Traffic flow trajectory alignment method and system for large-scale live broadcast testing

Abstract

The present invention belongs to the field of intelligent transportation technology, and specifically discloses a traffic flow trajectory alignment method and system for large-scale live broadcast testing. The method includes the following steps: HVs upload scene information in real time and form trajectory data, while using a laser radar to shoot regional traffic flow information to obtain a regional traffic flow trajectory dataset; using a network prediction model to predict the future position information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset; comparing the HV trajectory data with the regional traffic flow trajectory dataset collected by the laser radar, and identifying the HV trajectory in the regional traffic flow trajectory dataset; matching the trajectory uploaded by the HV in real time with the HV trajectory collected by the laser radar, obtaining the time difference of the trajectory data, and determining the future live broadcast position information of the vehicles contained in the regional traffic flow to meet the live broadcast test requirements. By adopting this technical solution, the time difference between the real-time trajectory of the HV and the trajectory data collected by the laser radar is obtained, which meets the live broadcast test requirements.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation technology and relates to a traffic flow trajectory alignment method and system for large-scale live broadcast testing. Background Art

[0002] With the rapid development of Intelligent Transportation Systems (ITS), V2X (Vehicle-to-Everything) technology is becoming a key pillar for achieving autonomous driving. Through wireless communication, V2X enables vehicles to exchange information with transportation infrastructure, other vehicles, pedestrians, and the cloud, improving road safety, efficiency, and user experience. Cellular Vehicle-to-Everything (C-V2X) technology, a key component of ITS, leverages existing cellular networks to provide highly reliable, low-latency communications. It supports V2N (vehicle-to-network), V2V (vehicle-to-vehicle), V2P (vehicle-to-pedestrian), and V2I (vehicle-to-infrastructure) communications, particularly with the support of 5G.

[0003] Digital twin (DT) technology is increasingly being applied in large-scale testing, creating virtual road test environments to simulate vehicle driving scenarios. This technology synchronizes virtual models with the physical world in real time, effectively improving testing efficiency and reducing costs and risks.

[0004] As the requirements for large-scale testing increase, single virtual simulation testing can no longer meet complex demands. The increasing demand for real-time data processing has made live testing a new testing method. Live testing refers to the use of real-time captured road information to test the responsiveness and performance of C-V2X-related applications on open roads. In live testing, the integration of the C-V2X large-scale test platform and DT has become a key technical direction. Based on this, researchers can accurately capture the real-time status of road vehicles and express it in the form of live scenes. Through scene injection, scene information (including the location, speed, direction, etc. of road vehicles) is converted into OTA data packets through the C-V2X large-scale test platform and broadcast in the test area. Vehicles equipped with C-V2X modules (test object HV, Host Vehicle) trigger V2V application responses by receiving OTA data packets. The test system receives and analyzes the HV application response information to achieve a complete closed-loop test.

[0005] However, live testing faces many technical challenges. First, live testing is different from pre-set scenario testing. It requires converting the real-time traffic flow collected by the roadside LiDAR into live scene information. The accuracy and real-time nature of the live scene information in expressing real-time traffic flow depends largely on the effectiveness of real-time data processing. Generally speaking, the road traffic flow trajectory data obtained by LiDAR lags behind the actual traffic flow trajectory data. Therefore, trajectory prediction is required to make up for this time difference and ensure the real-time and accuracy of the live scene information. How to obtain the time difference between LiDAR data and the actual traffic flow vehicle trajectory data has become a core technical problem in improving the effectiveness of live testing. Summary of the Invention

[0006] The purpose of the present invention is to provide a traffic flow trajectory alignment method and system for large-scale live broadcast testing, which can obtain the time difference between the HV real-time trajectory and the trajectory data collected by the lidar to meet the needs of live broadcast testing.

[0007] To achieve the above objectives, the basic solution of the present invention is: a traffic flow trajectory alignment method for large-scale live broadcast testing, comprising the following steps:

[0008] HVs upload scene information in real time and generate trajectory data. At the same time, they use lidar to capture regional traffic flow information and obtain a regional traffic flow trajectory dataset.

[0009] Use the prediction network to predict the future location information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset;

[0010] Compare the HV trajectory data with the regional traffic flow trajectory dataset collected by LiDAR, and identify the HV trajectory in the regional traffic flow trajectory dataset;

[0011] Match the real-time HV trajectory uploaded by the HV with the HV trajectory collected by the LiDAR to obtain the time difference of the trajectory data;

[0012] Based on the time difference of trajectory data, the future live position information of vehicles included in the regional traffic flow is determined, and the road test data of road vehicles is aligned with the real data to meet the live test requirements.

