A calibration method for a signal-free intersection model based on UAV aerial photography data

Abstract

A calibration method for a signal - free intersection model based on UAV aerial photography data, belonging to the field of intelligent transportation technology. To improve the simulation results of the two - parameter gap - acceptance model, the present invention includes obtaining road network data, collecting vehicle trajectory data and intersection aerial images through UAVs; processing the obtained road network data, merging one - way lines of the road network, and then classifying intersections to be calibrated; processing the obtained vehicle trajectory data, removing missing or abnormal data to obtain processed vehicle trajectory data; associating the obtained processed vehicle trajectory data with the corresponding intersection turns; calibrating the expected speeds for different types of intersections and different turns; calibrating the critical gaps and critical waiting times for different types of intersections. The present invention optimizes the simulation results of the two - parameter gap - acceptance model.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent transportation, and particularly relates to a method for calibrating a signal-free intersection model based on unmanned aerial vehicle (UAV) aerial photography data. Background Art

[0002] There is a relatively common problem in many mid- and micro-scale simulation software. Regardless of how long the vehicles wait at the intersection during traffic flow simulation, the minimum acceptable gap is a fixed value, which does not conform to the actual situation. Generally, when the waiting time increases, motor vehicle drivers will become impatient and are willing to accept a smaller gap to pass through the intersection more quickly.

[0003] For example, a two-parameter gap acceptance model can simulate the impatience effect when a driver cannot enter a conflicting traffic flow. At the conflict point of the intersection, the decision of whether a low-priority vehicle will precede a high-priority vehicle is based on two variables, namely the available gap and the relative waiting time . These two variables are used together with the critical gap G and the critical waiting time W parameters to calculate the probability that a vehicle with a lower priority precedes a vehicle with a higher priority (i.e., the priority probability P ). The model form is as follows:

[0004] ;

[0005] Currently, there are few parameter calibration methods for such two-parameter gap acceptance models. Usually, they are set according to experience or default parameters, resulting in insufficient refinement of parameter calibration and affecting the simulation results. Summary of the Invention

[0006] The problem to be solved by the present invention is to improve the simulation results of the two-parameter gap acceptance model, and propose a method for calibrating a signal-free intersection model based on UAV aerial photography data.

[0007] To achieve the above object, the present invention is realized through the following technical solutions:

[0008] A method for calibrating a signal-free intersection model based on UAV aerial photography data, comprising the following steps:

[0009] S1. Obtain road network data, and collect vehicle trajectory data and intersection aerial photography maps through a UAV;

[0010] S2. Process the road network data obtained in step S1, merge the one-way lines of the road network, and then classify the intersections to be calibrated for parameters;

[0011] S3. Process the vehicle trajectory data obtained in step S1, remove the missing or abnormal data, and obtain the processed vehicle trajectory data;

[0012] S4. Associate the processed vehicle trajectory data obtained in step S3 with the corresponding intersection turns;

[0013] S5. Calibrate the expected speeds for different types of intersections and different turns;

[0014] S6. Calibrate the critical gaps and critical waiting times for different types of intersections.

[0015] Furthermore, the road network data in step S1 includes the unique segment number, length, direction, road grade, and number of lanes attributes; the vehicle trajectory data is extracted from the UAV aerial video with a time accuracy of 0.1 s, and includes the vehicle unique number, timestamp, lane number, longitude, latitude, and speed; the intersection aerial map includes the intersection area and the sign and marking information.

[0016] Furthermore, the specific implementation method of step S2 includes the following steps:

[0017] S2.1. Road network one-way line merging: Merge the two single lines representing the two directions of the same road into a single line representing the two directions of the road;

[0018] S2.2. Classification of intersections with parameters to be calibrated, and classify the intersections to be calibrated without signals into cross intersections, T-shaped intersections, roundabouts, and confluence openings.

[0019] Furthermore, the specific implementation method of step S3 includes the following steps:

[0020] S3.1. Remove the data with any one of the fields of the missing vehicle unique number, timestamp, lane number, longitude, latitude, and speed;

[0021] S3.2. Remove the data with a speed greater than 120;

[0022] S3.3. If the proportion of the removed data volume of a certain vehicle trajectory data exceeds 5%, then delete the vehicle trajectory data.

