Traffic violation identifying and reporting method and device

Abstract

The invention provides a traffic violation identification and reporting method and device. The method comprises the following steps: determining a target working mode of a sensor assembly in a first vehicle according to the speed and / or battery capacity of the first vehicle; wherein the power consumption of the sensor assembly in different working modes and the data type of the acquired multi-dimensional sensing data are different; the sensor assembly is controlled to work in the target working mode so as to obtain multi-dimensional sensing data, and the multi-dimensional sensing data comprises environment information of the surrounding environment and / or state information of a second vehicle around the first vehicle; according to the multi-dimensional sensing data, traffic violation recognition is carried out on the second vehicle based on a violation recognition algorithm of multi-source data fusion; and reporting illegal information of the second vehicle to a traffic violation supervision platform under the condition of determining that the second vehicle has the traffic violation behavior.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traffic monitoring and intelligent traffic management, and in particular to a traffic violation identification and reporting method and device. BACKGROUND

[0002] Traffic violation identification technology is an important part of the intelligent traffic system, aiming to improve the efficiency and accuracy of traffic law enforcement through automated means. Current traffic violation monitoring mainly relies on fixed electronic police and manual patrol, which has problems such as limited monitoring range, poor real-time performance, and lack of privacy protection. For example, traditional electronic police cannot identify temporary road violations, and irrelevant vehicle information is easily leaked in the evidence chain.

[0003] In related technologies, a method is provided for identifying and reporting traffic violations based on vehicle-mounted sensors collecting surrounding environment and vehicle state information. However, this type of method usually collects data by setting a fixed sensor working mode, and then uses a pre-set identification model to judge the violation. This mode has the problem of balancing identification accuracy and resource consumption, and the identification accuracy is low in complex scenes such as bad weather. SUMMARY

[0004] The embodiments of the present application provide a traffic violation identification and reporting method and device.

[0005] In a first aspect, a traffic violation identification and reporting method is provided, applied to a first vehicle, the method comprising: determining a target working mode of a sensor component in the first vehicle according to a speed and / or battery power of the first vehicle; wherein the power consumption and the data type of the multi-dimensional sensor data collected by the sensor component in different working modes are different; controlling the sensor component to work in the target working mode to obtain the multi-dimensional sensor data, the multi-dimensional sensor data including environmental information of a surrounding environment and state information of a second vehicle around the first vehicle; identifying traffic violations of the second vehicle based on a multi-source data fusion violation identification algorithm according to the multi-dimensional sensor data; and reporting violation information of the second vehicle to a traffic violation supervision platform if it is determined that the second vehicle has a traffic violation.

[0006] According to the above technical means, by dynamically adjusting the working mode of the sensor component according to the speed and / or battery power of the first vehicle, the running efficiency is ensured while the whole vehicle energy consumption is reduced, and resource waste is avoided. At the same time, combined with the multi-source data fusion violation identification algorithm, the traffic violation of the second vehicle can be more accurately identified, and the identification accuracy is improved. Furthermore, by reporting the violation information to the platform, timely processing and supervision of the violation behavior are realized.

[0007] In some embodiments, the environmental information includes visibility, light intensity, and precipitation level, for describing a current weather scene; the state information of the second vehicle includes speeds of the second vehicle measured by multiple sensors in the sensor assembly respectively; the illegal identification algorithm based on multi-source data fusion, for identifying traffic violations of the second vehicle, includes: determining a weight of each sensor according to a measurement error of each sensor under the current weather scene; determining a fused speed value of the second vehicle according to the speed measured by each sensor and the corresponding weight; and determining that the second vehicle has a speeding traffic violation behavior in a case where the fused speed value is greater than a speed limit value of a current environment.

[0008] According to the above technical means, by assigning weights according to measurement errors of each sensor under different weather scenes, the influence of errors on the identification result can be effectively reduced, and the accuracy of the fused speed value can be improved; thus, when judging whether there is a speeding violation behavior, a more reliable determination can be made, especially in harsh weather conditions, while maintaining a high identification accuracy.

[0009] In some embodiments, the method further includes: obtaining true value speed data provided by a satellite navigation system based on an interval of a preset distance or a preset time; correcting a measurement error of a corresponding sensor under a current weather scene based on a deviation of the true value speed data and a measured speed of each sensor; and reducing a weight of a target sensor under a corresponding weather scene in a case where a number of times that a measurement error of the target sensor in the sensor assembly continuously exceeds a preset error threshold is greater than a first number threshold.

[0010] According to the above technical means, by introducing true value speed data as a reference, the measurement error model of each sensor can be further calibrated to ensure the scientificity and real-time of weight assignment; when a sensor continuously has a large error, the weight of the sensor is automatically adjusted to prevent misjudgment, thereby enhancing the robustness and reliability of the method.

[0011] According to the above technical means, by introducing true value speed data as a reference, the measurement error model of each sensor can be further calibrated to ensure the scientificity and real-time of weight assignment; when a sensor continuously has a large error, the weight of the sensor is automatically adjusted to prevent misjudgment, thereby enhancing the robustness and reliability of the method.

[0012] In some embodiments, the environmental information includes weather conditions, lane line clarity, and traffic flow density; the violation identification algorithm based on multi-source data fusion identifies traffic violations by the second vehicle, including: determining a first value for the weather condition node, a second value for the lane line clarity node, and a third value for the traffic flow density node in the Bayesian network decision model based on the weather conditions, the lane line clarity, and the traffic flow density, respectively; determining a fourth value for the lane change intention probability node based on the causal relationships and conditional probability table between nodes in the Bayesian network decision model, according to the first value, the second value, and the third value; determining a fifth value for the violation risk level node based on the fourth value; and determining that the second vehicle has committed a traffic violation of illegally changing lanes if the fifth value is greater than a preset risk threshold.

[0013] Based on the aforementioned technical means, by constructing a Bayesian network decision model and combining it with environmental factors such as weather, lane line clarity, and traffic flow, the lane-changing intention of the second vehicle and its violation risk level can be comprehensively evaluated, making the violation identification process more logical and scientific. Especially in complex traffic environments, it significantly improves the accuracy and rationality of identification.

[0014] In some embodiments, the method further includes: determining the definition and value range of each node in the Bayesian network decision model; constructing causal relationships between nodes, wherein the causal relationships include at least: the weather condition node affects the lane line clarity node, the traffic flow density node and the lane line clarity node jointly affect the lane change intention probability node, and the lane change intention probability node affects the violation risk level node; establishing a conditional probability table among the multiple nodes based on the degree of influence of different values ​​of the cause nodes of the multiple nodes in the Bayesian network decision model on different values ​​of the result nodes; and establishing the Bayesian network decision model according to the definition of each node, the causal relationships, and the conditional probability table.

[0015] Based on the aforementioned technical means, the structured modeling of the Bayesian network decision model, including node definition, causal relationship and conditional probability table establishment, realizes systematic reasoning of traffic violations, improves the transparency and interpretability of the identification process, and facilitates subsequent optimization and maintenance.

[0016] In some embodiments, the environmental information includes traffic light status information and stop line coordinates. The traffic light status information includes the color and remaining duration of the traffic light. The status information of the second vehicle includes continuously collected multi-frame motion data of the second vehicle. The traffic violation identification algorithm based on multi-source data fusion identifies traffic violations by the second vehicle, including: predicting the time when the second vehicle will reach the stop line based on the multi-frame motion data of the second vehicle; and determining that the second vehicle has committed a traffic violation of running a red light when the traffic light is red and the time when the second vehicle reaches the stop line is less than the remaining duration.

