Road surface and dirty state detection method based on laser radar

Abstract

The invention discloses a road surface and dirty state detection method based on a laser radar, and the method comprises the following steps: scanning a road surface through the laser radar, obtaining a visual image and environment data of the road surface through a sensor, carrying out the fusion processing of the point cloud data of the laser radar and the data of the sensor, and generating a three-dimensional data set; (2) filtering, denoising, aligning and normalizing the point cloud data by using a computer vision technology; performing feature extraction on the data by using a computer vision algorithm; (3) according to the identified surface features, identifying an abnormal region by using a deep learning algorithm to judge whether the road surface is dirty or not; (4) if smudginess exists, further analyzing the type and the distribution condition of the smudginess; and (5) sending a detection result to a central control system through a wireless communication module. The geometrical morphology of the pavement can be accurately measured, and meanwhile the cleanliness of the pavement is evaluated.

Description

Technical Field

[0001] The present invention relates to the field of automatic driving assistance technology, and in particular to a method for detecting road surface and dirtiness based on laser radar. Background Art

[0002] Currently, road quality inspection methods primarily include manual inspection and on-board inspection systems. Manual inspection is inefficient and lacks safety. While on-board inspection systems offer improved automation, they are costly, have poor equipment integration, and require significant time for data processing, making them unable to fully meet the current needs of road maintenance inspections. With the advancement of autonomous driving technology, the demand for precise road condition monitoring is increasing, necessitating a high-resolution, fast-response inspection technology that is unaffected by environmental factors. Summary of the Invention

[0003] The present invention aims to provide a method for detecting road surface and dirtiness based on LiDAR (LiDAR), which can accurately measure road geometry and assess road cleanliness. By optimizing LiDAR's data processing algorithm, the present invention achieves rapid and accurate detection of road surface conditions, adapting to varying environmental conditions.

[0004] In one aspect of the present invention, a method for detecting road surface and dirtiness based on laser radar is provided. According to an embodiment of the present invention, the method comprises the following steps:

[0005] (1) Use LiDAR to scan the road surface, use the sensor to obtain visual images and environmental data of the road surface, and fuse the LiDAR point cloud data with data from other sensors to generate a complete three-dimensional data set;

[0006] (2) Using computer vision technology to filter, denoise, align, and normalize the point cloud data to eliminate noise and errors; then using computer vision algorithms to extract features from the data, identify the surface features of the road surface, analyze the height changes, surface flatness, and geometry of the road surface data, and identify the physical characteristics of the road surface and potential dirty areas;

[0007] (3) Determine whether the road surface is dirty based on the identified surface features;

[0008] (4) If there is dirt, further analyze the type and distribution of dirt;

[0009] (5) The detection results are sent to the central control system through the wireless communication module.

[0010] In addition, the method for detecting road surface and dirtiness based on laser radar according to the above embodiment of the present invention may also have the following additional technical features:

[0011] In some embodiments of the present invention, in step (1), the laser radar used has a scanning range of at least 360 degrees.

[0012] In some embodiments of the present invention, in step (2), the computer vision technology includes a deep learning model.

[0013] In some embodiments of the present invention, in step (3), if there is no difference between the surface characteristics of the road surface and the preset clean road surface characteristics, then the road surface is not dirty; otherwise, the road surface is dirty.

[0014] In some embodiments of the present invention, in step (4), the types of dirt are classified according to physical properties, sources, and impacts. According to physical properties, the dirt is divided into viscous dirt, granular dirt, and solid dirt; according to source, the dirt is divided into vehicle leakage dirt, natural environment dirt, and construction activity dirt; according to impact, the dirt is divided into dirt that affects the friction coefficient of the road surface, dirt that affects the smoothness and appearance of the road surface, and dirt that affects the environment.

[0015] In some embodiments of the present invention, in step (5), the wireless communication module supports at least one communication protocol.

[0016] In another aspect, the present invention provides a road surface and dirt condition detection system based on laser radar. According to an embodiment of the present invention, the system includes:

[0017] At least one detection unit, configured to execute step (1) of the method for detecting a road surface and dirtiness based on a laser radar;

[0018] At least one image processing unit, configured to execute step (2) of the method for detecting road surface and dirtiness based on laser radar;

[0019] At least one analysis unit, configured to execute steps (3) and (4) of the method for detecting road surface and dirtiness based on laser radar;

[0020] At least one communication module is used to execute step (5) of the road surface and dirt status detection method based on laser radar.

