Vehicle interaction quantification method based on driver spatial perception and relative motion mapping and electronic device

Abstract

The application relates to the technical field of intelligent traffic, and discloses a vehicle interaction quantification method based on driver spatial perception and relative motion mapping and an electronic device; the method establishes a local coordinate system of a target vehicle taking relative motion as a reference through coordinate transformation, constructs a parameterized approach resistance model, quantifies the discomfort of a driver to the intrusion of surrounding vehicles into his / her psychological space, optimizes model parameters by using historical trajectory data, finally calculates quantification indexes of quasi-density and quasi-flow based on the optimized model, and draws an interaction representation graph to dynamically represent a traffic flow state; the method breaks through the limitation of a traditional one-dimensional traffic model, quantifies driver psychological perception for the first time and embeds the driver psychological perception into interaction analysis, realizes fine modeling of a complex two-dimensional interaction scene such as a city intersection, and provides an innovative theoretical tool and a dynamic analysis framework for intelligent traffic control and automatic driving decision-making.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and more specifically to a vehicle interaction quantification method and electronic device based on driver spatial perception and relative motion mapping. Background Technology

[0002] Urban intersections are bottlenecks and high-incidence areas of conflict in road traffic networks. Their operational efficiency and safety level directly affect the traffic capacity of the entire road network. Accurately characterizing, quantifying, and predicting the dynamic interaction behavior between vehicles within the intersection area is a key theoretical foundation for implementing intelligent traffic proactive control, high-level autonomous driving decision-making, and real-time traffic safety early warning.

[0003] In existing technologies, traffic flow modeling typically focuses on one-dimensional driving. Microscopic models describe the characteristics of individual vehicles, while macroscopic models focus on the relationships between traffic states (such as speed, flow rate, and density). However, traditional flow models are based on fluid dynamics assumptions, but these assumptions often do not hold true in urban traffic. To describe more complex interactive behaviors, social force models and game theory models have been introduced into traffic research. However, these models still have significant shortcomings. Most are based on the vehicle's physical kinematics (position, speed, acceleration) and pre-defined mathematical or behavioral rules, lacking effective quantification and integration of the driver's subjective psychological perceptions (such as comfort, risk perception, and tolerance). In fact, driving behavior is not only constrained by physical factors but also profoundly influenced by the driver's instantaneous psychological state. The driver's definition of "safe space" or "comfort space" is dynamic, subjective, and directionally asymmetric (for example, the driver is usually more sensitive to approaching vehicles from the front and side than from the side and rear). This lack of a key "human factor" element limits the explanatory power and predictive accuracy of existing models when describing micro-level driving behavior in complex conflict scenarios, failing to fundamentally solve the core challenge of integrating two-dimensional interaction quantification with human perception modeling.

[0004] Therefore, there is an urgent need in this field for a method that can break through the limitations of one-dimensional models and map the two-dimensional interactions of various geometric directions at intersections onto a calculable and comparable quantitative index. This would enable dynamic and quantitative assessment and diagnosis of the overall "interaction intensity" and "interaction efficiency" of intersections, supporting the next generation of intelligent transportation systems to leap from "state monitoring" to "behavioral inference" and "active intervention," thus facilitating the implementation of proactive intelligent transportation management. Summary of the Invention

[0005] To overcome the aforementioned technical problems, this invention provides a vehicle interaction quantification method based on driver spatial perception and relative motion mapping. This method establishes a local coordinate system of the target vehicle based on relative motion through coordinate transformation, constructs a parameterized approach resistance model, quantifies the driver's discomfort at the intrusion of surrounding vehicles into their psychological space, optimizes model parameters using historical trajectory data, and finally calculates quantitative indicators of quasi-density and quasi-flow based on the optimized model, and draws an interaction representation map to dynamically represent the traffic flow state. This method breaks through the limitations of traditional one-dimensional traffic models, quantifies driver psychological perception and embeds it into interaction analysis for the first time, and realizes refined modeling of complex two-dimensional interaction scenarios such as urban intersections, providing innovative theoretical tools and dynamic analysis frameworks for intelligent traffic management and autonomous driving decision-making.

[0006] To achieve the above objectives, the present invention provides a vehicle interaction quantification method based on driver spatial perception and relative motion mapping, the method comprising: Construct a proximity resistance model based on spatial perception; Collect historical vehicle trajectory datasets and optimize the parameters of the approach resistance model; The interaction index is determined and quantified using the optimized approach resistance model to generate a vehicle interaction characterization map.

[0007] Preferably, a spatially-aware approach resistance model is constructed, including: Construct vehicle-based mapping coordinates; A spatially-aware proximity resistance model is constructed using the density function of the generalized Gaussian distribution.