[0017] Based on the aforementioned technical means, by introducing traffic light status information and stop line coordinates, and by combining traffic light status with vehicle trajectory prediction, it is possible to identify red light running behavior in advance before the vehicle actually crosses the line. In some embodiments, the method further includes: upon determining that the second vehicle has committed a traffic violation by running a red light, recording a violation video using the sensor components, the violation video including at least timestamp information and the location information of the second vehicle; capturing multiple keyframe images of the second vehicle at its current location, at its position about to cross the stop line, and at the moment of crossing the stop line; and generating violation information of the second vehicle based on the violation video and the multiple keyframe images.

[0018] By recording and capturing keyframe images using the aforementioned technical means, the integrity and credibility of evidence of illegality are enhanced, which helps to facilitate the smooth progress of subsequent law enforcement procedures.

[0019] In some embodiments, the sensor assembly includes at least a forward-facing camera, a millimeter-wave radar, a lidar, and a surround-view camera; determining the target operating mode of the sensor assembly of the first vehicle based on the speed and / or battery charge of the first vehicle includes: determining the operating mode of the sensor assembly as a low-speed operating mode when the speed of the first vehicle is less than a first speed threshold and / or the battery charge is less than a first charge threshold; determining the operating mode of the sensor assembly as a high-speed operating mode when the speed of the first vehicle is greater than a second speed threshold and the battery charge is greater than a second charge threshold; wherein the first speed threshold is less than the second speed threshold, and the first charge threshold is less than the second charge threshold; in the low-speed operating mode, the forward-facing camera and the millimeter-wave radar are active, while the lidar and the surround-view camera are inactive; in the high-speed operating mode, all sensors in the sensor assembly are active.

[0020] Based on the above-mentioned technical means, by setting different speed and power thresholds to trigger different sensor working modes, energy consumption can be effectively controlled and the device's battery life can be extended while ensuring recognition performance. This is particularly suitable for energy-sensitive application scenarios such as new energy vehicles.

[0021] In some embodiments, before reporting the violation information of the second vehicle to the traffic violation monitoring platform, the method further includes: masking the identity information of the non-violation vehicle contained in the violation information and deleting the identity information of the first vehicle to obtain the processed violation information; performing integrity verification on the processed violation information; if the integrity verification passes, compressing and encrypting the violation information; and reporting the compressed and encrypted violation information to the traffic violation monitoring platform.

[0022] Based on the aforementioned technical means, a pre-emptive privacy protection mechanism ensures that illegal information uploaded to the regulatory platform does not contain irrelevant vehicle or personal identity information, complies with personal information protection regulations, and enhances user trust and system compliance.

[0023] Secondly, a traffic violation identification device is provided, applied to a first vehicle. The device includes: a first determining unit, configured to determine a target operating mode of a sensor component in the first vehicle based on the vehicle's speed and / or battery charge; wherein the power consumption and data type of the multidimensional sensing data collected by the sensor component differ in different operating modes; a control unit, configured to control the sensor component to operate in the target operating mode to acquire the multidimensional sensing data, the multidimensional sensing data including environmental information of the surrounding environment and state information of second vehicles surrounding the first vehicle; an identification unit, configured to identify traffic violations by the second vehicle based on the multidimensional sensing data and a multi-source data fusion-based violation identification algorithm; and a reporting unit, configured to report the violation information of the second vehicle to a traffic violation monitoring platform when it is determined that the second vehicle has committed a traffic violation. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the traffic violation identification and reporting method provided in the embodiments of this application. Figure 1 ; Figure 2 This is a flowchart illustrating a method for identifying traffic violations based on a multi-source data fusion algorithm. Figure 1 ; Figure 3 This is a flowchart illustrating the traffic violation identification and reporting method provided in the embodiments of this application. Figure 2 ; Figure 4This is a flowchart illustrating a method for identifying traffic violations based on a multi-source data fusion algorithm. Figure 2 ; Figure 5 This is a flowchart illustrating the traffic violation identification and reporting method provided in the embodiments of this application. Figure 3 ; Figure 6 This is a flowchart illustrating a method for identifying traffic violations based on a multi-source data fusion algorithm. Figure 3 ; Figure 7 This is a flowchart illustrating the traffic violation identification and reporting method provided in the embodiments of this application. Figure 4 ; Figure 8 This is a flowchart illustrating the traffic violation identification and reporting method provided in the embodiments of this application. Figure 5 ; Figure 9 This is a schematic diagram of the traffic violation recognition device provided in the embodiments of this application; Figure 10 This is a block diagram of a traffic violation identification and reporting device provided in another embodiment of this application. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of the application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0026] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0027] In the following description, the terms "first," "second," and "third" are used only to distinguish different objects and do not represent a specific order of objects, nor are they constituting a chronological order. It is understood that "first," "second," and "third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0029] To address the limitations of existing traffic violation identification systems, such as limited monitoring range, insufficient privacy protection, and low accuracy in complex scenarios, this application provides a traffic violation identification and reporting method and device based on an intelligent dual-mode switching mechanism, a multi-source data fusion algorithm, and privacy protection pre-processing. The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0030] Figure 1 This application provides a method for identifying and reporting traffic violations. This method can be executed by the vehicle's onboard system, which may include sensor components, intelligent driving controller, mode switching module, privacy processing module, and communication module. Figure 1 The method includes steps S110-S140.

[0031] In step S110, the target operating mode of the sensor components in the first vehicle is determined based on the speed of the first vehicle and / or the battery charge. The power consumption and the data type of the multidimensional sensing data collected by the sensor components differ in different operating modes.

[0032] In this embodiment, the target operating mode of the sensor components is determined based on the current operating state of the first vehicle, and the operating mode of the sensor components is dynamically adjusted to achieve a balance between resource consumption and recognition accuracy.

[0033] For example, in low-speed scenarios or when the battery is low, the system can switch to a low-speed or low-battery operating mode. In this mode, only the forward-facing camera and millimeter-wave radar can be activated to reduce energy consumption and focus on monitoring nearby traffic violations, such as illegal lane cutting. In high-speed scenarios, the LiDAR and 360° surround-view cameras can be activated; in this mode, the sensor's acquisition range is larger (e.g., the LiDAR and 360° surround-view cameras can cover up to 300 meters), effectively identifying whether a second vehicle is speeding or using the emergency lane.

[0034] In step S120, the sensor assembly is controlled to operate in the target operating mode to acquire multidimensional sensing data.

[0035] After determining the target operating mode of the sensor assembly according to the aforementioned steps, the controller operates in that mode to acquire multidimensional sensing data corresponding to that operating mode. In this embodiment, multidimensional sensing data refers to a data set collected by various types of sensors in the sensor assembly. For example, it may include environmental information such as weather conditions, light intensity, and precipitation levels, as well as state information of the second vehicle, such as its speed, position, and trajectory. It should be noted that the second vehicle is any one of several vehicles surrounding the first vehicle.

[0036] In step S130, based on multi-dimensional sensor data and a violation identification algorithm based on multi-source data fusion, traffic violations of the second vehicle are identified.

[0037] The traffic violation identification algorithm based on multi-source data fusion mentioned in this application refers to the use of data from multiple sensors to identify traffic violations. This method takes into account the measurement errors of different sensors under specific environments and assigns corresponding weights to each sensor, thereby improving the accuracy and robustness of the final identification result.

[0038] For example, in rainy or foggy weather conditions, visual sensors are more susceptible to interference and have larger errors. Therefore, the system assigns a lower weight to visual sensors when calculating their weights. In contrast, millimeter-wave radar and lidar are more stable in this environment, so their weights can be increased accordingly.

[0039] For example, in rainy conditions, the weight of the millimeter-wave radar can be set to approximately 0.56, the weight of the visual sensor to 0.14, and the weight of the lidar to 0.30. A more reliable recognition result can be obtained by combining these weight values.

[0040] In this embodiment, the traffic violation identification algorithm based on multi-source data fusion can identify traffic violations by the second vehicle, including determining whether the second vehicle has committed violations such as speeding, illegal lane changing, and running red lights. Specific methods for determining different types of violations will be described in detail later.

