Expressway traffic flow real-time induction method based on intelligent perception and unmanned aerial vehicle

Abstract

This application relates to the technical field of traffic control systems for road vehicles, and proposes a real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles (UAVs). The method includes: collecting vehicle speeds and headway data, as well as traffic flow in the road segment where the vehicles are located, through roadside equipment; extracting standard traffic data for the road segment where the vehicles are located; using UAVs to collect continuous frame images of vehicle movement within the time period of congestion source location, and identifying vehicles at the congestion source; identifying the driving trajectory of vehicles at the congestion source, calculating the degree of deviation and fluctuation of these vehicles, selecting the location analysis segment and upstream interfering vehicles, calculating the segment disturbance change index and local congestion correction parameters for vehicles at the congestion source, and combining the data determined by the roadside equipment to provide real-time traffic flow guidance on the highway. This application aims to improve the accuracy of traffic guidance.

Description

Technical Field

[0001] This application relates to the field of traffic control systems for road vehicles, specifically to a real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles (UAVs). Background Technology

[0002] With the continuous and substantial increase in my country's car ownership, highway traffic volume has been steadily rising, exacerbating traffic congestion, especially during holidays and peak hours. Frequent localized congestion caused by concentrated traffic flow not only severely reduces road efficiency and increases travel time but also easily leads to traffic accidents, threatening road safety. Therefore, accurate and real-time traffic guidance solutions have become crucial for alleviating these problems. Traditional traffic flow guidance systems primarily rely on overall traffic flow data collected by roadside equipment. Data processing is delayed, resulting in sluggish responses and difficulty in capturing sudden traffic anomalies in real time, failing to meet dynamic guidance needs. The rise of edge computing technology provides reliable support for the real-time acquisition and rapid processing of multi-source traffic data. Furthermore, drones possess the advantages of high-altitude hovering, wide-angle viewing, and flexible movement, effectively integrating global macroscopic traffic flow data provided by base stations with local microscopic dynamic data captured by drones to construct quantitative indicators that accurately reflect the frequency and amplitude of traffic flow fluctuations. This has become a critical technical challenge that urgently needs to be overcome in this field.

[0003] In real-world traffic scenarios, when vehicles frequently change lanes or experience localized slowdowns, drones can detect an increase in localized traffic "disorder" in real time. However, the global average speed data collected by base stations may still be within the normal range, leading to a discrepancy between the two assessments of traffic congestion. Current technology cannot accurately determine the source of this discrepancy between the local micro-state observed by drones and the global macro-state retrieved by roadside equipment. This prevents the effective fusion of multi-source data, directly impacting the accuracy and real-time performance of traffic guidance schemes and hindering rapid congestion relief. Summary of the Invention

[0004] This application provides a real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles (UAVs) to address the problem of insufficient accuracy in traffic guidance caused by the discrepancy between microscopic data collected by UAVs and macroscopic data collected by roadside equipment. The specific technical solution adopted is as follows:

[0005] One embodiment of this application provides a real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles (UAVs), the method comprising the following steps:

[0006] The speed and headway of vehicles on the highway, as well as the traffic flow of the road segment where the vehicles are located, are collected through preset roadside equipment.

[0007] Extract standard traffic data of the road segment where the vehicle is located, use drones to collect continuous frame images of vehicles traveling within the time period of the congestion source location, and combine the standard traffic data to identify the vehicles at the source of congestion.

[0008] The system identifies the driving trajectories of vehicles originating from congestion sources. Based on the differences in lateral displacement of adjacent vehicles relative to the lane centerline within these trajectories, it calculates the degree of deviation fluctuation of these vehicles. It then selects and analyzes the location of these vehicles and upstream interfering vehicles. Based on the distance and lateral displacement differences between upstream interfering vehicles and the vehicles originating from congestion sources, as well as the number of upstream interfering vehicles, it calculates the road segment disturbance change index for these vehicles. Combining this with vehicle speed and traffic flow data determined by roadside equipment and drones within the location analysis segment, it calculates local congestion correction parameters for these vehicles. Finally, by combining this with traffic flow, vehicle speed, and headway data determined by roadside equipment, it provides real-time traffic flow guidance on highways.