[0023] Furthermore, the specific implementation method of step S4 includes the following steps:

[0024] S4.1. For the processed vehicle trajectory data obtained in step S3, filter out the vehicle trajectory data within 50 meters of the center point of each intersection based on the location;

[0025] S4.2. Group the vehicle trajectory data obtained in step S4.1 based on the vehicle ID, and sort the vehicle trajectory data in chronological order;

[0026] Based on the fact that the same vehicle may pass through an intersection multiple times and the turns when passing through the intersection at different times are different, grouping flags are set between adjacent trajectory points with a time difference greater than 600 s to further group the vehicle trajectory data, and grouping numbers PERIOD_ID are set respectively.

[0027] S4.3. Traverse each group of vehicle trajectory data based on vehicle ID and PERIOD_ID, select the first and last trajectory points of the group, associate the road section numbers LINK_ID corresponding to the trajectory points through location, the vehicle number corresponding to the first trajectory point, and the incoming road section FROM_LINK where the vehicle passes through the intersection, and the outgoing road section TO_LINK corresponding to the last trajectory point, so as to determine the turning number MOVEMENT_ID of the vehicle at the intersection and establish a one-to-one correspondence between the vehicle trajectory and the intersection turning.

[0028] Furthermore, the specific implementation method of step S5 includes the following steps:

[0029] S5.1. On the basis of step S4, filter the vehicle trajectory data with time from 10:00 to 16:00 for calibrating the turning speed.

[0030] S5.2. Group the vehicle trajectory data according to the intersection turning class and a 5-minute time interval, and calculate the average speed of vehicles in each group group , and the calculation formula is as follows: ;

[0031] ;

[0032] Among them, represents the average vehicle speed data of the class th group group of the intersection turning i , N represents the data volume of the class th group group of the intersection turning

[0033] S5.3. Sort the average vehicle speeds of each group in ascending order of speed, and take the data with a speed at the 50% quantile as the expected speed of the intersection turning.

[0034] Furthermore, the specific implementation method of step S6 includes the following steps:

[0035] S6.1. Set the turning type of the parameters to be calibrated and the corresponding high-priority turns according to the intersection type:

[0036] Set the turning type of the calibration parameters at the cross intersection to left turn on the approach lane. The straight movement on the oncoming approach lane has a higher priority than the left turn on the approach lane, and the left turn on the approach lane has a lower priority than the straight movement on the oncoming approach lane;

[0037] Set the turning type of the calibration parameters at the roundabout to the roundabout entrance. The straight movement in the outermost lane of the roundabout road has a higher priority than the roundabout entrance, and the roundabout entrance has a lower priority than the straight movement in the outermost lane of the roundabout road;

[0038] Set the turning type of the calibration parameters at the merge point to the ramp entrance. The straight movement on the main line has a higher priority than the ramp entrance, and the ramp entrance has a lower priority than the straight movement on the main line;

[0039] S6.2. Calibrate the critical gap and critical waiting time parameters of the low-priority turns in Step S6.1:

[0040] S6.2.1. Traverse the intersections with calibration parameters and obtain the vehicle trajectory point data within 50 meters of the center points of the intersections with calibration parameters;

[0041] S6.2.2. Identify the stopping positions of vehicles with different turns:

[0042] Divide the approach road section into m sections at 5-meter intervals. Divide the vehicle trajectory point data obtained in Step S6.2.1 into the corresponding sections according to the coordinates, and count the number of trajectory points with a speed < 5 km / h in each section. The section with the largest number of points is the straight or left-turn stopping position section ;

[0043] S6.2.3. Sort the vehicle trajectory points:

[0044] Filter the trajectory points of low-priority turning vehicles and sort the vehicles based on the time of the first trajectory point of the vehicle; filter the trajectory points of high-priority turning vehicles and sort the vehicles based on the time of the first trajectory point of the vehicle; let the set of available gap data of vehicles passing through the intersection be denoted as The form of each data in the set is For the high-priority turning vehicles and low-priority turning vehicles with conflicts at the intersection, let be the time difference between the arrival time of the high-priority turning vehicle at the stopping position and the arrival time of the low-priority turning vehicle at the stopping position. If the low-priority vehicle passes through the intersection first, let and add the data to If the high-priority vehicle passes through the intersection first, let and add the data to The set of relative waiting time data of vehicles passing through the intersection is denoted as , each piece of data in the set is in the form of , let be the difference between the waiting time of the low-priority vehicle at the parking position and the waiting time of the high-priority vehicle at the parking position. If the low-priority vehicle passes through the intersection, let , and add the data to . If the high-priority vehicle passes through the intersection first, let , and add the data to ;

[0045] S6.2.4. Identify the arrival and departure times of vehicles at the parking position:

[0046] Traverse the trajectory points of the low-priority vehicle , and find the point that enters the straight or left-turn parking position section and the point that leaves the straight or left-turn parking position section . and are the times of entering and leaving the straight or left-turn parking position section respectively; Traverse the trajectory points of the high-priority vehicle , and find the point that enters the straight or left-turn parking position section and the point that leaves the straight or left-turn parking position section . and are the times of entering and leaving the straight or left-turn parking position section respectively;

[0047] S6.2.5. Identify the order relationship of vehicles passing through the intersection:

[0048] S6.2.5.1. Traverse the low-priority vehicle ;

[0049] S6.2.5.2. Traverse the trajectory points of the low-priority vehicle. The time corresponding to this point is denoted as t , and the horizontal and vertical coordinates are denoted as , respectively. Let the next point be , where = t + 0.5, and the horizontal and vertical coordinates are denoted as , respectively;

[0050] S6.2.5.3. Traverse high-priority vehicles ;

[0051] S6.2.5.4. Traverse the trajectory points of high-priority vehicles , and record the corresponding time of this point as u , and record the horizontal and vertical coordinates as 、 , let the next point be , where = u + 0.5, and record the horizontal and vertical coordinates as 、 ;

[0052] S6.2.5.5. Identify line segments and whether there are conflicts in time and space: If the following formula holds, then line segments and are in spatio-temporal conflict, enter S6.2.5.6, otherwise return to S6.2.5.4 and continue to traverse the next trajectory point of high-priority vehicles. The formula is:

[0053] ;

[0054] ;

[0055] ;

[0056] S6.2.5.6. Identify the order in which conflicting vehicles pass through the intersection: If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, and enter S6.2.5.1 to continue traversing the next low-priority vehicle; If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, and enter S6.2.5.3 to continue traversing the next high-priority vehicle;

[0057] S6.2.6. Filter the set of available gap data of vehicles passing through the intersection 's data, denoted as , and group and count them at 0.5s time intervals respectively Frequency and , where is the total number of data in that meet is the total number of data in that meet Calculate the frequency of , and the expression is:

[0058] ;

[0059] Obtain the minimum value of and that meet ; ;

[0060] Obtain the maximum value of and that meet ; ;

[0061] According to the two-parameter gap acceptance model, let , , where G is the critical gap;

[0062] Calibrate the final critical gap G through the following formula, and the expression is:

[0063] ;

[0064] S6.2.7. Filter the relative waiting time data set in the data, denoted as , and group and count the frequency of respectively at 0.5s time intervals and , where is the total number of data in that meet is the total number of data in that meet Calculate the frequency of , and the expression is:

[0065] ;

[0066] Obtain the minimum value of and that meet ; ;

[0067] Obtain those that satisfy and of the maximum value ;

[0068] According to the two-parameter gap acceptance model, let , , where W is the critical waiting time;

[0069] Calculate the critical waiting time through the following formula W , and the expression is:

[0070] .