[0013] The working principle and beneficial effects of this basic solution are as follows: This technical solution uses LiDAR to collect regional traffic flow scene information and assigns a unique ID to each vehicle to form its own trajectory. HVs can upload their scene information in real time through the C-V2X module to generate corresponding trajectories. Due to the data errors between the two types of trajectories, in order to reduce the impact of this error, the LiDAR-collected traffic flow trajectory dataset is used for prediction processing to achieve a match between the two, thereby identifying HV trajectories in the LiDAR-collected traffic flow trajectory dataset.

[0014] Using the HV trajectory captured by the LiDAR as a reference trajectory, the HV's real-time trajectory is matched against the reference trajectory for similarity. Once a matching trajectory is found, the difference between the HV's real-time trajectory endpoint and the reference trajectory endpoint is calculated as the trajectory data time difference. Based on this time difference, the network model predicts the future position information of vehicles in the regional traffic flow. This future position information is used to implement live scene broadcasting, compensating for the delay caused by the transmission and processing of LiDAR data during data collection, thus achieving online live broadcasting.

[0015] Furthermore, the method for HV to upload scene information in real time and form trajectory data is as follows:

[0016] HV equipped with C-V2X module uploads the current scene information in real time HV , scene information point HV Includes longitude, latitude, time, speed and direction, and is stored in a collection trajectory HV (start HV ,end HV ), the HV real-time trajectory data is formed as:

[0017] point HV (t) = {lon t ,lat t ,v t ,heading t}

[0018] trajectory HV (start HV ,end HV )={point HV(t)|t∈[start HV ,end HV ]}

[0019] Where t represents the time when HV currently transmits data; lon t Indicates the longitude of HV at time t; lat t represents the latitude of HV at time t; v t represents the speed of HV at time t; heading t Indicates the direction of HV at time t; start HV and end HV Respectively represent the start time and end time of HV uploading data.

[0020] HV uploads scene information in real time and forms trajectory data, which is convenient for subsequent use.

[0021] Furthermore, the lidar captures the regional traffic flow information at time t, and the method for obtaining the regional traffic flow trajectory dataset is as follows:

[0022] Traffic flow information point in the laser radar shooting area vid (t), traverse the regional traffic flow scene information collected by the lidar, exclude invalid data outside the road track, use the unique vehicle ID (vid) as the identifier for each vehicle information, and store the regional traffic flow scene information in the collection trajectorys V , a regional traffic flow trajectory dataset is formed:

[0023]

[0024] trajectory vid (start V ,end V )={point vid (t V )|t V ∈[start V ,end V ]}

[0025] trajectories V

[0026] ={trajectory vid (start V ,end V )|vid∈[1,n]}

[0027] ={{point vid (t V )}|t V ∈[startv ,end v ],vid∈[1,n]}

[0028] Among them, t V Indicates the time when the current radar collects traffic flow data; vid indicates the ID of the current vehicle; Indicates t V The longitude of the vehicle at the moment vid; Indicates t V The latitude of the vehicle at the moment; Indicates t V The speed of the vehicle at the moment vid; Indicates t V The direction of the vehicle at the moment vid; start v and end v They represent the start and end time of the lidar data collection; trajectory vid Represents the regional traffic flow scene information of vehicle vid; n represents the number of vehicles contained in the current area.

[0029] Use lidar to capture regional traffic flow information for subsequent use.

[0030] Furthermore, we use the prediction network to predict the future position information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset. The specific method is as follows:

[0031] Let the time interval of HV uploading data be Δt HV , the time interval of traffic flow in the laser radar collection area is Δt V ; In the vehicle position data contained in the laser radar acquisition area, from the current time t cur Take the window size of x time points forward, and input the longitude and latitude of x time points as the input sequence into the prediction network, which is recorded as:

[0032]

[0033] Where I represents the input sequence of the prediction network; Indicates t cur The longitude of the vehicle at the moment, Indicates t cur The latitude of the vehicle at the moment, Indicates that from the current time t cur Take the vehicle longitude for x time intervals forward, Indicates that from the current time t cur Take the vehicle latitude x time intervals ahead; t V Indicates the time when the current radar collects traffic flow data;

[0034] O represents the output sequence of the prediction network, which is the longitude and latitude of the vehicles in the area at the next y moments after the calculation delay, recorded as:

[0035]

[0036] in, Indicates the future cur +Δt V The longitude of the vehicle at the moment, Indicates the future cur +Δt V The latitude of the vehicle at the moment, Indicates the future cur +yΔt V The longitude of the vehicle at the moment, Indicates the future cur +yΔt V The latitude of the vehicle at the moment;

[0037] Store the predicted points into the set trajectorys according to the vehicle vid V ', then trajectory V 'for:

[0038] trajectories V '

[0039] ={trajectory vid (t cur -(x-1)Δt V ,t cur +yΔt V )vid∈[1,n]}

[0040] ={{point vid (t)}|t∈[t cur -(x-1)Δt V ,t cur +yΔt V ],vid∈[1,n]}

[0041] Among them, vid represents the ID of the current vehicle; n represents the number of vehicles in the current area; trajectory vid Indicates the regional traffic flow scene information of vehicle vid, point vid (t) represents the regional traffic flow information captured by the lidar at time t.