[0009] Furthermore, the standard traffic data includes the average speed and average headway of the highway segment where the roadside equipment is located.

[0010] Furthermore, the method for identifying the vehicles originating from the congestion is as follows:

[0011] When a vehicle's speed is less than the average speed of the road segment in which the vehicle is located, a high-definition camera mounted on a drone is used to collect continuous frames of vehicle driving images within a preset time period. The vehicle's speed and headway are calculated. When the headway of a vehicle is greater than twice the average headway, or when the vehicle's speed is continuously lower than a preset slow-moving threshold within a preset time period, the vehicle is recorded as a congestion source vehicle. The preset time period is the congestion source location time period.

[0012] Furthermore, the method for calculating the degree of deviation fluctuation of vehicles at the source of congestion is as follows:

[0013] Based on the driving trajectory of vehicles at the source of congestion during the time period of the congestion source location, obtain the lateral displacement sequence of the lateral displacement of the vehicles relative to the lane centerline.

[0014] For any lateral displacement in the lateral displacement sequence, the ratio of the absolute value of the difference between the lateral displacement and the previous adjacent lateral displacement to the acquisition time interval of the vehicle driving image in the adjacent frame is recorded as the degree of lateral displacement offset. The sum of the offset degrees of all lateral displacements in the lateral displacement sequence is recorded as the degree of offset fluctuation of the vehicle at the source of congestion.

[0015] Furthermore, the method for selecting the location analysis section and upstream interfering vehicles is as follows:

[0016] The location analysis segment is a segment extending upstream from the source of the congestion for a first preset length;

[0017] The drone is used to collect continuous frame images of all vehicles in the location analysis section within a preset time period. The offset fluctuation of all vehicles in the location analysis section is calculated, and vehicles with offset fluctuation greater than or equal to the fluctuation threshold are recorded as upstream interfering vehicles.

[0018] Furthermore, the specific calculation method for the road segment disturbance change index of the vehicles at the source of congestion is as follows:

[0019] The upstream disturbance intensity of the congestion source vehicle is calculated based on the difference in distance and lateral displacement between the upstream interfering vehicle and the vehicle at the source of the congestion.

[0020] The ratio of the number of upstream interfering vehicles to the total number of vehicles in a road segment extending upstream from the congestion source is denoted as the first ratio of the congestion source vehicles. When the lateral displacement of the congestion source vehicles exceeds the width of one lane, the sum of the number of times the lateral displacement of all vehicles exceeds the width of one lane within a second preset length behind the congestion source vehicles and during the congestion source positioning time period is denoted as the number of lane changes. The ratio of the number of lane changes to the second preset length is denoted as the second ratio. The product of the second ratio, the upstream disturbance intensity of the congestion source vehicles, and the first ratio is denoted as the road segment disturbance change index of the congestion source vehicles.

[0021] Furthermore, the specific calculation method for the upstream disturbance intensity of the vehicles at the source of the congestion is as follows:

[0022] The distance between the upstream interfering vehicle and the vehicle at the source of congestion is denoted as the first distance between the upstream interfering vehicle and the vehicle at the source of congestion. The sum of the absolute value of the difference between the lateral displacement of the upstream interfering vehicle and the vehicle at the source of congestion relative to the lane centerline and the standard lane width, plus the number 1, is denoted as the second distance between the upstream interfering vehicle and the vehicle at the source of congestion. The negative correlation between the first distance and the second distance between the upstream interfering vehicle and all vehicles at the source of congestion is denoted as the upstream disturbance intensity of the vehicles at the source of congestion.

[0023] Furthermore, the specific calculation method for the local congestion correction parameters of the vehicles at the source of congestion is as follows:

[0024] During the congestion source location time period, the average speed of all vehicles in the location analysis section determined by the roadside equipment and the drone is calculated, and denoted as macro vehicle speed and micro vehicle speed, respectively. The ratio of the number of vehicles to the total lane area in the location analysis section of the roadside equipment and the drone is calculated, and denoted as macro vehicle density and micro vehicle density, respectively. The macro vehicle speed and macro vehicle density are normalized and weighted summed by a preset weighted fusion algorithm to obtain the macro congestion prediction result. The micro vehicle speed and micro vehicle density are processed by Kalman filtering to obtain the micro congestion prediction result.

