Forward collision warning method, device, apparatus and storage medium

Abstract

Embodiments of the present application provide a forward collision warning method, device, equipment and storage medium. Applied to the field of forward collision warning technology, the method collects vehicle motion information, front vehicle motion information and environment information through a plurality of preset sensors; the vehicle motion information, the front vehicle motion information and the environment information are time-aligned and space-aligned; the front vehicle motion state is determined according to the front vehicle motion information, and the front vehicle motion state is filtered; the vehicle motion information, the front vehicle motion state and the environment information are inferred by using a dynamic Bayesian network, and the risk probability value of vehicle collision is determined; and corresponding collision warning strategies are taken for the vehicle according to the risk probability value, thereby improving the accuracy of forward collision warning and providing strong support for ensuring vehicle driving safety.

Description

Technical Field

[0001] This application relates to the field of forward collision warning technology, and in particular to a forward collision warning method, apparatus, device and storage medium. Background Technology

[0002] Forward Collision Warning (FCW) is a core function of Advanced Driver Assistance Systems (ADAS). Existing technical solutions are mostly based on data fusion from sensors such as millimeter-wave radar, cameras, and lidar, combined with Time To Collision (TTC) calculations to achieve warnings.

[0003] Traditional TTC calculations assume the vehicle ahead moves at a constant speed, ignoring the effects of acceleration and road curvature. Existing solutions often use fixed-weight fusion, failing to dynamically adjust based on the environment. For example, camera performance degrades in rainy or foggy weather, but the system still assigns it high weight, leading to missed detections or false alarms. Summary of the Invention

[0004] This application provides a forward collision warning method, apparatus, device, and storage medium to address the problems existing in current forward collision warning systems, such as limitations in static model assumptions, rigid sensor fusion strategies, and insufficient processing capabilities for complex scenarios.

[0005] In a first aspect, this application provides a forward collision warning method, the method comprising:

[0006] The vehicle collects motion information from its own vehicle, the vehicle in front, and environmental information through multiple preset sensors.

[0007] Perform time and space alignment on the vehicle's motion information, the motion information of the vehicle in front, and environmental information;

[0008] The motion state of the vehicle in front is determined based on the motion information of the vehicle in front, and the motion state of the vehicle in front is filtered.

[0009] A dynamic Bayesian network is used to infer the vehicle's motion information, the motion state of the vehicle in front, and environmental information to determine the probability value of a vehicle collision.

[0010] The vehicle is given a corresponding collision warning strategy based on the risk probability value.

[0011] Optionally, filtering the motion state of the preceding vehicle includes:

[0012] Obtain sensor state parameters, and determine a state decay factor based on the sensor state parameters. The state decay factor is used to indicate the degree of influence of the sensor's own state on the sensor performance.

[0013] An environmental attenuation factor is determined based on environmental parameters, and the environmental attenuation factor is used to indicate the degree of influence of environmental conditions on sensor performance.

[0014] A comprehensive attenuation factor is determined based on the state attenuation factor and the environment attenuation factor. The comprehensive attenuation factor is used to indicate the total degree of attenuation of the current sensor performance.

[0015] The real-time confidence score of the sensor is calculated based on the comprehensive attenuation factor, and the measurement noise covariance of the sensor is adjusted based on the real-time confidence score.

[0016] Construct the measurement noise covariance matrix based on the measurement noise covariance;

[0017] The preceding vehicle's state is filtered based on the measured noise covariance matrix.

[0018] Optionally, the vehicle motion information includes: vehicle speed; the preceding vehicle motion state includes: preceding vehicle acceleration; the environmental information includes: road curvature; and before using a dynamic Bayesian network to infer the vehicle motion information, the preceding vehicle motion state, and the environmental information, the method further includes:

[0019] The vehicle speed is discretized to obtain the vehicle speed state, which is used to indicate the discrete level of the vehicle's current driving speed.

[0020] The acceleration of the vehicle in front is discretized to obtain the acceleration state of the vehicle in front, which is used to indicate the discrete level of the acceleration of the vehicle in front.