[0008] Preferably, constructing vehicle-based mapped coordinates includes: Obtain the coordinates of the target vehicle and surrounding vehicles in the defined global coordinate system, and calculate the relative velocity vectors of the target vehicle and surrounding vehicles using formulas (1)-(3). (1) (2) (3) in, For the target vehicle In the global coordinate system The following position coordinates, for surrounding vehicles In the global coordinate system The following position coordinates, For the target vehicle In the global coordinate system Below x Coordinates along the axis, For the target vehicle In the global coordinate system Below y Coordinates along the axis, for surrounding vehicles In the global coordinate system Below x Coordinates along the axis, for surrounding vehicles In the global coordinate system Below y Coordinates along the axis, For the target vehicle speed, For the target vehicle In the global coordinate system Below x Velocity components in the axial direction, For the target vehicle In the global coordinate system The velocity component along the y-axis. for surrounding vehicles speed, for surrounding vehicles In the global coordinate system Below x Velocity components in the axial direction, for surrounding vehicles In the global coordinate system The velocity component along the y-axis. It is a relative velocity vector; Construct a system based on the target vehicle's position coordinates as the origin and... Direction is The local coordinate system in the positive direction of the axis is transformed from the global coordinate system to the local coordinate system using formula (4). (4) in, for surrounding vehicles Coordinates in a local coordinate system with the target vehicle's position coordinates as the origin. It is a two-dimensional rotation matrix. To transform and rotate the y-axis of the global coordinate system so that it aligns with... The rotation angle required for orientation alignment. for surrounding vehicles Coordinates in the global coordinate system The translation vector required to transform the origin of the coordinate system from the global coordinate system to the local coordinate system; Based on the target vehicle being the origin in the local coordinate system and the endpoints of the relative velocity vector being in the local coordinate system... In the positive direction of the axis, construct the corresponding parameters, and obtain the transformation parameters corresponding to the translation vector and rotation angle according to formula (5). (5) in, To move the origin of the global coordinate system along x The components of the translation vector that translates the axis to the origin of the local coordinate system. To move the origin of the global coordinate system along y The components of the translation vector that translates the axis to the origin of the local coordinate system; The target vehicle and the surrounding vehicles are mapped using formulas (6)-(7). (6) (7) in, for surrounding vehicles In the local coordinate system x The coordinate values ​​of the axis are... , for surrounding vehicles In the local coordinate system y The coordinate values ​​of the axis are... , for surrounding vehicles Relative to the target vehicle The relative velocity scalar is... .

[0009] Preferably, a spatially-aware proximity resistance model is constructed using the density function of a generalized Gaussian distribution, including: Obtain the mapped coordinates of the target vehicle and surrounding vehicles; Based on the density function of the generalized Gaussian distribution, a symmetric approach resistance model in a two-dimensional scenario is constructed using formulas (8)-(9). (8) (9) in, To approach the resistance value, For parameter vectors, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters in the axial direction, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters along the axial direction; Based on the asymmetry of the driver's perception in the front-back and lateral directions, the parameter vectors of the symmetrical approach resistance model in the two-dimensional scenario are extended using formulas (10)-(15), and the two-dimensional asymmetric approach resistance model is obtained as the final approach resistance model. (10) (11) (12) (13) (14) (15) in, For the final parameter vector, For symbolic functions, for The dimensional parameters in the positive axis direction are the dimensional parameters on the driver's right side. for The dimensional parameters in the negative axis direction are the same as the dimensional parameters on the driver's left side. for The dimensional parameters in the positive axis direction are the dimensional parameters in front of the driver. for The dimensional parameters in the negative axis direction are the dimensional parameters behind the driver. for The speed parameter in the positive direction of the axis is the speed parameter on the driver's right side. for The speed parameter in the negative direction of the axis is the speed parameter on the driver's left side. for The speed parameter in the positive direction of the axis is the speed parameter in front of the driver. for The speed parameter in the negative direction of the axis is the speed parameter behind the driver.

[0010] Preferably, historical vehicle trajectory datasets are collected, and the approach resistance model is optimized for parameters, including: Collect historical vehicle trajectory datasets, obtain the required preset number of valid vehicle pairs, construct a local coordinate system based on the target vehicle in each vehicle pair, and construct the mapping coordinates of the surrounding vehicles corresponding to the target vehicle. Formula (16) is used to construct the overall likelihood function based on maximizing the probability of the presence of surrounding vehicles as the first objective function. At the same time, based on the modification of the approach resistance model, the first objective function is transformed and modified, and the log-likelihood function is obtained using formula (17) as the stable objective function. (16) (17) in, Let be the overall likelihood function, which is also the first objective function. The multiplication symbol is used. For the preset quantity, Surrounding vehicles based on parameter vectors Relative to the target vehicle Approaching the resistance value, Let be the log-likelihood function, which is also the stable objective function. To correct the parameters, It is the natural logarithm function; The stable objective function is iterated by adjusting the parameter vector using a numerical optimization algorithm to obtain the maximum value of the stable objective function; Determine whether at least one of the following conditions is met: the maximum value of the current stable objective function tends to be stable, the parameter vector tends to converge, and the number of iterations reaches the preset maximum value. Under the given conditions, output the parameter vector and the current optimized approximate resistance model.

