Road condition data collection and analysis method and system based on commercial vehicle network driving recorder

Abstract

This invention provides a method and system for road condition data acquisition and analysis based on a commercial vehicle connected driving recorder. The invention utilizes a connected driving recorder, a mandatory standard feature in commercial vehicles, to acquire road images in real time via an image acquisition module and perform preprocessing such as backlight compensation, distortion correction, noise reduction, and effective area cropping. An image recognition module employs a lightweight deep learning and rule-based algorithm to locally identify congestion and traffic accidents. A positioning module acquires latitude and longitude, vehicle speed, and timestamps via GPS / BeiDou dual-mode, achieving precise spatiotemporal binding through timestamp alignment and interpolation compensation. A 4G communication module uploads data using a strategy prioritizing event triggering and supplementing with periodic heartbeats. A data monitoring platform verifies the data, performs 3D deduplication, and then visualizes it on an electronic map. This invention requires no additional hardware, is low-cost, has wide coverage, and strong real-time performance, enabling automatic and precise monitoring of road conditions across the entire area, and is suitable for traffic management and logistics scheduling.

Description

Technical Field

[0001] This invention relates to the field of transportation and logistics, specifically to a method and system for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder. Background Technology

[0002] Current road condition data collection mainly relies on manual on-site photography and subsequent annotation and analysis, which has drawbacks such as poor real-time performance, low collection efficiency, high labor costs, and limited coverage.

[0003] Commercial vehicle driving recorders are standard hardware, characterized by nationwide operation, large numbers, and wide driving routes. Currently, there is no mature solution to use them as road condition collection terminals to achieve automated, large-scale, and low-latency road condition monitoring. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder, the specific steps of which are as follows: S1: The connected driving recorder captures images of the road in front of the vehicle in real time at a fixed frame rate and performs local image preprocessing. The driving recorder is installed in the driver's cab of the commercial vehicle.

[0006] The local image preprocessing specifically includes backlight compensation, distortion correction, noise reduction, and effective region cropping preprocessing.

[0007] First, backlight compensation is performed. By statistically analyzing the brightness of different regions of the image, bright areas are dynamically suppressed and dark details are enhanced. Local adaptive histogram equalization is used to enhance road and vehicle features and eliminate the effects of backlight, overexposure, and uneven brightness. The global average brightness formula is used to calculate the average brightness of the image and determine the exposure compensation coefficient. ; in This represents the average brightness of the entire image frame. Image width × height; coordinates Brightness value; Then, wide-angle lens distortion correction is performed. Based on the distortion coefficient matrix pre-calibrated by the equipment at the factory, the image is inversely transformed and mapped to correct the barrel distortion and edge curvature caused by the wide-angle lens, so that the lane lines and roads are restored to the true straight line shape. A general distortion correction formula using Zhang Zhengyou's calibration method; Radial distortion: ; ; Tangential distortion: ; ; in, Radial distortion coefficient; The tangential distortion coefficient; and These represent the x and y coordinates of any pixel in the original distorted image in the image coordinate system, that is, the pixel plane coordinates with the top left corner of the image as the origin. and This represents the corrected true geometric position of the pixel. For a point on the image The mathematical expression for the distance to the optical center (distortion center) of the image is: ;but This represents the square of the distance from the point to the optical center; Next, image denoising and sharpening are performed. Random noise is removed by Gaussian filtering, impulse interference is filtered out by median filtering, and motion-blurred images are appropriately sharpened to make the image clean, clear, and the target outline complete. Denoising is achieved using the Gaussian filtering formula: ; in, For the Gaussian kernel in coordinates Weighting coefficients at each location; These represent the horizontal and vertical offsets of pixels within the filter kernel relative to the kernel center, respectively. The standard deviation of the Gaussian filter is used to control the smoothing intensity. It is a natural constant; Pi is a constant. The square of the distance from the pixel to the center of the kernel; The larger the value, the stronger the filter, and the blurrier the image; The smaller the value, the weaker the filtering, but the more details are preserved. Finally, effective area cropping is performed to remove invalid areas such as the sky, roof, and hood, retaining only the core recognition area of ​​the road ahead, and outputting standardized image frames with hardware timestamps for subsequent road condition recognition.

[0008] Finally, after preprocessing, standard image frames are generated that are distortion-free, noise-free, have appropriate brightness, and contain only the effective road area, and are then stamped with a hardware timestamp. The data is then sent to step S2 for road condition recognition algorithm analysis.