[0025] Based on the differences between macro-level and micro-level congestion prediction results, prediction heterogeneity is calculated.

[0026] The normalized value of the ratio of the road segment disturbance change index of vehicles at the source of congestion to the square of the predicted heterogeneity is denoted as the local congestion correction parameter for vehicles at the source of congestion.

[0027] Furthermore, the specific calculation method for predicting heterogeneity is as follows:

[0028] The ratio of the absolute value of the difference between the macro-level congestion prediction and the micro-level congestion prediction to the macro-level congestion prediction is denoted as prediction heterogeneity.

[0029] Furthermore, the specific steps for real-time traffic flow guidance on highways, combining traffic volume, vehicle speed, and headway determined by roadside equipment, include:

[0030] The traffic flow, vehicle speed, headway, and local congestion correction parameters of the vehicles at the source of congestion are collected by roadside equipment within the location analysis section of the road where the source of congestion is located and arranged sequentially into a feature vector. Kalman filtering is used to process the feature vector to obtain the congestion prediction result for the next adjacent time period after the location time of the congestion source.

[0031] When the congestion prediction result in the next adjacent time period after the congestion source location time period is greater than the preset congestion threshold, the drone is dispatched to reduce its altitude and provide driving guidance.

[0032] The beneficial effects of this application are:

[0033] This application considers that the high-definition camera on the drone can capture in real time non-steady-state driving behaviors such as frequent lane changes by vehicles in congested road sections, which are characterized by short duration, small displacement, and strong trajectory abrupt changes. Based on continuous frame vehicle driving images and standard traffic data collected by the drone, the application identifies vehicles that are the source of congestion; analyzes the impact of frequent lane changes by vehicles that are the source of congestion, calculates the degree of deviation fluctuation of vehicles that are the source of congestion and the road section disturbance change index, and realizes the quantification of the accumulation of micro-behavior to the road section level disturbance. The greater the degree of deviation fluctuation, the more significant the serpentine driving and frequent lane-changing behavior of vehicles that are the source of congestion, and the more obvious the disturbance of vehicles that are the source of congestion to traffic flow. Furthermore, by combining the vehicle speed and traffic flow determined by roadside equipment and drones within the location analysis section, local congestion correction parameters for vehicles at the source of congestion are calculated. The larger the local congestion correction parameters for vehicles at the source of congestion, the more severe the impact of local lane changes on the degree of congestion. Therefore, it is even more important to ensure that the micro-data collected by drones has a greater impact on the data fusion results. Finally, by combining the traffic flow, vehicle speed, and headway determined by roadside equipment, real-time traffic flow guidance is provided on highways to resolve the problem of insufficient accuracy in traffic guidance caused by the contradiction between the micro-data collected by drones and the macro-data collected by roadside equipment. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a schematic diagram of a real-time traffic flow guidance method for highways based on intelligent sensing and drones, provided as an embodiment of this application. Detailed Implementation

[0036] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0037] Please see Figure 1 The diagram illustrates a flowchart of a real-time highway traffic flow guidance method based on intelligent sensing and unmanned aerial vehicles (UAVs) according to an embodiment of this application. The method includes the following steps:

[0038] Step S001: Collect the speed and headway of vehicles on the highway, as well as the traffic flow of the road segment where the vehicles are located, through preset roadside equipment.

[0039] Roadside equipment is deployed at predetermined locations along the highway, and all roadside equipment forms a communication network covering key sections of the highway.

[0040] The roadside equipment includes cameras, radar, and ETC gantries. The cameras collect images of vehicles in motion, the radar detects vehicle speed and headway, and the ETC gantries calculate traffic flow on different sections of the highway.

[0041] At this point, the vehicle's speed and headway, as well as the traffic flow on the road segment, are obtained.