[0021] In addition, the road surface and dirt condition detection system based on laser radar according to the above embodiment of the present invention may also have the following additional technical features:

[0022] In some embodiments of the present invention, the lidar scanning unit includes multiple laser transmitters and receivers for omnidirectional scanning of at least 360 degrees.

[0023] In some embodiments of the present invention, the image processing unit includes a deep learning processor for performing image analysis tasks.

[0024] In another aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is configured to implement a method for detecting a road surface and dirtiness based on a laser radar when executed, according to an embodiment of the present invention.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] 1) High Precision: LiDAR has high-precision ranging capabilities and can provide accurate road surface morphology data;

[0027] 2) Fast response: LiDAR has a fast response speed and is suitable for real-time road condition monitoring;

[0028] 3) Strong environmental adaptability: Optimizing data processing through ambient light, temperature and humidity sensors improves the adaptability of the system;

[0029] 4) Wide range of applications: Applicable to many fields such as road maintenance, intelligent transportation systems and autonomous vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Schematic diagram of a road surface and dirt condition detection system based on laser radar in Example 1 of the present invention;

[0031] Figure 2 This is a flow chart of a method for detecting road surface and dirtiness based on laser radar in Example 2 of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0033] Example 1

[0034] See also Figure 1 As shown, a road surface and dirt status detection system based on laser radar includes a detection unit, an image processing unit, an analysis unit, a communication module, and a control and power supply module.

[0035] (1) The detection unit includes a lidar module, a sensor module, and an image acquisition module.

[0036] The LiDAR module consists of a LiDAR transmitter and a LiDAR receiver. These are installed on a vehicle or a fixed detection station. The LiDAR transmitter emits a laser beam toward the road, while the LiDAR receiver receives the reflected signal. By emitting laser beams and receiving the reflected signals, the LiDAR transmitter and LiDAR receiver measure the distance between the mobile platform and surrounding objects, providing the mobile platform with environmental awareness capabilities for obstacle detection, map building, and other purposes. While the mobile platform is in operation, the LiDAR continuously emits laser beams. When the laser beam hits an object, it reflects back and is received by the receiver. By measuring the laser signal's time of flight or phase difference, for example, the distance to surrounding objects can be calculated. This distance data is then transmitted to the image processing unit for further analysis and processing.

[0037] Sensor modules include ambient light sensors, temperature and humidity sensors, spectral sensors, infrared sensors, and contact sensors. Ambient light sensors are typically based on the photoelectric effect, using photosensors to detect light intensity and provide road surface data. Upon absorbing photons, photosensors generate a current proportional to the light intensity. For example, the resistance of a photoresistor changes with light intensity, and the current generated by a photodiode also changes accordingly. These changes are amplified and converted digitally to analog, then processed by an intelligent chip to monitor ambient light intensity. Light sensors are used to detect ambient light intensity and adjust the mobile platform's lighting system or optimize image acquisition based on lighting conditions. Environmental sensors can comprehensively monitor multiple environmental parameters, such as air quality and air pressure, providing more comprehensive environmental information for mobile platform operations and decision-making. Temperature and humidity sensors combine humidity and temperature sensors to measure ambient temperature and humidity, providing road surface temperature and humidity data to assess the suitability of the equipment's operating environment and provide reference for specific environmental parameters for specific tasks. Moisture-sensing components, such as humidity resistors or humidity-sensitive capacitors, will absorb moisture from the air, causing their resistance or capacitance to change. Temperature-sensing components, such as thermocouples, thermistors, or platinum resistors, will change their resistance with temperature. These changes are converted into electrical signals, processed, and output. These sensors continuously monitor environmental parameters and transmit data to the control system of the mobile platform, which can make corresponding adjustments based on this data. Spectral sensors can detect the spectral reflectance characteristics of the road surface to infer the material composition and possible contamination of the surface; infrared sensors can detect potential problems caused by temperature differences, such as oil or other pollutants, by detecting the temperature distribution on the road surface; contact sensors can directly measure the friction characteristics of the road surface, which may be affected by dirt.

[0038] The image acquisition module includes a camera or other image acquisition equipment, which can obtain images of the road surface through a high-resolution camera to observe surface features such as stains and cracks.