[0011] Preferably, the interaction index is determined and quantified using the optimized approach resistance model to generate a vehicle interaction characterization map, including: Obtain historical vehicle trajectory datasets and construct structured interactive analysis scenarios; Obtain the parameter vector of the approach resistance model based on each structured interaction analysis scenario, determine the approach resistance threshold and interaction space area corresponding to each structured interaction analysis scenario, and thus determine the interaction index; Based on the discrete points generated by the interaction indicators, curve fitting is performed to generate a vehicle interaction representation map.

[0012] Preferably, historical vehicle trajectory datasets are acquired, and structured interactive analysis scenarios are constructed, including: Obtain historical vehicle trajectory datasets, construct vehicle interaction datasets, and classify them by interaction type, identifying two interaction types: longitudinal interaction and lateral interaction. Based on the relative speed of the vehicles, they are continuously grouped into a preset number of interval groups, and the number of vehicles with effective interaction in each interval group is greater than or equal to a preset first effective interaction threshold. For each interaction type and each interval group, the corresponding speed-interaction type analysis scenario is obtained by combining them. For each analysis scenario, the corresponding vehicle pair data of effective interaction is added to form a structured interaction analysis scenario.

[0013] Preferably, the parameter vector of the approach resistance model based on each structured interaction analysis scenario is obtained, and the approach resistance threshold and interaction space area corresponding to each structured interaction analysis scenario are determined, thereby determining the interaction indicators, including: For each structured interaction analysis scenario, the parameter vector is optimized using the proximity resistance model to obtain the optimal parameter vector for each structured interaction analysis scenario. Based on the scale parameters of the proximity resistance model, a reasonable proximity resistance threshold is set. Formulas (18)-(19) are used to determine the interaction space area for each structured interaction analysis scenario based on the proximity resistance threshold and the optimal parameter vector for each scenario. (18) (19) in, For each structured interaction analysis scenario, the corresponding proximity resistance threshold is defined. for The critical distance in the positive axis direction is the critical distance to the driver's right side. for The critical distance in the negative direction of the axle is the critical distance to the driver's left. for The critical distance in the positive axis direction is the critical distance in front of the driver. for The critical clearance in the negative axle direction is the critical clearance behind the driver. The area of ​​the interaction space under effective interaction conditions; Using the interaction space area in each structured interaction analysis scenario, formula (20) is used to construct the quasi-density and quasi-flow interaction indicators for each structured interaction analysis scenario, and the structural state points for each structured interaction analysis scenario are obtained. (20) in, This refers to the total area of ​​interaction space equivalent to that occupied by vehicle interactions per unit road area at relative speed in a structured interaction analysis scenario. The effective interaction rate per unit time through a unit cross section at relative velocity. The relative speed for structured interactive analysis scenarios.

[0014] Preferably, a vehicle interaction representation map is generated by curve fitting based on discrete points generated from interaction metrics, including: Obtain the structural state points in all structured interaction analysis scenarios, and perform curve fitting on the discrete structural state points according to different interaction types; Based on the fitted curves, corresponding quasi-flow-relative velocity interaction plots and quasi-flow-quasi-density interaction plots are generated. Feature annotation is performed on the two interaction graphs to generate the final vehicle interaction representation graph.

[0015] A second aspect of the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method as described in any one of the preceding claims.

[0016] Through the above technical solutions, this method constructs a local coordinate system based on the relative speed direction between vehicles, realizing a standardized description of interactions in any direction. It also proposes a parameterized approach resistance model based on driver spatial perception. This model is based on a generalized Gaussian distribution and introduces directional parameters, which can accurately characterize the driver's asymmetric psychological perception of vehicle intrusion from different directions (front, back, left, and right). For the first time, it realizes a dynamic mapping from physical distance to psychological discomfort. By maximizing the likelihood function, it can learn the optimal model parameters for different interaction analysis scenarios from large-scale trajectory data, giving the model powerful data-driven and situation-adaptive capabilities. It defines two quantitative interaction indicators: quasi-density and quasi-flow. Through the drawn interaction representation map, it dynamically reveals the evolution law of traffic flow efficiency and state with relative speed, realizing accurate quantification of complex two-dimensional interactions and effective embedding of human perception. The generated representation map provides a direct basis for real-time diagnosis of traffic conditions, congestion warning, and proactive control, and can be extended to the anthropomorphic decision-making of autonomous driving, promoting the upgrade of traffic management from state monitoring to interactive inference. Attached Figure Description

