Vehicle longitudinal and transverse cooperative trajectory reconstruction method, device, equipment, medium and product

Abstract

The invention discloses a vehicle longitudinal and transverse cooperative trajectory reconstruction method, device and equipment, a medium and a product, and relates to the technical field of intelligent traffic. The method comprises the steps of firstly constructing a vehicle position matrix and a vehicle speed matrix at each moment, and then determining a vehicle driving state of a target HDV vehicle at the current moment and a longitudinal position and a longitudinal speed of the target HDV vehicle at the next moment based on a construction result. Meanwhile, the vehicle lane changing mode of the target HDV vehicle at the current moment is recognized through a vehicle lane changing mode recognition model based on a two-way gating circulation unit with a self-attention mechanism, and the transverse movement displacement of the target HDV vehicle at the next moment is estimated in combination with an HDV vehicle lane changing transverse movement model; and finally, fusing the longitudinal position and the transverse motion displacement to obtain a three-dimensional space-time trajectory of the target HDV vehicle, so that multi-source data fusion can be realized to solve the problem of trajectory reconstruction of the manually driven vehicle under low CAV permeability, and the vehicle trajectory estimation precision is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, and specifically relates to the field of vehicle trajectory reconstruction technology in mixed environments of connected autonomous vehicles (CAVs) and human-driven vehicles (HDVs). Specifically, it relates to a method, device, equipment, medium, and product for longitudinal and lateral collaborative trajectory reconstruction of vehicles. This invention can combine computer algorithms, traffic flow theory, and machine learning methods to accurately reconstruct HDV vehicle trajectories in complex traffic scenarios such as urban signalized intersections, improving the accuracy of traffic state estimation and signal control optimization. It is particularly suitable for real-time intelligent transportation applications such as dynamic traffic state estimation, signal optimization, and vehicle energy consumption and emission calculation in urban roads and intersections. Background Technology

[0002] Vehicle trajectory data contains rich spatiotemporal traffic information, providing crucial data for analyzing traffic flow dynamics and vehicle operating environments. Current research focuses on the transitional phase between connected autonomous vehicles and human-driven vehicles before the widespread adoption of CAVs. By reconstructing the real-time location, speed, and trajectory of HDV vehicles, traditional high-cost data collection methods such as video surveillance and drones can be replaced, effectively increasing the CAV equivalence ratio in the traffic environment and thus improving the performance of real-time intelligent transportation scenarios.

[0003] Currently, existing vehicle trajectory estimation and reconstruction schemes can be mainly divided into the following three categories: The first type of approach employs deterministic theoretical methods such as traffic shock wave models, variational theory, and micro-vehicle car-following models to estimate the individual HDV state and trajectory using CAV data. For example, based on partial travel time data, it estimates the intersection vehicle arrival rate by assuming uniform flow rate and queue dissipation speed; or it considers the Wiedemann car-following model for HDV position / vehicle estimation, dividing the approach road into stopping zones, deceleration zones, and free-flow zones for separate estimation. This type of approach performs well in scenarios such as emergency vehicle priority and variable speed limit strategies, but it cannot determine the number of HDVs and is prone to estimation bias in low CAV penetration or free-flow scenarios.

[0004] The second type of approach employs probabilistic theoretical methods such as Bayesian networks and maximum likelihood estimation. It proposes algorithms including a fusion model for predicting driving behavior intent based on Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM), Naive Bayes theory, and an adaptive unscented Kalman filter (AUKF) state observer design based on a three-degree-of-freedom model and the Sage-Husa algorithm. These algorithms aim to address the challenge of determining the number of HDVs under low CAV penetration. However, most of these approaches assume that HDV arrivals follow a certain distribution. The randomness of actual traffic flow leads to unstable estimation of the number of HDVs between CAVs, and the accuracy of trajectory estimation during acceleration and deceleration phases is limited.

[0005] The third approach integrates multi-source data, such as data from fixed-point detectors like loop coils and CAV data, and employs artificial intelligence algorithms such as Elman neural networks, multi-node joint optimization algorithms, and dynamic RNNs (Recurrent Neural Networks) to consider trajectory paths and spatiotemporal correlation characteristics, significantly improving the completeness and reliability of road network trajectory reconstruction.

[0006] However, existing technologies primarily focus on driving scenarios such as highways or smooth urban roads, paying less attention to scenarios with complex and variable driving behaviors, such as signalized intersections. Most existing methods assume vehicles maintain their lanes, neglecting the impact of lane-changing behavior on lateral movement, and lack sufficient accuracy in capturing vehicle trajectories during acceleration, deceleration, and slow-moving phases. For example, trajectory reconstruction methods based on Variational Theory (VT) and floating vehicles suffer from limited accuracy in low CAV penetration rates due to unstable HDV (High-Depth Vehicle) number estimation and error accumulation. Furthermore, existing methods struggle to identify HDV lane-changing opportunities and capture longitudinal and lateral trajectory changes, failing to meet the high-precision trajectory data requirements of connected autonomous vehicle applications. Summary of the Invention

[0007] The purpose of this invention is to provide a method, apparatus, computer device, computer-readable storage medium, and computer program product for reconstructing the longitudinal and lateral cooperative trajectory of a vehicle, in order to solve the problems of neglecting the influence of lane-changing behavior on lateral motion and insufficient accuracy of vehicle trajectory capture in existing vehicle trajectory estimation and reconstruction schemes.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a method for reconstructing the longitudinal and lateral cooperative trajectory of a vehicle is provided, including: Based on vehicle trajectory data and the timestamps of vehicles entering the mixed-traffic lane, a system of size [value] is constructed at each time point. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers; According to The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments; At any time The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. Vehicle lane-changing modes; According to the target HDV vehicle at time Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement; Couple the target HDV vehicle at time The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

[0009] Based on the above-mentioned invention, a novel scheme for reconstructing the longitudinal and lateral collaborative trajectories of vehicles in scenarios with low connected vehicle penetration and considering lane-changing behavior is provided. First, vehicle position and velocity matrices are constructed at each time step. Then, based on the constructed results, the driving state of the target HDV vehicle at the current time step, as well as its longitudinal position and longitudinal velocity at the next time step, are determined. Simultaneously, a vehicle lane-changing pattern recognition model based on a bidirectional gated cyclic unit with self-attention mechanism is used to identify the target HDV vehicle's lane-changing pattern at the current time step. This is combined with the HDV vehicle's lateral motion model to estimate the target HDV vehicle's lateral displacement at the next time step. Finally, the longitudinal position and lateral displacement are fused to obtain the target HDV vehicle's three-dimensional spatiotemporal trajectory. By integrating traffic flow theory, neural network algorithms, and trajectory planning methods, multi-source data fusion can be achieved to solve the challenge of reconstructing the trajectory of manually driven vehicles with low CAV penetration, significantly improving the accuracy of vehicle trajectory estimation and facilitating practical application and promotion.