[0039] (2) The image processing unit (IPU) is used to process data from the detection unit. The specific methods include: first, converting the radar data into a format suitable for processing by the image processing unit, such as converting the radar data into a radar image so that it can be analyzed using image processing technology. For lidar data, the IPU can perform point cloud processing and three-dimensional reconstruction to generate a detailed environmental model, which requires adapting specific algorithms to analyze the intensity and frequency characteristics of the radar signal and requires additional software or hardware support. The IPU can also fuse sensor data with image data to correct and compensate for data deviations. In some applications, the IPU works in conjunction with a dedicated radar data processing unit to comprehensively process image data and radar data to provide more comprehensive information. The IPU's processing of sensor data includes converting the raw sensor data into a format suitable for image analysis, such as visualizing the data as an image. Through data fusion, the sensor data is aligned with the image data to enhance the comprehensive information. Then, image processing technologies such as edge detection and feature extraction are applied to analyze these images. To improve data quality, the IPU also performs data enhancement and correction. The IPU is used to filter, denoise, align and normalize the point cloud data to eliminate noise and errors and ensure data quality. The IPU uses features extracted from the data (such as edge detection, texture analysis, and deep learning models) to identify surface features such as cracks, potholes, and stains. It also analyzes road surface data for height variations, surface flatness, and geometry to identify physical features and potentially contaminated areas. In short, the IPU converts and fuses sensor data, radar data, and image data to generate optimized image information, supporting decisions such as navigation planning, obstacle avoidance, and mission execution.

[0040] (3) The analysis unit includes a data processor, which is used to analyze the surface characteristics of the road surface, determine whether there is dirt on the road surface, and analyze the type and distribution of dirt. The image processing unit then provides important input to the analysis unit by converting sensor data, radar data, and image data into optimized visual information. The IPU first converts radar data and sensor data into radar images and visual images for detailed analysis. The lidar data undergoes point cloud processing and three-dimensional reconstruction to generate an environmental model. The IPU also fuses and corrects this data to improve the accuracy of the information. The analysis unit receives this processed data and performs further optimization, including data cleaning, format conversion, and data integration, extracting key information from multiple sensors, and performing feature selection and temperature compensation. Finally, the analysis unit uses this optimized data to analyze road surface characteristics, such as determining the presence, type, and distribution of dirt, thereby supporting decisions such as navigation planning, obstacle avoidance, and task execution.

[0041] The data optimization process of the analysis unit goes through the following steps: First, the data after IPU processing is cleaned to remove noise, missing values ​​and outliers to improve data quality, and then the raw data is converted into a format suitable for analysis. If the data comes from multiple sensors or data sources, they need to be integrated into a unified data set to improve the accuracy of the analysis. It is worth noting that in feature selection, the class feature selection technology in the relevant database is used to select the most important features, reduce the size of the data set, and retain the most critical information. Since the performance of the ambient light sensor and the temperature and humidity sensor may be affected by temperature changes, temperature compensation is considered. The deviation caused by temperature can be corrected by adding a temperature compensation circuit to the system or using an algorithm.

[0042] (4) The communication module sends the detection results to the central control system through the wireless communication module, analyzes the reflection intensity, angle and frequency of the signal, and calculates the geometric shape of the road surface.

[0043] (5) The control and power module is used to control the operation of each component.

[0044] Example 2

[0045] like Figure 2 As shown, a method for detecting road surface and dirtiness based on laser radar includes the following steps:

[0046] (1) Use LiDAR to scan the road surface, and combine multiple sensors and image acquisition mechanisms to obtain three-dimensional data of the road surface. The three types of data obtained - LiDAR data, sensor data and image data - undergo a specific processing process to extract useful road surface information. The three-dimensional point cloud data generated by the LiDAR module is first processed and three-dimensionally reconstructed by the image processing unit (IPU) to form a detailed environmental model. These data provide geometric information of the road surface, including height, slope and flatness, providing a basis for further analysis. The ambient light, temperature and humidity, spectrum, infrared and contact sensor data collected by the sensor module are converted into a format suitable for the image processing unit. These data are combined with LiDAR data and image data through data fusion technology to generate comprehensive environmental information. For example, the data of the ambient light sensor and temperature and humidity sensor are used to adjust the image acquisition settings and correct environmental factors in the image, while the data of the spectrum and infrared sensor help identify the surface material and pollution. The data of the contact sensor provides direct measurement results on the friction characteristics of the road surface. The high-resolution image data obtained by the image acquisition module is analyzed by image processing technologies such as edge detection and feature extraction to identify surface features such as stains and cracks on the road surface. The IPU also fuses, corrects, and enhances all sensor and image data, eliminating data bias and improving information accuracy. Ultimately, this processed data is fed into the analysis unit for further analysis, such as determining the presence, type, and distribution of contamination. This supports decisions such as navigation planning, obstacle avoidance, and mission execution. This multi-layered processing allows the system to comprehensively consider different types of data and provide a comprehensive understanding of road conditions.