[0041] In step S140, if it is determined that the second vehicle has committed a traffic violation, the violation information of the second vehicle is reported to the traffic violation monitoring platform.

[0042] When a traffic violation is confirmed for a second vehicle, an evidence package containing a timestamp, geographic location information, vehicle identification, and type of violation can be generated and uploaded to the monitoring platform.

[0043] In some embodiments, the identity information of non-illegal vehicles needs to be anonymized before uploading the evidence package to avoid disclosing the identity information of other unrelated vehicles.

[0044] Based on the aforementioned technical means, by dynamically adjusting the working mode of the sensor components according to the speed and / or battery level of the first vehicle, the energy consumption of the entire vehicle can be reduced while ensuring operational efficiency, thus avoiding resource waste. At the same time, by combining a violation identification algorithm based on multi-source data fusion, the traffic violations of the second vehicle can be identified more accurately, improving the identification accuracy. Furthermore, by reporting the violation information to the platform, the timely processing and supervision of violations can be achieved.

[0045] In some embodiments, the aforementioned environmental information includes visibility, light intensity, and precipitation level, used to describe the current weather scenario.

[0046] Visibility refers to the distance at which a driver or other perception systems can clearly see objects ahead, and is an important indicator for judging whether a road is suitable for driving.

[0047] Illumination intensity refers to the brightness level of natural or artificial light sources in the environment, which has a significant impact on the image quality captured by the camera.

[0048] Precipitation levels indicate the amount of rainfall or snowfall per unit time, and are usually classified as light rain, moderate rain, heavy rain, and rainstorm, which have a direct impact on the detection accuracy of radar and lidar.

[0049] Weather scenarios refer to typical driving conditions formed by the combination of the above environmental parameters, such as sunny days, rainy or foggy days, strong sunlight, and nighttime. These weather scenarios determine the measurement error characteristics of different sensors and affect the weighting and final recognition results.

[0050] The status information of the second vehicle includes its speed, measured separately by multiple sensors in the sensor assembly. By using multiple sensors to collect speed information of the same target, redundant protection is created, reducing the risk of identification failure due to the failure of a single sensor.

[0051] In practical applications, when measuring based on multiple sensors, it is also necessary to dynamically adjust the weight of each sensor according to different weather scenarios. For example, in rainy weather, the camera may not be able to accurately identify the speed due to moisture obscuring it, and the weight of the camera should be reduced in this case.

[0052] See Figure 2 The traffic violation identification algorithm based on multi-source data fusion in the aforementioned step S130 further includes steps S210-S230 for identifying traffic violations by the second vehicle.

[0053] In step S210, the weight of each sensor is determined based on the measurement error of each sensor under the current weather scenario.

[0054] The sensor assembly may include multiple sensors such as forward-looking millimeter-wave radar, surround-view cameras, and lidar, each with its own unique speed measurement principle and applicable range. For example, millimeter-wave radar is suitable for all-weather high-speed speed measurement, but is susceptible to interference from metal reflections; cameras rely on image recognition, but their performance degrades in low-light conditions; lidar has high resolution and anti-interference capabilities, but is more expensive.

[0055] Therefore, the reliability of each sensor can be dynamically adjusted according to different weather scenarios. For example, in rain and fog scenarios, the measurement error of the camera may be large due to water vapor obstruction. In this case, the weight of the camera should be reduced and the weight of the data from radar sensors should be increased.

[0056] In the technical solution of this application embodiment, the weight of each sensor in the sensor assembly can be determined based on the inverse error method. Taking a sensor assembly with n sensors as an example, the weight of the i-th sensor is: (1) According to the above formula (1), the weight of the i-th sensor is the ratio of the reciprocal of the error of the i-th sensor to the sum of the reciprocals of the errors of all sensors.

[0057] For example, in the aforementioned rainy / foggy weather scenario, the measurement error of the millimeter-wave radar is 0.8 km / h, the speed measurement error based on the camera is 3.2 km / h, and the measurement error of the lidar is 1.5 km / h. According to equation (1), the weight of the millimeter-wave radar is 0.56, the weight of the camera is 0.14, and the weight of the lidar is 0.3.

[0058] In step S220, the fused speed value of the second vehicle is determined based on the speed measured by each sensor and the corresponding weight.

[0059] In this embodiment, the fused velocity value is calculated by multiplying the measured values ​​of each sensor by their respective weights and then summing the results to obtain a velocity estimate with higher overall reliability.

[0060] Taking the aforementioned rainy or foggy weather as an example, assuming the millimeter-wave radar measurement result is 64 km / h, the camera measurement result is 67.5 km / h, and the lidar measurement result is 65 km / h, then the speed estimate is: V = 64 * 0.56 + 67.5 * 0.14 + 65 * 0.3 = 64.8 km / h.

[0061] In step S230, if the fusion speed value is greater than the speed limit of the current environment, it is determined that the second vehicle has committed a traffic violation of speeding.

[0062] In the technical solution of this application, the speed limit value of the current environment can be obtained based on map navigation data. Alternatively, as a possible implementation, the speed limit value of the current environment can be obtained by collecting speed limit signs on the side of the road from the vehicle's camera. If the fused speed determined according to the aforementioned steps is greater than the speed limit value, it can be determined that the target vehicle has committed a speeding traffic violation.

[0063] Based on the above technical means, by assigning weights to the measurement errors of each sensor under different weather scenarios, the impact of errors on the recognition results can be effectively reduced, and the accuracy of the fusion speed value can be improved. Thus, when judging whether there is a speeding violation, a more reliable judgment can be made, especially maintaining a high recognition accuracy under adverse weather conditions.

[0064] See Figure 3 In some embodiments, the aforementioned traffic violation identification and reporting method further includes steps S310-S330.

[0065] In step S310, true velocity data provided by the satellite navigation system is obtained based on a preset distance or preset time.

[0066] Satellite navigation systems can provide high-precision position and velocity. Examples of such systems include the BeiDou Navigation Satellite System, GLONASS, Galileo, or the Global Positioning System.

[0067] In this embodiment of the application, a satellite navigation system can be used as a reference benchmark for measurement. The speed data it provides is called true speed data, which is the data that is closest to the actual speed under ideal conditions. It can be used to calibrate the speed measurement results of other sensors.

[0068] In practice, the satellite navigation system module can be called to read the vehicle's actual speed at preset intervals (e.g., every 100 meters) or preset intervals (e.g., every 5 seconds). This speed value is not affected by weather, lighting, or road conditions, and a stable and accurate speed reference can be obtained.

[0069] In step S320, based on the deviation between the true velocity data and the measured velocity of each sensor, the measurement error of the corresponding sensor under the current weather scenario is corrected.

[0070] Because sensors such as cameras, millimeter-wave radar, and lidar exhibit measurement errors under varying weather conditions (e.g., blurred visual sensors in rain and fog, attenuated millimeter-wave radar signals), this application introduces a dynamic error correction mechanism. Specifically, by comparing the true velocity data provided by the satellite navigation system with the velocities measured by each sensor in real time, the measurement deviation of each sensor is calculated, and the sensor error model is adjusted according to the current weather scenario (e.g., sunny, rainy, bright light, nighttime). By dynamically correcting the error based on the deviation between the true velocity data and the sensor measurements, the accuracy of velocity measurement can be effectively improved under adverse weather conditions, further enhancing the robustness and adaptability of the system.

[0071] In step S330, if the number of times the measurement error of the target sensor in the sensor assembly exceeds the preset error threshold is greater than the first threshold, the weight of the target sensor in the corresponding weather scenario is reduced.

[0072] To further optimize the effect of multi-source data fusion, this application introduces a dynamic sensor weight adjustment mechanism. When the measurement error of a certain sensor in a specific weather scenario exceeds a set error threshold multiple times consecutively, it is determined that the sensor's measurement performance in the current environment has deteriorated, thereby reducing the weight value of that sensor in the subsequent data fusion process.