[0021] The road curvature is discretized to obtain the road curvature state, which is used to indicate the discrete level of the geometry of the current driving segment.

[0022] Optionally, the method further includes:

[0023] Acquire historical driving behavior data and determine the driver's reaction time level based on the historical driving behavior data;

[0024] Acquire driver eye images, determine driver fatigue state based on driver eye images, and the driver fatigue state is used to indicate the discrete level of the driver's real-time physiological state;

[0025] The probability of a vehicle collision is determined based on the vehicle's speed, the acceleration of the vehicle in front, the road curvature, the driver's reaction time level, and the driver's fatigue level.

[0026] Optionally, determining the probability value of a vehicle collision based on the vehicle's speed state, the acceleration state of the vehicle in front, the road curvature state, the driver's reaction time level, and the driver's fatigue state includes:

[0027] An evidence set is constructed based on the vehicle's speed status, the preceding vehicle's acceleration status, curvature status, the driver's reaction time level, and the driver's fatigue status.

[0028] The global joint probability distribution of the dynamic Bayesian network is determined based on the preset conditional probability table, and the joint probability of the collision risk probability and the evidence set is determined based on the global joint probability distribution.

[0029] By iterating through all state values ​​of the collision risk probability, and summing the joint probability of the collision risk probability and the evidence set with respect to all state values ​​of the collision risk probability, the marginal probability of the evidence set is obtained.

[0030] Based on the joint probability of the collision risk probability and the marginal probability of the evidence set, the posterior probability distribution of the collision risk probability is calculated to obtain the risk probability value of the vehicle collision.

[0031] Optionally, the step of adopting a corresponding collision warning strategy for the vehicle based on the risk probability value includes:

[0032] Determine whether the risk probability value is less than a first preset value;

[0033] If so, then no warning will be triggered;

[0034] If not, then determine whether the risk probability value is less than the second preset value;

[0035] If so, a Level 1 warning will be issued to the vehicle, which includes: triggering an information prompt or an audio / visual warning;

[0036] If not, a secondary warning will be issued for the vehicle, including triggering an emergency audible / visual warning and automatically applying braking measures.

[0037] Secondly, this application provides a forward collision warning device, comprising:

[0038] The acquisition module is used to collect vehicle motion information, forward vehicle motion information, and environmental information through multiple preset sensors;

[0039] The preprocessing module is used to perform time and space alignment on the vehicle's motion information, the motion information of the vehicle in front, and the environmental information;

[0040] The preprocessing module is also used to determine the motion state of the preceding vehicle based on the preceding vehicle motion information, and to filter the preceding vehicle motion state;

[0041] The processing module is used to infer the vehicle's motion information, the motion state of the vehicle in front, and environmental information using a dynamic Bayesian network to determine the probability value of a vehicle collision.

[0042] The warning module is used to take corresponding collision warning strategies for vehicles based on risk probability values.

[0043] Thirdly, this application provides a forward collision warning device, comprising:

[0044] Memory;

[0045] processor;

[0046] The memory stores computer-executed instructions;

[0047] The processor executes computer execution instructions stored in the memory to implement the forward collision warning method as described in the first aspect and various possible implementations of the first aspect above.

[0048] Fourthly, this application provides a computer storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the forward collision warning method as described in the first aspect and various possible implementations of the first aspect above.

[0049] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the forward collision warning method as described in the first aspect and various possible implementations of the first aspect.

[0050] This application provides a forward collision warning method, apparatus, device, and storage medium. The method collects vehicle motion information, preceding vehicle motion information, and environmental information through multiple preset sensors; performs time and spatial alignment on the vehicle motion information, preceding vehicle motion information, and environmental information; determines the preceding vehicle's motion state based on the preceding vehicle's motion information and filters the preceding vehicle's motion state; uses a dynamic Bayesian network to infer the vehicle motion information, preceding vehicle's motion state, and environmental information to determine the probability value of a vehicle collision; and adopts a corresponding collision warning strategy for the vehicle based on the probability value, thereby improving the accuracy of forward collision warning. Attached Figure Description

[0051] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0052] Figure 1 A flowchart illustrating the forward collision warning method provided in this application embodiment;

[0053] Figure 2This is a schematic diagram of the forward collision warning device provided in the embodiments of this application;

[0054] Figure 3 This is a schematic diagram of the structure of the forward collision warning device provided in the embodiments of this application.

