Automated vehicle infrastructure deployment method for roadside devices and system thereof

Abstract

The application relates to the technical field of vehicle wireless communication, and discloses an automatic vehicle-road cooperation deployment method for a roadside device and a system thereof, which comprises the following steps: acquiring intersection data and generating structured data of a road; inputting performance parameters of a vehicle-road cooperation roadside device; setting thresholds and weights influencing sensor sensing accuracy; configuring special service capabilities; setting limited conditions according to special service capabilities planned for an intersection; constructing a sensing capability scoring model according to weather conditions, a structured road scene, a target posture, RTK positioning accuracy, and pre-installed position dimensions; taking a deployment scheme of a vehicle-road cooperation roadside device and an intersection meeting S3 and S4 requirements as a selectable deployment scheme, substituting the selectable deployment scheme into the sensing capability scoring model for scoring; and outputting the selectable deployment scheme with the highest score as an optimal deployment scheme. The application can effectively solve the problems of resource waste and unsatisfactory detection effect in automatic deployment.

Description

Technical Field

[0001] The present invention relates to the field of vehicle wireless communication technology, and in particular to an automated vehicle-road collaborative deployment method and system for roadside equipment. Background Art

[0002] Vehicle-road collaboration uses advanced wireless communications and new-generation Internet technologies to implement all-round dynamic real-time information interaction between vehicles and roads, and carries out active vehicle safety control and road collaborative management based on the collection and integration of dynamic traffic information in all time and space, to achieve effective collaboration between people, vehicles and roads, ensure traffic safety, improve vehicle traffic efficiency, and thus form a safe, efficient and environmentally friendly road traffic system.

[0003] The development of 5G in my country and the rapid construction of communications infrastructure have accelerated the pace of road construction and renovation, enabling the implementation of vehicle-road collaboration. This technology can perceive traffic over a wider range and at longer distances, taking into account the real-time driving conditions of all vehicles on the road. This can effectively prevent obstacles from impacting vehicles, thereby improving the safety of autonomous driving.

[0004] However, during the planning and construction of the vehicle-road collaborative smart road network in the "Vehicle Network Application Pilot Demonstration Zone", there is no unified standard for roadside equipment deployment methods. The deployment plan mainly comes from the on-site survey conditions as the basis for equipment deployment decision-making. During the exploration period of the pilot demonstration zone, there is no deployment experience to draw on. Therefore, actively obtaining decision-making basis during the on-site survey has certain limitations. Only the intersection location, traffic volume, and whether to install signal control are surveyed as decision-making conditions for equipment deployment. After deployment, there is a waste of resources and the detection effect does not meet expectations. Summary of the Invention

[0005] The present invention aims to provide an automated vehicle-road collaborative deployment method for roadside equipment. Existing deployments waste resources and produce suboptimal detection results.

[0006] In order to solve the above problems, the present invention adopts the following technical solutions:

[0007] Solution 1: An automated vehicle-road collaborative deployment method for roadside equipment includes the following steps:

[0008] S1, obtain intersection data and generate structured data of the road;

[0009] S2, input the performance parameters of the vehicle-road cooperative roadside equipment;

[0010] S3, setting the threshold and weight that affect the sensor perception accuracy;

[0011] S4, configure special service capabilities: set limiting conditions based on the special service capabilities planned for the intersection;

[0012] S5, constructing a perception capability score model: constructing a perception capability score model according to weather conditions, structured road scenes, target postures, RTK positioning accuracy, and pre-installation position dimensions, taking the deployment scheme of the vehicle-road cooperative roadside device and intersection meeting the requirements of S3 and S4 as an optional deployment scheme, and substituting the optional deployment scheme into the perception capability score model for scoring;

[0013] S6, automatically generating a deployment scheme, taking the optional deployment scheme with the highest score as the best deployment scheme, and outputting a deployment scheme construction drawing.

[0014] Preferably, all optional deployment schemes include both hardware system deployment and software system deployment.

[0015] Preferably, in S2, the performance parameter of the device refers to the theoretical performance parameter and the actual performance parameter of the vehicle-road cooperative roadside device, including the broadcast range of the RSU roadside communication unit, the image processing capability of the MEC roadside computing unit, and the device performance of the perception device, such as memory, CPU occupancy, and unit price. Through the input of these performance parameters, it is helpful to more accurately match the corresponding road structure information, so that the formed optional deployment scheme is more accurate.

[0016] Preferably, in S4, the limiting conditions set according to the special service capability of the intersection include installing a long-focus camera in the direction of the intersection where the pedestrian crossing area exits the intersection to facilitate the appropriate addition of a long-focus camera for a specific application scenario to obtain more accurate signals.