[0009] S2: On the network-connected driving recorder side, the image is analyzed frame by frame through the road condition recognition algorithm to identify and confirm road congestion, vehicle traffic accident road condition events; The road condition recognition algorithm combines a lightweight deep learning detection model with a road condition rule judgment algorithm; the lightweight deep learning detection model is the YOLOv8n model.

[0010] After receiving the standardized image frame output in step S1, the connected driving recorder initializes the lightweight deep learning detection model deployed in the local NPU / MCU memory; simultaneously, it reads the hardware timestamp attached to the image frame. To ensure that all subsequent calculations are based on the same time reference, image features are extracted using deep learning algorithms to obtain underlying semantic information. The model extracts image features through convolutional layers and a feature pyramid network (FPN) to output multi-scale feature maps; Perform object detection and output the bounding box coordinates, class probability, and confidence score of all objects in the image; the detected objects include vehicles, lane lines, and road surface areas.

[0011] Based on the detection results output by deep learning, a road condition rule-based judgment algorithm is executed. The specific calculation process is as follows: To determine if there is congestion, calculate the vehicle occupancy density: ; in, This is the sum of the pixel areas of all detected vehicle bounding boxes in the current frame; The total area of ​​the effective road area in step S1 (i.e., the area after removing the sky and hood, which is a constant); The density of vehicles occupying the effective road area; Then, the vehicle speed is estimated by the target movement between consecutive frames; ; in, and For the current frame The pixel coordinates of the center point of the bounding box of a certain vehicle; and Previous frame The pixel coordinates of the center point of the corresponding vehicle; This refers to the frame interval time. The calibration constant for converting pixel coordinates to physical velocity; To estimate the speed of the vehicle ahead.

[0012] The steps for determining the characteristics of a traffic accident are as follows: For traffic accidents, a composite rule of geometry and state is established: Frame intersection-union ratio calculation: ; Among them, when When this occurs, it is determined to be a vehicle overlap / collision; The predicted bounding box (the detection box output by the model) is the rectangular region that the model believes the target is located in; The true bounding box is the rectangular area representing the actual location of the target manually labeled in the data annotation. This represents the area of ​​the intersection region between the predicted bounding box and the ground truth bounding box. The area of ​​the union of the predicted bounding box and the ground truth bounding box; Static determination: Calculate the displacement of a vehicle over N consecutive frames. If the sum of the squares of the displacements is less than a threshold, the vehicle is considered stationary. ; in, The static threshold; The number of consecutive frames is the length of the time window used to determine whether the vehicle is stationary. and They represent the first In the frame, the horizontal and vertical coordinates of the center point of the vehicle detection box; and They represent the first In the frame, the horizontal and vertical coordinates of the center point of the vehicle detection box; , ) is the first Within a frame, the vehicle's position in the image coordinate system (usually the center pixel coordinates of the detection box); The detection results from deep learning are logically ANDed with the calculated values ​​from the rule-based algorithm to confirm the final event. Congestion event confirmation logic: ; in, The density threshold is set to 0.4. The speed threshold is set to 10 km / h. This is the number of frames that continuously satisfy the above conditions; That is, when the density is greater than 0.4 and the speed is less than 10 km / h, and this state is maintained for 3 consecutive frames, it is confirmed as a valid congestion event.

[0013] Accident / Event Confirmation Logic: ; That is, if the frame intersection ratio is greater than 0.6 (collision) or meets one of the two conditions of "stationary and driving across lanes / parking across", and is valid for 3 consecutive frames, it is confirmed as a traffic accident; To determine the detection box Is the corresponding vehicle stationary? when < This explains If the vehicle's position changes very little within a frame, we determine that the vehicle is stationary (corresponding to...). = True); when ≥ This indicates that the vehicle's position has changed beyond the permissible range, and is therefore classified as a moving state (corresponding to...). =False); To determine the detection box Whether the corresponding vehicle crosses the lane (crosses the lane line / deviates from the lane).

[0014] After confirming valid traffic events, the algorithm generates and outputs the following structured data, including: Event type: ; for Traffic Jam The full name is (traffic congestion), which corresponds to the congestion event; for Accident The full name of the traffic accident is the corresponding accident event; and the valid timestamp of the road condition event is also included. = Confidence score .

[0015] S3: Obtains real-time vehicle latitude and longitude, speed and timestamp information through the positioning module, and accurately binds it with road condition events; Read the hardware timestamp output in step S1 With the valid timestamp of the road condition event generated in step S2 The recorder internally establishes a timestamp alignment queue to cache the location data and image frame timestamps within the most recent second in real time.