[0055] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0056] 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 with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0057] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0058] In this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0059] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0060] Figure 1 This is a flowchart illustrating a collision warning method provided in an embodiment of this application. Figure 1 As shown, the forward collision warning method provided in this embodiment includes:

[0061] S1: Collects vehicle motion information, forward vehicle motion information, and environmental information through multiple preset sensors.

[0062] Among them, the vehicle's motion information can include, for example, the vehicle's speed, acceleration, direction of travel, steering angle, and vehicle posture; the preceding vehicle's motion information can include, for example, the relative distance, relative speed, relative angle, and driving status of the preceding vehicle; and the environmental information can include, for example, road conditions, traffic signs, traffic light status, weather conditions, and the position and status of surrounding obstacles.

[0063] Understandably, in intelligent driving or vehicle-assisted driving systems, various types of sensors are pre-installed on the vehicle to acquire data on the vehicle's own motion status, the motion status of vehicles in front, and relevant data on the vehicle's surrounding environment.

[0064] Specifically, appropriate sensors are selected based on the type of information to be collected. For example, wheel speed sensors and inertial measurement units can be used to collect vehicle speed and acceleration; millimeter-wave radar and lidar can be used to detect the distance and speed of the vehicle ahead; and high-definition cameras can be used to identify traffic signs and traffic lights. The sensors collect data in real time and transmit the raw data to the vehicle's electronic control unit or central processing unit via the vehicle's communication bus.

[0065] S2: Perform time and space alignment on the vehicle's motion information, the motion information of the vehicle in front, and environmental information.

[0066] Temporal alignment refers to synchronizing the vehicle's motion information, the preceding vehicle's motion information, and environmental information along the time dimension, ensuring consistency in timestamps across data from different sources. Spatial alignment refers to unifying and calibrating the vehicle's motion information, the preceding vehicle's motion information, and environmental information within a spatial coordinate system, accurately mapping them to the same spatial reference frame. For example, information about the positions of the vehicle, the preceding vehicle, and obstacles collected by different sensors is transformed into a unified spatial coordinate system centered on the vehicle or another unified system, enabling an accurate description of their relative positional relationships.

[0067] Specifically, hardware devices are used to achieve time synchronization of sensor data acquisition. For example, a high-precision time synchronization signal generator sends synchronization pulse signals to each sensor. Upon receiving the synchronization pulse signal, the sensors simultaneously begin acquiring data and record the timestamp of the synchronization pulse signal in the acquired data. In subsequent data processing, the data from different sensors can be time-aligned based on the timestamps. In intelligent driving systems, the frequency and trigger time of data acquisition by various sensors may differ. Time alignment ensures the accuracy of subsequent data analysis and processing.

[0068] Based on the calibration parameters of the sensors, data collected by different sensors are transformed from their own coordinate systems to a unified coordinate system. For example, the coordinates of a target point detected by radar are transformed from the radar coordinate system to a coordinate system centered on the vehicle. This transformation process needs to consider the sensor's installation position, angle, and the transformation relationship between coordinate systems. Through coordinate transformation, spatial alignment of data from different sensors is achieved, thereby accurately describing the relative positional relationships between the vehicle, the vehicle in front, and surrounding objects.

[0069] S3: Determine the motion state of the vehicle in front based on the motion information of the vehicle in front, and filter the motion state of the vehicle in front.

[0070] This involves converting the motion information of the vehicle in front into state parameters that can accurately describe its motion characteristics, such as its absolute speed, acceleration, direction of travel, and trajectory, thus constituting the motion state of the vehicle in front.