[0017] Preferably, in S4, the limiting conditions set according to the special service capability of the intersection include adding a long-focus camera at the zebra crossing to detect the pedestrian crossing situation in the pedestrian crossing section of a specific area. The traffic volume in the specific area is more than twice the traffic volume of the crossing section in a certain time period, and the corresponding number of long-focus cameras can also be increased by a multiple when installed.

[0018] Preferably, in S5, the perception capability score model is integrated according to positioning accuracy , heading angle detection accuracy , size detection accuracy , and speed measurement accuracy

[0019] Perception capability score ,

[0020] In the formula, a, b, c, and d represent the corresponding index weights, all greater than 0 and less than 1, and can be weighted for comprehensive scoring.

[0021] ​This solution creatively calculates the perception capability score through a perception capability scoring model, which can more objectively and accurately evaluate various optional deployment plans, and can predictably present and select various optional deployment plans, so that the best deployment plan with the highest score will also be the most effective plan after actual deployment, making expectations achievable.

[0022] Preferably, in S5, the plane positioning accuracy The mean absolute error between the position detection value and the true value of the center point of all traffic participants perceived by the roadside perception system within the perception range and within a specific time in the local plane coordinate system can be estimated according to the following formula:

[0023]

[0024] Where:

[0025] ——the two-dimensional plane position vector of the i-th sample in the system under test;

[0026] ——The true position vector of the i-th sample in the two-dimensional plane;

[0027] ——Euclidean distance;

[0028] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0029] Preferably, in S5, the heading angle detection accuracy The mean absolute error between the heading angle detection value and the true value of all traffic participants sensed by the roadside perception system within the perception range and within a specific time period can be estimated according to the following formula:

[0030]

[0031] Where:

[0032] ——the heading angle of the system under test in the i-th sample;

[0033] ——The true heading angle in the i-th sample;

[0034] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0035] Preferably, in S5, the speed measurement accuracy , within the perception range, the mean absolute error between the scalar speed detection value and the true value of the center point of all traffic participants perceived by the roadside perception system can be estimated as follows:

[0036]

[0037] Where:

[0038] ——the speed of the system under test in the i-th sample;

[0039] ——The true speed of the i-th sample;

[0040] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0041] Solution 2: The present invention further provides an automated vehicle-road collaborative deployment system for roadside equipment, which is used to implement the automated vehicle-road collaborative deployment method for roadside equipment as described above, including a central processing unit, and an intersection data acquisition module, a device parameter acquisition module, a threshold weight setting module, a special service capability condition setting module, a model building module, and a solution generation module, each of which is communicatively connected to the central processing unit;

[0042] An initial perception capability scoring model is preset in the model construction module. The central processing unit obtains the threshold weight and the limiting conditions from the threshold weight setting module and the special capability condition setting module respectively, and adjusts the weight parameters of the initial perception capability scoring model to obtain the current perception capability scoring model; the central processing unit obtains the intersection information and the equipment information from the intersection data acquisition module and the equipment parameter acquisition module respectively, matches the intersection information with the equipment information, and forms a variety of optional deployment schemes for installing roadside equipment at corresponding intersections; the central processing unit substitutes the intersection information and equipment information corresponding to all optional deployment schemes into the current perception capability scoring model, calculates the corresponding perception capability score, and sends the optional deployment scheme with the highest score as the best deployment scheme to the scheme generation module to output the deployment scheme construction drawing.

[0043] The advantages of the present invention are:

[0044] The present invention automates deployment plans, primarily collecting intersection feature data online, supplemented by offline field surveys. The full amount of road structured data combined with equipment performance is used as a prerequisite for deployment plans, automatically generating deployment plans and reducing the workload of field surveys.

[0045] This invention makes detection accuracy predictable. By building a perception capability scoring model based on factors such as weather conditions, structured road scenes, target posture, RTK positioning accuracy, and pre-installed locations, intersection / road section detection accuracy can be estimated, providing a prerequisite for quantifiable evaluation of subsequent deployment plans.

[0046] This invention makes economic indicators controllable. The automated deployment solution can estimate construction costs, enabling comprehensive control of deployment economics. This avoids the waste of resources that are not used after construction. Compared to existing deployment methods, this invention effectively avoids resource waste, and deployment results can be quantified and evaluated, achieving the desired goals.

[0047] In addition, compared with the existing technology that only considers the optimal RSU deployment location, the present invention not only considers the RSU deployment location, but also considers the deployment location of the sensing device, so that the overall deployment situation is more in line with actual needs, providing the premise for the post-deployment detection effect to achieve the expected results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of embodiment 1 of the present invention.

[0049] Figure 2 This is the satellite map of intersection A in Example 1 of the present invention.

[0050] Figure 3 This is the structured map of intersection A in Example 1 of the present invention.