[0042] Using the prediction network to predict the future position information of vehicles included in the regional traffic flow to complete the regional traffic flow trajectory dataset can reduce the error between the lidar collected data and the HV uploaded data, thereby more accurately identifying the HV trajectories in the lidar collected regional traffic flow trajectory dataset.

[0043] Furthermore, the HV trajectory data is compared with the regional traffic flow trajectory dataset collected by the LiDAR to identify the HV trajectory in the regional traffic flow trajectory dataset. The specific steps are as follows:

[0044] trajectories V 'Each vehicle contains x+y time data trajectory, in trajectory HV (start HV ,end HV ) from the current time t cur Take the point data of the window size x+y moments forward to form the trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ), then trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) is expressed as:

[0045] trajectory HV '(t cur -(x+y-1)Δt HV ,t cur )={point HV (t)|t∈[t cur -(x+y-1)Δt HV ,t cur ]}

[0046] trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) and collection trajectories V The trajectory of vehicle r in ' r (t cur -(x-1)Δt V ,t cur +yΔt V ) is defined as:

[0047]

[0048] Among them, |point HV (i)-point r (j)| indicates point HV (i) with point r (j) Euclidean distance between

[0049] Then

[0050]

[0051] Calculate the trajectory set trajectories V All trajectories and trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ), then the LiDAR collects the traffic flow trajectory dataset trajectorys V 'Real-time vehicle trajectory uploaded by HV HV '(t cur -(x+y-1)Δt HV ,t cur ) The vehicle id of the trajectory with the highest similarity is the vid of the HV trajectory collected by the lidar, recorded as DUTvid, and its trajectory is trajectory DUTvid (start V ,end V ).

[0052] Compare the HV uploaded trajectory data with the regional traffic flow trajectory dataset collected by the lidar, and identify the HV trajectory in the regional traffic flow trajectory dataset.

[0053] Furthermore, the method for matching the real-time uploaded trajectory of the HV with the HV trajectory collected by the lidar to obtain the time difference of the trajectory data is as follows:

[0054] HV trajectory collected by lidar DUTvid (start V ,end V ) as a reference trajectory, in the real-time upload trajectory of HV HV (start HV ,end HV ) and set the sliding window size to z vehicle position coordinates. DUTvid (start V ,end V ) from the current moment tcur Get z points forward as the reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ), expressed as:

[0055]

[0056] From real-time trajectory HV (start HV ,end HV ) Current time t cur Slide forward to divide the trajectory segment set ts HV :

[0057]

[0058] Reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ) and the trajectory segment set ts HV Trajectory HV (t k -(z-1)Δt HV ,t k ) is defined as:

[0059]

[0060] Then

[0061]

[0062] From trajectory HV (start HV ,end HV ) Current time t cur When the point moves forward, the similarity generally shows a trend of increasing first and then decreasing. Find the point with the highest similarity, i.e., f(trajectory HV (t k -(z-1)Δt HV ,t k ),tra DUTvid (t cur -(z-1)Δt V ,t cur ))The lowest trajectory segment trajectory HV (t u -(z-1)Δt HV ,tu ), the starting time is t u -(z-1)Δt HV The end time is t u ,for:

[0063] trajectory HV (t u -(z-1)Δt HV ,t u )={point HV (t u -(z-1)Δt HV ),…,point HV (t u )}

[0064] Then slide forward one window size distance in turn. If these trajectory segments are consistent with the reference trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) is more similar than trajectory HV (t u -(z-1)Δt HV ,t u ) and tra DUTvid (t cur -(z-1)Δt V ,t cur ) is low, then it can be determined that the trajectory HV (t u -(z-1)Δt HV ,t u ) and trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) best matches;

[0065] The trajectory segment trajectory HV (t u -(z-1)Δt HV ,t u ) End point time t u With the tracktra DUTvid (t cur -(z-1)Δt V ,t cur ) End point time t cur The time difference T between the regional traffic flow trajectory data collected by the lidar and the HV real-time data is obtained by subtraction. offset :

[0066] Toffset =|t cur -t u |.