[0071] Advantages of the present invention:

[0072] A method for calibrating a signal-free intersection model based on UAV aerial photography data according to the present invention can quickly and comprehensively obtain high-precision vehicle trajectory data of the intersection by using UAV aerial photography data, thereby calibrating the parameters of the two-parameter gap acceptance model. Compared with traditional empirical settings and manual measurement methods, it greatly reduces the errors and workload of manual measurement, and greatly improves work efficiency and the accuracy of calibration results. In addition, the results of the present invention are applied to traffic simulation at signal-free intersections, which can better reflect the actual traffic conditions at signal-free intersections, contribute to optimizing traffic planning and design, and have good economic benefits. Description of the drawings

[0073] Figure 1 is a flowchart of a method for calibrating a signal-free intersection model based on UAV aerial photography data according to the present invention;

[0074] Figure 2 is a structural block diagram of a method for calibrating a signal-free intersection model based on UAV aerial photography data according to the present invention;

[0075] Figure 3 is a classification diagram of intersections with parameters to be calibrated according to the present invention. Detailed implementation manners

[0076] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners. It should be understood that the specific implementation manners described herein are only used to explain the present invention and are not used to limit the present invention, that is, the specific implementation manners described are only a part of the implementation manners of the present invention, rather than all of the specific implementation manners. The components of the specific implementation manners of the present invention usually described and shown in the drawings here can be arranged and designed in various different configurations, and the present invention can also have other implementation manners.

[0077] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected specific embodiments of the present invention. Based on the specific embodiments of the present invention, all other specific embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0078] In order to further understand the content, features and effects of the present invention, the following specific implementation methods are given as examples, and the attached Figure 1 -Attached Figure 3 The detailed instructions are as follows:

[0079] Embodiment 1:

[0080] A method for calibrating a model of an unsignalized intersection based on drone aerial photography data comprises the following steps:

[0081] S1. Obtain road network data and collect vehicle trajectory data and aerial photos of intersections through drones;

[0082] Furthermore, the road network data in step S1 includes the unique number of the road section, length, direction, road grade, and number of lanes; the vehicle trajectory data is extracted through drone aerial video with a time accuracy of 0.1s, including the unique vehicle number, timestamp, lane number, longitude, latitude, and speed; the aerial view of the intersection includes the intersection area and sign and marking information.

[0083] S2. Process the road network data obtained in step S1, merge the one-way lines of the road network, and then classify the intersections to be calibrated parameters;

[0084] Furthermore, the specific implementation method of step S2 includes the following steps:

[0085] S2.1. Merge one-way lines in road network: Merge two single lines representing two directions of the same road into a single line representing the two directions of the road;

[0086] S2.2. Classification of parameter intersections to be calibrated: divide the unsignalized intersections to be calibrated into cross intersections, T-intersections, roundabouts, and merges.

[0087] S3. The vehicle trajectory data obtained in step S1 is processed to remove missing or abnormal data to obtain processed vehicle trajectory data;

[0088] Furthermore, the specific implementation method of step S3 includes the following steps:

[0089] S3.1. Eliminate data that is missing any of the vehicle unique number, timestamp, lane number, longitude, latitude, and speed fields;

[0090] S3.2. Eliminate data with a speed greater than 120;

[0091] S3.3. If the proportion of the excluded data of a vehicle trajectory data exceeds 5%, the vehicle trajectory data shall be deleted.

[0092] S4. Associating the processed vehicle trajectory data obtained in step S3 with the corresponding intersection turn;

[0093] Furthermore, the specific implementation method of step S4 includes the following steps:

[0094] S4.1. The processed vehicle trajectory data obtained in step S3 is filtered out based on the location of the vehicle trajectory data within 50 meters of the center point of each intersection;

[0095] S4.2. The vehicle trajectory data obtained in step S4.1 is grouped based on the vehicle ID, and the vehicle trajectory data is sorted in chronological order;

[0096] Since the same vehicle may pass through an intersection multiple times and the turns taken at different times are different, a grouping flag is set between adjacent trajectory points with a time difference greater than 600s, and the vehicle trajectory data is further grouped, and the group numbers PERIOD_ID are set respectively;

[0097] S4.3. Traverse each group of vehicle trajectory data based on vehicle ID and PERIOD_ID, select the first and last trajectory points of the group, and associate the trajectory points with the link number LINK_ID through the position. The first trajectory point corresponds to the vehicle number and the entrance link FROM_LINK of the intersection through which the vehicle passes, and the last trajectory point corresponds to the exit link TO_LINK, thereby determining the turning number MOVEMENT_ID of the vehicle at the intersection and establishing a one-to-one correspondence between the vehicle trajectory and the intersection turn.