[0010] In one possible design, based on The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal velocity include: According to The target HDV vehicle on the aforementioned mixed lane at time The position and the car in front at the time The location of the target HDV vehicle at time [time] is determined. The distance between the front of the vehicle, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle; According to the target HDV vehicle at time The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, or free-flowing. Let the target HDV vehicle be the one in the first... The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. One vehicle, and according to the target HDV vehicle at time The vehicle's driving status, position, and speed, as well as the time of the preceding vehicle. The position and speed of the target HDV vehicle at time [time] are estimated according to the following methods (A1) to (A4). Longitudinal position and longitudinal velocity: (A1) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is red, the following formula is used to estimate the time of the target HDV vehicle. vertical position and longitudinal velocity :

[0011] In the formula, It represents the time difference between two consecutive moments; (A2) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is green, the target HDV vehicle's position at time [time] is estimated using the following formula. vertical position and longitudinal velocity :

[0012] In the formula, Indicates the driver's reaction time and is less than , Indicates that the target HDV vehicle is at time... acceleration, This represents the expected acceleration of a vehicle in free-flow conditions. This represents the expected speed of a vehicle in free-flow conditions. This indicates the preset speed index parameter; (A3) When the target HDV vehicle is at time When the vehicle is in a variable speed or following mode, the following formula is used to estimate the target HDV vehicle at time [time value missing]. vertical position and longitudinal velocity :

[0013] In the formula, This represents the minimum safe distance between vehicles when they are completely stationary. Indicates the time of the preceding vehicle speed, Indicates the time of the preceding vehicle Location, This represents the maximum acceleration parameter of the target HDV vehicle. This represents the maximum deceleration parameter of the target HDV vehicle. This indicates the length of the target HDV vehicle; (A4) When the target HDV vehicle is at time When the vehicle is in a free-flow state, the target HDV vehicle at time [time value missing] is estimated using the following formula. vertical position and longitudinal velocity : .

[0014] In one possible design, based on the target HDV vehicle at time... The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, and free-flowing. According to the target HDV vehicle at time The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed of the target HDV vehicle at time is determined according to the following logic (B1) to (B4). Vehicle driving status: (B1) If the target HDV vehicle is at time The speed is less than or equal to a preset first speed threshold or the target HDV vehicle and the preceding vehicle are at time... If the speed of the target HDV vehicle is less than or equal to the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicles are in a queue. (B2) If the target HDV vehicle is at time If the speed of the target HDV vehicle is less than a preset second speed threshold but greater than the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a variable speed driving state, wherein the second speed threshold is greater than the first speed threshold; (B3) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is less than or equal to a preset distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a following position. (B4) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is greater than the distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a free-flow state.

[0015] In one possible design, the target HDV vehicle is estimated at time [time value missing]. After determining the longitudinal position and longitudinal velocity and coupling the target HDV vehicle at time [time missing] Before determining the longitudinal position and lateral displacement, the method further includes: Statistics in All HDV vehicles on the aforementioned mixed-traffic lane at time The number of vehicles in various driving states is calculated, and the statistical results are imported into a pre-established error allocation model to output the position cumulative error and speed cumulative error. Applying the aforementioned position accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal position to obtain the target HDV vehicle at time [time]. The corrected longitudinal position; Applying the aforementioned speed accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal velocity to obtain the target HDV vehicle at time [time value missing]. The corrected longitudinal velocity.

[0016] In one possible design, the error allocation model is pre-constructed as follows: A linear regression equation containing a constant term and a linear term was established using regression analysis:

[0017] In the formula, Represents positive integers less than or equal to 2. Indicates the cumulative error, and in Specifically, it represents the cumulative position error and in Specifically, this represents the cumulative speed error. This indicates the number of HDV vehicles currently in the queue. This indicates the number of HDV vehicles in a variable speed state. This indicates the number of HDV vehicles in a following mode. This indicates the number of HDV vehicles in free-flow mode. , , , and They represent the regression coefficients, respectively. Based on multiple sample data, the regression coefficients are obtained using logistic regression techniques. , , , and The solution is then imported into the linear regression equation to obtain the error allocation model.

[0018] In one possible design, the bidirectional gated loop unit includes two independent GRU units, each belonging to a forward propagation layer and a backward propagation layer respectively. The forward propagation layer and the backward propagation layer are used to process the forward and backward information of the input sequence respectively to obtain bidirectional features of vehicle motion. The calculation of the GRU unit includes the following, and at time... Update Gate Reset door Candidate hidden state And the vehicle's final hidden state :

[0019] In the formula, This represents the Sigmoid activation function. This represents the hyperbolic tangent activation function. It represents the Hadamah accumulation. Indicates at time Vehicle input status, Indicates at time The vehicle was finally hidden. , , , , and These represent the weight matrices respectively; And / or, the self-attention mechanism is used to capture the dependencies between elements at different positions in the input sequence in such a way that the final hidden state of the forward-facing vehicle output by the bidirectional gated loop unit is obtained. And the rear-facing vehicle's final hidden state The assembly is shown as the final hidden state of the comprehensive vehicle. And generate the query matrix through linear transformation. Key matrix Sum matrix Then, the attention score used as the output of the self-attention layer is calculated. ,in, Represents the key matrix The number of dimensions, Represents the normalized exponential function, Represents the matrix transpose symbol; And / or, based on the target HDV vehicle at time Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement includes: When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement :

[0020] In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement :

[0021] In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0022] In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0023] In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0024] In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0025] In the formula, Represents a time variable.

[0026] Secondly, a vehicle longitudinal and lateral cooperative trajectory reconstruction device is provided, including a position velocity matrix construction unit, a longitudinal position estimation unit, a lane change pattern recognition unit, a lateral displacement estimation unit, and a longitudinal and lateral data coupling unit. The positional velocity matrix construction unit is used to construct a velocity matrix with a size equal to [value] at each time point based on vehicle trajectory data and the timestamp of the vehicle entering the mixed-traffic lane. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers; The longitudinal position estimation unit is communicatively connected to the position velocity matrix construction unit, and is used to estimate the position based on the position of the matrix. The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments; The lane change pattern recognition unit is communicatively connected to the bit velocity matrix construction unit, and is used to determine the time... The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. Vehicle lane-changing modes; The lateral displacement estimation unit is communicatively connected to the velocity matrix construction unit and the lane change pattern recognition unit, respectively, and is used to estimate the target HDV vehicle at time... Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement; The longitudinal and lateral data coupling unit is communicatively connected to the longitudinal position estimation unit and the lateral displacement estimation unit, respectively, and is used to couple the target HDV vehicle at time... The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

[0027] Thirdly, the present invention provides a computer device comprising a storage module, a processing module, and a transceiver module connected in sequence for communication, wherein the storage module is used to store a computer program, the transceiver module is used to send and receive messages, and the processing module is used to read the computer program and execute the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect or any possible design in the first aspect.