[0042] Step S002: Extract standard traffic data of the road segment where the vehicle is located, use a drone to collect continuous frame images of vehicles traveling within the time period of the congestion source location, and combine the standard traffic data to identify the vehicles at the source of congestion.

[0043] Frequent lane changes by vehicles in congested areas are non-steady-state, random driving behaviors characterized by short durations, small displacements, and highly abrupt trajectory changes. While high-definition cameras on drones can capture these instantaneous trajectory changes in real time, the macroscopic data collected by roadside equipment, due to its low sampling frequency and coarse positioning accuracy, cannot synchronously respond to these microscopic trajectory changes. This discrepancy leads to a spatiotemporal mismatch between the vehicle trajectories collected by drones and the macroscopic vehicle driving states acquired by base stations. This results in data inconsistencies in the congestion analysis of edge computing nodes, severely impacting the real-time performance and accuracy of drone-guided dispatching, and potentially exacerbating traffic chaos in localized areas. Therefore, it is necessary to focus on identifying and analyzing vehicle lane-changing behavior, utilizing drone trajectory data to determine the timing, duration, and direction of lane changes in real time. Furthermore, to address the limitation of base station-collected macroscopic data in capturing lane-changing behavior, lane-changing behavior compensation should be applied to the base station's macroscopic data to correct the macroscopic data and achieve dynamic matching between microscopic trajectories and macroscopic states.

[0044] Standard traffic data for all highway sections where roadside equipment is located is extracted. The standard traffic data includes average speed and average headway. Both the average speed and average headway are calculated by those skilled in the art. In this embodiment, the average speed and average headway are selected as the sliding average of the communication base station within 5 minutes prior to the current time to ensure the dynamic validity of the reference data.

[0045] When a vehicle's speed is lower than the average speed of the road segment it is on, a drone equipped with a high-definition camera is deployed to the vehicle's location. The high-definition camera captures consecutive frames of vehicle movement images within a preset time period. Based on these images, the vehicle's trajectory, timestamp, and location information are extracted, and its speed and headway are calculated. If a vehicle's headway is greater than twice the average headway, or if its speed remains consistently below a preset slow-moving threshold within the preset time period, the vehicle is identified as a congestion source. The time period for capturing these consecutive vehicle movement images is designated as the congestion source location time period.

[0046] The headway refers to the time interval between the front ends of two adjacent vehicles passing the same road reference point, measured in seconds; in this embodiment, the slow-moving threshold is set to 5 km / h.

[0047] Calculating the vehicle's trajectory, timestamp, and location information, as well as its speed and headway, from consecutive frames of vehicle driving images is a well-known technique and will not be elaborated further. In this embodiment, the preset time period is set to 5 seconds.

[0048] Using drones equipped with broadcast systems, guidance signs are projected onto vehicles at the source of congestion and surrounding vehicles, and voice reminders are played to help alleviate congestion.

[0049] At this point, the source of the congestion can be identified.

[0050] Step S003: Identify the driving trajectory of vehicles originating from congestion. Based on the difference in lateral displacement of adjacent vehicles relative to the lane centerline in the driving trajectory, calculate the degree of deviation fluctuation of vehicles originating from congestion. Select the location analysis section and upstream interfering vehicles through the vehicles originating from congestion. Based on the distance and lateral displacement difference between upstream interfering vehicles and vehicles originating from congestion, and the number of upstream interfering vehicles, calculate the road segment disturbance change index of vehicles originating from congestion. Combine the vehicle speed and traffic flow determined by roadside equipment and UAVs within the location analysis section, calculate the local congestion correction parameters of vehicles originating from congestion. Combine the traffic flow, vehicle speed, and headway determined by roadside equipment, and provide real-time guidance for highway traffic flow.

[0051] Existing congestion identification models primarily rely on longitudinal vehicle speed fluctuation analysis, making it difficult to accurately identify the micro-level driving behavior of vehicles originating from congestion sources. When upstream vehicles zigzag or frequently change lanes, they force vehicles at the congestion source to swerve or even stop. This lateral disturbance not only exacerbates the driving resistance of vehicles at the congestion source but also propagates across multiple lanes, causing wider traffic flow turbulence. However, vehicles that frequently change lanes exhibit relatively small speed changes, and traditional congestion identification models based on speed thresholds cannot effectively capture such lateral disturbances. This can easily lead to misjudgments or omissions of vehicles causing congestion, resulting in ineffective congestion mitigation solutions.