[0047] Autonomous vehicles are equipped with high-resolution LiDAR transmitters and receivers, ambient light sensors, temperature and humidity sensors, cameras, spectral sensors, infrared sensors, and contact sensors. These transmitters and receivers are capable of emitting and receiving laser pulses at an extremely rapid rate, enabling three-dimensional scanning of the road surface surrounding the vehicle. LiDAR has a scanning range of at least 360 degrees, meaning it can capture information about the surrounding environment from all angles. This all-around scanning capability is achieved through the LiDAR's internal mechanical structure, typically a rotating laser transmitter and receiver module that provides continuous 360-degree coverage. For example, the RPLIDAR A1 utilizes laser triangulation technology, coupled with a high-speed visual acquisition and processing mechanism, capable of performing over 8,000 ranging operations per second, enabling 360-degree scanning and ranging detection of the surrounding environment and obtaining a contour map of the surrounding environment.

[0048] When laser scanning technology acquires road surface data, it typically includes three-dimensional geometric information about the road surface, such as elevation, slope, and smoothness. This data can be processed and analyzed using the point cloud captured by the laser scanner. To determine road contamination, other types of sensors are often required. For example, image acquisition devices can use high-resolution cameras to capture images of the road surface and observe surface features such as stains and cracks. Image processing techniques can analyze and identify these features. Spectral sensors can measure the spectral reflectance characteristics of the road surface, thereby inferring the surface material composition and possible contamination. Infrared sensors can detect temperature differences on the road surface, such as oil or other contaminants, by measuring the temperature distribution. Contact sensors can directly measure the friction characteristics of the road surface, which can be affected by contamination. In practical applications, laser scanners can be integrated with other sensors on a mobile platform, such as a vehicle or drone, to achieve comprehensive road surface inspection. Data fusion techniques can combine data from different sensors to obtain a more comprehensive picture of road conditions.

[0049] In addition, 360-degree LiDAR scanning technology can optimize performance through different operating modes, such as providing point counts at different frequencies, such as 8,000 points at 10 Hz, 4,000 points at 10 Hz, and 2,000 points at 5 Hz, to adapt to different application scenarios and needs. This high-frequency sampling ensures that even fast-moving objects can be accurately captured and mapped.

[0050] LiDAR's all-around scanning capability is crucial for applications requiring precise environmental perception, such as self-driving cars, mobile robots, and urban landscape surveys. In the autonomous driving sector, LiDAR serves as the vehicle's "eyes," monitoring its surroundings in real time to ensure driving safety. This high-frequency scanning allows the vehicle to capture every subtle change in the road surface, including uneven surfaces, cracks, potholes, and various contaminants.

[0051] (2) The 3D point cloud data obtained by the LiDAR scan, the data collected by the sensors, and the image data are sent to the image processing unit (IPU). In the IPU, the point cloud data undergoes noise reduction, filtering, and 3D reconstruction to generate a detailed environmental model that provides geometric information about the road surface, such as height, slope, and flatness. At the same time, the sensor data is converted into a format suitable for image analysis and fused with the image data to generate comprehensive environmental information.

[0052] In the image processing unit, image data is analyzed using image processing techniques such as edge detection and feature extraction. The IPU also fuses and corrects all sensor data with the image data to improve data accuracy. This processed data is then fed into the analysis unit.

[0053] (3) Computer vision technology can effectively process three-dimensional data and identify the surface features of the road surface by converting this data into analyzable visual information. Specifically, three-dimensional data usually exists in the form of point clouds, and computer vision technology can use this data for feature extraction and analysis. First, the point cloud data is filtered and denoised to remove irrelevant reflections and errors to ensure the accuracy of the data. Then, through point cloud processing technologies such as segmentation and clustering, the geometric features of the road surface, including height, slope, and flatness, can be extracted. These processed data can be further applied with visual algorithms such as edge detection and feature matching to identify detailed features such as stains and cracks on the surface. Combining visual technology with three-dimensional data, through the processing and analysis of point cloud data, it can provide detailed road surface information, thereby supporting applications such as navigation planning, obstacle avoidance, and environmental monitoring. Computer vision technology is used to analyze the acquired three-dimensional data and identify the surface features of the road surface.

