] This application's embodiments, through the linkage of graded collision levels and dynamic alarm strategies, achieve precise response to collision accidents and efficient resource utilization, significantly improving the vehicle's active safety performance. Specifically:
[0108] (1) Accurate graded response: Based on a quantitative scoring model with multi-dimensional features, it can distinguish the severity of collisions (slight / moderate / severe) and avoid "one-size-fits-all" alarms.
[0109] (2) Alarm resource optimization: minor collisions are only alerted inside the vehicle, reducing the pressure on cloud storage and communication; moderate collisions upload key data to the cloud to assist insurance loss assessment; severe collisions automatically call for rescue and notify the contact person, shortening the response time and forming a resource allocation chain of "lightweight-medium-heavyweight".
[0110] (3) Improved user experience: Differentiated content for graded prompts reduces user anxiety; emergency contact notification function enhances collaborative handling capabilities after an accident.
[0111] In some embodiments of this application, the preset collision level classification model is trained in the following manner: collecting multi-dimensional collision feature data of the vehicle; labeling the multi-dimensional collision feature data with corresponding collision level labels using preset collision level classification conditions based on the multi-dimensional collision feature data; training the collision level classification model based on the multi-dimensional collision feature data and the corresponding collision level labels to obtain the preset collision level classification model.
[0112] When training the collision level classification model, multi-dimensional collision feature data is first collected. Acceleration data during the collision is collected through a three-axis accelerometer (IMU), velocity changes before and after the collision are obtained through a GPS module, and vehicle attitude angles are collected through a gyroscope. In addition, test track collision test data (such as using collision dummies to simulate collisions at different speeds / angles) and real accident case data can be combined to expand the diversity of data.
[0113] The collected raw data undergoes a series of preprocessing steps. For example, Kalman filtering is applied to the raw acceleration data to eliminate high-frequency noise such as road surface disturbances. Acceleration, velocity, and attitude angle data are synchronized by timestamps to ensure feature correlation, such as the peak acceleration at the moment of collision corresponding to the moment of velocity change. Min-Max normalization is performed on the acceleration, velocity change, and attitude angle data to map the data to the [0,1] interval and eliminate dimensional differences.
[0114] Labeling of the multi-dimensional feature data after the above processing is based on predefined collision level classification conditions. For example, the collision level classification conditions can be set in the following form:
[0115] Level 1 (Minor Collision):
[0116] Acceleration condition: a < 2.5g, such as the force exerted on the bumper during a low-speed scrape;
[0117] Speed condition: Δv < 10km / h, such as when slowly reversing in a parking lot and encountering an obstacle;
[0118] Attitude conditions: Vehicle attitude angle θ < 5°, with no risk of rollover or rollover.
[0119] Level 2 (Medium Collision):
[0120] Acceleration condition: 2.5g≤a<5g, such as when the car door is dented due to a side impact from an electric vehicle;
[0121] Speed condition: Δv≥10km / h, such as a rear-end collision at medium speed causing deformation of the rear of the vehicle;
[0122] Attitude conditions: θ < 30°, the vehicle body is slightly tilted but not out of control.
[0123] Level 3 (Severe Collision):
[0124] Acceleration condition: a≥5g, such as when a high-speed head-on collision with a guardrail causes the airbags to deploy;
[0125] Velocity condition: No strict upper limit, usually accompanied by high Δv;
[0126] Attitude conditions: θ≥30°, such as rollover or high-intensity collision causing the vehicle to roll over.
[0127] If the data simultaneously meets the conditions of a certain level of acceleration, velocity, attitude, etc., it is labeled as the corresponding level. For example, if a=4g, Δv=15km / h, θ=10°, it is labeled as Level2. If the data spans multiple level conditions, such as a=6g but θ=20°, it is labeled as a higher level (Level3) according to the "highest principle" to ensure safety.
[0128] Of course, it should be noted that the collision level classification conditions set above are only one example in the embodiments of this application. How to set them specifically can be flexibly adjusted by those skilled in the art according to actual needs, and no specific limitation is made here.