[0098] S5. Calibrate the expected speeds for different types of intersections and different turns;

[0099] Furthermore, the specific implementation method of step S5 includes the following steps:

[0100] S5.1. Based on step S4, the vehicle trajectory data of 10-16 points is selected for calibrating the steering speed;

[0101] S5.2. Turn vehicle trajectory data according to intersection class , group by 5-minute time intervals and calculate each group The average speed of vehicles , the calculation formula is as follows:

[0102] ;

[0103] Among them, represents the grouping of intersection turns class in the i n-th average vehicle speed data, N represents the grouping of intersection turns class in the data volume; S5.3. Sort the average vehicle speeds of each group in ascending order of speed, and take the data at the 50% quantile as the expected speed of intersection turns.

[0104] S6. Calibrate the critical gap and critical waiting time of different types of intersections;

[0105] Furthermore, the specific implementation method of step S6 includes the following steps:

[0106] S6.1. Set the turn type of the parameter to be calibrated and the corresponding high-priority turn according to the intersection type:

[0107] Set the turn type of the parameter to be calibrated at a cross intersection as a left turn at the approach lane. The straight-ahead movement at the oncoming approach lane is a high-priority turn relative to the left turn at the approach lane, and the left turn at the approach lane is a low-priority turn relative to the straight-ahead movement at the oncoming approach lane;

[0108] Set the turn type of the parameter to be calibrated at a roundabout as an entry to the roundabout. The straight-ahead movement in the outermost lane of the roundabout road is a high-priority turn relative to the entry to the roundabout, and the entry to the roundabout is a low-priority turn relative to the straight-ahead movement in the outermost lane of the roundabout road;

[0109] Set the turn type of the parameter to be calibrated at a merging intersection as an entrance ramp. The straight-ahead movement on the main line is a high-priority turn relative to the entrance ramp, and the entrance ramp is a low-priority turn relative to the straight-ahead movement on the main line;

[0110] S6.2. Calibrate the critical gap and critical waiting time parameters of each low-priority turn in step S6.1:

[0111] S6.2.1. Traverse the intersections with parameters to be calibrated, and obtain the data of each vehicle trajectory point within 50 meters of the center point of the intersections with parameters to be calibrated;

[0112] S6.2.2. Identify the parking positions of vehicles with different turns:

[0113] Divide the approach road section into m sections at 5-meter intervals, assign the vehicle trajectory point data obtained in step S6.2.1 to the corresponding sections according to the coordinates, and count the number of trajectory points with a speed < 5 km / h in each section. The section with the largest number of points is the straight-ahead or left-turn parking position section ;

[0114] S6.2.3. Sorting vehicle trajectory points:

[0115] Filter low-priority turning vehicle trajectory points and sort the vehicles based on the time of the first vehicle trajectory point; filter high-priority turning vehicle trajectory points and sort the vehicles based on the time of the first vehicle trajectory point; let the available gap data set of vehicles passing through the intersection be recorded as , each data in the collection is in the form of , for the conflicting high-priority turning vehicles and low-priority turning vehicles at the intersection, let is the difference between the time when the high-priority turning vehicle reaches the parking position and the time when the low-priority turning vehicle reaches the parking position. If the low-priority vehicle passes the intersection first, , and the data Add to If the high priority vehicle passes the intersection first, , and the data Add to The relative waiting time data set of the intersection is recorded as , each data in the collection is in the form of ,make is the difference between the waiting time of low-priority vehicles at the parking location and the waiting time of high-priority vehicles at the parking location. If a low-priority vehicle passes through the intersection, , and the data Add to If the high priority vehicle passes the intersection first, , and the data Add to middle;

[0116] S6.2.4. Identify the moment a vehicle arrives at and leaves a parking location:

[0117] Iterate over low priority vehicles The trajectory point , find the parking position for entering straight ahead or turning left point and leaving the straight ahead or left turn parking position segment point , and The time of entering and leaving the straight or left-turn parking position segment respectively; traversing high priority vehicles The trajectory point , find the parking position for entering straight ahead or turning left point and leaving the straight ahead or left turn parking position segment points , and are the moments of entering and leaving the straight or left-turn stop position segments respectively;