[0067] By matching the real-time HV uploaded trajectory with the HV trajectory collected by the lidar and obtaining the time difference of the trajectory data, the similarity between the trajectories can be accurately and quickly determined, meeting the need to match the two trajectories and find the time difference between the regional traffic flow trajectory data obtained by the lidar and the real-time HV data.

[0068] Furthermore, the method for determining the future position information of vehicles included in the regional traffic flow based on the time difference of trajectory data is as follows:

[0069] If the time difference of trajectory data is T offset ≤yΔt V , then in the prediction network output sequence Find the offset Recent data Expressed as:

[0070]

[0071] T offset '=iΔt V ,i∈[1,7]

[0072] point vid (t cur +T offset ') is the future position information of the vehicle to be determined; T offset ' represents the time difference of the corrected trajectory data; i represents an index value;

[0073] If T offset >yΔt V , then use the prediction network model to predict T offset The area after the vehicle contains the future position of the vehicle, which is the future position information of the vehicle that needs to be determined.

[0074] Based on the time difference of trajectory data, the future location information of vehicles included in the regional traffic flow is determined, meeting the needs of live broadcast testing.

[0075] The present invention also provides a traffic flow trajectory alignment system for large-scale live broadcast testing, including a C-V2X module, a laser radar, and a processing module;

[0076] The C-V2X module is installed on the HV to upload HV trajectory data in real time, and the lidar is used to capture regional traffic flow information;

[0077] The input end of the processing module is connected to the output end of the C-V2X module and the laser radar respectively. The processing module executes the method described in the present invention to obtain the time difference of the trajectory data and determine the future position information of the vehicles included in the regional traffic flow.

[0078] This system obtains the time difference between the lidar data and the actual traffic flow vehicle trajectory data to improve the live broadcast test effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 It is a flow chart of the traffic flow trajectory alignment method for large-scale live broadcast testing of the present invention. DETAILED DESCRIPTION

[0080] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0081] In the description of the present invention, it should be understood that the terms "longitudinal", "transverse", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0082] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal communication between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0083] The present invention discloses a traffic flow trajectory alignment method for large-scale live broadcast testing, such as Figure 1 As shown, the following steps are included:

[0084] The HV (host vehicle) uploads scene information in real time and generates trajectory data. It also uses lidar to capture regional traffic flow information and obtain a regional traffic flow trajectory dataset.

[0085] Use prediction networks (such as BP neural network (Backpropagation Neural Network, BPNN), long short-term memory network (Long Short-Term Memory, LSTM), Transformer model, etc.) to predict the future position information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset; in actual scenarios, data collection is multivariate. LiDAR collects regional traffic flow scene information and assigns a unique ID to each vehicle to form its own trajectory, while HV can upload its scene information in real time through the C-V2X module to generate the corresponding trajectory. Since there is an error in the data between the two types of trajectories, in order to reduce the impact of the error, the traffic flow trajectory dataset collected by the LiDAR needs to be predicted and processed to achieve matching between the two, so as to identify the HV trajectory in the traffic flow trajectory dataset collected by the LiDAR.

[0086] Compare the HV trajectory data with the regional traffic flow trajectory dataset collected by LiDAR, and identify the HV trajectory in the regional traffic flow trajectory dataset;

[0087] Match the real-time HV trajectory uploaded by the HV with the HV trajectory collected by the LiDAR to obtain the time difference of the trajectory data;

[0088] Based on the time difference of trajectory data, the future live position information of vehicles included in the regional traffic flow is determined, and the road test data of road vehicles is aligned with the real data to meet the live test requirements.

[0089] Because the regional traffic flow trajectory data collected by the LiDAR needs to be processed, the processing time is necessarily longer than the real-time data uploaded by the HV. Therefore, there is a certain delay between the traffic flow trajectory data collected by the LiDAR and the real-time data uploaded by the HV.

[0090] The HV trajectory captured by the LiDAR is used as the reference trajectory, and a sliding window algorithm is applied to the HV's real-time uploaded trajectory. By setting a fixed window size, the window slides forward from the HV's real-time trajectory endpoint to perform a similarity match with the reference trajectory. After finding a matching trajectory, the difference between the HV's real-time trajectory endpoint and the reference trajectory endpoint is the trajectory data time difference. Based on this time difference, the future position information of vehicles included in the traffic flow in the network model prediction area is obtained. This future position information is used to realize live scene broadcasting to compensate for the delay caused by data transmission and processing during the LiDAR data acquisition process.