[0028] Fourthly, the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, perform the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect or any possible design in the first aspect.

[0029] Fifthly, the present invention provides a computer program product, including a computer program or instructions, wherein when the computer program or instructions are executed by a computer, they implement the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect or any possible design in the first aspect.

[0030] The beneficial effects of the above scheme are: (1) This invention creatively provides a new scheme for reconstructing the longitudinal and lateral collaborative trajectory of vehicles in scenarios with low CAV penetration and considering lane-changing behavior. That is, firstly, the vehicle position matrix and vehicle speed matrix at each time are constructed. Then, based on the construction results, the vehicle driving state of the target HDV vehicle at the current time and the longitudinal position and longitudinal speed at the next time are determined. At the same time, the vehicle lane-changing pattern recognition model based on the bidirectional gating loop unit with self-attention mechanism is used to identify the vehicle lane-changing pattern of the target HDV vehicle at the current time. Combined with the lateral motion model of HDV vehicle lane changing, the lateral motion displacement of the target HDV vehicle at the next time is estimated. Finally, the longitudinal position and the lateral motion displacement are fused to obtain the three-dimensional spatiotemporal trajectory of the target HDV vehicle. In this way, by integrating traffic flow theory, neural network algorithm and trajectory planning method, multi-source data fusion can be achieved to solve the problem of trajectory reconstruction of manually driven vehicles with low CAV penetration and significantly improve the accuracy of vehicle trajectory estimation. (2) By combining the trajectory estimation algorithm with the cumulative error allocation model, and by using the multi-scale IDM model and multivariate nonlinear fitting, high-precision estimation of HDV longitudinal trajectory can be achieved, effectively solving the error accumulation problem. (3) A method combining spatiotemporal self-attention mechanism and bidirectional gated recurrent unit network for HDV lane-changing behavior detection is proposed. The self-attention mechanism captures sequence dependencies, and the bidirectional gated recurrent unit network is combined to improve the accuracy of lane-changing timing identification, thus overcoming the limitations of existing methods in processing lane-changing behavior. (4) By dividing the lane change trajectory curves according to the speed range and fitting them, and by constructing a three-dimensional spatiotemporal trajectory through motion coupling, the longitudinal and lateral dynamic characteristics can be effectively integrated, thereby improving the integrity and accuracy of trajectory reconstruction. (5) A three-dimensional HDV spatiotemporal trajectory reconstruction scheme that is automated, highly accurate, and can be applied to urban roads and intersections where CAV and HDV are mixed, as well as considering lane-changing behavior, is proposed. It has the characteristics of novel, stable and versatile algorithms, and can provide a technical breakthrough in lane-level vehicle trajectory acquisition, which is convenient for practical application and promotion. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart illustrating the vehicle longitudinal and lateral cooperative trajectory reconstruction method provided in this application embodiment.

[0033] Figure 2 This is an example diagram illustrating vehicle trajectory reconstruction analysis in a mixed CAV and HDV scenario, provided as an embodiment of this application.

[0034] Figure 3 Example diagram showing the difference in trajectory estimation results of the target HDV vehicle when using different minimum time intervals, as provided in the embodiments of this application.

[0035] Figure 4 Example diagrams provided for embodiments of this application, which subdivide 14 vehicle lane-changing modes based on the direction of lane change, the number of HDVs in front of the target vehicle, and the number of HDVs to be changed.

[0036] Figure 5 This is a schematic diagram of the structure of the vehicle lane change pattern recognition model provided in the embodiments of this application.

[0037] Figure 6An example diagram illustrating the curve fitting process for vehicle lane-changing behavior provided in this application embodiment.

[0038] Figure 7 This is an example diagram of the HDV trajectory lane change fitting curves under different initial speed ranges provided in the embodiments of this application.

[0039] Figure 8 An example diagram illustrating the HDV three-dimensional spatiotemporal trajectory reconstruction effect provided in this application embodiment.

[0040] Figure 9 This is a schematic diagram of the vehicle longitudinal and lateral cooperative trajectory reconstruction device provided in an embodiment of this application.

[0041] Figure 10 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these embodiments without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0043] It should be understood that although the terms "first" and "second", etc., may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another. For example, the first object may be referred to as the second object, and similarly, the second object may be referred to as the first object, without departing from the scope of the exemplary embodiments of the invention.

[0044] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, or A and B exist simultaneously. Another example is A, B and / or C, which can mean that any one of A, B, and C or any combination thereof exists. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone or A and B exist simultaneously. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0045] Example like Figure 1As shown, the vehicle longitudinal and lateral cooperative trajectory reconstruction method provided in the first aspect of this embodiment can be executed, but is not limited to, by computer devices with certain computing resources, such as servers, edge computers, and personal computers (PCs, which are multi-purpose computers of a size, price, and performance suitable for personal use; desktop computers, laptops, mini-laptops, tablets, and ultrabooks are all considered personal computers). Figure 1 As shown, the vehicle longitudinal and lateral cooperative trajectory reconstruction method includes, but is not limited to, the following steps S1 to S5.

[0046] S1. Based on vehicle trajectory data and the timestamps of vehicles entering the mixed-traffic lane, construct a system of size [value missing] at each time point. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers.

[0047] In step S1, the vehicle trajectory data has two sources: real-time trajectory data of the connected autonomous vehicle and vehicle trajectory data stored in the storage space, which can be conventionally obtained based on existing technologies. The mixed-traffic lane is exemplified, but not limited to, an entrance lane at an urban intersection (e.g., Figure 2The timestamp of any vehicle (including CAV or HDV vehicles) entering the mixed-traffic lane can be, but is not limited to, acquired by a fixed-point sensor installed in the detection area of ​​the mixed-traffic lane (this fixed-point sensor can also routinely capture other instantaneous state data of CAV and HDV vehicles within the detection area); such as Figure 2 As shown, when all vehicles pass through the detection area sequentially, the time of vehicle detection signal sensed by the specific lane induction coil (i.e., the timestamp of the corresponding vehicle entering the mixed-traffic lane) can be recorded. Simultaneously, the intelligent agent can routinely receive information from all CAV vehicles within the wireless communication range (e.g., real-time vehicle trajectory data). Since the vehicle trajectory data packet contains information such as the position and speed of each vehicle at each collection time, the coil-recorded data can be matched with the CAV and HDV trajectories by sorting vehicles in the same lane according to the time they pass through the detection area, thus forming the vehicle position matrix and the vehicle speed matrix. The maximum number of vehicles per lane... This refers to the maximum number of vehicles that the mixed-traffic lane can accommodate when the traffic density is at a congestion density and all vehicles are queuing at the minimum safe distance (which is a preset spacing value). In detail, this can be discretized. Each of the aforementioned mixed-traffic lanes is divided into lanes of equal length (which may be equal to vehicle length plus minimum safety distance). There are several road blocks, ensuring that at any given time, each road block can accommodate at most one vehicle. Based on the vehicle trajectory data, if a vehicle is found in a road block, its position and speed are written into the corresponding row and column elements of the vehicle position matrix and vehicle speed matrix, respectively. If no vehicle is found in a road block, the corresponding row and column elements of the vehicle position matrix and vehicle speed matrix are filled with 0 (i.e., if the element...). Then it means in the first... The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. (The vehicle in question is a virtual vehicle that does not exist). Furthermore, the first... The vehicle can be a CAV vehicle or an HDV vehicle.