[0051] Figure 4 This is a satellite map of the marked pole positions at intersection A in Example 1 of the present invention.

[0052] Figure 5 This is a construction drawing of the automatic deployment solution for the intersection in Example 1 of the present invention. DETAILED DESCRIPTION

[0053] The following is further described in detail through specific implementation methods:

[0054] Example 1

[0055] Vehicle-road collaboration uses advanced wireless communications and new-generation Internet technologies to implement all-round dynamic real-time information interaction between vehicles and roads, and carries out active vehicle safety control and road collaborative management based on the collection and integration of dynamic traffic information in all time and space, to achieve effective collaboration between people, vehicles and roads, ensure traffic safety, improve vehicle traffic efficiency, and thus form a safe, efficient and environmentally friendly road traffic system.

[0056] The development of 5G in my country and the rapid construction of communications infrastructure have accelerated the pace of road construction and renovation, enabling the implementation of vehicle-road collaboration. This technology can perceive traffic over a wider range and at longer distances, taking into account the real-time driving conditions of all vehicles on the road. This can effectively prevent obstacles from impacting vehicles, thereby improving the safety of autonomous driving.

[0057] However, during the planning and construction of the vehicle-road collaborative smart road network in the "Vehicle Network Application Pilot Demonstration Zone", there is no unified standard for roadside equipment deployment methods. The deployment plan mainly comes from the on-site survey conditions as the basis for equipment deployment decision-making. During the exploration period of the pilot demonstration zone, there is no deployment experience to draw on. Therefore, actively obtaining decision-making basis during the on-site survey has certain limitations. Only the intersection location, traffic volume, and whether to install signal control are surveyed as decision-making conditions for equipment deployment. After deployment, there is a waste of resources and the detection effect does not meet expectations.

[0058] The present invention focuses on how to quickly generate an optimal deployment plan based on specific intersection or road conditions. In contrast, current deployment of roadside equipment only requires that information be properly transmitted to the intersection or road section where it is located, with no restrictions on the deployment location or number of roadside equipment. Consequently, to ensure signal transmission stability and coverage, more roadside equipment is often deployed, resulting in a waste of resources. This deployment approach, which involves laying out or even stacking a large number of roadside equipment, can be described as an oversaturated deployment. This oversaturated deployment approach not only wastes equipment costs and resources but also increases installation and maintenance workload. Furthermore, the effectiveness achieved does not scale linearly with the number of roadside equipment, resulting in detection results often falling short of expectations relative to the investment cost. This solution, however, achieves greater cost savings without compromising performance. It allows for more proactive and targeted deployment of roadside equipment, avoiding the blind oversaturation of deployments by stacking too many roadside equipment. This effectively improves the input-output ratio, conserves resources, and ensures that the final detection results meet expectations.

[0059] like Figure 1 As shown, the automated vehicle-road collaborative deployment method for road test equipment includes the following steps:

[0060] Step 1: Obtain intersection data and generate structured data of roads

[0061] (1) Rapidly obtain intersection feature data through road-level maps, panoramic maps, and image data, including intersection type, angle of deviation from true north in each direction, slope, number of lanes, etc.

[0062] (2) Through on-site surveys, the positions of the original poles and the number of lanes that cannot be obtained online are supplemented to generate structured data of the road.

[0063] Step 2: Enter the performance parameters of the equipment. This includes the theoretical and actual performance parameters of the roadside equipment for vehicle-road cooperative systems, including the broadcast range of the RSU roadside communication unit, the image processing capability of the MEC roadside computing unit, the operating performance of the perception equipment (camera, millimeter-wave radar, lidar), memory usage, CPU usage, and unit price.

[0064] Step 3: Set the thresholds and weights that affect sensor perception accuracy

[0065] (1) Set the thresholds and weights of the network camera's detection accuracy due to weather conditions (rain, snow, fog, etc.), structured road scenes (slopes, curves, overpasses, tunnels with light and dark changes), target posture (vehicle longitudinal and lateral posture changes, pedestrian posture changes, target color and texture changes), target occlusion (pedestrian occlusion, vehicle occlusion, blind spots), and target speed (vehicle speed, pedestrian movement).

[0066] (2) Set the threshold and weight of the millimeter wave radar that affects the target detection accuracy due to factors such as target movement speed, target geometry, target posture changes, and signal multi-path reflection interference.

[0067] (3) Set the threshold and weight of the impact of weather and obstruction on RTK positioning accuracy: including weather (fog, rain, snow, etc.) and obstruction (tunnel, tree shade, etc.).

[0068] (4) Set the threshold and weight of the impact of the sensor’s own installation position, angle, and focal length on target detection accuracy.

[0069] The corresponding thresholds and weights are adjusted according to the corresponding experience tables and the classification and comparison of each parameter item. This article does not elaborate on them.