[0091] In a preferred embodiment of the present invention, the method for HV to upload scene information in real time and form trajectory data is as follows:

[0092] HV equipped with C-V2X module uploads the current scene information in real time HV , scene information pointHV Includes longitude, latitude, time, speed and direction, and is stored in a collection trajectory HV (start HV ,end HV ), the HV real-time trajectory data is formed as:

[0093] point HV (t) = {lon t ,lat t ,v t ,heading t}

[0094] trajectory HV (start HV ,end HV )={point HV (t)|t∈[start HV ,end HV ]}

[0095] Where t represents the time when HV currently transmits data; lon t Indicates the longitude of HV at time t; lat t represents the latitude of HV at time t; v t represents the speed of HV at time t; heading t Indicates the direction of HV at time t; start HV and end HV Respectively represent the start time and end time of HV uploading data.

[0096] In a preferred embodiment of the present invention, a method for obtaining a regional traffic flow trajectory dataset by photographing regional traffic flow information using a laser radar is as follows:

[0097] LiDAR captures regional traffic flow information at time t point vid (t), traverse the regional traffic flow scene information collected by the lidar, exclude invalid data outside the road track, use the unique vehicle ID (vid) as the identifier for each vehicle information, and store the regional traffic flow scene information in the collection trajectorys V , a regional traffic flow trajectory dataset is formed:

[0098]

[0099] trajectory vid (start V ,end V )={point vid (t V )tV ∈[start V ,end V ]}

[0100] trajectories V

[0101] ={trajectory vid (start V ,end V )vid∈[1,n]}

[0102] ={{point vid (t V )}t V ∈[start v ,end v ],vid∈[1,n]}

[0103] Among them, t V Indicates the time when the current radar collects traffic flow data; vid indicates the ID of the current vehicle; Indicates t V The longitude of the vehicle at the moment vid; Indicates t V The latitude of the vehicle at the moment; Indicates t V The speed of the vehicle at the moment vid; Indicates t V The direction of the vehicle at the moment vid; start v and end v They represent the start and end time of the lidar data collection; trajectory vid Represents the regional traffic flow scene information of vehicle vid; n represents the number of vehicles contained in the current area.

[0104] In a preferred embodiment of the present invention, since the real-time data uploaded by HVs is necessarily more timely than the data collected by LiDAR, in order to reduce the influence of the error between the LiDAR collected data and the HV uploaded data, a prediction network is used to predict the future position information of vehicles included in the regional traffic flow information and to complete the regional traffic flow trajectory dataset. The specific method is as follows:

[0105] Let the time interval of HV uploading data be Δt HV , the time interval of traffic flow in the laser radar collection area is Δt V ; In the vehicle position data contained in the laser radar acquisition area, from the current time t cur Take the points with a window size of x (e.g. 10) time points in the past, and input the longitude and latitude of x time points as the input sequence into the network prediction model, which is recorded as:

[0106]

[0107] Where I represents the input sequence of the prediction network; Indicates t cur The longitude of the vehicle at the moment, Indicates t cur The latitude of the vehicle at the moment, Indicates that from the current time t cur Take the vehicle longitude for x time intervals forward, Indicates that from the current time t cur Take the vehicle latitude x time intervals ahead; t V Indicates the time when the current radar collects traffic flow data;

[0108] O represents the output sequence of the prediction network, which is the longitude and latitude of the vehicles in the area at the next y (e.g., 7) moments after the calculation delay, and is recorded as:

[0109]

[0110] in, Indicates the future cur +Δt V The longitude of the vehicle at the moment, Indicates the future cur +Δt V The latitude of the vehicle at the moment, Indicates the future cur +yΔt V The longitude of the vehicle at the moment, Indicates the future cur +yΔt V The latitude of the vehicle at the moment;

[0111] Store the predicted points into the set trajectorys according to the vehicle vid V ', then trajectory V 'for:

[0112] trajectories V '

[0113] ={trajectory vid (t cur -(x-1)Δt V ,t cur +yΔt V )|vid∈[1,n]}

[0114] ={{point vid (t)}|t∈[t cur -(x-1)ΔtV ,t cur +yΔt V ],vid∈[1,n]}

[0115] Among them, vid represents the ID of the current vehicle; n represents the number of vehicles in the current area; trajectory vid Indicates the regional traffic flow scene information of vehicle vid, point vid (t) represents the regional traffic flow information captured by the lidar at time t.