[0017] Figure 1 This is a connection block diagram of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Figure 2 This is a connection block diagram for constructing a proximity resistance model according to an embodiment of the present invention, which is a vehicle interaction quantification method based on driver spatial perception and relative motion mapping. Figure 3 This is a connection block diagram illustrating the implementation of a vehicle interaction quantization method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention, which generates mapped coordinates through coordinate system transformation. Figure 4 This is a connection block diagram of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention, based on a spatial perception proximity resistance model. Figure 5This is a connection block diagram for optimizing the parameters of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Figure 6 This is a connection block diagram for generating a vehicle interaction representation diagram, which is an implementation of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Figure 7 This is a connection diagram of a structured interaction analysis scenario for a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Figure 8 This is a connection diagram for constructing interaction indicators of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Figure 9 This is a connection diagram of an interaction representation graph based on interaction indicators, which is a method for quantifying vehicle interaction based on driver spatial perception and relative motion mapping according to an embodiment of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0019] like Figure 1 The diagram shown is a connection block diagram of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention; Figure 1 In this quantification method, the following are included: In step S10, a proximity resistance model based on spatial perception is constructed; In step S11, historical vehicle trajectory datasets are collected, and the parameters of the approach resistance model are optimized. In step S12, the interaction index is determined and quantified using the optimized approach resistance model to generate a vehicle interaction characterization map.

[0020] Step S10 establishes an approach resistance model that quantifies the driver's psychological discomfort by using the relative positions between vehicles; Step S11 involves using historical vehicle trajectory data to identify effective interactive vehicle pairs and then using an optimization iterative method to determine the parameters in the model that best reflect real driving behavior. Step S12 generates interactive indicators based on the trained model and draws an interactive representation map that reveals the traffic flow status.

[0021] This method establishes a local coordinate system for the target vehicle based on relative motion through coordinate transformation, constructs a parameterized approach resistance model, quantifies the driver's discomfort at the intrusion of surrounding vehicles into their psychological space, optimizes model parameters using historical trajectory data, and finally calculates quantitative indicators of quasi-density and quasi-flow based on the optimized model, and draws an interaction representation map to dynamically represent traffic flow status. It breaks through the limitations of traditional one-dimensional traffic models, and for the first time quantifies and embeds driver psychological perception into interaction analysis, realizing refined modeling of complex two-dimensional interaction scenarios such as urban intersections, and providing innovative theoretical tools and dynamic analysis frameworks for intelligent traffic management and autonomous driving decision-making.

[0022] like Figure 2 The diagram shown is a connection block diagram for constructing a proximity resistance model according to an embodiment of the present invention, which is a method for quantifying vehicle interaction based on driver spatial perception and relative motion mapping; Figure 2 In order to ensure the basis for building a spatially-aware approach resistance model and to achieve accurate quantification of subsequent interaction indicators, in one embodiment of the present invention, building a spatially-aware approach resistance model may include the following steps: In step S20, vehicle-based mapped coordinates are constructed; In step S21, a spatially-aware proximity resistance model is constructed using the density function of the generalized Gaussian distribution.

[0023] Transforming surrounding vehicles from a global coordinate system to a local coordinate system with the target vehicle as its origin allows for the analysis of interactions in different directions. The approach resistance model is constructed using the density function of the generalized Gaussian distribution as its mathematical basis. This model is suitable for simulating the continuous transition process of psychological perception, closely aligns with the driver's spatial perception, and forms the foundation for the accurate quantification of interaction indicators.

[0024] like Figure 3 The diagram shown is a connection block diagram illustrating the implementation of a vehicle interaction quantization method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention, involving coordinate system transformation to generate mapped coordinates; Figure 3 In one embodiment of the present invention, constructing vehicle-based mapping coordinates may include the following steps: In step S30, the coordinates of the target vehicle and surrounding vehicles in the defined global coordinate system are obtained, and the relative velocity vectors of the target vehicle and surrounding vehicles are calculated using formulas (1)-(3). (1) (2) (3) in, For the target vehicle In the global coordinate system The following position coordinates, for surrounding vehicles In the global coordinate system The following position coordinates, For the target vehicle In the global coordinate system Below x Coordinates along the axis, For the target vehicle In the global coordinate system Below y Coordinates along the axis, for surrounding vehicles In the global coordinate system Below x Coordinates along the axis, for surrounding vehicles In the global coordinate system Below y Coordinates along the axis, For the target vehicle speed, For the target vehicle In the global coordinate system Below x Velocity components in the axial direction, For the target vehicle In the global coordinate system The velocity component along the y-axis. for surrounding vehicles speed, for surrounding vehicles In the global coordinate system Below x Velocity components in the axial direction, for surrounding vehicles In the global coordinate system The velocity component along the y-axis. It is a relative velocity vector; In step S31, a system is constructed based on the position coordinates of the target vehicle as the origin and... Direction is The local coordinate system in the positive direction of the axis is transformed from the global coordinate system to the local coordinate system using formula (4). (4) in, for surrounding vehicles Coordinates in a local coordinate system with the target vehicle's position coordinates as the origin. It is a two-dimensional rotation matrix. To transform and rotate the y-axis of the global coordinate system so that it aligns with... The rotation angle required for orientation alignment. for surrounding vehicles Coordinates in the global coordinate system The translation vector required to transform the origin of the coordinate system from the global coordinate system to the local coordinate system; In step S32, based on the target vehicle being the origin in the local coordinate system and the endpoints of the relative velocity vector being in the local coordinate system... In the positive direction of the axis, construct the corresponding parameters, and obtain the transformation parameters corresponding to the translation vector and rotation angle according to formula (5). (5) in, To move the origin of the global coordinate system along x The components of the translation vector that translates the axis to the origin of the local coordinate system. To move the origin of the global coordinate system along y The components of the translation vector that translates the axis to the origin of the local coordinate system; In step S33, the mapping coordinates of the target vehicle and surrounding vehicles are constructed using formulas (6)-(7). (6) (7) in, for surrounding vehicles In the local coordinate system x The coordinate values ​​of the axis are... , for surrounding vehicles In the local coordinate system y The coordinate values ​​of the axis are... , for surrounding vehicles Relative to the target vehicle The relative velocity scalar is... .