[0071] Understandably, sensor-collected data may contain noise, errors, or interference, leading to inaccurate and unstable determinations of the preceding vehicle's motion. By filtering the preceding vehicle's status, noise and interference are removed, data fluctuations are smoothed, and the obtained preceding vehicle's motion status becomes closer to reality, improving the reliability and accuracy of the data.

[0072] Specifically, state estimation algorithms are used to determine the motion state of the vehicle in front. For example, based on the relative distance and relative speed between the vehicle in front and the vehicle itself, combined with the motion state of the vehicle itself, the absolute speed and acceleration of the vehicle in front are estimated using a kinematic model. Commonly used filtering algorithms include Kalman filtering, extended Kalman filtering, and unscented Kalman filtering. Among them, Kalman filtering is suitable for linear systems and Gaussian noise; when the system is nonlinear, extended Kalman filtering or unscented Kalman filtering can be used.

[0073] Taking the Extended Kalman Filter (EKF) as an example, its process includes two main steps: state prediction and state update. In the state prediction stage, based on the previous time-series estimate of the preceding vehicle's motion state and the system's motion model, the current time-series state value and covariance matrix are predicted. In the state update stage, using the observation data collected by sensors at the current time, combined with the observation model, the predicted state value is corrected to obtain a more accurate state estimate. By continuously repeating these two steps, real-time filtering of the preceding vehicle's motion state is achieved, removing noise and interference to obtain smooth and accurate preceding vehicle motion state data.

[0074] Understandably, to overcome the limitations of traditional fixed-weight fusion strategies, this approach involves real-time, dynamic evaluation of the reliability of each sensor under its specific environment and state, and adjusting the sensor's influence weight in the state estimation algorithm based on the evaluation results. This allows the fusion processing of sensor data to prioritize data from the best-performing sensor and suppress poorly performing or potentially misleading data, thereby significantly improving the accuracy and robustness of target state estimation in complex and changing environments.

[0075] Specifically, sensor state parameters are acquired, and a state attenuation factor is determined based on these parameters. The state attenuation factor indicates the degree of influence of the sensor's own state on its performance. An environmental attenuation factor is determined based on environmental parameters, indicating the degree of influence of the environmental state on the sensor's performance. A comprehensive attenuation factor is determined based on the state attenuation factor and the environmental attenuation factor, indicating the total attenuation of the current sensor performance. The real-time confidence score of the sensor is calculated based on the comprehensive attenuation factor, and the measurement noise covariance corresponding to the sensor is adjusted based on the real-time confidence score. A measurement noise covariance matrix is ​​constructed based on the measurement noise covariance matrix. The preceding vehicle's state is filtered based on the measurement noise covariance matrix.

[0076] More specifically, a degradation factor (DF) is defined for the degree of influence of each environmental factor and sensor state on the performance of a specific sensor. Its value range is usually set between [0,1] (1 represents no influence or ideal state, and 0 represents complete failure or maximum influence).

[0077] For example, the rainfall attenuation factor of the camera can be calculated based on the rainfall intensity (Intensity_rain):

[0078] DF_rain_cam=max(0,1-k_rain_cam*Intensity_rain)

[0079] Wherein, DF_rain_cam is the rainfall attenuation factor, and k_rain_cam is the camera's sensitivity coefficient to rainfall.

[0080] The state attenuation factor DF_status is determined based on the sensor health diagnostic information Health_sensor. It is obtained by multiplying all relevant environmental attenuation factors and state attenuation factors affecting a specific sensor, resulting in the sensor's current total attenuation factor.

[0081] DF_total_sensor(t)=DF_env_sensor(t)*DF_status_sensor(t)

[0082] Wherein, DF_total_sensor(t) is the total attenuation factor, DF_env_sensor(t) is the product of all applicable environmental attenuation factors, and DF_status_sensor(t) is the status attenuation factor.

[0083] Based on the total attenuation factor mentioned above, the current real-time confidence score C_sensor(t) of each sensor is calculated. This score quantifies the reliability of the sensor's current data in the fusion process.