[0054] After passing through the image processing unit, the 3D data is transmitted to the vehicle's analysis unit, where it undergoes noise reduction and filtering to remove irrelevant reflections and errors. The image processing unit is primarily responsible for preliminary data processing and fusion, while the analysis unit conducts in-depth analysis and optimization of the data after the image processing unit completes its tasks. This collaborative approach enables the system to provide comprehensive road condition information and support complex application requirements.

[0055] OpenCV provides general algorithms for image processing and computer vision, and supports interfaces for deep learning models, allowing users to load pre-trained models for tasks such as image recognition and object detection. It is particularly useful for image filtering and principal component analysis. The basic operating formula is as follows:

[0056] A. Image filtering (Gaussian blur), used for image filtering processing.

[0057]

[0058] Where dst is the filtered image, src is the original image, G(x,y) is the Gaussian kernel, N is the total area of ​​the kernel, and (x,y) are the pixel coordinates in the output image.

[0059] B. Hough transform, used for line detection.

[0060]

[0061] Where ρ and θ are the radius and angle in the polar coordinate system, respectively. o ,y0) is the center point of the line.

[0062] C. Feature matching, used for template comparison.

[0063] score=Σ x,y (template(x,y)-image(x,y) 2

[0064] Among them, score is the matching score, template is the template image, and image is the target image area.

[0065] D. Principal Component Analysis (PCA)

[0066]

[0067] Where W is the principal component, x i is a data point, W i is the projection coefficient.

[0068] (4) Determine whether the road surface is dirty based on the identified surface features.

[0069] Road contamination is determined based on the differences between road surface characteristics and pre-defined clean road surface characteristics. Using deep learning algorithms, vehicles can identify different types of road contamination, such as oil stains, mud, and water. Deep learning algorithms primarily involve building and training deep convolutional neural networks (DNNs) to learn and recognize complex patterns in image data. These algorithms are trained on large amounts of sample data, automatically extracting features and performing intelligent recognition and classification. Through training and learning from a large number of road images and contamination samples, these algorithms can accurately distinguish different types of road surface characteristics and anomalies.

[0070] Deep learning models, such as convolutional neural networks (CNNs), encompass image classification, object detection, semantic segmentation, and super-resolution reconstruction. For example, in object detection, deep learning uses a DNN to extract target features, uses ROI mapping and IoU to determine thresholds, and trains the model using a Region Proposal Network (RPN). Furthermore, in semantic segmentation, deep learning uses learned features to classify images at the pixel level, enabling the distinction of different objects.

[0071] In practical applications, deep learning platforms such as Huawei's ModelArts provide one-stop services, automatically performing model design, parameter tuning, model training, model compression, and model deployment based on user-labeled data.

[0072] In terms of model training, ModelArts provides a variety of evaluation metrics and calculation formulas to measure model performance. For example, in image classification tasks, metrics such as accuracy evaluation, confusion matrix, recall rate, precision rate, F1 value, and ROC curve are all very important. Specifically:

[0073] The calculation formula for recall is: Here TP stands for the number of true positive examples and FN stands for the number of false negative examples.

[0074] The formula for calculating the accuracy is: Where FP stands for the number of false positives.

[0075] The F1 value is the harmonic mean of precision and recall, and the calculation formula is:

[0076] For object detection tasks, evaluation metrics include accuracy, PR curves, and mAP at different target framework comparison thresholds. For image semantic segmentation tasks, evaluation metrics include intersection over union (IoU) and the Dice coefficient. Trained models are converted using ModelArts into formats suitable for different execution environments, such as TensorFlow's SavedModel format, and deployed to the cloud or edge devices for efficient inference.

[0077] (5) If dirt exists, further analyze the type and distribution of the dirt.