[0073] For example, if a sensor's measurement error exceeds 5 km / h in three consecutive rainy / foggy weather conditions, the weight of that visual sensor in rainy / foggy scenarios will be reduced to decrease its impact on the final speed estimation. This mechanism of dynamically adjusting weights based on sensor performance can ensure the overall stability of the system while avoiding reliance on poorly performing sensors.

[0074] In some embodiments, the aforementioned environmental information includes weather conditions, lane line clarity, and traffic flow density.

[0075] Weather conditions, including clear skies, rain, fog, strong sunlight, and nighttime, are used to determine visibility and road surface adhesion; lane line clarity indicates whether road markings are clearly identifiable, directly affecting whether drivers can correctly judge lane boundaries; traffic flow density reflects the number of vehicles on the road and is used to determine whether there is congestion or smooth traffic.

[0076] By collecting the above three types of environmental information, the current road conditions can be assessed more comprehensively, thus helping to determine whether a second vehicle is likely to make an illegal lane change. For example, in rainy or foggy weather, reduced visibility can make lane lines unclear, which may cause drivers to misjudge their lane position and ultimately lead to illegal lane changes.

[0077] See Figure 4 The traffic violation identification algorithm based on multi-source data fusion in the aforementioned step S130 further includes steps S410-S430 for identifying traffic violations by the second vehicle.

[0078] In step S410, the first value of the weather condition node, the second value of the lane line clarity node, and the third value of the traffic flow density node in the Bayesian network decision model are determined based on the weather conditions, lane line clarity, and traffic flow density, respectively.

[0079] Bayesian network decision models are graphical models based on probabilistic reasoning, used to express causal relationships between variables and perform uncertain reasoning. A Bayesian network decision model consists of multiple nodes, each representing a variable, with directed edges representing causal relationships between nodes. In this application, the weather condition node, lane line clarity node, and traffic flow density node are the key input nodes in the Bayesian network decision model.

[0080] The first value represents the numerical state of the weather condition node at a specific moment. For example, sunny day can be defined as 1, rain / fog as 2, strong light as 3, and night as 4. The second value represents the state of the lane line clarity node, with clear as 1 and blurry as 2. The third value represents the state of the traffic flow density node, with smooth flow as 1 and congestion as 2.

[0081] By mapping environmental information to corresponding nodes in a Bayesian network decision model and assigning specific values ​​to these nodes, foundational data can be provided for subsequent causal inference. For example, in rainy or foggy weather (first value = 2), when lane lines are blurred (second value = 2), and during traffic congestion (third value = 2), a violation identification algorithm based on multi-source data fusion will consider the current road environment unfavorable for safe driving, thereby increasing its sensitivity to violations.

[0082] In step S420, based on the causal relationships and conditional probability tables between nodes in the Bayesian network decision model, the fourth value of the lane change intention probability node is determined according to the first value, the second value, and the third value.

[0083] In Bayesian network decision models, causal relationships refer to the logical dependencies between nodes. For example, weather conditions affect lane line clarity, while traffic flow density and lane line clarity together influence whether a driver intends to change lanes. Conditional probability tables are used to quantify the strength of causal relationships, that is, the probability that other nodes will take a specific value given that the values ​​of some nodes are known.

[0084] The lane change intention probability node is an intermediate node in the Bayesian network decision model, used to indicate whether the second vehicle has a strong intention to change lanes in the current environment. The fourth value is the probability value of the lane change intention probability node derived from the Bayesian network decision model under given conditions, usually ranging from 0 to 1, representing the strength of the lane change intention.

[0085] By substituting the first, second, and third values ​​into the Bayesian network decision model and combining causal relationships and conditional probability tables for reasoning, a fourth value for the lane change intention probability node can be derived. For example, in rainy / foggy weather (first value = 2), when lane lines are blurred (second value = 2), and during traffic congestion (third value = 2), the Bayesian network decision model might derive a lane change intention probability of 0.75, indicating a strong possibility of illegal lane changes.

[0086] Step 430: Determine the fifth value of the violation risk level node based on the fourth value. If the fifth value is greater than the preset risk threshold, it is determined that the second vehicle has committed a traffic violation of illegally changing lanes.

[0087] The violation risk level node is the output node in the Bayesian network decision model, used to comprehensively determine whether the second vehicle has engaged in illegal lane changing. The fifth value is the risk level derived from the inference by the violation risk level node, typically represented by high, medium, or low. When the fifth value is higher than a preset risk threshold (e.g., 0.8), the violation identification algorithm based on multi-source data fusion determines that the second vehicle has engaged in illegal lane changing.

[0088] By taking the fourth value of the lane change intention probability node as input and combining it with pre-defined risk level classification rules, the fifth value of the violation risk level node can be finally determined. For example, if the fourth value is 0.75, which is lower than the threshold of 0.8, it is judged as medium risk; if the fourth value is 0.85, which is higher than the threshold of 0.8, it is judged as high risk, and the second vehicle is marked as having engaged in illegal lane change behavior.

[0089] Based on the aforementioned technical means, by constructing a Bayesian network decision model and combining it with environmental factors such as weather, lane line clarity, and traffic flow, the lane-changing intention of the second vehicle and its violation risk level can be comprehensively evaluated, making the violation identification process more logical and scientific. Especially in complex traffic environments, it significantly improves the accuracy and rationality of identification.

[0090] See Figure 5 In some embodiments of this application, the aforementioned traffic violation identification and reporting method further includes steps S510-S540.

[0091] In step S510, the definition and value range of each node in the Bayesian network decision model are determined.

[0092] As mentioned earlier, Bayesian networks are graphical models based on probabilistic reasoning, used to express causal relationships between variables and to explain uncertainties.

[0093] The Bayesian network in this application consists of multiple nodes, each representing an influencing factor or state variable, such as weather condition nodes, lane line clarity nodes, traffic flow density nodes, lane change intention probability nodes, and violation risk level nodes.

[0094] Node definition refers to assigning a clear semantic meaning and quantifiable standard to each node. For example, the weather condition node may include four values: sunny, rainy / foggy, bright light, and nighttime; the lane line clarity node is divided into clear (≥70%) and blurry (<70%). By clearly defining the nodes and their value ranges, logical consistency and interpretability in the subsequent modeling process can be ensured. Table 1 below shows the definition of each node and its corresponding value range.

[0095] Table 1

[0096] In the step of defining the nodes and their value ranges in a Bayesian network decision model, the Bayesian network decision model is a higher-level concept, and its specific implementation can be a dynamic Bayesian network, a static Bayesian network, or a hybrid Bayesian network. Taking a dynamic Bayesian network as an example, a dynamic Bayesian network is suitable for scenarios that change over time and can simulate the dynamic evolution of the traffic environment; while a static Bayesian network is more suitable for reasoning and judgment in fixed scenarios.

[0097] In step S520, the causal relationships between the nodes are constructed.

[0098] The causal relationships in this application embodiment include: weather conditions node affects lane line clarity node, traffic flow density node and lane line clarity node jointly affect lane change intention probability node, and lane change intention probability node affects violation risk level node.

[0099] In Bayesian network models, causal relationships are used to describe the dependencies between different nodes. For example, in this application, the weather conditions node directly affects the lane line clarity node—when the weather conditions node is in a rainy or foggy state, lane lines are easily obscured, leading to a decrease in lane line clarity. Simultaneously, when the traffic flow density node is at a high value and the lane line clarity node is in a blurry state, drivers are more likely to have the intention to forcibly change lanes. The value of the lane change intention probability node further determines the risk level of whether it constitutes a violation. Table 2 below is an example of a causal relationship table.

[0100] Table 2

[0101] In step S530, a conditional probability table is established among multiple nodes based on the degree of influence of different values ​​of the cause nodes on different values ​​of the result nodes in the Bayesian network decision model.