[0025] This complete and computable set of coordinate transformation equations can construct the mapped coordinates of surrounding and target vehicles for an effective interactive vehicle pair, encode the relative positions of surrounding and target vehicles, and determine whether the surrounding and target vehicles are moving away from or approaching each other based on the determined relative speed. The final output is... Triples serve as standardized inputs for subsequent model calculations, ensuring that all interactive analyses are performed within a unified motion reference frame.

[0026] like Figure 4The diagram shown is a connection block diagram of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping according to an embodiment of the present invention, based on a spatial perception proximity resistance model; Figure 4 In this context, considering the asymmetry in the driver's spatial perception of the surrounding directions, the basic symmetrical model needs to be extended to better reflect real driving psychology. In one embodiment of the invention, constructing a spatial perception-based approach resistance model using the density function of a generalized Gaussian distribution may include the following steps: In step S40, the mapped coordinates of the target vehicle and surrounding vehicles are obtained; In step S41, a symmetric approach resistance model based on a two-dimensional scenario is constructed using formulas (8)-(9) based on the density function of the generalized Gaussian distribution. (8) (9) in, To approach the resistance value, For parameter vectors, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters in the axial direction, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters along the axial direction; In step S42, based on the asymmetry of the driver's perception in the forward and backward directions and the lateral direction, the parameter vector of the symmetrical approach resistance model in the two-dimensional scene is extended using formulas (10)-(15), and the two-dimensional asymmetric approach resistance model is obtained as the final approach resistance model. (10) (11) (12) (13) (14) (15) in, For the final parameter vector, For symbolic functions, for The dimensional parameters in the positive axis direction are the dimensional parameters on the driver's right side. for The dimensional parameters in the negative axis direction are the same as the dimensional parameters on the driver's left side. for The dimensional parameters in the positive axis direction are the dimensional parameters in front of the driver. for The dimensional parameters in the negative axis direction are the dimensional parameters behind the driver. for The speed parameter in the positive direction of the axis is the speed parameter on the driver's right side. for The speed parameter in the negative direction of the axis is the speed parameter on the driver's left side. for The speed parameter in the positive direction of the axis is the speed parameter in front of the driver. for The speed parameter in the negative direction of the axis is the speed parameter behind the driver.

[0027] By introducing a basic two-dimensional symmetric model, which assumes that drivers have equal sensitivity to left / right and front / rear directions, but real driving psychology exhibits significant directional asymmetry (for example, drivers are usually more sensitive to oncoming vehicles than to those to the side and rear), the model expands each scale parameter in the basic model by introducing a sign function sgn() as a logical switch. This allows the model to dynamically select the corresponding directional parameter for calculation based on the sign of the input coordinates. This enables the model to accurately characterize the differentiated perception intensity of drivers regarding vehicle intrusions in different quadrants such as left front and right rear, greatly enhancing the model's realism, explanatory power, and adaptability to complex traffic scenarios.

[0028] like Figure 5 The diagram shown is a connection block diagram for optimizing the parameters of a vehicle interaction quantification method based on driver spatial perception and relative motion mapping, according to an embodiment of the present invention; Figure 5 In order to solve and optimize the parameter vector, in one embodiment of the present invention, collecting historical vehicle trajectory datasets and optimizing the parameters of the approach resistance model may include the following steps: In step S50, historical vehicle trajectory datasets are collected, a preset number of valid vehicle pairs with interaction are obtained, a local coordinate system based on the target vehicle in each vehicle pair is constructed, and the mapping coordinates of the surrounding vehicles corresponding to the target vehicle are constructed. In step S51, the overall likelihood function based on maximizing the probability of the presence of surrounding vehicles is constructed using formula (16) as the first objective function. At the same time, based on the modification of the approach resistance model, the first objective function is transformed and modified, and the log-likelihood function is obtained using formula (17) as the stable objective function. (16) (17) in, Let be the overall likelihood function, which is also the first objective function. The multiplication symbol is used. For the preset quantity, Surrounding vehicles based on parameter vectors Relative to the target vehicle Approaching the resistance value, Defined as close to tolerance, Let be the log-likelihood function, which is also the stable objective function. To correct the parameters, It is the natural logarithm function; In step S52, the parameter vector is adjusted using a numerical optimization algorithm to iterate over the stable objective function and obtain the maximum value of the stable objective function. In step S53, it is determined whether at least one of the following conditions is met: the maximum value of the current stable objective function tends to be stable, the parameter vector tends to converge, and the number of iterations reaches the preset maximum value. In step S54, if the conditions are met, the output parameter vector and the current optimized approach resistance model are output.