[0129] After labeling the dataset, machine learning algorithms such as XGBoost (Gradient Boosting Decision Tree) are used for training. XGBoost is chosen because it is highly compatible with multi-dimensional features and can automatically handle feature importance ranking, such as recognizing the crucial role of acceleration in classification. The trained XGBoost model is then compressed into the vehicle's ECU for real-time inference.
[0130] In addition, new accident data, such as false alarms / missed alarms reported by users, can be collected regularly, re-labeled, and added to the training set. The model parameters can be updated through incremental learning to adapt to different vehicle types or road conditions, such as urban roads and highway scenarios.
[0131] This application's embodiments achieve accurate quantification and dynamic response of collision severity through a data-driven collision level classification model, significantly improving the intelligence level of vehicle active safety systems.
[0132] In some embodiments of this application, before acquiring the vehicle's IMU data and CAN bus data, the vehicle collision alarm method further includes: determining the vehicle's current driving state; and if the current driving state is a stationary state, acquiring the IMU data in the stationary state and establishing a gravity direction reference.
[0133] Before performing real-time collision detection based on IMU and CAN bus data, the IMU data can be preprocessed. This preprocessing mainly refers to dynamic baseline calibration of the IMU data. Specifically, the current driving state of the vehicle is determined first. For example, wheel speed sensor data (such as the rotational speed of the four wheels), engine speed (RPM) or motor speed (electric vehicle), gear position signal (such as P / R / N / D gear), and vehicle speed signal (VSS) can be obtained through the vehicle CAN bus.
[0134] If the rotational speed of all four wheels is below the threshold (e.g., 0.5 km / h, to eliminate low-speed creep interference), the CAN bus vehicle speed signal VSS=0 (or remains below 0.5 km / h for more than 3 seconds), and the gear is in P (park) or N (neutral) (for electric vehicles, this can be determined by the motor speed = 0), then the vehicle is considered stationary. If any condition is not met, the vehicle is considered to be in motion.
[0135] When the vehicle is stationary, the vertical acceleration Az measured by the IMU should be equal to the gravitational acceleration g, with no lateral (Ay) or longitudinal (Ax) acceleration components. By statistically analyzing the long-term average value of Az under stationary conditions, a gravitational direction reference can be established, eliminating zero-bias drift of the IMU caused by temperature changes or mechanical vibrations.
[0136] With the vehicle stationary, continuously collect IMU data (e.g., sample every 10ms). The collection duration is recommended to exceed a certain threshold, such as 30 seconds, to balance calibration accuracy and response speed. If the vehicle starts during the data collection process, immediately terminate the calibration and retain the collected data. Calculate the arithmetic mean of the collected Az data as the gravity direction reference value (G_ref); use the 3σ principle to remove data that deviates from the mean by more than three standard deviations, such as sudden changes in Az caused by external impacts.
[0137] During subsequent driving, the real-time measured Az is compared with G_ref, the zero bias error Δg = Az - G_ref is calculated, and this error is subtracted from the total acceleration during collision detection. For example, the corrected collision acceleration a_corrected = a_raw -Δg is obtained to avoid false alarms (such as road bumps being mistakenly judged as collisions).
[0138] Each time the vehicle comes to a standstill (such as after parking and turning off the engine), the calibration process can be re-executed to update the G_ref value. This allows for adaptation to long-term IMU performance changes.
[0139] The calibrated gravity reference G_ref is stored in the non-volatile memory of the vehicle ECU and can be retained even after power failure. During the collision detection phase, the real-time IMU data (Ax, Ay, Az) must first undergo zero-bias compensation (based on G_ref) before being input into the collision level classification model.
[0140] The embodiments of this application significantly improve the reliability and environmental adaptability of the vehicle collision warning system through the collaborative design of driving status detection and dynamic baseline calibration.
