A trajectory data driven highway litter detection and avoidance warning method

Abstract

The present application relates to a kind of trajectory data driven highway litter detection and avoidance early warning method, belong to traffic intelligent perception technical field, by traffic data acquisition module, real-time collection highway vehicle position information using non-vision sensor, data processing module carries out edge processing to vehicle trajectory data, data transmission module sends edge processing data to cloud, litter judgment module is responsible for according to trajectory feature judgment litter position, variable speed limit information release module uses RSU to release litter position information to road user.The highway litter detection and avoidance early warning method provided by the present application uses the real-time trajectory data of vehicle obtained using non-vision sensor, gets rid of the dependence of litter detection on visual sensor, uses vehicle trajectory data to achieve litter judgment and position estimation from statistical significance, solves the problem of low detection precision and high false alarm rate of traditional litter detection.The present application has the characteristics of reproducible, strong robustness.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of intelligent traffic perception technology, in particular to a trajectory data driven highway litter detection and avoidance warning method. BACKGROUND

[0002] With the continuous growth of highway mileage, the number of motor vehicles is increasing. The number of accidents induced by highway litter is also increasing. These litters include not only the goods dropped from trucks due to insufficient bundling, but also the debris dropped from cars, and some illegal litter thrown by drivers. These items, large or small, pose a great threat to highway users. Therefore, real-time and accurate detection of highway litter has become a problem to be solved.

[0003] Unlike motor vehicles, non-motor vehicles, pedestrians and other targets on the highway, litters do not have general features in images. At the same time, litter perception based on visual sensors may produce false positives due to camera shaking, brightness changes, and shadow shaking. SUMMARY

[0004] The present application is based on radar and other non-vision sensors to collect trajectories, which has become a trend for the development of intelligent highways in the future. Vehicles inherently have the behavior of avoiding litters. Based on historical vehicle trajectory data and real-time vehicle trajectory distribution, litter judgment and identification can be achieved in a statistical sense, thereby realizing indirect judgment of litters without relying on visual sensors. The highway litter detection and avoidance warning method provided by the present application uses real-time trajectory data of vehicles obtained by non-vision sensors (for example, CN 202110589275.3 A vehicle trajectory splicing method based on millimeter wave radar data; CN 202011179544.0 A lane line shape detection method based on millimeter wave radar data), which eliminates the dependence of litter detection on visual sensors. The vehicle trajectory data is used to achieve litter judgment and position estimation in a statistical sense, solving the problem of low detection accuracy and high false positive rate of traditional litter detection.

[0005] The technical scheme adopted by the present application to achieve its purpose is as follows:

[0006] A trajectory data driven highway litter detection and avoidance warning method, a trajectory data driven highway litter detection and avoidance warning system is constructed, including a traffic data acquisition module, an edge data processing module, a data transmission module, a litter judgment module and a litter information publishing module; comprising the following steps:

[0007] (1) The traffic data collection module uses a non-visual high-precision vehicle position sensor to collect real-time expressway vehicle lane-level position trajectory information, transmits the collected traffic data to the edge data processing module for edge processing, and then transmits the processed data to the cloud through the data transmission module;

[0008] (2) The litter judgment module is set in the cloud, the litter judgment module constructs a road grid network, and constructs a vehicle trajectory score distribution map on the road grid network according to the edge processing data in step (1), constructs a trajectory decay function according to the distribution of historical vehicle trajectories on the road grid network, judges the real-time vehicle trajectory distribution according to the trajectory decay function, and obtains the abnormal distribution of the real-time vehicle trajectory, and judges the position of the litter according to the abnormal distribution of the real-time vehicle trajectory;

[0009] (3) The litter information publishing module is used to broadcast the position of the litter to individual vehicles and issue warning information.

[0010] Further, the non-visual high-precision vehicle position sensor in step (1) is a millimeter wave radar or a laser radar.

[0011] Further, the frequency of vehicle position collection and transmission in step (1) should be millisecond level, and the vehicle position positioning accuracy should be decimeter level and above, which can adapt to vehicle position and speed monitoring under all-weather and various weather conditions.

[0012] Further, in step (2), the judgment of the position of the litter includes the following steps:

[0013] 1) The road is gridded with a unit length of 0.1m to obtain a road gridded point array;

[0014] 2) Project the historical vehicle trajectory according to the actual position on the road gridded point array, take the vehicle trajectory point as the center, select the nearest road grid point, and project the vehicle trajectory point to the road grid point;

[0015] 3) Project all vehicle trajectories in a month during the device operation to the road grid points according to the method in 2), and count the hour average projection point number N i of each road grid point corresponding to the morning and evening peak and flat peak in weekdays and weekends in this period;

[0016] 4) Use real-time trajectory data to count the hour average projection point number D i of each hour projected to each road grid point, and calculate the point projection coefficient:

[0017]

[0018] 5) If P i<0.3 then mark the point;

[0019] 6) If P i <0.3 for more than 2 hours, it is considered that the point has a spill or an obstacle.

[0020] Further, the position of the spill is broadcast to the individual vehicles using the RSU in step (3).

[0021] The beneficial effects of the present application are:

[0022] The data used by the expressway spill detection and avoidance warning system constructed by the method of the present application is collected by fixed millimeter wave radar and laser radar detection equipment on the roadside, and real-time radar data is used, which has the characteristics of high detection accuracy and fast detection speed. Through distributed computing, real-time vehicle position information sensing can be realized, and it does not depend on visual sensors.

[0023] The real-time trajectory data of the vehicle obtained by the non-visual sensor breaks the dependence of spill detection on visual sensors, and the vehicle trajectory data is used to realize spill judgment and position estimation from a statistical point of view, solving the problem of low detection accuracy and high false alarm rate of traditional spill detection. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is the schematic diagram of the trajectory data driven expressway spill detection and avoidance warning method in the embodiment.