[0048] S2. According to in The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments.

[0049] In step S2, the target HDV vehicle is the object of vehicle trajectory reconstruction. Considering that it has different motion trajectories under different vehicle driving states, it is necessary to first identify and determine its vehicle driving state, and then select an appropriate vehicle trajectory estimation algorithm to estimate its trajectory at time [time]. The longitudinal position and longitudinal velocity. Since both the target HDV vehicle and the preceding vehicle are real vehicles, their elements in the vehicle position matrix are all non-zero. Specifically, according to... The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal velocity include, but are not limited to, the following steps S21 to S23.

[0050] S21. According to... The target HDV vehicle on the aforementioned mixed lane at time The position and the car in front at the time The location of the target HDV vehicle at time [time] is determined. The headway, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle.

[0051] In step S21, the target HDV vehicle and the preceding vehicle can be calculated at time... The position difference is then subtracted from the known length of the target HDV vehicle to obtain the position of the target HDV vehicle at time [time value missing]. The distance between the front and rear of the vehicles. Furthermore, the preceding vehicle can be a CAV (Continuous Vehicle) or an HDV (High-Density Vehicle).

[0052] S22. Based on the target HDV vehicle at time... The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, or free-flowing.

[0053] In step S22, this embodiment divides the vehicle driving state into four types: queuing state, shifting state, following state, and free-flowing state; specifically, based on the target HDV vehicle at a given time... The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, and free-flowing, including but not limited to: based on the target HDV vehicle at time... The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed of the target HDV vehicle at time is determined according to the following logic (B1) to (B4). The vehicle's driving status.

[0054] (B1) If the target HDV vehicle is at time The speed is less than or equal to a preset first speed threshold or the target HDV vehicle and the preceding vehicle are at time... If the speed of the target HDV vehicle is less than or equal to the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicles are in a queue. The first speed threshold can be determined based on traffic flow theory, for example, 0.1 meters per second.

[0055] (B2) If the target HDV vehicle is at time If the speed of the target HDV vehicle is less than a preset second speed threshold but greater than the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle's driving state is a variable speed state, wherein the second speed threshold is greater than the first speed threshold. The variable speed state is also the vehicle's acceleration / deceleration state. The second speed threshold can also be determined based on traffic flow theory, for example, as 80% of the free-flow speed (i.e., 13.34 meters per second).

[0056] (B3) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is less than or equal to a preset distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle's driving state is in a following state. The distance threshold can also be determined based on traffic flow theory, for example, 100 meters.

[0057] (B4) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is greater than the distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a free-flow state.

[0058] S23. Let the target HDV vehicle be the one in the... The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. One vehicle, and according to the target HDV vehicle at time The vehicle's driving status, position, and speed, as well as the time of the preceding vehicle. The position and speed of the target HDV vehicle at time [time] are estimated according to the following methods (A1) to (A4). The longitudinal position and longitudinal velocity.

[0059] (A1) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is red, the following formula is used to estimate the time of the target HDV vehicle. vertical position and longitudinal velocity :

[0060] In the formula, This indicates the time difference between two consecutive moments. Because the traffic lights are at different times... A red light indicates that the target HDV vehicle will be in the future. The vehicle will still be in a stopped or low-speed queuing state, so the vehicle's acceleration is assumed to be zero during this period, hence the above calculation formula.

[0061] (A2) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is green, the target HDV vehicle's position at time [time] is estimated using the following formula. vertical position and longitudinal velocity :

[0062] In the formula, Indicates the driver's reaction time and is less than , Indicates that the target HDV vehicle is at time... acceleration, This represents the expected acceleration of a vehicle in free-flow conditions. This represents the expected speed of a vehicle in free-flow conditions. This indicates the preset speed index parameter. Because the traffic light is at a certain time... A green light indicates that the target HDV vehicle is the first vehicle in the queue (i.e., During the driver's reaction time, it will be within the time limit. The vehicle accelerates later (because there are no vehicles ahead), then gradually approaches free-flow speed; the target HDV vehicle is not the first vehicle in the queue (i.e., When ), its future time period The vehicle will remain stopped or in a low-speed queue, therefore its acceleration during this period is assumed to be zero; based on the aforementioned analysis, the above calculation formula is derived. Furthermore, the vehicle's expected acceleration... The desired speed of the vehicle and the speed exponent parameter It can also be determined based on traffic flow theory, such as the vehicle's expected acceleration. The desired speed of the vehicle is 2.1 meters per square second. The velocity exponent parameter is 16.67 meters per second. .

[0063] (A3) When the target HDV vehicle is at time When the vehicle is in a variable speed or following mode, the following formula is used to estimate the target HDV vehicle at time [time value missing]. vertical position and longitudinal velocity :

[0064] In the formula, This represents the minimum safe distance between vehicles when they are completely stationary. Indicates the time of the preceding vehicle speed, Indicates the time of the preceding vehicle Location, This represents the maximum acceleration parameter of the target HDV vehicle. This represents the maximum deceleration parameter of the target HDV vehicle. This indicates the length of the target HDV vehicle.

[0065] In the aforementioned method (A3), when the target HDV vehicle is in a shifting or following state, it can be assumed that it will be in a future time period. The acceleration within is constant, thus the longitudinal position is as described. and the longitudinal velocity The calculation can be performed based on the following ballistic scheme:

[0066] Among them, acceleration The calculation can be performed based on the following improved IDM (Intelligent Driver Model) model:

[0067] In the formula, This indicates that the target HDV vehicle is in a future time period. The acceleration value inside, And it represents the speed difference between the target HDV vehicle and the preceding vehicle (if the target HDV vehicle is the leading vehicle, i.e.) Since the only factor affecting the acceleration of the first vehicle is whether the light is red or green, the first vehicle is released from restrictions when the light turns green, and then... (The sentence is incomplete and ends abruptly.) Perform acceleration calculations to obtain the acceleration. This is equivalent to the absence of speed difference and headway parameters in the formula; while at a red light, it can be assumed that there is a completely stopped vehicle at the minimum safe distance before the stop line as the vehicle in front, whose speed and acceleration are fixed at 0, and the distance before the stop line is... At this point, the speed difference and headway of the first vehicle can still be calculated using the car-following model. and These are the maximum acceleration and maximum deceleration parameters of the target HDV vehicle, respectively. Indicates the time when the target HDV vehicle and the preceding vehicle are... The actual headway (subtracting the vehicle length) is calculated as follows: (If the target HDV vehicle is the first vehicle, i.e.) : Based on the foregoing analysis, the above calculation formula exists. Furthermore, to improve accuracy, [the following can be added]: Further divided into A short segment, To replace the original interval period And calculate the acceleration value within each small segment, that is:

[0068] In the formula, This indicates that the target HDV vehicle is in a future time period. The acceleration value within the vehicle, the acceleration value of an HDV vehicle or a CAV vehicle in Internal update Next, calculate the acceleration segment by segment. and These represent the preset control coefficients (their values ​​are 0.23 and 0.27, for example). This represents the expected headway of the ACC (Adaptive Cruise Control) following model (1.1 seconds in this example). Figure 3 The minimum time intervals are shown as follows: and The trajectory estimation results of the target HDV vehicle differ at that time.

[0069] (A4) When the target HDV vehicle is at time When the vehicle is in a free-flow state, the target HDV vehicle at time [time value missing] is estimated using the following formula. vertical position and longitudinal velocity : .

[0070] In the aforementioned method (A4), since the preceding vehicle has no effect on its motion, the target HDV vehicle can be calculated according to the free-flow formula until it tends to accelerate to the desired speed, hence the above calculation formula.

[0071] In step S2, to minimize the estimation error, the HDV data between any two CAVs needs to be calculated starting from the first HDV following the preceding CAV (i.e., the preceding CAV in any two CAVs) and extending to the last HDV before the following CAV (i.e., the last CAV in any two CAVs). Additionally, it is necessary to... Estimated lane Upper The car is acceleration values ​​during the period Then, based on its dynamic following relationship with the vehicle in front, it estimates its position at the next moment. vertical position and longitudinal velocity And continue until the end time. .

[0072] S3. At time The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. The vehicle lane-changing mode.

[0073] In step S3, the signal status includes, but is not limited to, red, green, and yellow lights. The vehicle lane-changing mode can be broadly categorized into three types: no lane change, left lane change, and right lane change, or it can be like... Figure 4 Based on the lane change direction, the number of HDVs in front of the target vehicle, and the number of HDVs intending to change lanes, it is further subdivided into 14 types. Specifically, the bidirectional gated recurrent unit includes two independent GRU (Gated Recurrent Unit) units, each belonging to the forward propagation layer and the backward propagation layer respectively. The forward propagation layer and the backward propagation layer are used to process the forward and reverse information of the input sequence respectively to obtain the bidirectional features of vehicle motion. The calculation of the GRU unit includes the following, and at time... Update Gate Reset door Candidate hidden state And the vehicle's final hidden state :

[0074] In the formula, This represents the Sigmoid activation function. This represents the hyperbolic tangent activation function. It represents the Hadamah accumulation. Indicates at time Vehicle input status, Indicates at time The vehicle was finally hidden. , , , , and These represent the weight matrices, respectively. The Sigmoid activation function generates probability values ​​between 0 and 1 to control the amount of information transmitted; the reset gate controls the amount of information retained from the previous time step, and the update gate controls the amount of information in the current candidate state and the output of the vehicle's previous hidden state. Thus, compared to a single GRU unit, the bidirectional gated recurrent unit provided in this embodiment can simultaneously consider the positive and negative relationships of elements in the input sequence, so as to capture the information in each vehicle observation sequence more completely.

[0075] In step S3, specifically, the self-attention mechanism is used to capture the dependencies between elements at different positions in the input sequence in the following manner: the final hidden state of the forward-facing vehicle output by the bidirectional gated loop unit. And the rear-facing vehicle's final hidden state The assembly is shown as the final hidden state of the comprehensive vehicle. And generate the query matrix through linear transformation. Key matrix Sum matrix Then, the attention score used as the output of the self-attention layer is calculated. ,in, Represents the key matrix The number of dimensions, Represents the normalized exponential function, The symbol represents the matrix transpose. The self-attention mechanism enables the model to simultaneously focus on local and global relationships in the data. Its structure is shown in Figure 5(a). The technical idea is: given a concatenated vector (i.e., the final hidden state of the integrated vehicle)... The input matrix is ​​transformed into the query matrix through various matrix transformations. Key matrix Sum matrix Then through the query matrix AND key matrix The similarity is calculated by multiplying the transposes of the matrix to obtain the correlation matrix. Then, the correlation matrix is ​​normalized using the Softmax operation; finally, the normalized correlation matrix and the value matrix are compared. Multiplying these results yields the result of the self-attention layer. The final output of the self-attention mechanism is then converted into a label for the lane-changing mode. Therefore, based on the aforementioned self-attention mechanism and bidirectional gated recurrent unit, a system can be constructed as follows: Figure 5 The neural network structure shown in (c) is then used to process a certain amount of sample data (its model input is at time 1). The vehicle position matrix and vehicle speed matrix, as well as the traffic lights in the lane where the sample HDV vehicle is located at time... The signal state, the model output term is the sample HDV vehicle at time t. Vehicle lane-changing mode labels (e.g., no lane change, changing lanes to the left, or changing lanes to the right) are imported into this neural network structure and processed sequentially through forward and backward propagation layers. The BiGRU encoder... The final hidden state can be obtained by splicing the hidden states of the forward propagation layer. and the hidden state of the backpropagation layer To obtain more comprehensive trajectory features, a self-attention mechanism is introduced to dynamically capture the dependencies between elements at different positions in the sequence, taking into account the dependencies between vehicles in front and behind and vehicles in adjacent lanes. Based on these dependencies, a new sequence representation is generated, and finally, the vehicle lane change pattern recognition model is output after training.

[0076] S4. According to the target HDV vehicle at time Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement.

[0077] In step S4, after identifying the lane-changing timing of the target HDV vehicle, it is necessary to fit and calibrate the lane-changing trajectory curve of the target HDV vehicle based on actual cases. Geometry-based trajectory planning typically uses parametric curves to depict the vehicle's lane-changing trajectory. This method is commonly used due to its intuitiveness, accuracy, and low computational load. To address the limitations of traditional curve fitting methods (such as the sine offset function ignoring the influence of vehicle speed changes on the lane-changing trajectory, or the high order of sixth-order polynomials leading to low computational efficiency), this embodiment introduces a fifth-order polynomial-based HDV lane-changing trajectory fitting method to construct a fifth-order polynomial lane-changing trajectory model. This fifth-order polynomial lane-changing trajectory model only requires the vehicle's initial and target states to calculate and derive the lane-changing trajectory. This embodiment uses a function... Describe the time variable during the HDV lane change process. The subsequent longitudinal motion trajectory and lateral movement trajectory This function clearly displays the kinematic characteristics of the vehicle from its initial position to its target position. Its lane change trajectory function expression is as follows:

[0078] In the formula, Represents integers, and These represent the polynomial coefficients.