[0029] Based on the principle of maximum likelihood estimation, the optimal parameter vector should enable the proximity tolerance calculated by the model to explain the actual observed distribution of surrounding vehicle positions with the highest probability. To this end, a population likelihood function is constructed using all preset numbers of effective interactive vehicle pairs as samples. For numerical stability, its logarithmic form is taken as the stability objective function. To prevent parameter estimation bias caused by conflicts between the theoretical assumptions of the model and the actual situation where vehicles may be very close in the data, a small positive constant ε is introduced into the stability objective function for correction. During optimization, initial values ​​are set for the parameters (e.g., the scale parameter is taken as the nearest percentile and the rate parameter is taken as 2). Then, numerical optimization algorithms such as gradient descent are used to iteratively adjust the parameter vector to maximize the stability objective function. The optimization stops when the parameter vector converges, the maximum value of the stability objective function tends to stabilize, or the maximum number of iterations is reached. The learned optimal parameter vector and proximity resistance model are output.

[0030] like Figure 6 The diagram shown is a connection block diagram illustrating the generation of a vehicle interaction representation graph, based on a vehicle interaction quantification method according to an embodiment of the present invention, which utilizes driver spatial perception and relative motion mapping. Figure 6 In order to generate a vehicle interaction characterization map, in one embodiment of the present invention, the interaction index is determined and quantified using an optimized approach resistance model. Generating the vehicle interaction characterization map may include the following steps: In step S60, historical vehicle trajectory datasets are acquired, and a structured interactive analysis scenario is constructed; In step S61, the parameter vector of the approach resistance model based on each structured interaction analysis scenario is obtained, and the approach resistance threshold and interaction space area corresponding to each structured interaction analysis scenario are determined, thereby determining the interaction index. In step S62, curve fitting is performed on the discrete points generated by the interaction index to generate a vehicle interaction representation map.

[0031] The system systematically cleans, classifies, and groups massive amounts of historical trajectory data, organizing them into multiple structured interactive analysis scenarios with clear traffic semantics. Each interactive analysis scenario represents a specific interaction mode and relative speed range. For each constructed interactive analysis scenario, using its dedicated dataset and optimized model parameters, the system calculates corresponding macroscopic statistical indicators, quasi-density and quasi-flow, reflecting the interaction efficiency under that scenario. This aggregates microscopic interactive behaviors into macroscopic state points. The state points calculated from all analysis scenarios are categorized according to their interaction types, and through curve fitting and plotting, a visual and intuitive interactive representation map revealing traffic flow patterns is generated.

[0032] like Figure 7 The diagram shown is a connection block diagram for constructing a structured interaction analysis scenario based on a vehicle interaction quantification method according to an embodiment of the present invention; Figure 7 In order to classify interaction types, a structured interaction analysis scenario for statistical analysis is constructed. In one embodiment of the present invention, acquiring a historical vehicle trajectory dataset and constructing a structured interaction analysis scenario may include the following steps: In step S70, a historical vehicle trajectory dataset is obtained, a vehicle interaction dataset is constructed, and the interaction types are classified to determine two interaction types: longitudinal interaction and lateral interaction. In step S71, the vehicles are continuously grouped according to their relative speed, and a preset number of interval groups are divided. The number of vehicles with effective interaction in each interval group is greater than or equal to a preset first effective interaction threshold. In step S72, each interaction type and each interval group are combined to obtain the corresponding speed-interaction type analysis scenario, and the corresponding effective interaction vehicle pair data are added to each analysis scenario to form a structured interaction analysis scenario.

[0033] Based on the kinematic geometry of vehicle pairs, each concurrently existing vehicle pair interaction is categorized as either longitudinal interaction (primarily car-following behavior) or lateral interaction (primarily turning and weaving behavior). The relative speed between each vehicle pair is calculated, and all effective interacting vehicle pairs are divided into multiple consecutive intervals based on their relative speed values. To ensure the statistical reliability of subsequent parameter estimation, the number of effective interaction samples within each speed interval must reach a preset threshold (e.g., tens of thousands of pairs). To maintain speed resolution, the median difference between adjacent intervals must be greater than a minimum value. The cross-combination of each interaction type (longitudinal / lateral) and each speed interval forms an independent, physically meaningful, structured analysis scenario, containing sample data of all vehicle pairs within that scenario, laying the foundation for subsequent refined modeling of different scenarios.