[0052] First, the impact of frequent lane changes by vehicles at the source of congestion is analyzed by using continuous frame images of vehicle driving.

[0053] Optical flow tracing is used to identify the driving trajectories of vehicles originating from congestion during the time period in which the congestion origin is located. The lateral displacement of these vehicles relative to the lane centerline is obtained, and a lateral displacement sequence is constructed. For any lateral displacement in the sequence, the ratio of the absolute value of the difference between the lateral displacement and the previous adjacent lateral displacement to the acquisition time interval of the adjacent frame vehicle driving image is recorded as the degree of lateral displacement offset. The sum of the offset degrees of all lateral displacements in the sequence is recorded as the degree of offset fluctuation of the congestion origin vehicle.

[0054] The time interval between acquiring adjacent frames of vehicle driving images is 0.05 seconds; the optical flow tracing method is a well-known technique for identifying driving trajectories in images and will not be elaborated further.

[0055] The greater the deviation and fluctuation of vehicles at the source of congestion, the more significant their serpentine driving and frequent lane-changing behavior, and the more obvious the disturbance to traffic flow caused by them.

[0056] A road segment extending upstream from the source of congestion for a predetermined length is selected and designated as the location analysis segment. A drone equipped with a high-definition camera is deployed to the location analysis segment to capture continuous frame images of all vehicles within the segment over a predetermined time period. The offset fluctuation of all vehicles within the location analysis segment is calculated using the same method as that used to calculate the offset fluctuation of vehicles at the source of congestion. Vehicles with offset fluctuation values ​​greater than or equal to the fluctuation threshold are designated as upstream interfering vehicles.

[0057] In this embodiment, the value of the fluctuation threshold is 0.5, and the value of the first preset length is 200 meters.

[0058] Next, the impact of lane-changing fluctuations by upstream interfering vehicles on the driving conditions within the road segment is analyzed. The impact of lane-changing fluctuations by upstream interfering vehicles on the driving pressure of the road segment where the congestion source vehicle is located follows three major transmission rules: decreasing pressure from near to far, lane association, and chain diffusion. Therefore, the closer the spatial distance between the upstream interfering vehicle and the congestion source vehicle, the greater the pressure of interference in the same lane, and the more likely there is to be frequent lane changes, the greater the degree of interference from the upstream interfering vehicle on the driving conditions of the congestion source vehicle.

[0059] The distance between the upstream interfering vehicle and the vehicle at the source of congestion is denoted as the first distance between the upstream interfering vehicle and the vehicle at the source of congestion. The sum of the absolute value of the difference between the lateral displacement of the upstream interfering vehicle and the vehicle at the source of congestion relative to the lane centerline and the standard lane width, plus the number 1, is denoted as the second distance between the upstream interfering vehicle and the vehicle at the source of congestion. The negative correlation between the first distance and the second distance between the upstream interfering vehicle and all vehicles at the source of congestion is denoted as the upstream disturbance intensity of the vehicles at the source of congestion.

[0060] It is understandable that a negative correlation is applied to the first and second distances between the upstream interfering vehicles and all vehicles originating from the congestion source, ensuring that both distances are negatively correlated with the upstream disturbance intensity of the congestion source vehicles. It is also understood that the negative correlation in this application refers to the relationship between the independent and dependent variables. The independent variables are the first and second distances between the upstream interfering vehicles and the vehicles originating from the congestion source, and the dependent variable is the upstream disturbance intensity of the congestion source vehicles. The negative correlation means that the dependent variable decreases (increases) as the independent variable increases (decreases), and can be an inverse relationship, a subtraction relationship, etc.