[0079] In step S4, the polynomial coefficients and The following method can be used to fit the data: Record multi-lane vehicle trajectories using aerial video, divide the video into single frames, and identify the coordinates of vehicles in each frame using an object detection algorithm; then, through... Figure 6 The smooth curve shown connects the vehicle coordinates to generate an instance trajectory, which in turn allows the determination of the polynomial coefficients. and Furthermore, considering the significant differences in the trajectories generated by vehicles during forced lane changes and free lane changes, this embodiment also considers the initial speed during lane changes. (Based on the observed phenomena, thresholds of 10 m / s and 15 m / s were set) and the lane change direction was used to divide the observed trajectory into 6 groups. Then, by overlapping the midpoints of the lane change trajectories, the following results were obtained: Figure 7 The results are shown, and the lane change trajectories are fitted according to six different classifications. This is achieved by using six different sets of the aforementioned polynomial coefficients. and Substituting the lane change trajectory function expression, we can obtain the six pairs of HDV vehicle lane change longitudinal motion models shown in (C1) to (C6) below. HDV vehicle lane change lateral motion model : (C1) The longitudinal velocity of the HDV at the initial lane change moment satisfy Furthermore, when changing lanes to the left, its longitudinal motion model and lateral motion model for: ; (C2) When Furthermore, when changing lanes to the right, its longitudinal motion model and lateral motion model for: ; (C3) When Furthermore, when changing lanes to the left, its longitudinal motion model and lateral motion model for: ; (C4) When Furthermore, when changing lanes to the right, its longitudinal motion model and lateral motion model for: ; (C5) When Furthermore, when changing lanes to the left, its longitudinal motion model and lateral motion model for: ; (C6) When Furthermore, when changing lanes to the right, its longitudinal motion model and lateral motion model for: .

[0080] In step S4, based on the lateral motion model described in (C1) to (C6) above... Specifically, based on the target HDV vehicle at a given time... Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement includes, but is not limited to, the displacement of the target HDV vehicle at time [time value missing]. The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement :

[0081] In the formula, Represents a time variable; when the target HDV vehicle is at time... The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement :

[0082] In the formula, Represents a time variable; when the target HDV vehicle is at time... The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0083] In the formula, Represents a time variable; when the target HDV vehicle is at time... The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0084] In the formula, Represents a time variable; when the target HDV vehicle is at time... The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0085] In the formula, Represents a time variable; when the target HDV vehicle is at time... The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement :

[0086] In the formula, Represents a time variable.

[0087] S5. Couple the target HDV vehicle at time... The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

[0088] In step S5, the longitudinal position and lateral displacement of the target HDV vehicle at each moment can be used as the trajectory points of the target HDV vehicle at each moment, thereby obtaining the three-dimensional spatiotemporal trajectory of the target HDV vehicle, such as... Figure 8 As shown. Considering that trajectory estimation errors accumulate with the increase in the number of vehicles, 1894 HDV trajectory data estimation cases were collected through simulation experiments to track the trends of position and speed errors. These included 602 cases of queuing, 484 cases of shifting, 542 cases of following, and 266 cases of free-flowing HDVs. The estimated values ​​obtained in step S2 were compared with the actual data, and the accumulated position and speed errors were sorted by vehicle state and error sign. Experimental results show that the accumulated error generally increases with the increase in the number of estimated vehicles, and the accumulated errors of the four vehicle states differ significantly. Based on the above findings, this embodiment also needs to establish an error allocation model to correct the estimated values ​​obtained in step S2. Specifically, preferably, when estimating the target HDV vehicle at time... After determining the longitudinal position and longitudinal velocity and coupling the target HDV vehicle at time [time missing] Before determining the longitudinal position and lateral displacement, the method further includes, but is not limited to, the following steps S501 to S503.

[0089] S501. Statistics in All HDV vehicles on the aforementioned mixed-traffic lane at time The system calculates the number of vehicles in various driving states and imports the statistical results into a pre-established error allocation model, outputting the cumulative position error and cumulative speed error.

[0090] In step S501, all HDV vehicles at time The specific process for determining the vehicle's driving state can be found in the conventional derivation of step S2 mentioned above, and will not be repeated here. The error allocation model can be constructed, but is not limited to, using regression analysis. That is, to quantitatively characterize the relationship between the cumulative error and the HDV quantity under each driving state, the error allocation model is pre-constructed as follows: First, a linear regression equation containing a constant term and a linear term is established using regression analysis:

[0091] In the formula, Represents positive integers less than or equal to 2. Indicates the cumulative error, and in Specifically, it represents the cumulative position error and in Specifically, this represents the cumulative speed error. This indicates the number of HDV vehicles currently in the queue. This indicates the number of HDV vehicles in a variable speed state. This indicates the number of HDV vehicles in a following mode. This indicates the number of HDV vehicles in free-flow mode. , , , and Let each represent the regression coefficient; then, based on multiple sample data, logistic regression techniques are applied to obtain the regression coefficients. , , , and The solution results are then imported into the linear regression equation to obtain the error allocation model. The multiple sample data can specifically refer to the 1894 HDV trajectory data estimation cases collected through simulation experiments. Logistic regression can be applied using Python to obtain the quantitative relationships between the cumulative position error and cumulative velocity error and the number of HDVs in each state, as shown below: .

[0092] S502. Apply the position accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal position to obtain the target HDV vehicle at time [time]. The corrected longitudinal position.

[0093] S503. Apply the speed accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal velocity to obtain the target HDV vehicle at time [time value missing]. The corrected longitudinal velocity.

[0094] After step S5, the three-dimensional spatiotemporal trajectory can be fed back to steps S2 and S501-S503 for error calibration, ultimately achieving iterative improvement in trajectory reconstruction accuracy.

[0095] Therefore, based on the vehicle longitudinal and lateral cooperative trajectory reconstruction method described in steps S1 to S5 above, a new scheme for vehicle longitudinal and lateral cooperative trajectory reconstruction is provided in scenarios with low connected vehicle penetration and considering lane-changing behavior. First, the vehicle position matrix and vehicle velocity matrix at each time step are constructed. Then, based on the construction results, the vehicle driving state of the target HDV vehicle at the current time step, as well as its longitudinal position and longitudinal velocity at the next time step, are determined. Simultaneously, a vehicle lane-changing pattern recognition model based on a bidirectional gated cyclic unit with self-attention mechanism is used to identify the target HDV vehicle's lane-changing pattern at the current time step. Combined with the HDV vehicle's lateral motion model for lane changing, the lateral displacement of the target HDV vehicle at the next time step is estimated. Finally, the longitudinal position and lateral displacement are fused to obtain the three-dimensional spatiotemporal trajectory of the target HDV vehicle. By integrating traffic flow theory, neural network algorithms, and trajectory planning methods, multi-source data fusion can be achieved to solve the problem of trajectory reconstruction for manually driven vehicles with low CAV penetration, significantly improving the accuracy of vehicle trajectory estimation and facilitating practical application and promotion.