[0070] Step 4: Configure special service capabilities: Set limiting conditions based on the special service capabilities planned for the intersection.

[0071] According to the service capacity design requirements of the corresponding intersection section, especially the geometric area of ​​the congestion area that can cause pedestrian gathering and vehicle congestion, the installation limit area and number of roadside equipment including RSU and sensor equipment are reasonably set.

[0072] (1) At intersections where sudden gatherings are detected and where the length of queues needs to be detected, telephoto cameras are installed in the direction of the pedestrian crossing area exiting the intersection. The number of telephoto cameras is related to the width of the exit, the direction of the intersection, and the length of the detection section. Generally, only one telephoto camera is required within 100 meters in the width direction, and two telephoto cameras are required within 300 meters in the length direction. When a sudden gathering occurs in a scenic area or shopping mall within a certain period of time, the telephoto cameras can be used to monitor the gathering congestion area in real time. The congestion signal can be sent to the roadside equipment through the telephoto cameras in the corresponding area, so that the monitoring needs of sudden events can be met without installing too much equipment.

[0073] (2) At sections of the road where car jams often occur at roundabouts, it is necessary to monitor the traffic flow at each entrance and exit, monitor the illegal parking situation in the long-term illegal parking area, calculate the length and width of the congested area based on the traffic flow changes or illegal parking situation, and determine whether additional cameras are needed in the length and width directions of the congested area, with one telephoto camera within 80 meters and two telephoto cameras within 320 meters. The distances between multiple cameras are not necessarily equal, that is, all cameras are not necessarily evenly distributed. By calculating the geometric area of ​​the congested area, the corresponding camera is installed at the corresponding optimal shooting position, generally at the inflection point of the geometric shape, so that the minimum number of cameras can be used to completely cover the congested area and achieve the purpose of real-time monitoring.

[0074] (3) In specific areas where pedestrians cross the street, it is necessary to add a telephoto camera at the zebra crossing to detect pedestrians crossing the street.

[0075] The specific pedestrian crossing service mentioned here refers to a specific area: in specific areas, such as school crossings, right-turning vehicles are not subject to traffic signals, resulting in mixed pedestrian traffic at the zebra crossings. Through precise deployment (device type, quantity, and device coverage), target activity in the zebra crossing area is monitored, providing blind spot warnings for right-turning vehicles, thus reducing safety risks in these areas. Of course, there are also other application scenarios in areas such as markets, schools, and exhibition halls.

[0076] Pedestrian traffic in these specific areas can exceed by more than twice the volume of regular street traffic during a specific time period. This setup allows for a limited number of telephoto cameras to more clearly monitor crowded areas near these locations, particularly near zebra crossings, and dynamically adjust traffic lights based on the number of pedestrians waiting.

[0077] Step 5. Build a perception capability scoring model: Build a perception capability scoring model based on weather conditions, structured road scenes, target posture, RTK positioning accuracy, pre-installation location, and other dimensions. For the deployment plans of vehicle-road cooperative roadside equipment and intersections that meet the thresholds and weights affecting sensor perception accuracy in step 3 and the limiting conditions set according to the special service capabilities of intersection planning in step 4, they are considered as optional deployment cases and substituted into the perception capability scoring model for scoring.

[0078] Step 6: Automatically generate a deployment plan and output the highest-scoring optional deployment plan as the optimal deployment plan.

[0079] (1) Hardware system deployment

[0080] Hardware deployment is based on the functional requirements of traffic system scenarios. Network cameras and millimeter-wave radars are deployed on roadside poles in specific areas. The network cameras' field of view covers road sections closer to the camera installation location, while the radars cover longer distances. RSUs are mounted on poles to broadcast target information detected by the cameras and radars. Floor-standing cabinets are installed near traffic lights on these sections. These contain industrial computers, switches, power supply systems, and Ethernet low-voltage communication systems. The industrial computers control sensors across the road section via switches and establish full-duplex communication with the RSUs. The RSUs subscribe to target information sent by the industrial computers via wired communication, and the industrial computers subscribe to traffic light information sent by the RSUs.

[0081] (2) Software system deployment

[0082] Software system deployment leverages the powerful computing power of industrial computers to automate software system deployment through IP, port, and other parameterized configurations. The software starts with a single click, reads configuration file parameters, and matches these parameters to complete software system initialization. Multi-process synchronous acquisition and processing of sensor data and traffic light signals completes automated deployment of real-time data processing within the software system.