[0078] Dirt can be classified according to its physical properties, sources, and impacts. According to its physical properties, it can be divided into viscous dirt, such as oil stains. This type of dirt has high viscosity, is not easily washed away by water, and easily adheres to the road surface; granular dirt, such as mud and gravel, which are composed of smaller solid particles and may be spread by vehicle movement; solid dirt, such as snow and ice, which are formed under low temperature conditions and will significantly change the friction characteristics of the road surface. According to its source, it can be divided into vehicle leakage, such as oil stains often come from vehicle oil or gasoline leaks; natural environment, such as organic matter such as fallen leaves and branches, which are usually naturally shed by trees in autumn; construction activities, dirt and dust generated in construction areas, and unconsolidated materials carried by transportation vehicles. According to the impact classification, it is divided into the impact of dirt on the friction coefficient of the road surface, such as oil stains and ice layers will reduce the static friction coefficient of the road surface, increasing the risk of slips and traffic accidents; the impact of dirt on the smoothness and appearance of the road surface, such as mud and garbage may accumulate on the road surface, affecting the smoothness of the road and the appearance of the urban environment; the impact of dirt on the environment, such as the use of chemical cleaning agents may cause pollution to the surrounding environment.

[0079] (6) The detection results are sent to the central control system through the wireless communication module for decision-making assistance.

[0080] Detected road contamination information is fed into the autonomous vehicle's decision-making system. The system assesses the potential impact of contamination on driving safety and automatically formulates response strategies, such as slowing down, avoiding contaminated areas, or finding alternative routes. Specifically, OpenCV is used to visualize the results, marking the detected contaminated areas on the image. This information is then presented to the end user or the automated decision-making system through a user interface, allowing appropriate action to be taken.

[0081] (7) Dynamic adjustment

[0082] Based on the dirt detection results, the autonomous vehicle can adjust driving parameters in real time. For example, if oil is detected, the vehicle may automatically slow down, increase the distance from the vehicle in front, or adjust the driving trajectory to avoid potential slipping risks.

[0083] (8) Information Sharing

[0084] Vehicles upload road contamination information to a cloud platform in real time via wireless communication technology. This information can be shared with other vehicles, helping them prepare for action in advance, thereby improving the safety and efficiency of the entire transportation system.

[0085] (9) System self-learning

[0086] Over time, the system continuously collects new road contamination data. This data is used to train and optimize the contamination detection algorithm, making it more accurate and efficient. Through a self-learning mechanism, the system can adapt to changing road conditions and environmental characteristics. Specifically, using the continuously collected detection data and the ModelArts model tuning feature, the deep learning model is continuously optimized and updated, improving detection accuracy and robustness.

[0087] Example 3

[0088] The LiDAR system deployed at fixed inspection sites is being further expanded to improve the efficiency and effectiveness of road maintenance work. The specific steps are as follows:

[0089] (1) Regular testing plan

[0090] After deploying the lidar system at a fixed site, a detailed periodic inspection plan should be developed. The plan will set the inspection frequency based on factors such as traffic volume, historical maintenance records, and seasonal changes in the road. For example, high-traffic or accident-prone areas may require more frequent inspections.

[0091] (2) Data Collection

[0092] At the scheduled time, the LiDAR system automatically activates and performs high-precision road scans, collecting three-dimensional data including road surface texture, cracks, potholes, and various types of dirt. This data provides detailed basic information for subsequent analysis.

[0093] (3) Analysis and evaluation

[0094] The collected point cloud data is transmitted to a central processing system and analyzed using specialized software to assess the type, distribution, and severity of road contamination. The software automatically identifies problem areas and categorizes and quantifies the data.

[0095] (4) Maintenance guidance

[0096] Based on the analysis results, targeted road maintenance and cleaning plans are developed. These plans will include priorities, required resources, expected results, and timelines to guide road maintenance teams to complete their work efficiently.

[0097] (5) Resource Scheduling

[0098] Based on the severity and type of road contamination and the requirements of the maintenance plan, cleaning and maintenance resources are rationally allocated, including cleaning vehicles, maintenance equipment, personnel, and materials, to ensure the most efficient use of resources.

[0099] (6) Long-term monitoring

[0100] Establishing a long-term monitoring mechanism not only tracks current road conditions but also analyzes trends in road contamination. This helps predict future maintenance needs and optimize maintenance strategies.

[0101] (7) Report generation

[0102] Detailed road condition reports will be generated regularly, including detailed analysis of road contamination, records of maintenance activities, and effectiveness evaluations. These reports will be provided to policymakers and resource allocation managers as a basis for decision-making.