[0016] By calling the GPS / BeiDou dual-mode positioning module and reading the NMEA-0183 protocol data output by the positioning module through the SPI communication bus, the following 5 core physical quantities are obtained: longitude: ;latitude: Ground speed: UTC timestamp: And location status indicators: (0 indicates no location, 1-3 indicate valid location); Then, through precise alignment, the slight time difference between image acquisition, algorithm processing, and data transmission is eliminated. The specific steps are as follows: Define alignment time difference : ; Hardware timestamps for confirming valid road condition events in step S2; UTC timestamps obtained for the positioning module; The time difference between the time the event occurred and the time of location sampling; The judgment rules are as follows: like If the alignment is successful, the current location data will be used directly. like This triggers linear interpolation compensation. in, The maximum acceptable time difference set for the system.

[0017] When the time difference exceeds the threshold, interpolation is performed using location data from consecutive frames: ; ;

[0018] in, , This is the positioning data from the previous frame; , , Locate data for the current frame; The sampling period is for positioning. , , The latitude, longitude, and speed are compensated and perfectly matched to the moment of the event; By integrating the GPS / BeiDou dual-mode positioning module into the main unit of the connected driving recorder and soldering it onto the main control PCB board, the vehicle's latitude and longitude and time information are obtained and accurately bound to road condition events according to timestamps.

[0019] After alignment and compensation are completed, the system generates the final spatiotemporal binding data packet, which contains the image features of S1 and the event results of S2, and directly sends it to step S4 for uploading. .

[0020] S4: Through the 4G mobile communication network, road condition events, corresponding video clips, location and time data are synchronously uploaded to the data monitoring platform; S5: The data monitoring platform verifies, deduplicates, and parses the received data, and displays it visually on an electronic map to achieve real-time monitoring of road traffic.

[0021] Preferably, in step S3, latitude and longitude, vehicle speed, timestamp, and road condition events form an inseparable structured data packet, which is then uploaded to the data monitoring platform for accurate tracing.

[0022] Preferably, in step S4, the 4G network synchronous upload adopts a transmission strategy that prioritizes event triggering and is supplemented by periodic heartbeats; when a valid road condition event is identified, it is uploaded immediately, and when there is no event, the device online status is uploaded at a fixed period of 5 seconds / upload.

[0023] Preferably, in step S5, after the data monitoring platform receives data from multiple terminals, it performs three-dimensional deduplication according to latitude and longitude, event type, and occurrence time, retaining only one valid record for events of the same location, type, and time.

[0024] Preferably, a road condition data acquisition and analysis system based on a commercial vehicle connected driving recorder, the analysis system including a connected driving recorder and a data monitoring platform; The connected driving recorder is equipped with an image acquisition module, an image recognition module, a positioning module, a 4G communication module, and a data processing module. The image acquisition module is used to acquire real-time image data of the road ahead of the vehicle; the image recognition module has a built-in road condition recognition algorithm to analyze the image data locally and identify road condition events, including road congestion and vehicle traffic accidents; the positioning module is used to obtain the real-time geographical location of the vehicle; the 4G communication module is used to synchronously upload road condition events, real-time video footage, and positioning data to the data monitoring platform. The data monitoring platform is used to receive, store, and visualize real-time traffic information across the country.

[0025] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention fully utilizes the connected driving recorders of commercial vehicles, eliminating the need for additional hardware deployment and significantly reducing the equipment and manpower costs for road condition data collection; relying on the large-scale, multi-route operation of a vast number of commercial vehicles nationwide, it achieves full-domain coverage and seamless collection of road condition data, solving the problems of low efficiency and limited coverage in traditional manual collection. Simultaneously, image preprocessing and road condition recognition are completed locally on the device side, employing a lightweight YOLOv8n model combined with a rule-based judgment algorithm fusion scheme, balancing low computational power consumption and high recognition accuracy; it can quickly and accurately determine two core road condition events: congestion and traffic accidents, avoiding the delay of centralized cloud processing and achieving second-level recognition and response to road condition events. Furthermore, through GPS / BeiDou dual-mode positioning and precise timestamp alignment and linear interpolation compensation, it eliminates time deviations between images, algorithms, and positioning, achieving strong binding between road condition events and latitude, longitude, vehicle speed, and time, generating structured data packets, and providing reliable data support for road condition tracing and precise positioning. Finally, a 4G transmission strategy prioritizing event triggers and supplementing with periodic heartbeats is adopted. Key data is uploaded only when a valid event occurs, and only device status is uploaded when there is no event, which significantly reduces network transmission traffic and platform pressure. On the platform side, duplicate data is avoided by deduplicating data in three dimensions: latitude and longitude, event type, and time, thereby improving data processing and display efficiency. Attached Figure Description

[0026] Figure 1 This is a schematic diagram illustrating the specific process of the present invention. Detailed Implementation

[0027] To provide a further understanding of the purpose, structure, features, and functions of the present invention, detailed descriptions are provided below with reference to specific embodiments.