[0141] This application embodiment also provides a vehicle collision alarm device 200, such as Figure 2 The diagram shows a structural schematic of a vehicle collision alarm device according to an embodiment of this application. The vehicle collision alarm device 200 includes: an acquisition unit 210, an extraction unit 220, a first determination unit 230, and an alarm unit 240, wherein:
[0142] Acquisition unit 210 is used to acquire vehicle IMU data and CAN bus data;
[0143] Extraction unit 220 is used to extract multi-dimensional collision features of the vehicle based on the IMU data and the CAN bus data using a multi-dimensional collision feature extraction strategy;
[0144] The first determining unit 230 is used to determine the collision level of the vehicle based on the multi-dimensional collision features of the vehicle and a preset collision level classification model, when the multi-dimensional collision features of the vehicle are extracted.
[0145] The alarm unit 240 is used to determine the alarm strategy corresponding to the vehicle based on the collision level of the vehicle, and to execute the alarm strategy.
[0146] In some embodiments of this application, the acquisition unit 210 is specifically used to: determine the current driving scenario of the vehicle; and acquire the vehicle's IMU data and CAN bus data when the current driving scenario triggers preset scenario conditions.
[0147] In some embodiments of this application, the multi-dimensional collision features include at least two of time-domain features, frequency-domain features, kinematic features, and energy features. The extraction unit 220 is specifically used to: determine the time-domain features and the frequency-domain features based on the IMU data; and determine the kinematic features and the energy features based on the IMU data and the CAN bus data.
[0148] In some embodiments of this application, the IMU data includes acceleration data and angular velocity data, the CAN bus data includes vehicle speed data and steering angle data, the time domain features include at least one of acceleration peak value and the duration of acceleration peak value, the frequency domain features include the frequency components of the collision signal, the kinematic features include at least one of velocity change features, attitude angle change features, and trajectory change features, and the energy features include collision vibration energy.
[0149] In some embodiments of this application, the collision level is divided into minor collision, moderate collision, and severe collision. The alarm unit 240 is specifically used for: if the collision level of the vehicle is minor collision, then determining the alarm strategy corresponding to the vehicle as a minor collision alarm strategy, the minor collision alarm strategy including the vehicle generating a first collision warning; if the collision level of the vehicle is moderate collision, then determining the alarm strategy corresponding to the vehicle as a moderate collision alarm strategy, the moderate collision alarm strategy including the vehicle generating a second collision warning and reporting the collision information to the cloud; if the collision level of the vehicle is severe collision, then determining the alarm strategy corresponding to the vehicle as a severe collision alarm strategy, the severe collision alarm strategy including dialing an emergency call and simultaneously notifying emergency contacts.
[0150] In some embodiments of this application, the preset collision level classification model is trained in the following manner: collecting multi-dimensional collision feature data of the vehicle; labeling the multi-dimensional collision feature data with corresponding collision level labels using preset collision level classification conditions based on the multi-dimensional collision feature data; training the collision level classification model based on the multi-dimensional collision feature data and the corresponding collision level labels to obtain the preset collision level classification model.
[0151] In some embodiments of this application, the vehicle collision alarm device 200 further includes: a second determining unit, used to determine the current driving state of the vehicle before acquiring the vehicle's IMU data and CAN bus data; and a calibration unit, used to acquire IMU data in a stationary state and establish a gravity direction reference when the current driving state is a stationary state.
[0152] It is understood that the above-mentioned vehicle collision alarm device can realize all the steps of the vehicle collision alarm method provided in the foregoing embodiments. The relevant explanations of the vehicle collision alarm method are applicable to the vehicle collision alarm device, and will not be repeated here.
[0153] Figure 3 This is a schematic diagram of the structure of a device according to an embodiment of this application. For example... Figure 3 As shown, the device includes one or more processors (or processing units), and may also include one or more memories coupled to the processors, and may also include a communication module coupled to the processors.