[0083] Among them, the perception ability scoring model constructed in the fifth step is:

[0084] Accuracy scores are set for plane positioning, heading, size, speed, and other aspects, and the perception results are scored comprehensively. The score calculation formula is as follows:

[0085] Comprehensive perception result score

[0086] In the formula, a, b, c, and d represent the weights of the corresponding indicators, all greater than 0 and less than 1, and can be weighted to create a comprehensive score. The values ​​of a, b, c, and d are determined based on the corresponding intersection structured information. The less obstruction and interference the corresponding intersection structured information parameter has on perception, the greater the corresponding indicator weight.

[0087] For positioning accuracy, is the heading angle detection accuracy, For size detection accuracy, For speed measurement accuracy.

[0088] Specifically, in order to enable the roadside perception system to better serve different application scenarios, the technical indicator evaluation requirements of the roadside perception system can be divided into three sensing levels, namely SL1-SL3. The basis for the perception level division is as follows:

[0089] Perception Level 1 (SL1): This level of perception capability is targeted at platform data applications. Roadside perception system indicators meeting this level should be able to support corresponding data statistics and platform control applications.

[0090] Perception Level 2 (SL2): This level of perception capability is targeted at driver assistance applications. Roadside perception systems meeting this level should be able to provide real-time road information to drivers and assist them in making on-the-spot decisions.

[0091] Perception Level 3 (SL3): A level of perception capability targeted at autonomous driving applications. Roadside perception systems meeting this level should be able to function as off-vehicle sensors, providing valuable real-time road information to support autonomous driving systems in decision-making applications.

[0092] 1 Positioning accuracy

[0093] Plane positioning accuracy

[0094] The mean absolute error between the position detection value and the true value of the center point of all traffic participants perceived by the roadside perception system within the perception range and within a specific time in the local plane coordinate system can be estimated according to the following formula:

[0095]

[0096] Where:

[0097] ——the two-dimensional plane position vector of the i-th sample in the system under test;

[0098] ——The true position vector of the i-th sample in the two-dimensional plane;

[0099] ——Euclidean distance;

[0100] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0101] The corresponding traffic participant positioning accuracy classification requirements are shown in Table 1.

[0102] Table 1

[0103]

[0104] 2 Heading Angle Detection Accuracy

[0105] The mean absolute error between the heading angle detection value and the true value of all traffic participants sensed by the roadside perception system within the perception range and within a specific time period can be estimated using the following formula:

[0106]

[0107] Where:

[0108] ——The heading angle of the system under test in the i-th sample;

[0109] ——The true heading angle in the i-th sample;

[0110] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0111] The corresponding traffic participant heading angle detection accuracy classification requirements are shown in Table 2.

[0112] Table 2

[0113]

[0114] 3. Dimension detection accuracy

[0115] The average value of the difference between the size detection value and the actual value of all traffic participants perceived by the roadside perception system within the perception range and a specific time. That is, if the actual size of a traffic participant is (long), (width) and (High), the measured value is , and , the size detection accuracy is:

[0116]

[0117] Where:

[0118] —— Average the number of all traffic participants detected at all perception moments within a specific time.

[0119] The corresponding traffic participant size detection accuracy classification requirements are shown in Table 3.

[0120] Table 3

[0121]

[0122] 4 Speed ​​measurement accuracy

[0123] Within the perception range, the mean absolute error between the scalar speed detection value and the true value of the center point of all traffic participants perceived by the roadside perception system can be estimated as follows:

[0124]

[0125] Where:

[0126] ——the speed of the system under test in the i-th sample;

[0127] ——The true speed of the i-th sample;

[0128] N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

[0129] The corresponding traffic participant speed detection accuracy classification requirements are shown in Table 4.

[0130] Table 4

[0131]

[0132] In summary, the accuracy of plane positioning, heading, size, speed measurement, etc. is scored, and the perception results are scored comprehensively. The score calculation formula is as follows:

[0133]

[0134] In the formula, a, b, c, and d represent the weights of the corresponding indicators, which can be weighted to form a comprehensive score. In this embodiment, a, b, c, and d can be set in a hierarchical table based on the parameters of the corresponding intersection structure information. The corresponding weight values ​​in the table are selected based on the matching content of the actual intersection parameters. This table is not further described here.

[0135] This method has the following characteristics:

[0136] 1. Comprehensive road structured data: Leveraging existing resources and on-site surveys, we acquire comprehensive intersection feature data, including intersection type, deflection angles in each direction, number of lanes in each direction, slope, and the location of existing poles. This data is then transformed into structured road data. This allows for more accurate extraction and identification of road features, enabling the optimal deployment plan to be tailored to specific road conditions.

[0137] 2. Equipment performance as a prerequisite for deployment: The theoretical and actual performance parameters of V2X roadside equipment, including the broadcast range of the RSU roadside communication unit, the image legend capability of the MEC roadside computing unit, the sensor's sensing range and detection accuracy, as well as power supply performance, operating environment, and unit price, are essential for equipment deployment. This approach not only considers road characteristics but also comprehensively considers the compatibility of equipment with the road based on equipment performance, providing the prerequisites for providing the optimal deployment plan.