[0084] C_sensor(t)=C_base_sensor*DF_total_sensor(t)

[0085] Typically, the base confidence level C_base_sensor is set to 1, so it simplifies to:

[0086] C_sensor(t)=DF_total_sensor(t)

[0087] At this point, the value range of C_sensor(t) is [0,1].

[0088] Furthermore, the sensor's measurement noise covariance is adjusted based on the real-time confidence score. Specifically, the higher the sensor's confidence level, the lower its measurement uncertainty (variance) should be; conversely, the lower the confidence level, the higher the uncertainty. The adjusted measurement variance σ_adjusted^2(t) can be calculated using the following formula:

[0089] σ_adjusted^2(t)=σ_base^2 / (C_sensor(t)+ε)

[0090] Where σ_base^2 is the base variance of the sensor's corresponding measurement under ideal conditions (usually obtained through offline calibration); C_sensor(t) is the sensor's current real-time confidence score; ε is a very small positive number (e.g., 1e-6) used to prevent the denominator from being zero when the confidence score approaches zero, ensuring that the variance is bounded.

[0091] The adjusted variance σ_adjusted^2(t) of the corresponding measured values ​​of each sensor is filled into the corresponding position of the measurement noise covariance matrix to construct and output the dynamically adjusted measurement noise covariance matrix.

[0092] S4: A dynamic Bayesian network is used to infer the vehicle's motion information, the motion state of the vehicle in front, and environmental information to determine the probability value of a vehicle collision.

[0093] Understandably, the core of a dynamic Bayesian network is a directed acyclic graph, where nodes represent input variables, intermediate variables, and the final output variable, and directed edges represent direct probabilistic dependencies between variables. The parameters of a dynamic Bayesian network are composed of conditional probability tables. These parameters can be set using domain expert knowledge or learned from large-scale real-world driving data through machine learning methods such as maximum likelihood estimation and Bayesian estimation. Furthermore, the network supports periodic optimization and updates via an edge-cloud collaborative architecture.

[0094] Before using a dynamic Bayesian network to infer the vehicle's motion information, the preceding vehicle's motion state, and environmental information to determine the probability of a collision, this method further includes discretizing the vehicle's motion information, the preceding vehicle's motion state, and the environmental information. The vehicle's motion information includes the vehicle's speed, the preceding vehicle's motion state includes the preceding vehicle's acceleration, and the environmental information includes the road curvature. Discretization can be defined based on statistical analysis such as data clustering, equal-frequency / equal-width partitioning, or domain expert knowledge.

[0095] Specifically, the vehicle speed is discretized to obtain the vehicle speed state, which is used to indicate the discrete level of the vehicle's current driving speed.

[0096] The acceleration of the vehicle in front is discretized to obtain the acceleration state of the vehicle in front, which is used to indicate the discrete level of the acceleration of the vehicle in front.

[0097] The road curvature is discretized to obtain the road curvature state, which is used to indicate the discrete level of the geometry of the current driving segment.

[0098] In an optional embodiment, the method further includes: acquiring historical driving behavior data and determining the driver's reaction time level based on the historical driving behavior data; acquiring driver's eye images and determining the driver's fatigue state based on the driver's eye images, wherein the driver's fatigue state is used to indicate the discrete level of the driver's real-time physiological state.

[0099] For example, assume that the historical response data follows a normal distribution N(μ,σ). 2 The sample mean μ and sample standard deviation σ are calculated. Based on the calculated μ and σ, or the percentiles of the reaction time distribution, they are compared with a set threshold. The driver's reaction time is then classified into discrete levels based on the comparison results. By processing the driver's eye images captured by the vehicle-mounted camera, the eyelid closure percentage PERCLOS is calculated: PERCLOS = (number of closure frames / t) × 100%, where t is the number of consecutive frames captured by the vehicle-mounted camera. The calculated eyelid closure percentage PERCLOS is compared with a preset threshold to determine the driver's fatigue state.

[0100] Furthermore, the obtained driver reaction time and fatigue state are input into a dynamic Bayesian network.