[0116] In a preferred embodiment of the present invention, the trajectory data of the HV is compared with the regional traffic flow trajectory dataset collected by the lidar, and the HV trajectory in the regional traffic flow trajectory dataset is identified. The specific steps are as follows:

[0117] trajectories V 'Each vehicle contains x+y time data trajectory, in trajectory HV (start HV ,end HV ) from the current time t cur Take the point data of the window size x+y moments forward to form the trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ), then trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) is expressed as:

[0118] trajectory HV '(t cur -(x+y-1)Δt HV ,t cur )={point HV (t)|t∈[t cur -(x+y-1)Δt HV ,t cur ]}

[0119] trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) and collection trajectories V The trajectory of vehicle r in ' r (t cur -(x-1)ΔtV ,t cur +yΔt V ) is defined as:

[0120]

[0121] Among them, |point HV (i)-point r (j)| indicates point HV (i) with point r (j) Euclidean distance between

[0122] Then

[0123]

[0124] Calculate the trajectory set trajectories V All trajectories and trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ), then the LiDAR collects the traffic flow trajectory dataset trajectorys V 'Real-time vehicle trajectory uploaded by HV HV '(t cur -(x+y-1)Δt HV ,t cur ) The vehicle id of the trajectory with the highest similarity is the vid of the HV trajectory collected by the lidar, recorded as DUTvid, and its trajectory is trajectory DUTvid (start V ,end V ).

[0125] In a preferred embodiment of the present invention, a method for matching the trajectory uploaded by the HV in real time with the HV trajectory collected by the lidar to obtain the time difference of the trajectory data is as follows:

[0126] Since the real-time data uploaded by HV is necessarily more timely than the data collected by LiDAR, the trajectory of HV collected by LiDAR is DUTvid (start V ,end V ) as a reference trajectory, in the real-time upload trajectory of HV HV (start HV ,end HV) and set the sliding window size to z (e.g. 10) vehicle position coordinates. DUTvid (start V ,end V ) from the current moment t cur Get forward z point as the reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ), expressed as:

[0127]

[0128] From real-time trajectory HV (start HV ,end HV ) Current time t cur Slide forward to divide the trajectory segment set ts HV :

[0129]

[0130] Reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ) and the trajectory segment set ts HV Trajectory HV (t k -(z-1)Δt HV ,t k ) is defined as:

[0131]

[0132] Then

[0133]

[0134] From trajectory HV (start HV ,end HV ) Current time t cur When the point moves forward, the similarity generally shows a trend of increasing first and then decreasing. Find the point with the highest similarity, i.e., f(trajectory HV (t k -(z-1)Δt HV ,t k ),tra DUTvid (tcur -(z-1)Δt V ,t cur ))The lowest trajectory segment trajectory HV (t u -(z-1)Δt HV ,t u ), the starting time is t u -(z-1)Δt HV The end time is t u ,for:

[0135] trajectory HV (t u -(z-1)Δt HV ,t u )={point HV (t u -(z-1)Δt HV ),…,point HV (t u )}

[0136] Then slide forward one window size distance in turn. If these trajectory segments are consistent with the reference trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) is more similar than trajectory HV (t u -(z-1)Δt HV ,t u ) and tra DUTvid (t cur -(z-1)Δt V ,t cur ) is low, then it can be determined that the trajectory HV (t u -(z-1)Δt HV ,t u ) and trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) best matches;

[0137] The trajectory segment trajectory HV (t u -(z-1)Δt HV ,t u ) End point time t u With the tracktra DUTvid (t cur -(z-1)Δt V,t cur ) End point time t cur The time difference T between the regional traffic flow trajectory data collected by the lidar and the HV real-time data is obtained by subtraction. offset :

[0138] T offset =|t cur -t u |.

[0139] In a preferred embodiment of the present invention, a method for determining the future position information of vehicles included in regional traffic flow based on the time difference of trajectory data is as follows:

[0140] If the time difference of trajectory data is T offset ≤yΔt V , then in the prediction network model output sequence Find the offset Recent data Expressed as:

[0141]

[0142] T offset '=iΔt V ,i∈[1,7]

[0143] point vid (t cur +T offset ') is the future position information of the vehicle to be determined; T offset ' represents the time difference of the corrected trajectory data; i represents an index value;

[0144] If T offset >yΔt V , then use the prediction network model to predict T offset The area after the vehicle contains the future position of the vehicle, which is the future position information of the vehicle that needs to be determined.

[0145] This embodiment achieves accurate real-time updates of vehicle positions in online live testing scenarios, thereby meeting the needs of large-scale testing.

[0146] The present invention also provides a traffic flow trajectory alignment system for large-scale live broadcast testing, including a C-V2X module, a lidar and a processing module. The C-V2X module is set on the HV and is used to upload HV trajectory data in real time. The lidar is used to shoot regional traffic flow information.

[0147] The input end of the processing module is electrically connected to the output end of the C-V2X module and the laser radar respectively. The processing module executes the method described in the present invention to obtain the time difference of the trajectory data and determine the future position information of the vehicles included in the regional traffic flow.

[0148] This system obtains the time difference between the lidar data and the actual traffic flow vehicle trajectory data to improve the live broadcast test effect.