[0102] Conditional probability tables are a key component of Bayesian networks, used to quantify the strength of causal relationships. For example, for the causal relationship between the weather condition node and the lane line clarity node, historical data statistics show that when the weather condition node is in a rainy or foggy state, the probability of the lane line clarity node being in a blurred state is 70%, and the probability of it being clear is 30%. Similarly, for the causal relationship between the traffic flow density node and the lane line clarity node and the lane change intention probability node, it can be set that when the traffic flow density node is in a high value state and the lane line clarity node is in a blurred state, the probability of the lane change intention probability node being high is 50%, medium is 30%, and low is 20%. By establishing such conditional probability tables, the system can quickly calculate the probability distribution of each node based on the current input conditions during actual operation and assess whether illegal behavior exists based on the conditional probability tables. Tables 3 and 4 are examples of conditional probability tables in this application.

[0103] Table 3

[0104] Table 4

[0105] In step S540, a Bayesian network decision model is established based on the definition of each node, causal relationships, and conditional probability tables.

[0106] This step integrates the aforementioned node definitions, causal relationships, and conditional probability tables into a complete Bayesian network model, forming a structured and logically clear reasoning system. This model can automatically perform probabilistic reasoning upon receiving multi-source sensor data and output a violation risk assessment result for a specific vehicle.

[0107] For example, when a vehicle is detected driving in rain or fog at the weather condition node, in a blurred state at the lane line clarity node, and in a high value at the traffic flow density node, the model will infer, based on existing causal relationships and conditional probability tables, that the vehicle has a high probability of intending to change lanes, and thus determine that the violation risk level node is high.

[0108] By constructing such a complete Bayesian network decision model, it is possible to efficiently and accurately identify and judge illegal behaviors in complex traffic scenarios, significantly improving the intelligence and automation level of illegal behavior identification.

[0109] Based on the aforementioned technical means, the structured modeling of the Bayesian network decision model, including node definition, causal relationship and conditional probability table establishment, realizes systematic reasoning of traffic violations, improves the transparency and interpretability of the identification process, and facilitates subsequent optimization and maintenance.

[0110] In some embodiments, the environmental information also includes traffic light status information and stop line coordinate information.

[0111] Traffic light status information is used to determine whether the current light is red, while image recognition algorithms are used to read how much time remains before the light turns green. Traffic light status information is typically captured by cameras and analyzed by image recognition algorithms; for example, color recognition determines the current light is red, and a countdown module reads the remaining time. Stop line coordinates are a virtual line set at the road intersection, indicating the position where vehicles should stop when the light is red. Stop line coordinates can be obtained using LiDAR or a high-precision map system and are used to calculate whether a vehicle has crossed the line.

[0112] See Figure 6 In some embodiments of this application, the traffic violation identification algorithm based on multi-source data fusion in the aforementioned step S130 for identifying traffic violations by the second vehicle also includes steps S610-S620.

[0113] In step S610, the time for the second vehicle to reach the stop line is predicted based on the multi-frame motion data of the second vehicle.

[0114] Multi-frame motion data refers to the continuous acquisition of the motion trajectory of a second vehicle over a period of time. Each frame of data includes information such as the position, speed, and direction of the second vehicle. By acquiring multi-frame motion data, a motion model of the second vehicle corresponding to the multi-frame motion data can be constructed.

[0115] For example, Table 5 shows multi-frame motion data of the second vehicle collected at 0.2s intervals before the current time t. In this example, the position coordinates of the stop line are x=110m. From the data in Table 5, it can be seen that the second vehicle is accelerating and moving towards the stop line.

[0116] Table 5

[0117] Based on the data in the table above, the LSTM model predicts 10 coordinate positions within the next two seconds (one every 0.2 seconds). According to the prediction mechanism, 1.0 second after the current moment, the predicted position coordinate of the second vehicle is x=109.8, which means that the second vehicle is about to reach the stop line. In other words, the LSTM model predicts that the second vehicle will reach the stop line at time t+1s.

[0118] In step S620, if the traffic light is red and the time it takes for the second vehicle to reach the stop line is less than the remaining time, it is determined that the second vehicle has committed a traffic violation of running a red light.

[0119] For example, if the aforementioned environmental information indicates that the traffic light is red and there is 1.5 seconds remaining, and the prediction in step S610 is that the second vehicle will arrive at the stop line in 1.0 second, then since the vehicle arrival time of 1.0 second is less than the remaining time of 1.5 seconds, it can be determined that the traffic light will still be red when the second vehicle arrives at the stop line.

[0120] Based on the aforementioned technical means, by introducing traffic light status information and stop line coordinates, and by combining traffic light status with vehicle trajectory prediction, it is possible to identify red light running behavior in advance before the vehicle actually crosses the line. In some embodiments, see Figure 7 The above-mentioned illegal identification and reporting method also includes steps S710-S730.

[0121] In step S710, a video of the violation is recorded using a sensor assembly. The video includes at least timestamp information and the location information of the second vehicle.

[0122] In this embodiment, the violation video is used to record the process of the violation by the second vehicle. The violation video includes timestamp information, i.e., the specific time when the violation occurred, and the location information of the second vehicle when the violation occurred. This information can ensure the integrity of the chain of evidence.

[0123] In step S710, multiple keyframe images of the second vehicle at its current position, the position about to cross the stop line, and the moment it crosses the stop line are acquired.

[0124] The aforementioned keyframe images can intuitively and completely reflect the entire process of the second vehicle running the red light, facilitating law enforcement agencies to conduct review and evidence collection.

[0125] In step S730, violation information of the second vehicle is generated based on the illegal video and multiple keyframe images.

[0126] Violation information is a structured description of a violation, including the type of violation, time, location, and relevant vehicle information. It is generated based on violation videos and keyframe images; analysis of this data ensures the objectivity and accuracy of the information.

[0127] By integrating illegal videos and keyframe images into a system, standardized violation information can be generated, which can help improve law enforcement efficiency and is more conducive to the construction of intelligent transportation systems.

[0128] By recording and capturing keyframe images using the aforementioned technical means, the integrity and credibility of evidence of illegality are enhanced, which helps to facilitate the smooth progress of subsequent law enforcement procedures.

[0129] In some embodiments, the sensor assembly includes a forward-facing camera, a millimeter-wave radar, a lidar, and a surround-view camera.

[0130] The aforementioned determination of the target operating mode of the sensor components in the first vehicle based on the speed and / or battery charge of the first vehicle further includes: If the speed of the first vehicle is less than a first speed threshold and / or the battery charge is less than a first charge threshold, the operating mode of the sensor assembly is determined to be a low-speed operating mode; if the speed of the first vehicle is greater than a second speed threshold and the battery charge is greater than a second charge threshold, the operating mode of the sensor assembly is determined to be a high-speed operating mode.

[0131] The aforementioned first speed threshold is typically set to a low value (e.g., 30 km / h). When the vehicle is traveling at low speeds in urban areas, the need for recognizing distant targets is lower, so only some sensors can be activated. The first battery level threshold is set below normal usage levels (e.g., 20%). When the battery level is below the first battery level threshold, priority is given to ensuring the operation of core functions, and some sensors are turned off to extend the power supply time.

[0132] In low-speed operation mode, the forward-facing camera and millimeter-wave radar are active, while the lidar and surround-view cameras are inactive. At this time, the sensor components retain only their core perception functions. The forward-facing camera is used to identify traffic conditions ahead, and the millimeter-wave radar is used to assist in ranging. Due to the high power consumption of the lidar and surround-view cameras, the system temporarily disables them in low-battery or low-speed scenarios to conserve energy and maintain the continuous operation of critical functions.