[0101] The probability of a vehicle collision is determined based on the vehicle's speed, the acceleration of the vehicle in front, the road curvature, the driver's reaction time level, and the driver's fatigue level. Specifically, this includes:

[0102] An evidence set is constructed based on the vehicle's speed status, the preceding vehicle's acceleration status, curvature status, the driver's reaction time level, and the driver's fatigue status.

[0103] The global joint probability distribution of the dynamic Bayesian network is determined based on the preset conditional probability table, and the joint probability of the collision risk probability and the evidence set is determined based on the global joint probability distribution.

[0104] By iterating through all state values ​​of the collision risk probability, and summing the joint probability of the collision risk probability and the evidence set with respect to all state values ​​of the collision risk probability, the marginal probability of the evidence set is obtained.

[0105] Based on the joint probability of the collision risk probability and the marginal probability of the evidence set, the posterior probability distribution of the collision risk probability is calculated to obtain the risk probability value of the vehicle collision.

[0106] Specifically, the posterior probability distribution of the collision risk probability is calculated, satisfying the following formula:

[0107] P(P_risk|e)=P(P_risk,e) / P(e)

[0108] Where P(P_risk|e) is the posterior probability distribution of the collision risk probability; P(P_risk,e) is the joint probability of the collision risk probability and the evidence set, which can be obtained from the global joint probability distribution defined by the dynamic Bayesian network by marginalizing non-query and non-evidence variables; P(e) is the marginal probability of the evidence set, which is a normalization constant and can be obtained by summing the joint probability P(P_risk,e) with respect to all possible states of the collision risk probability P_risk.

[0109] By performing the above calculation process, the numerical result of P(P_risk|e) is finally obtained, and this result is used as the risk probability value of a vehicle collision.

[0110] S5: Take corresponding collision warning strategies for vehicles based on risk probability values.

[0111] Specifically, it is determined whether the risk probability value is less than a first preset value; if so, no warning is triggered; if not, it is determined whether the risk probability value is less than a second preset value; if so, a first-level warning is issued to the vehicle, wherein the first-level warning includes triggering an information prompt or an audio / visual warning; if not, a second-level warning is issued to the vehicle, wherein the second-level warning includes triggering an emergency audio / visual warning and automatically taking braking measures.

[0112] For example, if the risk probability value is less than 0.3, it is determined to be safe or low risk and no warning is triggered; if the risk probability value is greater than or equal to 0.3 and less than 0.7, an information prompt or sound / visual warning is triggered; if the risk probability value is greater than or equal to 0.7, an emergency sound / visual warning is triggered, and the pre-braking system or automatic emergency braking system can be linked.

[0113] The forward collision warning method provided in this application collects vehicle motion information, preceding vehicle motion information, and environmental information through multiple preset sensors; performs time and spatial alignment on the vehicle motion information, preceding vehicle motion information, and environmental information; determines the preceding vehicle motion state based on the preceding vehicle motion information and filters the preceding vehicle motion state; uses a dynamic Bayesian network to infer the vehicle motion information, preceding vehicle motion state, and environmental information to determine the risk probability value of a vehicle collision; and adopts a corresponding collision warning strategy for the vehicle based on the risk probability value, thereby improving the accuracy of forward collision warning.

[0114] Figure 2 This is a schematic diagram of the forward collision warning device provided in an embodiment of this application. Figure 2 As shown, the forward collision warning device 200 provided in this embodiment includes:

[0115] The acquisition module 201 is used to collect vehicle motion information, forward vehicle motion information, and environmental information through multiple preset sensors;

[0116] Preprocessing module 202 is used to perform time and space alignment on the vehicle's motion information, the preceding vehicle's motion information, and environmental information;

[0117] The preprocessing module 202 is also used to determine the motion state of the preceding vehicle based on the preceding vehicle motion information, and to filter the preceding vehicle motion state;

[0118] Processing module 203 is used to use a dynamic Bayesian network to reason about the vehicle's motion information, the motion state of the vehicle in front, and environmental information to determine the probability value of a vehicle collision.

[0119] The warning module 204 is used to take corresponding collision warning strategies for vehicles based on risk probability values.