[0138] 3. Perception accuracy can be evaluated: A perception capability scoring model is built to score deployment plans based on weather conditions, structured road scenes, target posture, RTK positioning accuracy, pre-installation location, and other dimensions to generate the optimal deployment plan.

[0139] 4. Differentiated deployment at special intersections / road sections: Based on the special service capacity requirements of intersection planning, such as intersections with sudden and concentrated events, sections with large vehicles passing through, and extra-large intersections, differentiated deployment is carried out by setting limiting conditions.

[0140] This method automates deployment plans. Online intersection feature data collection is the primary method, supplemented by offline field surveys. Comprehensive road structured data combined with equipment performance is a prerequisite for deployment plans. This automatically generates deployment plans, reducing the workload of field surveys.

[0141] This method makes detection accuracy predictable. By building a perception capability scoring model based on factors such as weather conditions, structured road scenes, target pose, RTK positioning accuracy, and pre-installed locations, intersection / road section detection accuracy can be estimated, providing a prerequisite for quantifiable evaluation of subsequent deployment plans.

[0142] This method makes economic indicators controllable. The automated deployment solution can estimate construction costs, enabling comprehensive control of deployment economics. This avoids wasted resources after construction. Compared to existing deployment methods, this method effectively avoids resource waste, and deployment results can be quantified and evaluated, effectively achieving the desired goals.

[0143] In addition, compared with the existing technology that only considers the optimal RSU deployment location, this method not only considers the RSU deployment location, but also the deployment location of the sensing device, making the overall deployment situation more in line with actual needs and providing the premise for the post-deployment detection effect to achieve the expected results.

[0144] Taking a single intersection as an example, the automated deployment of roadside equipment at the intersection using the aforementioned method includes the following:

[0145] 1. Obtain intersection data and generate road structured data.

[0146] (1) Identify intersection areas in high-definition remote sensing image maps through the U-Net network.

[0147] (2) Use the Canny edge detection algorithm to extract the lane contours at the intersection.

[0148] (3) Based on the parallel characteristics of lane lines, the intersection road is matched from the lane contour.

[0149] (4) Use yolov5 to identify the intersection poles in the image.

[0150] (5) Manually adjust the identified structural data according to actual conditions.

[0151] The above five steps are common steps for extracting structured road data. However, if the structured road data can be directly obtained, then these steps are not necessary. For example, the structured information of intersections can be directly obtained through Baidu / Amap.

[0152] In this embodiment, Figure 2 When extracting road structured data from intersection A, directly use Figure 3 The structured map provided by the Baidu / AutoNavi map program shown in the figure can be used to obtain the structured information of the intersection shown in Table 5. The structured data shown in Table 5 can be directly obtained through the pictures taken by the camera. Based on this structured data, a deployment plan that can be implemented and has predictable detection accuracy can be quickly and accurately given, without the need for field surveys to obtain the optimal deployment plan in the subsequent steps.

[0153] Table 5

[0154]

[0155] 2. Enter the performance parameters of the equipment and select a group of qualified equipment from the database based on the number of roads, pole locations, and service capabilities.

[0156] like Figure 4 As shown, the locations where sensors can be installed at intersections: The figure shows the locations where poles are already in place, and there is no need to re-erect the poles.

[0157] 3. Set the threshold and weight that affects the sensor perception accuracy

[0158] (1) Factors affecting perception accuracy: The slope in the AE direction will affect radar detection and the camera illumination area.

[0159] (2) Intersection perception accuracy requirements:

[0160] Perception coverage range ≥ 99%, time accuracy ≤ 10ms, delay ≤ 200ms, message output frequency ≥ 5HZ, recognition accuracy, longitudinal positioning error in the module direction ± 2cm, car ≥ 99%, bike ≥ 99%, person ≥ 99%, tracking accuracy ≥ 80%.

[0161] Able to detect illegal parking, wrong-way driving, and lane change light events

[0162] Expected cost: 200,000

[0163] 4. Input the device and intersection data into the perception capability scoring model to obtain the perception capability score.

[0164] The intersection structured information in the precondition table 1, (1) the factors affecting perception accuracy, and (2) the intersection perception accuracy requirements are input into the automated deployment program and the scoring model parameters are manually adjusted:

[0165] Manually adjust the weights of each influencing factor in the perception scoring model: Plane positioning * weight + Dimension detection accuracy * weight + Speed ​​detection value * weight + Dimension detection accuracy * weight + Perception range coverage * weight + Cost * weight + Equipment selection * weight = Score

[0166] In this embodiment, the manually adjusted weight parameters of the perception scoring model for intersection A are as follows:

[0167] Plane positioning detection accuracy * 0.80 + dimensional detection accuracy * 0.83 + speed detection value * 0.95 + dimensional detection accuracy * 0.95 + sensing range coverage * 0.87 + cost * 0.86 + equipment selection * 0.76 = 87.89

[0168] 5. Repeat steps 2-4 until all valid combinations have been traversed.