[0061] Preferably, as an embodiment of this application, the negative of the sum of the products of the first distance and the second distance between the upstream interfering vehicle and all congestion source vehicles is used as the exponent of an exponential function with the natural constant as the base, and the calculated value of the exponential function is recorded as the upstream disturbance intensity of the congestion source vehicles.

[0062] The greater the distance between the upstream interfering vehicle and the vehicle at the source of congestion, and the greater the difference in lateral displacement between the upstream interfering vehicle and the vehicle at the source of congestion relative to the lane centerline, the less the vehicle at the source of congestion is affected by the movement of the vehicle at the source of congestion. In this case, the upstream disturbance intensity of the vehicle at the source of congestion is smaller.

[0063] The serpentine movements or frequent lane changes of vehicles will increase the avoidance actions of other vehicles in the road segment. Frequent lane changes by upstream interfering vehicles will force multiple surrounding vehicles to make continuous avoidance and chain lane changes, forming a disturbance amplification effect. Therefore, it is necessary to comprehensively consider the lateral disturbances of all vehicles in the road segment and the effective number of lane changes to quantify the degree of traffic flow disorder perceived by the drone, directly reflecting the intensity of frequent lane changes by vehicles.

[0064] The ratio of the number of upstream interfering vehicles to the total number of vehicles in a road segment extending upstream from the congestion source is denoted as the first ratio of the congestion source vehicles. When the lateral displacement of the congestion source vehicles exceeds the width of one lane, the sum of the number of times the lateral displacement of all vehicles exceeds the width of one lane within a second preset length behind the congestion source vehicles and during the congestion source positioning time period is denoted as the number of lane changes. The ratio of the number of lane changes to the second preset length is denoted as the second ratio. The product of the second ratio, the upstream disturbance intensity of the congestion source vehicles, and the first ratio is denoted as the road segment disturbance change index of the congestion source vehicles.

[0065] In this embodiment, the values ​​for the first preset length and the second preset length are 200 meters and 100 meters, respectively.

[0066] The road segment disturbance change index of vehicles at the source of congestion realizes the quantification of micro-behavior accumulation to road segment-level disturbance.

[0067] During the time period for locating the congestion source, the average speed of all vehicles detected by radar within the location analysis segment of the congestion source vehicle is calculated and recorded as the macro vehicle speed. The ratio of the number of vehicles counted by the ETC gantry within the location analysis segment to the total lane area is calculated and recorded as the macro vehicle density. The macro vehicle speed and macro vehicle density are normalized and weighted and summed using a preset weighted fusion algorithm to obtain the macro congestion prediction result.

[0068] Within the time period for locating the congestion source, the average speed of all vehicles within the location analysis segment determined by the high-definition camera mounted on the drone is calculated and denoted as the micro vehicle speed. The ratio of the number of vehicles within the location analysis segment determined by the high-definition camera mounted on the drone to the total lane area is calculated and denoted as the micro vehicle density. Kalman filtering is used to process the micro vehicle speed and micro vehicle density to obtain the micro congestion prediction results.

[0069] Both the micro-level and macro-level congestion prediction results are congestion prediction results. The congestion prediction result is a percentage greater than or equal to 0 and less than or equal to 1. It is used to assess the degree of congestion in the location analysis section determined by the congestion source vehicle. The larger the congestion prediction result, the greater the difference between the congestion status determined by roadside equipment and drones in the location analysis section determined by the congestion source vehicle, and the less reliable the congestion status determined by the data of roadside equipment.

[0070] The ratio of the absolute value of the difference between the macro-level congestion prediction and the micro-level congestion prediction to the macro-level congestion prediction is denoted as prediction heterogeneity.

[0071] The normalized value of the ratio of the road segment disturbance change index of vehicles at the source of congestion to the square of the predicted heterogeneity is denoted as the local congestion correction parameter for vehicles at the source of congestion.

[0072] It should be noted that this embodiment uses the sigmoid function to calculate the normalized value. In practical applications, implementers may use other methods of existing technology, such as the tanh function or the maximum-minimum normalization method, to calculate the normalized value, and no limitation is made here.

