Research on Early Warning Technology for High-Grade Highway Pavement Damage

Road surface damage on high-grade highways has a serious impact on driving safety. Conventional road surface damage detection mainly adopts manual detection technology. The detection efficiency is low, and the detection results are easily disturbed by human factors, which is not conducive to objective and accurate evaluation of road damage. To solve the problems of accurate collection, efficient identification, and correct positioning of high-grade road surface damage data, the research group has studied the early warning technology of road surface damage. The proposed early warning technology scheme enables drivers to take preventive driving action in a timely manner, avoids the occurrence of traffic accidents, and provides drivers with reasonable avoidance schemes to effectively prevent the occurrence of traffic accidents.


Introduction
Road quality has an important impact on driving safety, comfort, economy, and road life cycle.With the development of the highway transportation industry, higher demands have been placed on the quality and maintenance of highway pavement.By the end of 2022, the national highway mileage had reached 5,534,800 kilometers, an increase of 74,100 kilometers over the end of the previous year; the highway density was 55.78 kilometers per 100 square kilometers, an increase of 0.77 kilometers per 100 square kilometers; the highway maintenance mileage was 5,530,300 kilometers, accounting for 99.9% of highway mileage [1].For a long time, for a lack of systematic data collection, evaluation of scientific maintenance decisions and quantitative evaluation of road damage, the waste of maintenance funds, the decline of road quality and the increase of use costs.To change this situation, many traffic management departments have implemented road management systems to meet the requirements of modern, largescale, high-speed, and high-quality highway maintenance management [2].However, the road surface damage data collected by the road surface management system is still mainly used for manual detection.It is difficult to evaluate road surface damage objectively and accurately [3].Therefore, it is of great significance to study the automatic detection of road damage, early warning, and other technologies, and combine high and new technologies to develop and design an efficient, fast, and accurate automatic road damage detection system.In this regard, many researchers at home and abroad have conducted indepth research.It has become a particularly hot research topic in this field, photography or videography technology is used to develop a road surface damage information collection system, and image recognition and other technologies are also used to process road surface damage information.

System Composition
With the advancement of the transportation industry, due to the increasing traffic volume of road transportation, the degree of crushing, damage, and frequency of damage of roads continues to increase, the road surface produces pits, sinking, peeling, cracking, mesh cracking, and other damage problems.These road damages will pose a serious threat to driving safety [4].Therefore, an urgent problem needs to be solved in how to make the vehicle get the road damage in time and get the avoidance plans in time according to the situation [5].
To solve the above problems, this paper proposes an early warning method for road damage, which is applied to an early warning system for road damage.The system includes an image processing module, a vehicle environment perception module, and a road damage warning module.Among them, the image processing module is responsible for collecting the image information of the road surface, according to the image [6].The information analyzes the damage to the road surface to obtain the damaged area image and damaged location information, and sends the damaged area image and damaged location information to the road environment perception module; the road environment perception module is responsible for collecting the movement parameters of the vehicle, determining the driving route of the vehicle according to the collected movement parameters, and root the early warning information of the vehicle is determined according to the damage situation and the driving route, and the early warning information is sent to the road damage warning module; the road damage early warning module projects the early warning information to the vehicle's head-view system and provides the driver with an avoidance scheme through the head-view system [7].The structure of the early warning system of road damage is shown in figure 1.

Image Acquisition and Processing Subsystem
The image acquisition and processing subsystem is responsible for collecting the image information of the road surface and analyzing the damage of the road surface according to the image information, so as to obtain the image of the damaged area and the damaged location information and sending the image of the damaged area and the damaged position information to the road environment perception module [8].
The image processing module mainly includes drones and control centers.The image acquisition system of multiple drones obtains the video image information of the road surface and transmits the image information to the control center [9].The image analysis tool in the control center, that is, the road damage recognition system, analyzes and identifies the video image data information, the identified road damage area image, and the damage location information.The corresponding information is stored in the server, and the information is synchronously sent to the roadside unit of the road environment perception subsystem [10].
Image acquisition and processing specifically includes the following steps: First of all, the drone collects road video image data.Secondly, the drone transmits the road video image information back to the control center.Finally, the image of the road damage area and the damage location information are sent to the roadside unit of the road environment perception module.After the road damage information is analyzed and stored in the ground control center, the 5G communication system is applied and transmitted synchronously to the processor in the road test unit of the vehicle environment perception subsystem.