[0169] Based on the selection of multiple device combinations, a total of four deployment solutions are obtained as shown in Table 6:

[0170] Table 6

[0171]

[0172] 6. Filter out the device combination with the highest perception capability score, and the corresponding deployment plan is the optimal deployment plan. Automatically generate deployment plan construction drawings, and the actual deployment can be completed directly according to the construction drawings.

[0173] After comparing the perception ability scores of the four solutions in Table 6, Solution 1 obtained the highest score, so Solution 1 was selected as the best deployment solution for intersection A.Figure 5 The construction drawing of the deployment plan is shown.

[0174] This method extracts road structure data solely through camera capture, without requiring or minimizing on-site inspections. It automatically matches road structure information with roadside equipment, generating multiple deployment options. By calculating a comprehensive perception score, the highest-scoring option is selected as the optimal one, and construction drawings are directly generated. This entire process is simplified and automated, while maintaining a certain level of accuracy. This effectively saves costs and resources, makes actual deployment plans predictable, and ensures that detection results meet expectations.

[0175] Example 2

[0176] Different from Example 1, in addition to the intersection structured information in Table 5 of Example 1, this embodiment also adds other parameter items: intersection type, whether there is a curve, and more rich structured data of the intersection, which can enable the intersection and the connected road sections to be more accurately matched with the roadside equipment, provide more optional deployment plans, and make it easier to screen out the best deployment plan from these optional deployment plans.

[0177] Example 3

[0178] In this embodiment, an automated vehicle-road collaborative deployment system for roadside equipment is used to implement the automated vehicle-road collaborative deployment method for roadside equipment as described above, and includes a central processing unit, and an intersection data acquisition module, a device parameter acquisition module, a threshold weight setting module, a special service capability condition setting module, a model building module, and a solution generation module, each of which is communicatively connected to the central processing unit.

[0179] An initial perception capability scoring model is preset in the model construction module. The central processing unit obtains the threshold weight and the limiting conditions from the threshold weight setting module and the special capability condition setting module respectively, and adjusts the weight parameters of the initial perception capability scoring model to obtain the current perception capability scoring model; the central processing unit obtains the intersection information and the equipment information from the intersection data acquisition module and the equipment parameter acquisition module respectively, matches the intersection information with the equipment information, and forms a variety of optional deployment schemes for installing roadside equipment at corresponding intersections; the central processing unit substitutes the intersection information and equipment information corresponding to all optional deployment schemes into the current perception capability scoring model, calculates the corresponding perception capability score, and sends the optional deployment scheme with the highest score as the best deployment scheme to the scheme generation module for output.

[0180] Through this system, the automated vehicle-road collaborative deployment of roadside equipment can be completed at any intersection. By generating construction drawings of the optimal deployment plan, the actual construction effect can be restored to the design purpose as much as possible, the detection results can achieve the expected purpose, and resources and costs can be greatly saved.

[0181] The above is only an embodiment of the present invention, and the common knowledge such as the specific technical solutions and / or characteristics in the solution are not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An automated vehicle-road collaborative deployment method for roadside equipment, characterized in that: The following steps are involved: S1. Obtain intersection data. Using road-level maps, panoramic maps, and imagery, obtain intersection feature data, including intersection type, angles of deviation from true north in each direction, slope, and number of lanes. On-site surveys supplement the locations of existing poles and lane numbers that cannot be obtained along the route, generating structured road data. S2: Enter the performance parameters of the VIS roadside equipment. This includes the theoretical and actual performance parameters of the VIS roadside equipment, including the broadcast range of the RSU roadside communication unit, the image processing capability of the MEC roadside computing unit, the performance of the perception equipment, memory usage, CPU usage, and unit price. S3, setting the threshold and weight that affect the sensor perception accuracy; S4, configure special service capabilities: set limiting conditions based on the special service capabilities planned for the intersection; S5, Build a Perception Capability Scoring Model: This model is constructed based on weather conditions, structured road scenes, target pose, RTK positioning accuracy, and pre-installed location. Deployment plans for VIS roadside equipment and intersections that meet the requirements of S3 and S4 are considered optional deployment scenarios and substituted into the perception capability scoring model for scoring. S6, automatically generate a deployment plan, select the optional deployment plan with the highest score as the best deployment plan, and output the deployment plan construction drawing; In S5, the perception ability scoring model is based on the positioning accuracy. , heading angle detection accuracy , dimensional detection accuracy , speed measurement accuracy Comprehensive points: Perception score , In the formula, a, b, c, and d represent the weights of the corresponding indicators, all of which are greater than 0 and less than 1, and can be weighted for comprehensive scoring.