[0118] S6.2.5. Identify the sequential relationship of vehicles passing through the intersection:

[0119] S6.2.5.1. Traverse the low-priority vehicles ;

[0120] S6.2.5.2. Traverse the trajectory points of the low-priority vehicles , the time corresponding to this point is denoted as t , and the horizontal and vertical coordinates are denoted as 、 , let the next point be , where = t +0.5, and the horizontal and vertical coordinates are denoted as 、 ;

[0121] S6.2.5.3. Traverse the high-priority vehicles ;

[0122] S6.2.5.4. Traverse the trajectory points of the high-priority vehicles , the time corresponding to this point is denoted as u , and the horizontal and vertical coordinates are denoted as 、 , let the next point be , where = u +0.5, and the horizontal and vertical coordinates are denoted as 、 ;

[0123] S6.2.5.5. Identify whether the line segments and conflict in time and space: If the following formula holds, then the line segments and conflict in space-time, enter S6.2.5.6, otherwise return to S6.2.5.4 and continue to traverse the next high-priority vehicle trajectory point. The formula is:

[0124] ;

[0125] ;

[0126] ;

[0127] S6.2.5.6. Identify the passing order of conflicting vehicles through the intersection: If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, enter S6.2.5.1, and continue to traverse the next low-priority vehicle; If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, enter S6.2.5.3, and continue to traverse the next high-priority vehicle;

[0128] S6.2.6. Filter the set of available gap data of vehicles passing through the intersection in , denoted as . Group and count the frequencies of at 0.5s time intervals respectively, where is the total number of data in that satisfy , and is the total number of data in that satisfy ; Calculate the frequency of , and the expression is:

[0129] ;

[0130] Obtain the minimum value of that satisfies and ;

[0131] Obtain the maximum value of that satisfies and ;

[0132] According to the two-parameter gap acceptance model, let , , where G is the critical gap;

[0133] Calibrate the final critical gap G through the following formula, and the expression is:​

[0134] ;

[0135] S6.2.7. Screen the relative waiting time data set in the data, denoted as , group and count the frequencies at 0.5 s time intervals respectively , where is the total number of data in that satisfy , is the total number of data in that satisfy ; calculate the frequency

[0136] ;

[0137] Obtain the minimum value and of that satisfy ;

[0138] Obtain the maximum value and of that satisfy ;

[0139] According to the two-parameter gap acceptance model, let , , where W is the critical waiting time;

[0140] Calculate the critical waiting time W through the following formula, and the expression is:

[0141] .

[0142] It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0143] Although the present application has been described above with reference to specific embodiments, various improvements can be made thereto and components thereof can be replaced with equivalents without departing from the scope of the present application. In particular, as long as there is no structural conflict, the various features in the specific embodiments disclosed in the present application can be combined with each other in any manner, and the exhaustive description of these combinations is not given in this specification only for the sake of saving space and resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for calibrating a signal-free intersection model based on drone aerial photography data, characterized in that The steps include: S1. Obtain road network data and collect vehicle trajectory data and aerial photos of intersections through drones; S2. Process the road network data obtained in step S1, merge the one-way lines of the road network, and then classify the intersections to be calibrated parameters; S3. The vehicle trajectory data obtained in step S1 is processed to remove missing or abnormal data to obtain processed vehicle trajectory data; S4. Associating the processed vehicle trajectory data obtained in step S3 with the corresponding intersection turn; S5. Calibrate the expected speeds for different types of intersections and different turns; S6. Determine the critical gap and critical waiting time for different types of intersections.

2. The calibration method of the signal-free intersection model based on UAV aerial photography data according to claim 1, characterized in that, The road network data in step S1 includes the unique number of the road section, length, direction, road grade, and number of lanes; the vehicle trajectory data is extracted through drone aerial video with a time accuracy of 0.1s, including the unique vehicle number, timestamp, lane number, longitude, latitude, and speed; the aerial view of the intersection includes the intersection area and sign and marking information.

3. A method for calibrating a signal-free intersection model based on UAV aerial photography data according to claim 1 or 2, characterized in that, The specific implementation method of step S2 includes the following steps: S2.