[0028] Please refer to the reference. Figure 1 This invention provides a method and system for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder.

[0029] Specific Implementation Example 1: A method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder.

[0030] S1: The connected driving recorder captures images of the road in front of the vehicle in real time at a fixed frame rate and performs local image preprocessing.

[0031] Once the connected driving recorder acquires a 1280×720 resolution road image, it immediately performs the following four real-time preprocessing steps in sequence in the local ISP image processor, as follows: S11: Backlight / High-Light Compensation Processing: Perform brightness region statistics on the current frame image and divide the image into bright areas (sky, direct sunlight), mid-gray areas (main road), and dark areas (shadows, under vehicle bodies). Brightness is suppressed in high-brightness areas to prevent overexposure and whitening; dark areas are moderately brightened to preserve details.

[0032] Local adaptive histogram equalization is used to enhance the contrast of key areas where roads, vehicles, and lane lines are located, making the target outlines clearly distinguishable. After processing, the images are free from overexposure and black shadows, regardless of backlighting, tunnel entrances and exits, or direct headlights at night, making them suitable for algorithm recognition.

[0033] S12: Wide-angle lens distortion correction processing: Call the distortion coefficient matrix (including radial distortion parameters and tangential distortion parameters) that was pre-stored at the factory of the recorder.

[0034] Perform inverse distortion mapping transformation on each pixel in the image to correct barrel distortion, edge curvature, and line offset caused by wide-angle lenses to a standard front view.

[0035] After correction, the originally curved lane lines and sloping road edges were restored to straight lines, and the position coordinates of vehicles and obstacles were consistent with the real road.

[0036] This ensures that subsequent road condition recognition algorithms will not misjudge location due to image distortion.

[0037] S13: Image denoising and sharpening enhancement processing: First, Gaussian filtering is applied to the image to smooth out and remove grain noise and high-frequency noise in nighttime and low-light environments.

[0038] Then, median filtering is performed to remove impulse interference and salt-and-pepper noise, while preserving hard edges such as vehicle edges, lane lines, and road surface contours from being blurred.

[0039] The image, which is slightly blurred due to the high-speed movement of the vehicle, is lightly sharpened and enhanced to strengthen the edges of the target and improve the clarity of features.

[0040] The processed image is clean, noise-free, and ghost-free, with clear target features.

[0041] S14: Effective area clipping processing: Based on the installation height and angle of the recorder, a preset fixed cropping window is used to crop a region of the image.

[0042] Vertically, the upper sky area and the lower hood / body area are removed; horizontally, the full lane width is retained.

[0043] After cropping, only the core areas that are effective for road condition identification, such as the road ahead, vehicles, road surface, and traffic facilities, are retained.

[0044] Reducing invalid pixel input lowers the computational load of the algorithm, improves recognition speed, and reduces false recognition.

[0045] Finally, after preprocessing, standard image frames with no distortion, no noise, appropriate brightness, and containing only effective road areas are generated, and hardware timestamps are added. These frames are then sent to step S2 for road condition recognition algorithm analysis.

[0046] S2: On the network-connected driving recorder side, the image is analyzed frame by frame through the road condition recognition algorithm to identify and confirm road congestion, vehicle traffic accident road condition events; After step S1 completes the preprocessing and outputs a standard image frame without distortion and containing an effective region, it is sent to the road condition recognition algorithm system in step S2.

[0047] The road condition recognition algorithm described in this invention adopts a solution that deeply integrates a lightweight deep learning detection algorithm with a road condition rule judgment algorithm.

[0048] First, the lightweight YOLOv8n deep learning model deployed on the recorder side is invoked to perform object detection on standard image frames, extract the bounding boxes and pixel coordinates of vehicles and lane lines, and calculate the vehicle density parameter Density.

[0049] Subsequently, the core calculation is performed using a rule-based judgment algorithm: through the formula: Estimate the speed of the vehicle ahead; calculate the overlap between vehicles using the IoU (Intersection over Union) formula.