[0133] The second speed threshold is typically higher than the first speed threshold (e.g., 60 km / h). When the vehicle enters highways or other high-speed scenarios, the ability to identify distant targets is crucial, thus requiring the activation of LiDAR and surround-view cameras to ensure recognition accuracy. The second battery level threshold is usually set to a higher value (e.g., 80%). Setting the second battery level to a higher value indicates that the vehicle is fully charged, enabling the operation of all sensors to guarantee the system's comprehensive perception capabilities. In other words, in high-speed operating mode, all sensors in the sensor assembly are active.

[0134] Based on the above-mentioned technical means, by setting different speed and power thresholds to trigger different sensor working modes, energy consumption can be effectively controlled and the device's battery life can be extended while ensuring recognition performance. This is particularly suitable for energy-sensitive application scenarios such as new energy vehicles.

[0135] In some embodiments, see Figure 8 In the aforementioned step S140, if it is determined that the second vehicle has committed a traffic violation, the violation information of the second vehicle is reported to the traffic violation monitoring platform, which also includes steps S810-S840.

[0136] In step S810, the identity information of the non-violation vehicle contained in the violation information is masked, and the identity information of the first vehicle is deleted to obtain the processed violation information.

[0137] During the generation of illegal information, identity information of non-violation vehicles unrelated to the illegal activity may be collected, such as license plate numbers and owner names. Uploading this information directly without processing could lead to privacy leaks and violate relevant laws and regulations on personal information protection.

[0138] Therefore, this information needs to be masked before reporting, for example, by using mosaics, blurring, or replacing it with virtual identifiers, so that the identity information of non-illegal vehicles cannot be identified. At the same time, deleting the identity information of the first vehicle ensures its anonymity and prevents the owner of the first vehicle from suffering retaliation or other risks.

[0139] In step S820, the integrity of the processed illegal video is verified.

[0140] Integrity verification refers to checking processed illegal information to determine whether it meets the basic format and content requirements for uploading. This includes checking for the presence of necessary evidence data (such as videos, images, timestamps, and GPS coordinates), whether key fields are complete, and whether there are any missing or outlier values. Integrity verification can be achieved through validation algorithms, data structure comparison, and metadata checks. Only after passing integrity verification will processed illegal information be deemed legitimate and valid evidence, and then received by the platform for further processing.

[0141] In step S830, if the integrity verification passes, the illegal information is compressed and encrypted.

[0142] Once the illegal information passes integrity verification, it is compressed to reduce transmission bandwidth usage and storage space consumption. Compression methods can employ general standards such as ZIP and GZIP, or proprietary algorithms, with the appropriate compression strategy selected based on the data type.

[0143] Meanwhile, to prevent data from being tampered with or stolen during transmission, the compressed data must be encrypted, for example using encryption algorithms such as AES or RSA, to ensure that the data can only be decrypted and read by authorized parties. Implementing compression and encryption effectively improves data transmission efficiency, enhances data security and tamper resistance, and protects the authenticity and confidentiality of illegal information.

[0144] In step S840, the compressed and encrypted information is reported to the traffic violation monitoring platform.

[0145] After compression and encryption, the violation information is sent to the traffic violation monitoring platform according to a preset communication protocol (such as HTTPS, MQTT, etc.). The traffic violation monitoring platform usually has a dedicated interface to receive violation data from different sources and to perform subsequent review, classification, archiving, and enforcement processing.

[0146] Based on the aforementioned technical means, a pre-emptive privacy protection mechanism ensures that illegal information uploaded to the regulatory platform does not contain irrelevant vehicle or personal identity information, complies with personal information protection regulations, and enhances user trust and system compliance.

[0147] The above text combined Figures 1-8 The method embodiments of this application have been described in detail. The device embodiments of this application will now be described in detail with reference to the accompanying drawings. It should be understood that the description of the device embodiments corresponds to the method embodiments; therefore, any parts not described in detail can be referred to the method embodiments above.

[0148] Figure 9 This is a schematic structural diagram of the traffic violation recognition device 900 provided in the embodiments of this application. Figure 9 The traffic violation identification device 900 includes: The first determining unit 910 is used to determine the target operating mode of the sensor components in the first vehicle based on the speed of the first vehicle and / or the battery charge; wherein the power consumption and the data type of the multidimensional sensing data collected by the sensor components are different in different operating modes.

[0149] The control unit 920 is used to control the sensor assembly to operate in a target operating mode to acquire multidimensional sensing data, including environmental information of the surrounding environment and / or status information of a second vehicle around the first vehicle.

[0150] The identification unit 930 is used to identify traffic violations by the second vehicle based on multi-dimensional sensor data and a violation identification algorithm based on multi-source data fusion.

[0151] Reporting unit 940 is used to report the traffic violation information of the second vehicle to the traffic violation monitoring platform when it is determined that the second vehicle has committed a traffic violation.

[0152] In some embodiments, environmental information includes visibility, light intensity, and precipitation level to describe the current weather scenario; the status information of the second vehicle includes the speed of the second vehicle measured by multiple sensors in the sensor assembly. The identification unit 930 is further configured to: determine the weight of each sensor based on the measurement error of each sensor in the current weather scenario; determine the fused speed value of the second vehicle based on the speed measured by each sensor and its corresponding weight; and determine that the second vehicle has committed a speeding traffic violation if the fused speed value is greater than the speed limit of the current environment.

[0153] In some embodiments, the traffic violation identification device further includes: a true speed acquisition unit, configured to acquire true speed data provided by a satellite navigation system based on a preset distance or preset time interval; a correction unit, configured to correct the measurement error of the corresponding sensor in the current weather scenario based on the deviation between the true speed data and the measurement speed of each sensor; and an adjustment unit, configured to reduce the weight of the target sensor in the corresponding weather scenario if the number of times the measurement error of the target sensor in the sensor assembly continuously exceeds a preset error threshold is greater than the first threshold.

[0154] In some embodiments, environmental information includes weather conditions, lane line clarity, and traffic flow density; the identification unit 930 is further configured to: determine a first value for the weather condition node, a second value for the lane line clarity node, and a third value for the traffic flow density node in the Bayesian network decision model based on the weather conditions, lane line clarity, and traffic flow density, respectively; determine a fourth value for the lane change intention probability node based on the causal relationships and conditional probability table between nodes in the Bayesian network decision model, according to the first, second, and third values; determine a fifth value for the violation risk level node based on the fourth value; and determine that the second vehicle has committed a traffic violation of illegally changing lanes if the fifth value is greater than a preset risk threshold.

[0155] In some embodiments, the traffic violation identification device further includes: a second determining unit, used to determine the definition and value range of each node in the Bayesian network decision model; a causal relationship construction unit, used to construct causal relationships between nodes, the causal relationships including at least: weather condition nodes affecting lane line clarity nodes, traffic flow density nodes and lane line clarity nodes jointly affecting lane change intention probability nodes, and lane change intention probability nodes affecting violation risk level nodes; a conditional probability establishment unit, used to establish a conditional probability table between multiple nodes based on the degree of influence of different values ​​of cause nodes on different values ​​of result nodes in the Bayesian network decision model; and a network model establishment unit, used to establish a Bayesian network decision model based on the definition of each node, causal relationships, and conditional probability table.

[0156] In some embodiments, the environmental information includes traffic light status information and stop line coordinates. The traffic light status information includes the color and remaining duration of the traffic light, and the status information of the second vehicle includes continuously collected multi-frame motion data of the second vehicle. The identification unit 930 is further configured to: predict the time when the second vehicle will reach the stop line based on the multi-frame motion data of the second vehicle; and determine that the second vehicle has committed a traffic violation of running a red light if the traffic light is red and the time when the second vehicle reaches the stop line is less than the remaining duration.

[0157] In some embodiments, the traffic violation identification device further includes: a recording unit, configured to record a violation video using sensor components when it is determined that the second vehicle has committed a traffic violation by running a red light; the violation video includes at least timestamp information and the location information of the second vehicle; an image processing unit, configured to capture multiple keyframe images of the second vehicle at its current location, its position before crossing the stop line, and its moment of crossing the stop line; and a violation information generation unit, configured to generate violation information of the second vehicle based on the violation video and the multiple keyframe images.