2. The automated vehicle-road collaborative deployment method for roadside equipment according to claim 1, characterized in that: All optional deployment solutions include hardware system deployment and software system deployment.

3. The method for automated vehicle-road collaborative deployment of roadside equipment according to claim 2, characterized in that: Software system deployment utilizes the computing power of industrial computers to automatically deploy the software system through IP, port and other parameterized configurations. The software is started with one click, reads configuration file parameters, and matches parameters to complete software system initialization. It uses multi-process synchronous collection and processing of sensor data and traffic light signals to complete the automated deployment of real-time data processing of the software system.

4. The method for automated vehicle-road collaborative deployment of roadside equipment according to claim 1, characterized in that: In S4, based on the special service capabilities of the intersection planning, the limiting conditions set include installing a telephoto camera in the direction of exiting the intersection in the pedestrian crossing area at the intersection where sudden cluster events are detected and the intersection where the queue length needs to be detected.

5. The method for automated vehicle-road collaborative deployment of roadside equipment according to claim 1, characterized in that: In S4, based on the special service capabilities of the intersection plan, the limiting conditions set include adding a telephoto camera at the zebra crossing to detect pedestrian crossings in specific areas.

6. The automated vehicle-road collaborative deployment method for roadside equipment according to claim 1, characterized in that: Configuring special service capabilities: According to the special service capabilities planned for intersections, when setting limiting conditions, set the installation area and number of roadside equipment including RSU and sensor equipment based on the geometric area of ​​the congestion area that can cause pedestrian gathering and vehicle congestion; at intersections where sudden cluster events are detected and where the queue length needs to be detected, install telephoto cameras in the direction of exiting the intersection in the pedestrian crossing area; the number of telephoto cameras is related to the width of the exit direction, the direction of the intersection and the length of the detection section. Only one telephoto camera is required within 100 meters in the width direction, and two telephoto cameras are required within 300 meters in the length direction.

7. The automated vehicle-road collaborative deployment method for roadside equipment according to claim 1, characterized in that: In S5, plane positioning accuracy The mean absolute error between the position detection value and the true value of the center point of all traffic participants perceived by the roadside perception system within the perception range and within a specific time in the local plane coordinate system can be estimated according to the following formula: Where: ——the two-dimensional plane position vector of the i-th sample in the system under test; ——The true position vector of the i-th sample in the two-dimensional plane; ——Euclidean distance; N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

8. The method for automated vehicle-road collaborative deployment of roadside equipment according to claim 1, characterized in that: In S5, heading angle detection accuracy The mean absolute error between the heading angle detection value and the true value of all traffic participants sensed by the roadside perception system within the perception range and within a specific time period can be estimated according to the following formula: Where: ——the heading angle of the system under test in the i-th sample; ——The true heading angle in the i-th sample; N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

9. The method for automated vehicle-road collaborative deployment of roadside equipment according to claim 1, characterized in that: In S5, speed measurement accuracy , within the perception range, the mean absolute error between the scalar speed detection value and the true value of the center point of all traffic participants perceived by the roadside perception system can be estimated as follows: Where: ——the speed of the system under test in the i-th sample; ——The true speed of the i-th sample; N is the total number of samples of all traffic participants detected at all perception moments within a specific time.

10. An automated vehicle-road collaborative deployment system for roadside equipment, characterized in that: The method for implementing the automated vehicle-road cooperative deployment for roadside equipment according to claim 1 comprises a central processing unit, and an intersection data acquisition module, an equipment parameter acquisition module, a threshold weight setting module, a special service capability condition setting module, a model building module, and a solution generation module, each of which is in communication with the central processing unit; An initial perception capability scoring model is preset in the model construction module. The central processing unit obtains the threshold weight and the limiting conditions from the threshold weight setting module and the special capability condition setting module respectively, and adjusts the weight parameters of the initial perception capability scoring model to obtain the current perception capability scoring model; the central processing unit obtains the intersection information and the equipment information from the intersection data acquisition module and the equipment parameter acquisition module respectively, matches the intersection information with the equipment information, and forms a variety of optional deployment schemes for installing roadside equipment at corresponding intersections; the central processing unit substitutes the intersection information and equipment information corresponding to all optional deployment schemes into the current perception capability scoring model, calculates the corresponding perception capability score, and sends the optional deployment scheme with the highest score as the best deployment scheme to the scheme generation module to output the deployment scheme construction drawing.

Images