[0073] The larger the local congestion correction parameter for vehicles at the source of congestion, the more severe the impact of local lane changes on the degree of congestion, and the greater the influence of micro-data collected by drones on the data fusion results should be.

[0074] The traffic flow collected by the ETC gantry, the speed of vehicles detected by radar, the headway, and the local congestion correction parameters of the vehicles at the source of congestion within the identified road segment are arranged sequentially into a feature vector. Kalman filtering is used to process the feature vector to obtain the congestion prediction result for the next adjacent time period after the time period of the congestion source location.

[0075] When the congestion prediction result in the next adjacent time period after the congestion source location time period is greater than the preset congestion threshold, the drone is dispatched to reduce its altitude, identify the license plate of the vehicle at the congestion source, and hover at a preset safe height. The drone broadcasts voice reminders and sends guidance information to the communication base station to control the information boards along the route, so as to realize real-time driving guidance, help alleviate congestion, and reduce the frequency of vehicles changing lanes frequently.

[0076] When performing calculations involving division, such as the first ratio, the second ratio, predicted heterogeneity, and local congestion correction parameters, a preset minimum positive number is added to the denominator by default to prevent calculation failure due to a zero denominator under extreme conditions. In this embodiment, the preset minimum positive number is 0.0001; extreme conditions include a total number of vehicles of 0, predicted heterogeneity of 0, and macroscopic congestion prediction result of 0.

[0077] In this embodiment, the congestion threshold is set to 75%.

[0078] This enables real-time traffic flow guidance on highways.

[0079] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for real-time highway traffic flow guidance based on intelligent sensing and unmanned aerial vehicles, characterized in that, The method includes the following steps: The speed and headway of vehicles on the highway, as well as the traffic flow of the road segment where the vehicles are located, are collected through preset roadside equipment. Extract standard traffic data of the road segment where the vehicle is located, use drones to collect continuous frame images of vehicles traveling within the time period of the congestion source location, and combine the standard traffic data to identify the vehicles at the source of congestion. The system identifies the driving trajectories of vehicles originating from congestion sources. Based on the differences in lateral displacement of adjacent vehicles relative to the lane centerline within these trajectories, it calculates the degree of deviation fluctuation of these vehicles. It then selects and analyzes the location of these vehicles and upstream interfering vehicles. Based on the distance and lateral displacement differences between upstream interfering vehicles and the vehicles originating from congestion sources, as well as the number of upstream interfering vehicles, it calculates the road segment disturbance change index for these vehicles. Combining this with vehicle speed and traffic flow data determined by roadside equipment and drones within the location analysis segment, it calculates local congestion correction parameters for these vehicles. Finally, by combining this with traffic flow, vehicle speed, and headway data determined by roadside equipment, it provides real-time traffic flow guidance on highways.

2. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The standard traffic data includes the average speed and average headway of the highway segment where the roadside equipment is located.

3. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 2, characterized in that, The method for identifying the vehicles that are the source of the congestion is as follows: When a vehicle's speed is less than the average speed of the road segment in which the vehicle is located, a high-definition camera mounted on a drone is used to collect continuous frames of vehicle driving images within a preset time period. The vehicle's speed and headway are calculated. When the headway of a vehicle is greater than twice the average headway, or when the vehicle's speed is continuously lower than a preset slow-moving threshold within a preset time period, the vehicle is recorded as a congestion source vehicle. The preset time period is the congestion source location time period.

4. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The method for calculating the degree of deviation fluctuation of vehicles at the source of congestion is as follows: Based on the driving trajectory of vehicles at the source of congestion during the time period of the congestion source location, obtain the lateral displacement sequence of the lateral displacement of the vehicles relative to the lane centerline. For any lateral displacement in the lateral displacement sequence, the ratio of the absolute value of the difference between the lateral displacement and the previous adjacent lateral displacement to the acquisition time interval of the vehicle driving image in the adjacent frame is recorded as the degree of lateral displacement offset. The sum of the offset degrees of all lateral displacements in the lateral displacement sequence is recorded as the degree of offset fluctuation of the vehicle at the source of congestion.

5. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The method for selecting the location analysis road segment and upstream interfering vehicles is as follows: The location analysis segment is a segment extending upstream from the source of the congestion for a first preset length; The drone is used to collect continuous frame images of all vehicles in the location analysis section within a preset time period. The offset fluctuation of all vehicles in the location analysis section is calculated, and vehicles with offset fluctuation greater than or equal to the fluctuation threshold are recorded as upstream interfering vehicles.

6. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The specific calculation method for the road segment disturbance change index of vehicles at the source of congestion is as follows: The upstream disturbance intensity of the congestion source vehicle is calculated based on the difference in distance and lateral displacement between the upstream interfering vehicle and the vehicle at the source of the congestion. The ratio of the number of upstream interfering vehicles to the total number of vehicles in a road segment extending upstream from the source of the congestion is denoted as the first ratio of the vehicles at the source of congestion. When the lateral displacement of a vehicle at the source of congestion exceeds the width of one lane, the sum of the number of times the lateral displacement of all vehicles exceeds the width of one lane within the second preset length range behind the vehicle at the source of congestion and during the time period when the vehicle is located at the source of congestion is recorded as the number of lane changes. The ratio of the number of lane changes to the second preset length is recorded as the second ratio. The product of the second ratio, the upstream disturbance intensity of the vehicle at the source of congestion, and the first ratio is recorded as the road segment disturbance change index of the vehicle at the source of congestion.

7. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 6, characterized in that, The specific calculation method for the upstream disturbance intensity of the vehicles at the source of the congestion is as follows: The distance between the upstream interfering vehicle and the vehicle at the source of congestion is denoted as the first distance between the upstream interfering vehicle and the vehicle at the source of congestion. The sum of the absolute value of the difference between the lateral displacement of the upstream interfering vehicle and the vehicle at the source of congestion relative to the lane centerline and the standard lane width, plus the number 1, is denoted as the second distance between the upstream interfering vehicle and the vehicle at the source of congestion. The negative correlation between the first distance and the second distance between the upstream interfering vehicle and all vehicles at the source of congestion is denoted as the upstream disturbance intensity of the vehicles at the source of congestion.

8. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The specific calculation method for the local congestion correction parameters of the vehicles at the source of congestion is as follows: During the congestion source location time period, the average speed of all vehicles in the location analysis section determined by the roadside equipment and the drone is calculated, and denoted as macro vehicle speed and micro vehicle speed, respectively. The ratio of the number of vehicles to the total lane area in the location analysis section of the roadside equipment and the drone is calculated, and denoted as macro vehicle density and micro vehicle density, respectively. The macro vehicle speed and macro vehicle density are normalized and weighted summed by a preset weighted fusion algorithm to obtain the macro congestion prediction result. The micro vehicle speed and micro vehicle density are processed by Kalman filtering to obtain the micro congestion prediction result. Based on the differences between macro-level and micro-level congestion prediction results, prediction heterogeneity is calculated. The normalized value of the ratio of the road segment disturbance change index of vehicles at the source of congestion to the square of the predicted heterogeneity is denoted as the local congestion correction parameter for vehicles at the source of congestion.

9. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 8, characterized in that, The specific calculation method for predicting heterogeneity is as follows: The ratio of the absolute value of the difference between the macro-level congestion prediction and the micro-level congestion prediction to the macro-level congestion prediction is denoted as prediction heterogeneity.

10. The real-time traffic flow guidance method for highways based on intelligent sensing and unmanned aerial vehicles according to claim 1, characterized in that, The real-time guidance of highway traffic flow based on traffic volume, vehicle speed, and headway determined by roadside equipment includes the following specific steps: The traffic flow, vehicle speed, headway, and local congestion correction parameters of the vehicles at the source of congestion are collected by roadside equipment within the location analysis section of the road where the source of congestion is located and arranged sequentially into a feature vector. Kalman filtering is used to process the feature vector to obtain the congestion prediction result for the next adjacent time period after the location time of the congestion source. When the congestion prediction result in the next adjacent time period after the congestion source location time period is greater than the preset congestion threshold, the drone is dispatched to reduce its altitude and provide driving guidance.

Images