1. Merge one-way lines in road network: Merge two single lines representing two directions of the same road into a single line representing the two directions of the road; S2.

2. Classification of parameter intersections to be calibrated: divide the unsignalized intersections to be calibrated into cross intersections, T-intersections, roundabouts, and merges.

4. A method for calibrating a signal - free intersection model based on UAV aerial photography data according to claim 3, characterized in that, The specific implementation method of step S3 includes the following steps: S3.

1. Eliminate data that is missing any of the vehicle unique number, timestamp, lane number, longitude, latitude, and speed fields; S3.

2. Eliminate data with a speed greater than 120; S3.

3. If the proportion of the excluded data of a vehicle trajectory data exceeds 5%, the vehicle trajectory data shall be deleted.

5. A method for calibrating a signal-free intersection model based on UAV aerial photography data according to claim 4, characterized in that, The specific implementation method of step S4 includes the following steps: S4.

1. The processed vehicle trajectory data obtained in step S3 is filtered out based on the location of the vehicle trajectory data within 50 meters of the center point of each intersection; S4.

2. The vehicle trajectory data obtained in step S4.1 is grouped based on the vehicle ID, and the vehicle trajectory data is sorted in chronological order; Since the same vehicle may pass through an intersection multiple times and the turns taken at different times are different, a grouping flag is set between adjacent trajectory points with a time difference greater than 600s, and the vehicle trajectory data is further grouped, and the group numbers PERIOD_ID are set respectively; S4.

3. Traverse each group of vehicle trajectory data based on vehicle ID and PERIOD_ID, select the first and last trajectory points of the group, and associate the trajectory points with the link number LINK_ID through the position. The first trajectory point corresponds to the vehicle number and the entrance link FROM_LINK of the intersection through which the vehicle passes, and the last trajectory point corresponds to the exit link TO_LINK, thereby determining the turning number MOVEMENT_ID of the vehicle at the intersection and establishing a one-to-one correspondence between the vehicle trajectory and the intersection turn.

6. The calibration method of the signal-free intersection model based on UAV aerial photography data according to claim 5, characterized in that The specific implementation method of step S5 includes the following steps: S5.

1. Based on step S4, filter the vehicle trajectory data from 10:00 to 16:00 for calibrating the steering speed; S5.

2. Group the vehicle trajectory data according to intersection turns class and 5-minute time intervals, and calculate the average vehicle speed of each group group . The calculation formula is as follows: ​ ； Among them, represents the grouping of intersection turns class and group the i th N average vehicle speed data of the class grouping of intersection turns group indicates the data volume; S5.

3. Sort the average speeds of each group of vehicles in ascending order of speed, and take the data at the 50% quantile of the speed as the expected intersection steering speed.

7. A method for calibrating a signal-free intersection model based on UAV aerial photography data according to claim 6, characterized in that The specific implementation method of step S6 includes the following steps: S6.

1. Set the steering type of the parameter to be calibrated and the corresponding high-priority steering according to the intersection type: Set the steering type of the parameter to be calibrated at the cross intersection as left turn on the approach lane. Straight ahead on the oncoming approach lane is a high-priority steering relative to the left turn on the approach lane, and the left turn on the approach lane is a low-priority steering relative to the straight ahead on the oncoming approach lane; The steering type of the parameter to be calibrated at the roundabout is the roundabout entrance. Straight ahead in the outermost lane of the roundabout road is a high-priority steering relative to the roundabout entrance, and the roundabout entrance is a low-priority steering relative to the straight ahead in the outermost lane of the roundabout road; The steering type of the parameter to be calibrated at the merge intersection is the ramp entrance. Straight ahead on the main line is a high-priority steering relative to the ramp entrance, and the ramp entrance is a low-priority steering relative to the straight ahead on the main line; S6.

2. Calibrate the critical gap and critical waiting time parameters of the low-priority steering in step S6.1: S6.2.

1. Traverse the intersections with parameters to be calibrated, and obtain the data of each vehicle trajectory point within 50 meters of the center point of the intersections with parameters to be calibrated; S6.2.