[0158] In some embodiments, the sensor assembly includes at least a forward-facing camera, a millimeter-wave radar, a lidar, and a surround-view camera.

[0159] The first determining unit is further configured to: determine the operating mode of the sensor assembly as a low-speed operating mode when the speed of the first vehicle is less than a first speed threshold and / or the battery charge is less than a first charge threshold; and determine the operating mode of the sensor assembly as a high-speed operating mode when the speed of the first vehicle is greater than a second speed threshold and the battery charge is greater than a second charge threshold; wherein the first speed threshold is less than the second speed threshold and the first charge threshold is less than the second charge threshold; in the low-speed operating mode, the forward-facing camera and millimeter-wave radar are activated while the lidar and surround-view camera are deactivated; and in the high-speed operating mode, all sensors in the sensor assembly are activated.

[0160] In some embodiments, the traffic violation identification device further includes: a first processing unit, configured to mask the identity information of non-violation vehicles contained in the violation information and delete the identity information of the first vehicle to obtain processed violation information; a verification unit, configured to verify the integrity of the processed violation information; an encryption and compression unit, configured to compress and encrypt the violation information if the integrity verification passes; and an uploading unit, configured to report the compressed and encrypted violation information to the traffic violation supervision platform.

[0161] The following is combined Figure 10 The technical solution of this application will be further described. Figure 10This is a block diagram of an exemplary traffic violation identification and reporting device provided in this embodiment. The device is applied to a first vehicle. Figure 10 The traffic violation identification and reporting device 1000 includes: a data acquisition unit 1010, an intelligent driving controller 1020, a behavior recognition unit 1030, a privacy processing unit 1040, an evidence judgment unit 1050, and a data upload unit 1060.

[0162] The data acquisition unit 1010 further includes a camera assembly 1011, a millimeter-wave radar 1012, a lidar 1013, and an environmental data acquisition module 1040.

[0163] The camera assembly 1011 may include a forward-facing camera assembly and a surround-view camera assembly. The forward-facing camera is used to capture images of the area in front of the first vehicle, while the surround-view camera is used to capture images of the area surrounding the first vehicle. The images captured by the cameras may include features such as second vehicles, obstacles, signs and markings, and traffic lights around the first vehicle.

[0164] The millimeter-wave radar 1020 is used to transmit millimeter waves in a specific frequency band. Based on the received reflected waves, it calculates the time difference and frequency offset to determine the motion parameters of other objects around the vehicle. These motion parameters can be, for example, distance, azimuth angle, and relative speed.

[0165] The LiDAR 1013 is used to emit laser pulses to the surroundings through a laser emitter, receive reflected pulses and calculate distances to form a laser point cloud map, thereby accurately reconstructing the three-dimensional contours, positions and distances of objects in the vehicle's environment.

[0166] The environmental data acquisition unit 1014 is used to collect environmental information about the environment in which the first vehicle is located. For example, the environmental data acquisition unit 1014 may include a humidity sensor, a temperature sensor, a barometric pressure sensor, a rainfall sensor, a light sensor, etc. The data collected by the environmental data acquisition unit 1014 can accurately describe the weather conditions of the environment in which the first vehicle is located.

[0167] The intelligent driving controller 1020 is connected to the aforementioned data acquisition unit 1010 to receive data collected by various components in the data acquisition unit 1010 and execute corresponding control logic based on the received data. In this embodiment, one of the functions of the intelligent driving controller 1020 is to control the mode switching of the data acquisition unit. More specifically, the intelligent driving controller 1020 is used to determine the vehicle's current speed and battery level, and to determine the mode to be activated based on this information. In different operating modes, the power consumption of each sensor in the data acquisition unit 1010 and the data type of the multi-dimensional sensor data collected are different. For example, in low-speed scenarios or when the battery level is low, it can switch to a low-speed or low-battery operating mode. At this time, only the forward-facing camera and millimeter-wave radar can be activated to reduce energy consumption and focus on monitoring traffic violations at close range, such as illegal lane cutting. In high-speed scenarios, the lidar and 360° surround-view camera can be activated; at this time, the sensor's acquisition range is large (for example, the lidar and 360° surround-view camera can cover up to 300 meters), which can effectively identify whether a second vehicle is speeding or occupying the emergency lane.

[0168] The behavior recognition unit 1030 is connected to the intelligent driving controller 1020 and is used to identify traffic violations of surrounding second vehicles based on multi-dimensional sensor data collected by the data acquisition unit 1010 and a violation recognition algorithm based on multi-source data fusion.

[0169] Traffic violation identification algorithms based on multi-source data fusion refer to the use of data from multiple sensors to identify traffic violations. This method takes into account the measurement errors of different sensors under specific environments and assigns corresponding weights to each sensor, thereby improving the accuracy and robustness of the final identification results.

[0170] For example, in rainy or foggy weather conditions, visual sensors are more susceptible to interference and have larger errors. Therefore, the system assigns a lower weight to visual sensors when calculating their weights. In contrast, millimeter-wave radar and lidar are more stable in this environment, so their weights can be increased accordingly.

[0171] For example, in rainy conditions, the weight of the millimeter-wave radar can be set to approximately 0.56, the weight of the visual sensor to 0.14, and the weight of the lidar to 0.30. A more reliable recognition result can be obtained by combining these weight values.

[0172] In this embodiment, the traffic violation identification algorithm based on multi-source data fusion can be used to identify traffic violations by a second vehicle, including determining whether the second vehicle has committed violations such as speeding, illegal lane changing, or running a red light. The traffic violation identification method based on multi-source data fusion has been described in detail in the preceding method embodiments and will not be repeated here.

[0173] The privacy processing unit 1040 is connected to the intelligent driving controller 1030 and is used to mask the identity information of non-illegal vehicles contained in the illegal information and delete the identity information of the first vehicle to obtain the processed illegal information.

[0174] During the generation of illegal information, identity information of non-illegal vehicles unrelated to the illegal activity may be collected, such as license plate numbers and owner names. Uploading this information directly without processing could lead to privacy leaks and violate relevant laws and regulations on personal information protection. Therefore, this information needs to be masked before reporting, for example, by using mosaics, blurring, or replacing it with virtual identifiers, to make the identity information of non-illegal vehicles unidentifiable. Simultaneously, deleting the identity information of the primary vehicle ensures its anonymity and prevents its owner from suffering retaliation or other risks.

[0175] The evidence assessment unit 1050 includes a data processing module 1051 and a data storage module 1052. The data processing module 1051 is used to perform integrity verification on the processed illegal video, and, if the integrity verification passes, to compress and encrypt the illegal information.

[0176] Integrity verification refers to checking processed illegal information to determine whether it meets the basic format and content requirements for uploading. This includes checking for the presence of necessary evidence data (such as videos, images, timestamps, and GPS coordinates), whether key fields are complete, and whether there are any missing or outlier values. Integrity verification can be achieved through validation algorithms, data structure comparison, and metadata checks. Only after passing integrity verification will processed illegal information be deemed legitimate and valid evidence, and then received by the platform for further processing.

[0177] Once the illegal information passes integrity verification, it is compressed to reduce bandwidth consumption and storage space usage. Compression methods can employ common standards such as ZIP and GZIP, or specialized algorithms, selecting a suitable strategy based on the data type. Simultaneously, to prevent data tampering or theft during transmission, the compressed data must be encrypted, for example using AES or RSA encryption algorithms, ensuring that the data can only be decrypted and read by authorized parties. By implementing compression and encryption, data transmission efficiency can be effectively improved, data security and tamper resistance enhanced, and the authenticity and confidentiality of the illegal information guaranteed.