[0050] Finally, a logical judgment is made: when the vehicle density (Density) ≥ 0.4 and the estimated vehicle speed (vest) ≤ 10 km / h is maintained for 3 consecutive frames, it is judged as a congestion event; when the vehicle bounding box IoU is greater than 0.6 or a stationary vehicle is detected crossing lanes, it is judged as a traffic accident. Through the above dual-algorithm mechanism, accurate identification of road condition events with low computing power and high anti-interference capability is achieved.

[0051] S3: Obtains real-time vehicle latitude and longitude, speed and timestamp information through the positioning module, and accurately binds it with road condition events; Read the hardware timestamp output in step S1 With the valid timestamp of the road condition event generated in step S2 The recorder internally establishes a timestamp alignment queue to cache the location data and image frame timestamps within the most recent second in real time.

[0052] By calling the GPS / BeiDou dual-mode positioning module and reading the NMEA-0183 protocol data output by the positioning module through the SPI communication bus, the following 5 core physical quantities are obtained: longitude: (Unit: degrees, range [−180, 180]); Latitude: (Unit: degrees, range [−90, 90]); Ground velocity: (Unit: km / h); UTC timestamp: And location status indicators: (0 indicates no location, 1-3 indicate valid location); Then, through precise alignment, the slight time difference between image acquisition, algorithm processing, and data transmission is eliminated. The specific steps are as follows: Define alignment time difference : ; Hardware timestamps for confirming valid road condition events in step S2; UTC timestamps obtained for the positioning module; The time difference between the time the event occurred and the time of location sampling; The judgment rules are as follows: like (This embodiment takes) If the time interval is 10ms, the alignment is considered successful, and the current positioning data is used directly. like Then linear interpolation compensation is triggered.

[0053] When the time difference exceeds the threshold, interpolation is performed using location data from consecutive frames: ;

[0054] in, , This is the positioning data from the previous frame; , , Locate data for the current frame; The sampling period is 1 second in this embodiment. , , The latitude, longitude, and speed are compensated and perfectly matched to the moment of the event; By integrating the GPS / BeiDou dual-mode positioning module into the main unit of the connected driving recorder and soldering it onto the main control PCB board, the vehicle's latitude and longitude and time information are obtained and accurately bound to road condition events according to timestamps.

[0055] After alignment and compensation are completed, the system generates the final spatiotemporal binding data packet, which contains the image features of S1 and the event results of S2, and directly sends it to step S4 for uploading. .

[0056] S4: Through the 4G mobile communication network, road condition events, corresponding video clips, location and time data are synchronously uploaded to the data monitoring platform; S5: The data monitoring platform verifies, deduplicates, and parses the received data, and displays it visually on an electronic map to achieve real-time monitoring of road traffic.

[0057] Example 2: A road condition data acquisition and analysis system based on a commercial vehicle connected driving recorder.

[0058] A road condition data acquisition and analysis system based on a commercial vehicle connected driving recorder. The analysis system includes a connected driving recorder and a data monitoring platform. The connected driving recorder is equipped with an image acquisition module, an image recognition module, a positioning module, a 4G communication module, and a data processing module. The image acquisition module is used to acquire real-time image data of the road ahead of the vehicle; the image recognition module has a built-in road condition recognition algorithm to analyze the image data locally and identify road condition events, including road congestion and vehicle traffic accidents; the positioning module is used to obtain the real-time geographical location of the vehicle; the 4G communication module is used to synchronously upload road condition events, real-time video footage, and positioning data to the data monitoring platform. The data monitoring platform is used to receive, store, and visualize real-time traffic information across the country.

[0059] This system mainly consists of two parts: a connected driving recorder and a data monitoring platform. The connected driving recorder integrates an image acquisition module, an image recognition module, a positioning module, a 4G communication module, and a data processing module. The image acquisition module is directly connected to the image recognition module via a hardware interface, transmitting pre-processed standard image frames to the image recognition module. The image recognition module interacts with the data processing module via an internal bus, outputting road condition event recognition results. The GPS / BeiDou dual-mode positioning module is soldered to the main control PCB board and directly connected to the data processing module via an SPI communication bus, transmitting positioning data such as latitude, longitude, vehicle speed, and timestamp. The data processing module serves as the core of the system. The control unit coordinates the work of each module, completes the spatiotemporal binding and structured processing of image data, event data, and location data, and then connects to the 4G communication module via a serial port or PCIe interface. The bound data packets are sent to the 4G communication module according to the specified transmission strategy. The 4G communication module establishes a wireless connection with the data monitoring platform through the 4G mobile communication network, uploading road condition events, video clips, location and time data to the platform. After the data monitoring platform completes data verification, deduplication and parsing, it realizes the visualization of road condition information on the electronic map. The whole process forms a complete data and control link from front-end acquisition, end-side identification, spatiotemporal fusion, wireless transmission to cloud application.