[0149] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0150] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A traffic flow trajectory alignment method for large-scale live broadcast testing, characterized by: The steps include: HVs upload scene information in real time and generate trajectory data. At the same time, they use lidar to capture regional traffic flow information and obtain a regional traffic flow trajectory dataset. Use the prediction network to predict the future location information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset; Compare the HV trajectory data with the regional traffic flow trajectory dataset collected by LiDAR, and identify the HV trajectory in the regional traffic flow trajectory dataset; Match the real-time HV trajectory uploaded by the HV with the HV trajectory collected by the LiDAR to obtain the time difference of the trajectory data; Based on the time difference of trajectory data, the future live position information of vehicles included in the regional traffic flow is determined, and the road test data of road vehicles is aligned with the real data to meet the live test requirements; The method for HV to upload scene information in real time and generate trajectory data is as follows: HV equipped with C-V2X module uploads the current scene information in real time HV , scene information point HV Includes longitude, latitude, time, speed and direction, and is stored in a collection trajectory HV (start HV ,end HV ), the HV real-time trajectory data is formed as: point HV (t)={lon t ,lat t ,v t ,heading t } trajectory HV (start HV ,end HV )={point HV (t)|t∈[start HV ,end HV ]} Where t represents the time when HV currently transmits data; lon t Indicates the longitude of HV at time t; lat t represents the latitude of HV at time t; v t represents the speed of HV at time t; heading t Indicates the direction of HV at time t; start HV and end HV Respectively represent the start time and end time of HV uploading data; The method of obtaining regional traffic flow trajectory dataset by shooting regional traffic flow information by lidar is as follows: LiDAR captures regional traffic flow information at time t point vid (t), traverse the regional traffic flow scene information collected by the lidar, exclude invalid data outside the road track, use the unique vehicle ID as the identifier for each vehicle, and store the regional traffic flow scene information in the collection trajectorys V In the above example, a regional traffic flow trajectory dataset is formed: trajectory vid (start V ,end V )={point vid (t V )|t V ∈[start V ,end V ]} trajectorys V ={trajectory vid (start V ,end V )|vid∈[1,n]} ={{point vid (t V )}|t V ∈[start v ,end v ],vid∈[1,n]} Among them, t V Indicates the time when the current radar collects traffic flow data; vid indicates the ID of the current vehicle; Indicates t V The longitude of the vehicle at the moment vid; Indicates t V The latitude of the vehicle at the moment; Indicates t V The speed of the vehicle at the moment; Indicates t V The direction of the vehicle at the moment vid; start v and end v They represent the start and end time of the lidar data collection; trajectory vid Represents the regional traffic flow scene information of the vehicle; n represents the number of vehicles contained in the current area; Use the prediction network to predict the future position information of vehicles contained in the regional traffic flow information and complete the regional traffic flow trajectory dataset. The specific method is as follows: Let the time interval of HV uploading data be Δt HV , the time interval of traffic flow in the laser radar collection area is Δt V ; In the vehicle position data contained in the laser radar acquisition area, from the current time t cur Take the window size of x time points forward, and input the longitude and latitude of x time points as the input sequence into the prediction network, which is recorded as: Where I represents the input sequence of the prediction network; Indicates t cur The longitude of the vehicle at the moment, Indicates t cur The latitude of the vehicle at the moment, Indicates that from the current time t cur Take the vehicle longitude for x time intervals forward, Indicates that from the current time t cur Take the vehicle latitude x time intervals ahead; t V Indicates the time when the current radar collects traffic flow data; O represents the output sequence of the prediction network, which is the longitude and latitude of the vehicles in the area at the next y moments after the calculation delay, recorded as: in, Indicates the future cur +Δt V The longitude of the vehicle at the moment, Indicates the future cur +Δt V The latitude of the vehicle at the moment, Indicates the future cur +yΔt V The longitude of the vehicle at the moment, Indicates the future cur +yΔt V The latitude of the vehicle at the moment; Store the predicted points into the set trajectorys according to the vehicle vid V ′, then trajectory V 'for: trajectorys V ′ ={trajectory vid (t cur -(x-1)Δt V ,t cur +yΔt V )|vid∈[1,n]} ={{point vid (t)}|t∈[t cur -(x-1)Δt V ,t cur +yΔt V ],vid∈[1,n]} Among them, vid represents the ID of the current vehicle; n represents the number of vehicles in the current