[0120] The forward collision warning device provided in this embodiment can execute the forward collision warning method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0121] Figure 3 This is a schematic diagram of the structure of a forward collision warning device provided in an embodiment of this application. Figure 3 As shown in the embodiment of this application, the forward collision warning device 300 includes: a receiver 301, a transmitter 302, a processor 303, and a memory 304.

[0122] Receiver 301 is used to receive instructions and data;

[0123] Transmitter 302 is used to send commands and data;

[0124] Memory 304 is used to store computer-executed instructions;

[0125] The processor 303 is used to execute computer execution instructions stored in the memory 304 to implement the various steps of the forward collision warning method in the above embodiments. For details, please refer to the relevant descriptions in the foregoing embodiments of the forward collision warning method.

[0126] Optionally, the memory 304 can be either standalone or integrated with the processor 303.

[0127] When the memory 304 is set up independently, the electronic device also includes a bus for connecting the memory 304 and the processor 303.

[0128] This application also provides a computer storage medium storing computer execution instructions. When the processor executes the computer execution instructions, it implements the forward collision warning method as described above by the forward collision warning device.

[0129] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned forward collision warning method.

[0130] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0131] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0132] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A forward collision warning method, characterized in that, The method includes: The vehicle motion information, the motion information of the vehicle in front, and the environmental information are collected by multiple preset sensors. The vehicle motion information includes the vehicle speed, and the environmental information includes the road curvature. Perform time and space alignment on the vehicle's motion information, the motion information of the vehicle in front, and environmental information; The motion state of the vehicle in front is determined based on the motion information of the vehicle in front, and the motion state of the vehicle in front is filtered. The motion state of the vehicle in front includes: the acceleration of the vehicle in front. The vehicle speed is discretized to obtain the vehicle speed state, which is used to indicate the discrete level of the vehicle's current driving speed. The acceleration of the vehicle in front is discretized to obtain the acceleration state of the vehicle in front, which is used to indicate the discrete level of the acceleration of the vehicle in front. The road curvature is discretized to obtain the road curvature state, which is used to indicate the discrete level of the geometry of the current driving segment. Acquire historical driving behavior data and determine the driver's reaction time level based on the historical driving behavior data; Acquire driver eye images, determine driver fatigue state based on driver eye images, and the driver fatigue state is used to indicate the discrete level of the driver's real-time physiological state; An evidence set is constructed based on the vehicle's speed status, the acceleration status of the vehicle in front, the road curvature status, the driver's reaction time level, and the driver's fatigue status. The global joint probability distribution of the dynamic Bayesian network is determined based on the preset conditional probability table, and the joint probability of the collision risk probability and the evidence set is determined based on the global joint probability distribution. By iterating through all state values ​​of the collision risk probability, and summing the joint probability of the collision risk probability and the evidence set with respect to all state values ​​of the collision risk probability, the marginal probability of the evidence set is obtained. The posterior probability distribution of the collision risk probability is calculated based on the joint probability of the collision risk probability and the evidence set, and the marginal probability of the evidence set. The posterior probability distribution for calculating the collision risk probability satisfies the following formula: ； in, Let be the posterior probability distribution of the collision risk probability; It is the joint probability of collision risk probability and evidence set, obtained from the global joint probability distribution defined by dynamic Bayesian network by marginalizing non-query, non-evidence variables; It is the marginal probability of the evidence set, which is a normalization constant, obtained by adjusting the joint probability. Regarding the probability of collision risk The summation of all possible states is obtained; The posterior probability distribution of the calculated collision risk probability The numerical result is used as the probability value of a vehicle collision; The vehicle is given a corresponding collision warning strategy based on the risk probability value.