[0096] like Figure 9 As shown, the second aspect of this embodiment provides a virtual device for implementing the vehicle longitudinal and lateral cooperative trajectory reconstruction method described in the first aspect, including a position velocity matrix construction unit, a longitudinal position estimation unit, a lane change pattern recognition unit, a lateral displacement estimation unit, and a longitudinal and lateral data coupling unit. The positional velocity matrix construction unit is used to construct a velocity matrix with a size equal to [value] at each time point based on vehicle trajectory data and the timestamp of the vehicle entering the mixed-traffic lane. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers; The longitudinal position estimation unit is communicatively connected to the position velocity matrix construction unit, and is used to estimate the position based on the position of the matrix. The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments; The lane change pattern recognition unit is communicatively connected to the bit velocity matrix construction unit, and is used to determine the time... The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. Vehicle lane-changing modes; The lateral displacement estimation unit is communicatively connected to the velocity matrix construction unit and the lane change pattern recognition unit, respectively, and is used to estimate the target HDV vehicle at time... Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement; The longitudinal and lateral data coupling unit is communicatively connected to the longitudinal position estimation unit and the lateral displacement estimation unit, respectively, and is used to couple the target HDV vehicle at time... The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

[0097] The working process, working details and technical effects of the aforementioned device provided in the second aspect of this embodiment can be found in the vehicle longitudinal and lateral cooperative trajectory reconstruction method described in the first aspect, and will not be repeated here.

[0098] like Figure 10 As shown, the third aspect of this embodiment provides a computer device for executing the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect. The device includes a storage module, a processing module, and a transceiver module connected in sequence. The storage module stores a computer program, the transceiver module sends and receives messages, and the processing module reads the computer program and executes the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect. Specifically, the storage module may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or first-in-last-out (FILO) memory, etc.; the processing module may, but is not limited to, use a microprocessor of the STM32F105 series. Furthermore, the computer device may also include, but is not limited to, a power supply module, a display screen, and other necessary components.

[0099] The working process, working details and technical effects of the aforementioned computer equipment provided in the third aspect of this embodiment can be found in the vehicle longitudinal and lateral cooperative trajectory reconstruction method described in the first aspect, and will not be repeated here.

[0100] This fourth aspect of the embodiment provides a computer-readable storage medium storing instructions comprising the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect. Specifically, the computer-readable storage medium stores instructions that, when executed on a computer, perform the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0101] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided in the fourth aspect of this embodiment can be found in the vehicle longitudinal and lateral cooperative trajectory reconstruction method described in the first aspect, and will not be repeated here.

[0102] This fifth aspect of the embodiment provides a computer program product, including a computer program or instructions, which, when executed by a computer, implements the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in the first aspect. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0103] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for reconstructing the longitudinal and lateral cooperative trajectory of a vehicle, characterized in that, include: Based on vehicle trajectory data and the timestamps of vehicles entering the mixed-traffic lane, a system of size [value] is constructed at each time point. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers; According to The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments; At any time The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. Vehicle lane-changing modes; According to the target HDV vehicle at time Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement; Couple the target HDV vehicle at time The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

2. The vehicle longitudinal and lateral cooperative trajectory reconstruction method according to claim 1, characterized in that, According to The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal velocity include: According to The target HDV vehicle on the aforementioned mixed lane at time The position and the car in front at the time The location of the target HDV vehicle at time [time] is determined. The distance between the front of the vehicle, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle; According to the target HDV vehicle at time The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, or free-flowing. Let the target HDV vehicle be the one in the first... The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. One vehicle, and according to the target HDV vehicle at time The vehicle's driving status, position, and speed, as well as the time of the preceding vehicle. The position and speed of the target HDV vehicle at time [time] are estimated according to the following methods (A1) to (A4). Longitudinal position and longitudinal velocity: (A1) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is red, the following formula is used to estimate the time of the target HDV vehicle. vertical position and longitudinal velocity : In the formula, It represents the time difference between two consecutive moments; (A2) When the target HDV vehicle is at time When the vehicles are in a queue, if the traffic light of the lane where the target HDV vehicle is located is at time... If the light is green, the target HDV vehicle's position at time [time] is estimated using the following formula. vertical position and longitudinal velocity : In the formula, Indicates the driver's reaction time and is less than , Indicates that the target HDV vehicle is at time... acceleration, This represents the expected acceleration of a vehicle in free-flow conditions. This represents the expected speed of a vehicle in free-flow conditions. This indicates the preset speed index parameter; (A3) When the target HDV vehicle is at time When the vehicle is in a variable speed or following mode, the following formula is used to estimate the target HDV vehicle at time [time value missing]. vertical position and longitudinal velocity : In the formula, This represents the minimum safe distance between vehicles when they are completely stationary. Indicates the time of the preceding vehicle speed, Indicates the time of the preceding vehicle Location, This represents the maximum acceleration parameter of the target HDV vehicle. This represents the maximum deceleration parameter of the target HDV vehicle. This indicates the length of the target HDV vehicle; (A4) When the target HDV vehicle is at time When the vehicle is in a free-flow state, the target HDV vehicle at time [time value missing] is estimated using the following formula. vertical position and longitudinal velocity : 。 3. The vehicle longitudinal and lateral cooperative trajectory reconstruction method according to claim 2, characterized in that, According to the target HDV vehicle at time The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed is used to determine the target HDV vehicle at time [time]. The vehicle's driving state is one of the following: queuing, shifting, following, and free-flowing. According to the target HDV vehicle at time The distance between the front of the vehicles and their speed / and the time of the vehicle ahead The speed of the target HDV vehicle at time is determined according to the following logic (B1) to (B4). Vehicle driving status: (B1) If the target HDV vehicle is at time The speed is less than or equal to a preset first speed threshold or the target HDV vehicle and the preceding vehicle are at time... If the speed of the target HDV vehicle is less than or equal to the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicles are in a queue. (B2) If the target HDV vehicle is at time If the speed of the target HDV vehicle is less than a preset second speed threshold but greater than the first speed threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a variable speed driving state, wherein the second speed threshold is greater than the first speed threshold; (B3) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is less than or equal to a preset distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a following position. (B4) If the target HDV vehicle is at time The speed is greater than or equal to the second speed threshold and the target HDV vehicle is at time [time value missing]. If the distance between the front ends of the vehicles is greater than the distance threshold, then the target HDV vehicle is determined to be at time [time value missing]. The vehicle is in a free-flow state.