[0154] A communication module can be used to communicate with other devices or apparatuses, such as sending or receiving data and / or signals. A communication module may have at least one communication module for communication. A communication module may include any interface necessary for communicating with other devices. Exemplarily, a communication module may be a transceiver, circuit, bus, module, or other type of communication module.
[0155] The processor may include, but is not limited to, one or more of the following: a general-purpose computer, a special-purpose computer, a microcontroller, a digital signal processor (DSP), or a controller-based multi-core controller architecture. The device may have multiple processors, such as application-specific integrated circuit (ASIC) chips, which are time-dependent on a clock synchronized with the main processor.
[0156] The memory may include one or more non-volatile memories and one or more volatile memories. Examples of non-volatile memories include, but are not limited to, at least one of the following: read-only memory (ROM), electrically programmable read-only memory (EPROM), flash memory, hard disk, compact disc (CD), digital video disc (DVD), or other magnetic and / or optical storage. Examples of volatile memories include, but are not limited to, at least one of the following: random access memory (RAM), or other volatile memories that do not persist during the duration of a power outage.
[0157] A computer program consists of computer-executable instructions that are executed by an associated processor. Programs can be stored in ROM. A processor can perform any appropriate action and processing by loading the program into RAM.
[0158] Possible implementations of this application can be achieved through a program, enabling the communication device to execute any of the processes discussed in the foregoing embodiments. Possible implementations of this application can also be achieved through hardware or a combination of software and hardware.
[0159] In some implementations, the program may be tangibly contained in a computer-readable storage medium, which may include in a device (such as in memory) or other storage device accessible by the device. The program may be loaded from the computer-readable storage medium into RAM for execution. The computer-readable storage medium may include any type of tangible non-volatile memory, such as ROM, EPROM, flash memory, hard disk, CD, DVD, etc.
[0160] This application also provides a computer-readable storage medium storing computer instructions or program code thereon, which, when executed by a processor, causes the processor to perform the methods and functions involved in any of the above embodiments. A computer-readable medium can be any tangible medium that contains or stores a program for or relating to an instruction execution system, apparatus, or device. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. More detailed examples of computer-readable storage media include electrical connections with one or more wires, magnetic media (e.g., disks, floppy disks, hard disks, magnetic tapes, magnetic storage devices), optical media (e.g., optical storage devices, DVDs), semiconductor media (e.g., solid-state drives), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), or any suitable combination thereof.
[0161] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. Embodiments of this application also provide at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. This computer program product includes one or more computer-executable instructions, such as instructions included in a program module, which execute in a device on a target real or virtual processor to perform the processes, methods, and functions involved in any of the above embodiments. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.
[0162] This application also proposes a computer program product, including a computer program or instructions that, when run on a computer, cause the computer to perform the processes, methods, and functions described in the above embodiments. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided as needed. The machine-executable instructions for the program modules can be executed locally or in a distributed device. In a distributed device, the program modules can reside in both local and remote storage media.
[0163] Generally, the various embodiments of this application can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software, which can be executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of this disclosure are shown and described as block diagrams, flowcharts, or represented using some other illustration, it should be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as, as non-limiting examples, in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.
[0164] It should be noted that although embodiments of this application have been described above with reference to the accompanying drawings, these embodiments are not independent of each other, and they can be combined to obtain other embodiments. The methods, situations, categories, and classifications of embodiments in this application are only for the convenience of description and should not constitute a special limitation. Various methods, categories, situations, and features in embodiments can be combined with each other if logically consistent. The various embodiments of this application can be arbitrarily combined to achieve different technical effects. The embodiments of this application will not list various combinations.
[0165] Furthermore, although the operation of the methods of this disclosure is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. Rather, the steps depicted in the flowcharts may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps. It should also be noted that the features and functions of two or more devices according to this disclosure may be embodied in one device. Conversely, the features and functions of one device described above may be further divided and embodied by multiple devices.
[0166] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0167] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.