[0060] Freight trucks are equipped with standard connected driving recorders for commercial vehicles. These recorders capture real-time images of the road ahead while the vehicles are in motion. Algorithms identify road congestion and upload congestion events, real-time video, and location data to a monitoring platform via 4G. Management personnel can then view the road conditions in real time, enabling precise traffic control. Multiple commercial vehicles across the country form a data collection network, operating simultaneously on roads of different grades in different provinces, each uploading its own road condition information. The data monitoring platform aggregates the data to generate a nationwide real-time road condition heatmap, displaying the distribution of congested and accident-prone sections, providing decision-making support for traffic management departments and logistics companies.

[0061] The present invention has been described in the above-described embodiments; however, these embodiments are merely examples for implementing the present invention. It must be noted that the disclosed embodiments do not limit the scope of the present invention. Conversely, any modifications and refinements made without departing from the spirit and scope of the present invention are within the scope of patent protection of the present invention.

Claims

1. A method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder, characterized in that: The specific steps of the method are as follows: S1: The connected driving recorder captures images of the road in front of the vehicle in real time at a fixed frame rate and performs local image preprocessing. The driving recorder is installed in the driver's cab of the commercial vehicle; S2: On the network-connected driving recorder side, the image is analyzed frame by frame through the road condition recognition algorithm to identify and confirm road congestion, vehicle traffic accident road condition events; S3: Obtains real-time vehicle latitude and longitude, speed and timestamp information through the positioning module, and accurately binds it with road condition events; By integrating the GPS / BeiDou dual-mode positioning module into the main unit of the connected driving recorder and soldering it onto the main control PCB board, the vehicle's latitude and longitude and time information are obtained and accurately bound to road condition events according to timestamps. S4: Through the 4G mobile communication network, road condition events, corresponding video clips, location and time data are synchronously uploaded to the data monitoring platform; S5: The data monitoring platform verifies, deduplicates, and parses the received data, and displays it visually on an electronic map to achieve real-time monitoring of road traffic.

2. The method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder as described in claim 1, characterized in that: In step S1, the local image preprocessing specifically includes backlight compensation, distortion correction, noise reduction, and effective region cropping preprocessing.

3. The method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder as described in claim 1, characterized in that: In step S2, the road condition recognition algorithm combines a lightweight deep learning detection algorithm with a road condition rule judgment algorithm.

4. The method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder as described in claim 1, characterized in that: In step S3, latitude and longitude, vehicle speed, timestamp, and road condition events form an inseparable structured data packet, which is uploaded to the data monitoring platform for accurate tracing.

5. The method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder as described in claim 1, characterized in that: In step S4, the 4G network synchronous upload adopts a transmission strategy that prioritizes event triggering and is supplemented by periodic heartbeats; when a valid road condition event is identified, it is uploaded immediately, and when there is no event, the device online status is uploaded at a fixed period of 5 seconds / upload.

6. The method for collecting and analyzing road condition data based on a commercial vehicle connected driving recorder as described in claim 1, characterized in that: In step S5, after receiving data from multiple terminals, the data monitoring platform performs three-dimensional deduplication based on latitude and longitude, event type, and occurrence time. Only one valid record is retained for events of the same location, type, and similar time.

7. A road condition data acquisition and analysis system based on a commercial vehicle connected driving recorder, characterized in that: The analysis system includes a connected driving recorder and a data monitoring platform; The connected driving recorder is equipped with an image acquisition module, an image recognition module, a positioning module, a 4G communication module, and a data processing module. The image acquisition module is used to acquire real-time image data of the road ahead of the vehicle; the image recognition module has a built-in road condition recognition algorithm to analyze the image data locally and identify road condition events, including road congestion and vehicle traffic accidents; the positioning module is used to obtain the real-time geographical location of the vehicle. The 4G communication module is used to synchronously upload road condition events, real-time video footage, and location data to the data monitoring platform. The data monitoring platform is used to receive, store, and visualize real-time traffic information across the country.

Images