[0103] (8) Predictive maintenance

[0104] Leveraging collected historical data and trend analysis, we can predict potential pavement problems and maintenance needs. Predictive maintenance can reduce the frequency of emergency repairs and lower long-term costs.

[0105] (9) Public Information Release

[0106] Release summary information of road conditions through public channels to inform the public of conditions that may affect traffic, such as construction areas, dirty road surfaces, etc., in order to improve the public's awareness and understanding of road conditions.

[0107] (10) Regulatory compliance checks

[0108] Ensure that all maintenance activities comply with local regulations and standards, and conduct regular self-inspections and third-party audits to ensure the compliance of road maintenance work.

[0109] (11) Technology upgrade and maintenance

[0110] As technology develops, lidar systems and analysis software are regularly upgraded to maintain the advanced nature and accuracy of detection and analysis technology.

[0111] Through these steps, the LiDAR system at fixed inspection sites can not only provide accurate road contamination detection, but also provide comprehensive support for road maintenance work, ensuring the long-term stability and safety of roads.

[0112] Example 4

[0113] A computer-readable storage medium stores a computer program, which is configured to implement the laser radar-based road surface and dirt status detection method of embodiment 2 when executed.

[0114] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in similar ways. As long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, they shall fall within the scope of protection of the present invention.

Claims

1. A method for detecting road surface and dirtiness based on laser radar, It is characterized in that The following steps are involved: (1) Use LiDAR to scan the road surface, use sensors to obtain visual images and environmental data of the road surface, and fuse the LiDAR point cloud data with data from other sensors to generate a complete 3D data set; (2) Use computer vision technology to filter, denoise, align, and normalize the point cloud data to eliminate noise and errors. Then use computer vision algorithms to extract features from the data, identify the surface features of the road surface, analyze the height changes, surface flatness, and geometry of the road surface data, and identify the physical features of the road surface and potential dirty areas. (3) Based on the identified surface features, a deep learning algorithm is used to identify abnormal areas and determine whether the road surface is dirty; (4) If there is dirt, further analyze the type and distribution of dirt; (5) The detection results are sent to the central control system through the wireless communication module.

2. A method for detecting road surface and dirtiness based on laser radar according to claim 1, Features: In step (1), the laser radar used has a scanning range of at least 360 degrees.

3. The method for detecting road surface and dirtiness based on laser radar according to claim 1, Features: In step (2), computer vision technology includes a deep learning model.

4. The method for detecting road surface and dirtiness based on laser radar according to claim 1, Features: In the step (3), if there is no difference between the road surface characteristics and the preset clean road surface characteristics, then the road surface is not dirty; otherwise, the road surface is dirty.

5. The method for detecting road surface and dirtiness based on laser radar according to claim 1, Features: In the step (4), the types of dirt are classified according to physical properties, sources, and impacts. According to the physical properties, the dirt is classified into viscous dirt, granular dirt, and solid dirt; according to the source, the dirt is classified into vehicle leakage dirt, natural environment dirt, and construction activity dirt; according to the impact, the dirt is classified into dirt that affects the friction coefficient of the road surface, dirt that affects the smoothness and appearance of the road surface, and dirt that affects the environment.

6. A method for detecting road surface and dirtiness based on laser radar according to claim 1, Features: In the step (5), the wireless communication module supports at least one communication protocol.

7. A road surface and dirt status detection system based on laser radar, It is characterized in that include: At least one detection unit, configured to execute step (1) of the method for detecting road surface and dirtiness based on laser radar as claimed in claim 1; At least one image processing unit, configured to execute step (2) of the method for detecting road surface and dirtiness based on laser radar as claimed in claim 1; At least one analysis unit, configured to execute steps (3) and (4) of the method for detecting road surface and dirtiness based on laser radar as claimed in claim 1; At least one communication module is used to execute step (5) in the road surface and dirtiness status detection method based on laser radar as described in claim 1.

8. A road surface and dirt status detection system based on laser radar according to claim 7, Features: The laser radar scanning unit includes multiple laser transmitters and receivers for all-round scanning of at least 360 degrees.

9. A road surface and dirt status detection system based on laser radar according to claim 7, Features: The image processing unit includes a deep learning processor for performing image analysis tasks.

10. A computer-readable storage medium having a computer program stored thereon, the program being configured to implement any one of the laser radar-based road surface and dirt status detection methods of claims 1-6 when executed.

Images