Vehicle Environment Perception Subsystem
The vehicle road environment perception subsystem is responsible for collecting the movement parameters of the vehicle, determining the vehicle's driving route according to the movement parameters, determining the vehicle's early warning information according to the damage situation and driving route, and sending the early warning information to the road damage warning module.
The road environment perception subsystem mainly includes roadside units and on-board units.The roadside unit is installed around the road, and each roadside unit is responsible for a section of the road.The on-board unit is installed on the vehicle.The roadside unit can receive the image of the road damage area from the control center and the corresponding damage position information, and at the same time receive the movement parameters from the vehicle, which includes vehicle basic information, vehicle speed information, vehicle position information, etc.First of all, the roadside unit accepts the movement parameters of the target vehicle and determines the basic information of the vehicle, of which the basic information includes the license plate number.The target vehicle identification system in the roadside unit identifies the target vehicle to determine whether the target vehicle has a damaged road surface in its driving lane, and whether its speed will affect the road damage, and then sends road damage warning information to these affected target vehicles, which will not be damaged.Vehicles affected by the lane do not send warning information.
The processor in the roadside unit has a target vehicle recognition function.Based on the embedded microprocessor, it realizes vehicle-road interconnection through the V2I terminal, that is, V2I remotely connects the roadside unit (R) through the edge cloud (Multi-access Edge Computing, MEC).Road Side Unit (RSU) and On Board Unit (OBU), the structure of the vehicle environment perception subsystem is shown in figure 2.  The work of the road environment perception subsystem includes the following steps: First of all, receive the image of road damage from the control center and its corresponding location information.Secondly, receive the vehicle motion parameter information from the on-board unit.Finally, the target vehicle identification system in the roadside unit identifies the target vehicle that will be affected by road damage.The roadside unit integrates the received vehicle high-precision position information, vehicle speed, driving direction and other vehicle movement parameters, as well as the position information of road damage, and determines whether the vehicle and road damage are in the same lane according to the identification results.If so, it is determined that the vehicle is the target vehicle.According to the relative distance X between the target vehicle and the road damage position and the obtained vehicle driving speed V, the time T for the target vehicle to reach the road damage position is calculated.

Road Damage Warning Subsystem
The road damage warning subsystem projects the early warning information on the head-to-view system of the vehicle to provide an avoidance solution to the driver through the head-to-eye system.
After the vehicle unit of the vehicle that is about to be affected by the road damage receives the early warning information from the roadside unit, the image information of the road damage is projected on the vehicle's level-up system so that the driver can intuitively understand the road damage in the front lane without affecting the driver's safe driving.The early warning process of the road damage warning module includes the following steps: First of all, for the target vehicles that will be affected by road damage, early warning information from the roadside unit will be received.The on-board unit on the target vehicle receives the road damage image information transmitted by the roadside unit, the time information of the target vehicle to reach the road damage location and other information through the V2I terminal, which can be transmitted to the head-up system and on-board voice system of the target vehicle respectively, as shown in figure 3.

An Analysis of Experimental Results
Two typical pavement damage images, namely potholes and cracks, are selected for research, and the recognition results are shown in table 1.The research shows that the improved YOLOv3 is superior to other algorithms in terms of accuracy, recall rate, mAP0.5, and processing speed, and can meet the requirements of pavement damage identification.The next step will be to conduct in-depth research on it and explore better improvement methods to improve recognition accuracy and practicability.

Conclusions
(1) Road damage warning technology can enable drivers to know the road condition information ahead of them in advance, so as to take reasonable avoidance plans and effectively prevent traffic accidents.
(2) There are many detection methods for pavement-damaged areas, among which the improved YOLOv3 has certain technical advantages and needs to be further studied.
(3) The application of advanced technology is helpful in solving the problems of accurate collection, efficient identification, and correct positioning of high-grade highway pavement damage data.

Figure 1 .
Figure 1.Schematic diagram of the structure of the early warning system for road damage.

Figure 2 .
Figure 2. Structure of vehicle-road environment awareness subsystem.The communication of V2I terminals mainly uses two technologies, one is a DSRC solution that uses the 5.9GHz band; the other is a solution based on cellular technology, such as LTE and 5G.V2I technology can guarantee transmission delay within 100ms, carry out direct communication without relying on base station coverage, and provide an efficient broadcasting mechanism.It is a very suitable technology for inter-vehicle communication.The work of the road environment perception subsystem includes the following steps: First of all, receive the image of road damage from the control center and its corresponding location information.Secondly, receive the vehicle motion parameter information from the on-board unit.Finally, the target vehicle identification system in the roadside unit identifies the target vehicle that will be affected by road damage.The roadside unit integrates the received vehicle high-precision position information, vehicle speed, driving direction and other vehicle movement parameters, as well as the position information of road damage, and determines whether the vehicle and road damage are in the same lane according to the identification results.If so, it is determined that the vehicle is the target vehicle.According to the relative distance X between the target vehicle and the road damage position and the obtained vehicle driving speed V, the time T for the target vehicle to reach the road damage position is calculated.

Table 1 .
Comparison of recognition results of different algorithms.