area; trajectory vid Indicates the regional traffic flow scene information of vehicle vid, point vid (t) represents the regional traffic flow information captured by the lidar at time t; Compare the HV trajectory data with the regional traffic flow trajectory dataset collected by LiDAR and identify the HV trajectory in the regional traffic flow trajectory dataset. The specific steps are as follows: trajectories V Each vehicle in ′ contains data trajectories at x+y moments, in trajectory HV (start HV ,end HV ) from the current time t cur Take the point data of the window size x+y moments forward to form the trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ), then trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) is expressed as: trajectory HV '(t cur -(x+y-1)Δt HV ,t cur )={point HV (t)|t∈[t cur -(x+y-1)Δt HV ,t cur ]} trajectory HV '(t cur -(x+y-1)Δt HV ,t cur ) and collection trajectories V The trajectory of vehicle r in ′ r (t cur -(x-1)Δt V ,t cur +yΔt V ) is defined as: Among them, |point HV (i)-point r (j)| indicates point HV (i) with point r (j) Euclidean distance between Then Calculate the trajectory set trajectories V All trajectories and trajectory in ′ HV '(t cur -(x+y-1)Δt HV ,t cur ), then the LiDAR collects the traffic flow trajectory dataset trajectorys V ′ and the real-time vehicle trajectory uploaded by HV HV '(t cur -(x+y-1)Δt HV ,t cur ) The vehicle id of the trajectory with the highest similarity is the vid of the HV trajectory collected by the lidar, recorded as DUTvid, and its trajectory is trajectory DUTvid (start V ,end V ); The method for matching the real-time HV uploaded trajectory with the HV trajectory collected by the lidar to obtain the time difference of the trajectory data is as follows: HV trajectory collected by lidar DUTvid (start V ,end V ) as a reference trajectory, in the real-time upload trajectory of HV HV (start HV ,end HV ) and set the sliding window size to z vehicle position coordinates. DUTvid (start V ,end V ) from the current moment t cur Get z points forward as the reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ), expressed as: From real-time trajectory HV (start HV ,end HV ) Current time t cur Slide forward to divide the trajectory segment set ts HV : Reference trajectory segment tra DUTvid (t cur -(z-1)Δt V ,t cur ) and the trajectory segment set ts HV Trajectory HV (t k -(z-1)Δt HV ,t k ) is defined as: Then From trajectory HV (start HV ,end HV ) Current time t cur When the point moves forward, the similarity generally shows a trend of increasing first and then decreasing. Find the point with the highest similarity, i.e., f(trajectory HV (t k -(z-1)Δt HV ,t k ),tra DUTvid (t cur -(z-1)Δt V ,t cur ))The lowest trajectory segment trajectory HV (t u -(z-1)Δt HV ,t u ), the starting time is t u -(z-1)Δt HV The end time is t u ,for: trajectory HV (t u -(z-1)Δt HV ,t u ) ={point HV (t u -(z-1)Δt HV ),…,point HV (t u )} Then slide forward one window size distance in turn. If these trajectory segments are consistent with the reference trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) is more similar than trajectory HV (t u -(z-1)Δt HV ,t u ) and tra DUTvid (t cur -(z-1)Δt V ,t cur ) is low, then the trajectory is considered HV (t u -(z-1)Δt HV ,t u ) and trajectory tra DUTvid (t cur -(z-1)Δt V ,t cur ) best matches; The trajectory segment trajectory HV (t u -(z-1)Δt HV ,t u ) End point time t u With the tracktra DUTvid (t cur -(z-1)Δt V ,t cur ) End point time t cur The time difference T between the regional traffic flow trajectory data collected by the lidar and the HV real-time data is obtained by subtraction. offset : T offset =|t cur -t u |。 2. The traffic flow trajectory alignment method for large-scale live broadcast testing according to claim 1 is characterized in that: The method for determining the future position information of vehicles included in the regional traffic flow based on the time difference of trajectory data is as follows: If the time difference of trajectory data is T offset ≤yΔt V , then in the prediction network output sequence Find the offset The most recent data point vid (t cur +T offset ′), expressed as: T offset '=iΔt V ,i∈[1,y] point vid (t cur +T offset ′) is the future position information of the vehicle to be determined; T offset ′ represents the time difference of the corrected trajectory data; i represents an index value; If T offset >yΔt V , then use the prediction network model to predict T offset The area after the vehicle contains the future position of the vehicle, which is the future position information of the vehicle that needs to be determined.

3. A traffic flow trajectory alignment system for large-scale live broadcast testing, characterized by: Includes C-V2X module, LiDAR and processing module; The C-V2X module is installed on the HV to upload HV trajectory data in real time, and the lidar is used to capture regional traffic flow information; The input end of the processing module is connected to the output end of the C-V2X module and the laser radar respectively. The processing module executes the method described in claim 1 or 2 to obtain the time difference of the trajectory data and determine the future position information of the vehicles included in the regional traffic flow.

Images