4. The vehicle longitudinal and lateral cooperative trajectory reconstruction method according to claim 1, characterized in that, The target HDV vehicle is estimated to be at time [time]. After determining the longitudinal position and longitudinal velocity and coupling the target HDV vehicle at time [time missing] Before determining the longitudinal position and lateral displacement, the method further includes: Statistics in All HDV vehicles on the aforementioned mixed-traffic lane at time The number of vehicles in various driving states is calculated, and the statistical results are imported into a pre-established error allocation model to output the position cumulative error and speed cumulative error. Applying the aforementioned position accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal position to obtain the target HDV vehicle at time [time]. The corrected longitudinal position; Applying the aforementioned speed accumulation error to the target HDV vehicle at time... Error correction is performed on the longitudinal velocity to obtain the target HDV vehicle at time [time value missing]. The corrected longitudinal velocity.

5. The vehicle longitudinal and lateral cooperative trajectory reconstruction method according to claim 4, characterized in that, The error allocation model is pre-constructed as follows: A linear regression equation containing a constant term and a linear term was established using regression analysis: In the formula, Represents positive integers less than or equal to 2. Indicates the cumulative error, and in Specifically, it represents the cumulative position error and in Specifically, this represents the cumulative speed error. This indicates the number of HDV vehicles currently in the queue. This indicates the number of HDV vehicles in a variable speed state. This indicates the number of HDV vehicles in a following mode. This indicates the number of HDV vehicles in free-flow mode. , , , and They represent the regression coefficients, respectively. Based on multiple sample data, the regression coefficients are obtained using logistic regression techniques. , , , and The solution is then imported into the linear regression equation to obtain the error allocation model.

6. The vehicle longitudinal and lateral cooperative trajectory reconstruction method according to claim 1, characterized in that, The bidirectional gated loop unit includes two independent GRU units, each belonging to the forward propagation layer and the backward propagation layer respectively. The forward propagation layer and the backward propagation layer are used to process the forward and backward information of the input sequence respectively to obtain bidirectional features of vehicle motion. The calculation of the GRU unit includes the following steps at time... Update Gate Reset door Candidate hidden state And the vehicle's final hidden state : In the formula, This represents the Sigmoid activation function. This represents the hyperbolic tangent activation function. It represents the Hadamah accumulation. Indicates at time Vehicle input status, Indicates at time The vehicle was finally hidden. , , , , and These represent the weight matrices respectively; And / or, the self-attention mechanism is used to capture the dependencies between elements at different positions in the input sequence in such a way that the final hidden state of the forward-facing vehicle output by the bidirectional gated loop unit is obtained. And the rear-facing vehicle's final hidden state The assembly is shown as the final hidden state of the comprehensive vehicle. And generate the query matrix through linear transformation. Key matrix Sum matrix Then, the attention score used as the output of the self-attention layer is calculated. ,in, Represents the key matrix The number of dimensions, Represents the normalized exponential function, Represents the matrix transpose symbol; And / or, based on the target HDV vehicle at time Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement includes: When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement : In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is less than or equal to 10 meters per second, the target HDV vehicle's lateral motion during lane changing is estimated by combining the pre-established HDV vehicle lane-changing lateral motion model. Lateral displacement : In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement : In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is between 10 and 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement : In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a left lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement : In the formula, Represents a time variable; When the target HDV vehicle is at time The vehicle lane-changing mode is a right lane change and the target HDV vehicle is at time When the speed is greater than or equal to 15 meters per second, the target HDV vehicle's lateral motion during lane change is estimated by combining the pre-established HDV vehicle lane change lateral motion model. Lateral displacement : In the formula, Represents a time variable.

7. A vehicle longitudinal and lateral cooperative trajectory reconstruction device, characterized in that, It includes a position velocity matrix construction unit, a longitudinal position estimation unit, a lane change pattern recognition unit, a lateral displacement estimation unit, and a longitudinal and lateral data coupling unit; The positional velocity matrix construction unit is used to construct a velocity matrix with a size equal to [value] at each time point based on vehicle trajectory data and the timestamp of the vehicle entering the mixed-traffic lane. The vehicle position matrix and vehicle speed matrix, wherein the mixed-traffic lane refers to a lane that allows CAV vehicles and HDV vehicles to travel together. This indicates the total number of the mixed-traffic lanes. Indicates in The maximum number of vehicles per lane on the mixed-traffic lane, and the vehicle position matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time The position of the vehicle speed matrix located at the th And the first Column elements Indicates the first The first one in each of the mixed-traffic lanes, arranged sequentially from the lane stop line. Each vehicle at time speed, Indicates less than or equal to positive integers, Indicates less than or equal to Positive integers; The longitudinal position estimation unit is communicatively connected to the position velocity matrix construction unit, and is used to estimate the position based on the position of the matrix. The target HDV vehicle on the aforementioned mixed lane at time Position and speed and the vehicle in front at the time The position and speed of the target HDV vehicle are used to determine the time. The vehicle's driving status, and based on that driving status and the target HDV vehicle at time... The position and velocity of the target HDV vehicle at time [time value missing] are estimated. The longitudinal position and longitudinal speed, wherein the preceding vehicle refers to the nearest vehicle in the same lane as the target HDV vehicle and located in front of the target HDV vehicle. It represents the time difference between two consecutive moments; The lane change pattern recognition unit is communicatively connected to the bit velocity matrix construction unit, and is used to determine the time... The vehicle position matrix, the vehicle speed matrix, and the traffic lights of the lane where the target HDV vehicle is located at time... The signal state is imported into a vehicle lane change pattern recognition model pre-trained based on a bidirectional gated recurrent unit with self-attention mechanism, and the output is the target HDV vehicle at time [time value missing]. Vehicle lane-changing modes; The lateral displacement estimation unit is communicatively connected to the velocity matrix construction unit and the lane change pattern recognition unit, respectively, and is used to estimate the target HDV vehicle at time... Based on the vehicle lane-changing pattern and speed, and combined with a pre-established HDV vehicle lane-changing lateral motion model, the target HDV vehicle's position at time [time value missing] is estimated. The lateral displacement; The longitudinal and lateral data coupling unit is communicatively connected to the longitudinal position estimation unit and the lateral displacement estimation unit, respectively, and is used to couple the target HDV vehicle at time... The longitudinal position and lateral displacement of the target HDV vehicle are used to form its three-dimensional spatiotemporal trajectory.

8. A computer device, characterized in that, It includes a storage module, a processing module, and a transceiver module that are sequentially connected in communication. The storage module is used to store a computer program, the transceiver module is used to send and receive messages, and the processing module is used to read the computer program and execute the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that... The computer-readable storage medium stores instructions that, when executed on a computer, perform the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or the instructions are executed by the computer, they implement the vehicle longitudinal and lateral cooperative trajectory reconstruction method as described in any one of claims 1 to 6.

Images