[0178] The data storage module 1052 is used to store the aforementioned illegal information that has been compressed and encrypted.

[0179] The data upload unit 1060 includes a communication module 1061, which establishes a communication connection with the traffic violation monitoring platform based on HTTPS, MQTT, etc. Violation information stored in the data storage module 1052 can be uploaded to the traffic violation monitoring platform through this communication module 1061.

[0180] This application provides a computer storage medium that stores one or more programs, which can be executed by one or more processors to implement the steps of the methods described in any of the above embodiments.

[0181] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0182] The aforementioned processor can be at least one of the following: application-specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field-programmable gate array (FPGA), central processing unit (CPU), controller, microcontroller, and microprocessor. It is understood that other electronic devices can also implement the functions of the aforementioned processor, and this application does not specifically limit the specific implementation.

[0183] The aforementioned computer storage media / memory can be read-only memory, programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD ROM), etc.

[0184] This application provides a computer program including computer-readable code. When the computer-readable code runs in an electronic device, the processor in the electronic device executes some or all of the steps in the above-described method.

[0185] This application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, it implements some or all of the steps in the above-described method. This computer program product can be implemented specifically through hardware, software, or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium; in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0186] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above steps / processes do not imply a sequential order of execution; the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above embodiments of this application are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0187] It should be noted that, in this document, 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. Unless otherwise specified, 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 that element.

[0188] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0189] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0190] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0191] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0192] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence or the part that contributes to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an in-vehicle terminal (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, magnetic disks, or optical disks.

[0193] The above are merely embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for identifying and reporting traffic violations, characterized in that, Applied to a first vehicle, the method includes: Based on the speed and / or battery level of the first vehicle, the target operating mode of the sensor components in the first vehicle is determined; wherein, the power consumption and the data type of the multidimensional sensing data collected by the sensor components are different in different operating modes. The sensor components are controlled to operate in the target operating mode to acquire the multidimensional sensing data, which includes environmental information of the surrounding environment and status information of the second vehicles around the first vehicle. Based on the multi-dimensional sensor data, and using a multi-source data fusion-based violation identification algorithm, traffic violations are identified in the second vehicle. If it is determined that the second vehicle has committed a traffic violation, the violation information of the second vehicle shall be reported to the traffic violation monitoring platform.

2. The method according to claim 1, characterized in that, The environmental information includes visibility, light intensity, and precipitation level, which are used to describe the current weather scenario; The status information of the second vehicle includes the speed of the second vehicle measured by multiple sensors in the sensor assembly; The traffic violation identification algorithm based on multi-source data fusion identifies traffic violations by the second vehicle, including: The weight of each sensor is determined based on the measurement error of each sensor under the current weather scenario. The fused speed value of the second vehicle is determined based on the speed measured by each sensor and the corresponding weight. If the fusion speed value is greater than the speed limit of the current environment, it is determined that the second vehicle has committed a traffic violation of speeding.

3. The method according to claim 2, characterized in that, The method further includes: Based on a preset interval distance or a preset interval time, obtain true velocity data provided by the satellite navigation system; Based on the deviation between the true velocity data and the measured velocity of each sensor, the measurement error of the corresponding sensor under the current weather scenario is corrected. If the number of times the measurement error of the target sensor in the sensor assembly exceeds a preset error threshold is greater than the first threshold, the weight of the target sensor in the corresponding weather scenario is reduced.

4. The method according to claim 1, characterized in that, The environmental information includes weather conditions, lane line clarity, and traffic flow density. The traffic violation identification algorithm based on multi-source data fusion identifies traffic violations by the second vehicle, including: The first value of the weather condition node, the second value of the lane line clarity node, and the third value of the traffic flow density node in the Bayesian network decision model are determined based on the weather conditions, the lane line clarity, and the traffic flow density, respectively. Based on the causal relationships and conditional probability table between nodes in the Bayesian network decision model, the fourth value of the lane change intention probability node is determined according to the first value, the second value, and the third value. The fifth value of the violation risk level node is determined based on the fourth value; If the fifth value is greater than the preset risk threshold, it is determined that the second vehicle has committed a traffic violation by illegally changing lanes.

5. The method according to claim 4, characterized in that, The method further includes: Determine the definition and value range of each node in the Bayesian network decision model; Construct causal relationships between nodes, including at least the following: the weather condition node affects the lane line clarity node; the traffic flow density node and the lane line clarity node jointly affect the lane change intention probability node; and the lane change intention probability node affects the violation risk level node. Based on the degree of influence of different values ​​of the cause nodes on different values ​​of the result nodes in the Bayesian network decision model, a conditional probability table is established among the multiple nodes. Based on the definitions of each node, the causal relationships, and the conditional probability table, the Bayesian network decision model is established.

6. The method according to claim 1, characterized in that, The environmental information includes traffic light status information and stop line coordinates. The traffic light status information includes the color and remaining duration of the traffic light. The status information of the second vehicle includes multi-frame motion data of the second vehicle collected continuously. The traffic violation identification algorithm based on multi-source data fusion identifies traffic violations by the second vehicle, including: Based on the multi-frame motion data of the second vehicle, predict the time when the second vehicle will reach the stop line; If the traffic light is red and the second vehicle reaches the stop line in less than the remaining time, it is determined that the second vehicle has committed a traffic violation by running a red light.

7. The method according to claim 6, characterized in that, The method further includes: If it is determined that the second vehicle has committed a traffic violation by running a red light, the sensor assembly is used to record a video of the violation, the video of the violation including at least timestamp information and the location information of the second vehicle; Capture multiple keyframe images of the second vehicle at its current position, at the position where it is about to cross the stop line, and at the moment it crosses the stop line; Based on the illegal video and the multiple keyframe images, the illegal information of the second vehicle is generated.

8. The method according to any one of claims 1-7, characterized in that, The sensor assembly includes at least a forward-facing camera, a millimeter-wave radar, a lidar, and a surround-view camera; Determining the target operating mode of the sensor components of the first vehicle based on the speed and / or battery charge of the first vehicle includes: If the speed of the first vehicle is less than a first speed threshold and / or the battery charge is less than a first charge threshold, the sensor assembly is determined to operate in a low-speed mode. If the speed of the first vehicle is greater than the second speed threshold and the battery charge is greater than the second charge threshold, the sensor component is determined to operate in a high-speed mode. Wherein, the first speed threshold is less than the second speed threshold, and the first power threshold is less than the second power threshold; In the low-speed operating mode, the forward-facing camera and the millimeter-wave radar are activated, while the lidar and the surround-view camera are deactivated. In the high-speed operating mode, all sensors in the sensor assembly are activated.

9. The method according to any one of claims 1-7, characterized in that, Before reporting the violation information of the second vehicle to the traffic violation monitoring platform, the method further includes: The identity information of the non-violation vehicle contained in the violation information is masked, and the identity information of the first vehicle is deleted to obtain the processed violation information; The integrity of the processed illegal information is verified. If the integrity verification passes, the processed illegal information is compressed and encrypted. The compressed and encrypted violation information will be reported to the traffic violation monitoring platform.

10. A traffic violation identification device, characterized in that, Applied to a first vehicle, the device includes: The first determining unit is configured to determine the target operating mode of the sensor components in the first vehicle based on the speed and / or battery charge of the first vehicle; wherein the power consumption and the data type of the multidimensional sensing data collected by the sensor components are different in different operating modes. A control unit is used to control the sensor components to operate in the target operating mode in order to acquire the multidimensional sensing data, which includes environmental information of the surrounding environment and status information of the second vehicles around the first vehicle. The identification unit is used to identify traffic violations by the second vehicle based on the multi-dimensional sensor data and a multi-source data fusion violation identification algorithm. The reporting unit is used to report the traffic violation information of the second vehicle to the traffic violation monitoring platform when it is determined that the second vehicle has committed a traffic violation.

Images