[0025] Figure 2 is the method diagram of the spill warning device layout in the trajectory data driven expressway spill detection and avoidance warning method in the embodiment. DETAILED DESCRIPTION

[0026] The present application will be described in detail below with reference to the drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor shall fall within the scope of protection of the present application.

[0027] The present application relates to a trajectory data driven expressway spill detection and avoidance warning method, which uses real-time trajectory data of vehicles obtained by non-visual sensors to break the dependence of spill detection on visual sensors, uses vehicle trajectory data to realize spill judgment and position estimation from a statistical point of view, and finally realizes spill warning information publishing through RSU. The method comprises the following steps:

[0028] A trajectory data-driven method for detecting and avoiding debris on highways is constructed, including a traffic data acquisition module, an edge data processing module, a data transmission module, a debris judgment module, and a debris information dissemination module.

[0029] Specifically, the traffic data acquisition module uses non-visual sensors such as millimeter-wave radar or lidar to collect real-time lane-level position trajectory information of vehicles on highways. The frequency of vehicle position acquisition and transmission should be at the millisecond level, and the vehicle position positioning accuracy should be at the decimeter level or above. It can adapt to vehicle position and speed monitoring under all weather conditions and various meteorological conditions. The collected traffic data is transmitted to the edge data processing module for edge processing, and then the processed data is transmitted to the cloud through the data transmission module.

[0030] The spillage detection module is located in the cloud. The spillage detection module constructs a road grid and builds a vehicle trajectory score distribution map on the road grid. Based on the distribution of historical vehicle trajectories on the road grid, a trajectory decay function is constructed to judge the real-time vehicle trajectory distribution and obtain the abnormal distribution of real-time vehicle trajectories.

[0031] Specifically, the road is rasterized in units of 0.1m to obtain the road rasterized point matrix.

[0032] Historical vehicle trajectories are projected onto the road grid based on their actual locations. Using the vehicle trajectory point as the center, the nearest road grid point is selected, and the vehicle trajectory point is projected onto that road grid point.

[0033] The method described above projects all vehicle trajectories onto road grid points during the month-long operation of the equipment, and calculates the hourly average number N of projected points for each road grid point during weekdays and weekends / holidays, corresponding to morning and evening peak hours and off-peak hours. i .

[0034] Calculate the hourly average number of projected points D on each road grid point using real-time trajectory data. i Calculate the projection coefficient of this point:

[0035]

[0036] If P i If the value is less than 0.3, then mark the point.

[0037] If P i If the duration of <0.3 exceeds 2 hours, the point is considered to have spilled material or an obstacle.

[0038] The location of spilled material is broadcast to individual vehicles using the RSU.

Claims

1. A trajectory data driven highway debris detection and avoidance warning method, characterized in that, The application discloses a trajectory data-driven highway litter detection and avoidance early warning system, which comprises a traffic data collection module, an edge data processing module, a data transmission module, a litter judgment module and a litter information publishing module. (1) The traffic data collection module uses a non-vision high-precision vehicle position sensor to collect highway vehicle lane-level position trajectory information in real time, and transmits the collected traffic data to the edge data processing module for edge processing, and then transmits the processed data to the cloud through the data transmission module; (2) The litter judgment module is arranged in the cloud, the litter judgment module constructs a road grid network, and constructs a vehicle trajectory score distribution graph on the road grid network, constructs a trajectory attenuation function according to the distribution of historical vehicle trajectories on the road grid network, judges the real-time vehicle trajectory distribution according to the trajectory attenuation function, and obtains the abnormal distribution of the real-time vehicle trajectory, and judges the position of the litter according to the abnormal distribution of the real-time vehicle trajectory; (3) The litter information publishing module is used to broadcast the position of the litter to individual vehicles and issue early warning information. In step (2), the judgment of the position of the litter comprises the following steps: 1) The road is rasterized with a unit length of 0.1 m to obtain a road raster point array; 2) The historical vehicle trajectory is projected on the road raster point array according to the actual position, the vehicle trajectory point is taken as the center, the nearest road grid point is selected, and the vehicle trajectory point is projected on the road grid point. 3) Project all vehicle trajectories within one month during the operation of the device onto the road grid points according to the method in 2) and count the average number of projected points per hour for each road grid point during the morning and evening rush hours and the flat peak hours on weekdays and weekends and holidays within the period ; 4) Using real-time trajectory data to count the hourly average number of projected points per road grid cell per hour , calculate the point projection coefficient: 5) If then mark the point; 6) If If the duration exceeds 2 hours, the point is considered to have a spill or an obstacle. 2.The trajectory data driven highway debris detection and avoidance warning method of claim 1, wherein: In step (1), the non-vision high-precision vehicle position sensor is a millimeter wave radar or a laser radar.

3. The trajectory data driven highway debris detection and avoidance warning method of claim 1, wherein: In step (1), the frequency of vehicle position collection and transmission should be millisecond level, and the vehicle position positioning accuracy should be decimeter level or above, which can adapt to vehicle position and speed monitoring under all-weather and various weather conditions.

4. The trajectory data driven highway debris detection and avoidance warning method of claim 1, wherein: In step (1), the edge processing is structural processing and analysis of the vehicle trajectory data by the edge computing unit of the data processing module.

5. The trajectory data driven highway debris detection and avoidance warning method of claim 1, wherein: In step (1), the data transmission module uses a fiber or a wireless signal transmitter to send.

6. The trajectory data driven highway debris detection and avoidance warning method of claim 1, wherein: In step (3), the RSU is used to broadcast the position of the litter to individual vehicles.

Images