2. The method according to claim 1, characterized in that, The filtering of the forward vehicle's motion state includes: Obtain sensor state parameters, and determine a state decay factor based on the sensor state parameters. The state decay factor is used to indicate the degree of influence of the sensor's own state on the sensor performance. An environmental attenuation factor is determined based on environmental parameters, and the environmental attenuation factor is used to indicate the degree of influence of environmental conditions on sensor performance. A comprehensive attenuation factor is determined based on the state attenuation factor and the environment attenuation factor. The comprehensive attenuation factor is used to indicate the total degree of attenuation of the current sensor performance. The real-time confidence score of the sensor is calculated based on the comprehensive attenuation factor, and the measurement noise covariance of the sensor is adjusted based on the real-time confidence score. Construct a measurement noise covariance matrix based on the measurement noise covariance; The motion state of the vehicle in front is filtered based on the measured noise covariance matrix.

3. The method according to claim 1, characterized in that, The method of adopting corresponding collision warning strategies for vehicles based on risk probability values ​​includes: Determine whether the risk probability value is less than a first preset value; If so, then no warning will be triggered; If not, then determine whether the risk probability value is less than the second preset value; If so, a Level 1 warning will be issued to the vehicle, which includes: triggering information prompts or sound / visual warnings; If not, a level two warning is issued for the vehicle, which includes triggering an emergency audible / visual warning and automatically taking braking measures.

4. A forward collision warning device, characterized in that, The device includes: The acquisition module is used to collect vehicle motion information, forward vehicle motion information, and environmental information through multiple preset sensors. The vehicle motion information includes vehicle speed, and the environmental information includes road curvature. The preprocessing module is used to perform time and space alignment on the vehicle's motion information, the motion information of the vehicle in front, and the environmental information; The preprocessing module is also used to determine the motion state of the preceding vehicle based on the preceding vehicle motion information, and to filter the preceding vehicle motion state, wherein the preceding vehicle motion state includes: preceding vehicle acceleration; The processing module is used to discretize the vehicle speed to obtain the vehicle speed state, which is used to indicate the discrete level of the vehicle's current driving speed. The acceleration of the vehicle in front is discretized to obtain the acceleration state of the vehicle in front, which is used to indicate the discrete level of the acceleration of the vehicle in front. The road curvature is discretized to obtain the road curvature state, which is used to indicate the discrete level of the geometry of the current driving segment. Acquire historical driving behavior data and determine the driver's reaction time level based on the historical driving behavior data; Acquire driver eye images, determine driver fatigue state based on driver eye images, and the driver fatigue state is used to indicate the discrete level of the driver's real-time physiological state; An evidence set is constructed based on the vehicle's speed status, the acceleration status of the vehicle in front, the road curvature status, the driver's reaction time level, and the driver's fatigue status. The global joint probability distribution of the dynamic Bayesian network is determined based on the preset conditional probability table, and the joint probability of the collision risk probability and the evidence set is determined based on the global joint probability distribution. By iterating through all state values ​​of the collision risk probability, and summing the joint probability of the collision risk probability and the evidence set with respect to all state values ​​of the collision risk probability, the marginal probability of the evidence set is obtained. The posterior probability distribution of the collision risk probability is calculated based on the joint probability of the collision risk probability and the evidence set, and the marginal probability of the evidence set. The posterior probability distribution for calculating the collision risk probability satisfies the following formula: ； in, Let be the posterior probability distribution of the collision risk probability; It is the joint probability of collision risk probability and evidence set, obtained from the global joint probability distribution defined by dynamic Bayesian network by marginalizing non-query, non-evidence variables; It is the marginal probability of the evidence set, which is a normalization constant, obtained by adjusting the joint probability. Regarding the probability of collision risk The summation of all possible states is obtained; The posterior probability distribution of the calculated collision risk probability The numerical result is used as the probability value of a vehicle collision; The warning module is used to take corresponding collision warning strategies for vehicles based on risk probability values.

5. A forward collision warning device, characterized in that, The device includes: Memory; processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the forward collision warning method as described in any one of claims 1-3.

6. A computer storage medium, characterized in that, The computer storage medium stores computer execution instructions, which, when executed by a processor, are used to implement the forward collision warning method as described in any one of claims 1-3.

7. A computer program product, characterized in that, It includes a computer program, which, when executed by a processor, is used to implement the forward collision warning method as described in any one of claims 1-3.

Images