2. Identify the parking positions of vehicles with different steers: Divide the imported road section into m sections at 5-meter intervals, distribute the vehicle trajectory point data obtained in step S6.2.1 to the corresponding sections according to the coordinates, count the number of trajectory points with a speed <5 km / h in each section, and the section with the largest number of points is the straight or left-turn stop position section ; S6.2.

3. Sort the vehicle trajectory points: Filter the vehicle trajectory points of the low-priority steering, and sort the vehicles based on the time of the first trajectory point of the vehicle; filter the vehicle trajectory points of the high-priority steering, and sort the vehicles based on the time of the first trajectory point of the vehicle; Let the set of available gap data of vehicles passing through the intersection be denoted as , and each data in the set is in the form of . For the high-priority turning vehicle and the low-priority turning vehicle with conflicts at the intersection, let be the time difference between the high-priority turning vehicle arriving at the stop position and the low-priority turning vehicle arriving at the stop position. If the low-priority vehicle passes through the intersection first, let , and add the data to . If the high-priority vehicle passes through the intersection first, let , and add the data to ; Let the set of relative waiting time data of vehicles passing through the intersection be denoted as , and each data in the set is in the form of . Let be the difference between the waiting time of the low-priority vehicle at the stop position and the waiting time of the high-priority vehicle at the stop position. If the low-priority vehicle passes through the intersection, let , and add the data to . If the high-priority vehicle passes through the intersection first, let , and add the data to ; S6.2.

4. Identify the moments when the vehicle arrives at and leaves the parking position: Traverse the low-priority vehicles 's trajectory points , and find the points entering the straight or left-turn parking position segment and the points leaving the straight or left-turn parking position segment . And are the moments of entering and leaving the straight or left-turn parking position segment respectively; Traverse the high-priority vehicles 's trajectory points , and find the points entering the straight or left-turn parking position segment and the points leaving the straight or left-turn parking position segment . And are the moments of entering and leaving the straight or left-turn parking position segment respectively; S6.2.

5. Identify the passing sequence relationship of vehicles through the intersection: S6.2.5.

1. Traverse low-priority vehicles ; S6.2.5.

2. Traverse the low-priority vehicle trajectory points , and record the corresponding time of this point as t , and record the horizontal and vertical coordinates as 、 , let the next point be , where = t + 0.5, and record the horizontal and vertical coordinates as 、 ; S6.2.5.

3. Traverse high-priority vehicles ; S6.2.5.

4. Traverse the high-priority vehicle trajectory points , record the corresponding time of this point as u , record the horizontal and vertical coordinates as 、 , let the next point be , where = u + 0.5, record the horizontal and vertical coordinates as 、 ; S6.2.5.

5. Identify line segments and whether there is a conflict in time and space: if the following formula holds, then the line segments and conflict in time and space, enter S6.2.5.6, otherwise return to S6.2.5.4 and continue to traverse the next high-priority vehicle trajectory point. The formula is: ； ； ； S6.2.5.

6. Identify the passing order of conflicting vehicles through the intersection: If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, enter S6.2.5.1, and continue to traverse the next low-priority vehicle; If , it indicates that vehicle passes through the intersection before vehicle . Add the available gap data and the relative waiting time data to and respectively, enter S6.2.5.3, and continue to traverse the next high-priority vehicle; S6.2.

6. Screen the set of available gap data of vehicles passing through the intersection of the data, denoted as , and group and count them at 0.5s time intervals for the frequency of , , where is the total number of data in that satisfy is the total number of data in that satisfy ; calculate the frequency of , and the expression is: ； Obtain the one that satisfies and of minimum value ; Obtain those satisfying and of the maximum value , According to the two-parameter gap acceptance model, let , , where G is the critical gap; Calibrate the final critical clearance by the following formula G , and the expression is: ； S6.2.

7. Screen the relative waiting time data set in the data, denoted as , and group and count the frequencies at 0.5s time intervals respectively , where is the total number of data in that meet the requirements , and is the total number of data in that meet the requirements; calculate the frequency of ； Obtain the and of the minimum value ; Obtain the one that satisfies and of maximum value ; According to the two-parameter gap acceptance model, let , , where W is the critical waiting time; Calculate the critical waiting time using the following formula W , and the expression is: 。

Images