Claims
1. A vehicle collision alarm method, characterized in that, The vehicle collision alarm method includes: Acquire vehicle IMU data and CAN bus data; Based on the IMU data and the CAN bus data, multi-dimensional collision features of the vehicle are extracted using a multi-dimensional collision feature extraction strategy. Once the multi-dimensional collision features of the vehicle are extracted, the collision level of the vehicle is determined using a preset collision level classification model based on the multi-dimensional collision features of the vehicle. The alarm strategy corresponding to the vehicle is determined based on the collision level of the vehicle, and the alarm strategy is executed.
2. The vehicle collision alarm method according to claim 1, characterized in that, The acquisition of vehicle IMU data and CAN bus data includes: Determine the current driving scenario of the vehicle; When the preset scenario conditions are triggered in the current driving scenario, the vehicle's IMU data and CAN bus data are acquired.
3. The vehicle collision alarm method according to claim 1, characterized in that, The multi-dimensional collision features include at least two of the following: time-domain features, frequency-domain features, kinematic features, and energy features. The step of extracting the multi-dimensional collision features of the vehicle using the multi-dimensional collision feature extraction strategy based on the IMU data and the CAN bus data includes: The time-domain features and the frequency-domain features are determined based on the IMU data; The kinematic characteristics and energy characteristics are determined based on the IMU data and the CAN bus data.
4. The vehicle collision alarm method according to claim 3, characterized in that, The IMU data includes acceleration data and angular velocity data; the CAN bus data includes vehicle speed data and steering angle data; the time-domain features include at least one of acceleration peak value and the duration of acceleration peak value; the frequency-domain features include the frequency components of the collision signal; the kinematic features include at least one of velocity change features, attitude angle change features, and trajectory change features; and the energy features include collision vibration energy.
5. The vehicle collision alarm method according to claim 1, characterized in that, The collision levels are categorized as minor, moderate, and severe. Determining the corresponding alarm strategy for the vehicle based on its collision level includes: If the collision level of the vehicle is minor collision, then the alarm strategy corresponding to the vehicle is determined to be the minor collision alarm strategy, which includes the vehicle generating a first collision warning. If the collision level of the vehicle is medium collision, then the alarm strategy corresponding to the vehicle is determined to be a medium collision alarm strategy, which includes the vehicle generating a second collision alert and reporting the collision information to the cloud. If the collision level of the vehicle is a severe collision, then the alarm strategy corresponding to the vehicle is determined to be a severe collision alarm strategy, which includes making an emergency call and simultaneously notifying emergency contacts.
6. The vehicle collision alarm method according to any one of claims 1 to 5, characterized in that, The preset collision level classification model is trained in the following manner; Collect multi-dimensional collision feature data of vehicles; Based on the multi-dimensional collision feature data, the multi-dimensional collision feature data is labeled with corresponding collision level tags using preset collision level classification conditions. The collision level classification model is trained based on the multi-dimensional collision feature data and the corresponding collision level labels to obtain the preset collision level classification model.
7. The vehicle collision alarm method according to any one of claims 1 to 5, characterized in that, Before acquiring the vehicle's IMU data and CAN bus data, the vehicle collision alarm method further includes: Determine the vehicle's current driving status; When the current driving state is stationary, acquire IMU data in the stationary state and establish a gravity direction reference.
8. A vehicle collision alarm device, characterized in that, The vehicle collision alarm device includes: The acquisition unit is used to acquire vehicle IMU data and CAN bus data; An extraction unit is used to extract multi-dimensional collision features of the vehicle based on the IMU data and the CAN bus data using a multi-dimensional collision feature extraction strategy. The first determining unit is used to determine the collision level of the vehicle based on the multi-dimensional collision features of the vehicle and a preset collision level classification model, after extracting the multi-dimensional collision features of the vehicle. An alarm unit is used to determine the alarm strategy corresponding to the vehicle based on the collision level of the vehicle, and to execute the alarm strategy.
9. An apparatus comprising: processor; And a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform any of the vehicle collision alarm methods of claims 1 to 7.
10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements any one of the vehicle collision alarm methods described in claims 1 to 7.