[0034] like Figure 8 The diagram shown is a connection block diagram for constructing interaction indicators according to an embodiment of the present invention, which is a method for quantifying vehicle interaction based on driver spatial perception and relative motion mapping; Figure 8 In this invention, considering that a reasonable approach resistance value corresponding to the parameter vector can define an effective interaction boundary, facilitating the driver's spatial perception, it can be transformed into the interaction space area in real space, forming a quantifiable interaction index; in one embodiment of the invention, obtaining the parameter vector of the approach resistance model based on each structured interaction analysis scenario, determining the corresponding approach resistance threshold and interaction space area under each structured interaction analysis scenario, and thus determining the interaction index may include the following steps: In step S80, each structured interaction analysis scenario is obtained, and the parameter vector is optimized using the proximity resistance model to obtain the optimal parameter vector for each structured interaction analysis scenario. In step S81, a reasonable proximity resistance threshold is set based on the scale parameters of the proximity resistance model. In step S82, the interaction space area for each structured interaction analysis scenario is determined using formulas (18)-(19) based on the proximity resistance threshold and the optimal parameter vector for each structured interaction analysis scenario. (18) (19) in, For each structured interaction analysis scenario, the corresponding proximity resistance threshold is defined. for The critical distance in the positive axis direction is the critical distance to the driver's right side. for The critical distance in the negative direction of the axle is the critical distance to the driver's left. for The critical distance in the positive axis direction is the critical distance in front of the driver. for The critical clearance in the negative axle direction is the critical clearance behind the driver. The area of ​​the interaction space under effective interaction conditions; In step S83, the interaction space area under each structured interaction analysis scenario is used to construct the quasi-density and quasi-flow interaction indicators under each structured interaction analysis scenario using formula (20), and the structured state points under each structured interaction analysis scenario are obtained. (20) in, The quasi-density is the total area of ​​interaction space equivalent to that occupied by vehicle interactions per unit road area at relative speed in a structured interaction analysis scenario. The effective interaction rate per unit time through a unit cross-section at relative velocity is called the quasi-flow rate. The relative speed for structured interactive analysis scenarios.

[0035] For each structured interaction analysis scenario, an optimal parameter vector is used to set a reasonable approach resistance threshold, which is used to define a clear effective interaction boundary from continuous psychological perception. Using the optimal parameter vector and the approach resistance threshold, the critical distances corresponding to reaching this threshold in the left, right, front, and rear directions can be calculated. Considering that oncoming vehicles are the main influencing factor in vehicle interaction, the rear is ignored. The product of the front critical distance and the sum of the left and right critical distances is used to approximate the average interaction space area occupied by a single effective interaction on a two-dimensional plane. Based on this, the quasi-density is the total equivalent interaction space area occupied by vehicle interactions per unit road area under the relative speed of the structured interaction analysis scenario. The quasi-flow is defined as the product of the quasi-density and the relative speed of the structured interaction analysis scenario, representing the effective interaction rate passing through a unit cross-section per unit time under the relative speed, intuitively reflecting the interaction throughput through the intersection per unit time. Simultaneously, a structural state point is generated for each scenario.

[0036] like Figure 9 The diagram shown is a connection block diagram of an interaction representation graph based on interaction indicators, constructed according to an embodiment of the present invention, for a vehicle interaction quantification method based on driver spatial perception and relative motion mapping; Figure 9 In order to dynamically display the vehicle flow status at traffic intersections and facilitate subsequent traffic management, in one embodiment of the present invention, curve fitting based on discrete points generated by interaction indicators to generate a vehicle interaction representation map may include the following steps: In step S90, structural state points under all structured interaction analysis scenarios are obtained, and curve fitting of discrete structural state points is performed according to different interaction types. In step S91, the corresponding quasi-flow-relative velocity interaction diagram and quasi-flow-quasi-density interaction diagram are generated based on the fitted curve; In step S92, feature annotation is performed on the two interaction graphs to generate the final vehicle interaction representation graph.

[0037] Discrete structural state points in all structured interaction analysis scenarios are obtained, and curve fitting is performed. All state points belonging to the same interaction type (vertical or horizontal) are used as datasets to fit continuous relationship curves of quasi-flow-quasi-density and quasi-flow-relative velocity. Based on the continuous relationship curves, an iFD interaction map is constructed. Key feature points need to be identified, especially the maximum quasi-flow when the flow reaches its peak and its corresponding critical quasi-density. Based on the feature points, the iFD map is divided into three typical state regions: free flow (low quasi-density, quasi-flow increases with quasi-density), saturated flow (based on critical quasi-density, quasi-flow approaches the maximum quasi-flow), and congested flow (high quasi-density, flow decreases with quasi-density). By comparing the curves of vertical and horizontal interactions, the spatial utilization efficiency of different interaction modes can be quantitatively analyzed.

[0038] A second aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of the preceding claims. Through the above technical solutions, this method constructs a local coordinate system based on the relative speed direction between vehicles, realizing a standardized description of interactions in any direction. It also proposes a parameterized approach resistance model based on driver spatial perception. This model is based on a generalized Gaussian distribution and introduces directional parameters, which can accurately characterize the driver's asymmetric psychological perception of vehicle intrusion from different directions (front, back, left, and right). For the first time, it realizes a dynamic mapping from physical distance to psychological discomfort. By maximizing the likelihood function, it can learn the optimal model parameters for different interaction analysis scenarios from large-scale trajectory data, giving the model powerful data-driven and situation-adaptive capabilities. It defines two quantitative interaction indicators: quasi-density and quasi-flow. Through the drawn interaction representation map, it dynamically reveals the evolution law of traffic flow efficiency and state with relative speed, realizing accurate quantification of complex two-dimensional interactions and effective embedding of human perception. The generated representation map provides a direct basis for real-time diagnosis of traffic conditions, congestion warning, and proactive control, and can be extended to the anthropomorphic decision-making of autonomous driving, promoting the upgrade of traffic management from state monitoring to interactive inference.

[0039] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solutions of the present invention, and these simple modifications all fall within the protection scope of the present invention. Furthermore, it should be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present invention will not describe the various possible combinations separately.

Claims

1. A method for quantifying vehicle interaction based on driver spatial perception and relative motion mapping, characterized in that, The method includes: Construct a proximity resistance model based on spatial perception; Collect historical vehicle trajectory datasets and optimize the parameters of the approach resistance model; The interaction index is determined and quantified using the optimized approach resistance model to generate a vehicle interaction characterization map.

2. The method according to claim 1, characterized in that, Constructing a spatially-aware approach resistance model, including: Construct vehicle-based mapping coordinates; A spatially-aware proximity resistance model is constructed using the density function of the generalized Gaussian distribution.

3. The method according to claim 2, characterized in that, Constructing vehicle-based mapping coordinates includes: Obtain the coordinates of the target vehicle and surrounding vehicles in the defined global coordinate system, and calculate the relative velocity vectors of the target vehicle and surrounding vehicles using formulas (1)-(3). ，（1） ，（2） ，（3） in, For the target vehicle In the global coordinate system The following position coordinates, for surrounding vehicles In the global coordinate system The following position coordinates, For the target vehicle In the global coordinate system Below x Coordinates along the axis, For the target vehicle In the global coordinate system Below y Coordinates along the axis, for surrounding vehicles In the global coordinate system Below x Coordinates along the axis, for surrounding vehicles In the global coordinate system Below y Coordinates along the axis, For the target vehicle speed, For the target vehicle In the global coordinate system Below x Velocity components in the axial direction, For the target vehicle In the global coordinate system The velocity component along the y-axis. for surrounding vehicles speed, for surrounding vehicles In the global coordinate system Below x Velocity components in the axial direction, for surrounding vehicles In the global coordinate system The velocity component along the y-axis. It is a relative velocity vector; Construct a system based on the target vehicle's position coordinates as the origin and... Direction is The local coordinate system in the positive direction of the axis is transformed from the global coordinate system to the local coordinate system using formula (4). ，（4） in, for surrounding vehicles Coordinates in a local coordinate system with the target vehicle's position coordinates as the origin. It is a two-dimensional rotation matrix. To transform and rotate the y-axis of the global coordinate system so that it aligns with... The rotation angle required for orientation alignment. for surrounding vehicles Coordinates in the global coordinate system The translation vector required to transform the origin of the coordinate system from the global coordinate system to the local coordinate system; Based on the target vehicle being the origin in the local coordinate system and the endpoints of the relative velocity vector being in the local coordinate system... In the positive direction of the axis, construct the corresponding parameters, and obtain the transformation parameters corresponding to the translation vector and rotation angle according to formula (5). ，（5） in, To move the origin of the global coordinate system along x The components of the translation vector that translates the axis to the origin of the local coordinate system. To move the origin of the global coordinate system along y The components of the translation vector that translates the axis to the origin of the local coordinate system; The target vehicle and the surrounding vehicles are mapped using formulas (6)-(7). ，（6） ，（7） in, for surrounding vehicles In the local coordinate system x The coordinate values ​​of the axis are... , for surrounding vehicles In the local coordinate system y The coordinate values ​​of the axis are... , for surrounding vehicles Relative to the target vehicle The relative velocity scalar is... .

4. The method according to claim 3, characterized in that, A spatially-aware proximity resistance model is constructed using the density function of the generalized Gaussian distribution, including: Obtain the mapped coordinates of the target vehicle and surrounding vehicles; Based on the density function of the generalized Gaussian distribution, a symmetric approach resistance model in a two-dimensional scenario is constructed using formulas (8)-(9). ，（8） ，（9） in, To approach the resistance value, For parameter vectors, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters in the axial direction, In local coordinate system Rate parameters in the axial direction, In local coordinate system Scale parameters along the axial direction; Based on the asymmetry of the driver's perception in the front-back and lateral directions, the parameter vectors of the symmetrical approach resistance model in the two-dimensional scenario are extended using formulas (10)-(15), and the two-dimensional asymmetric approach resistance model is obtained as the final approach resistance model. ，（10） ，（11） ，（12） ，（13） ，（14） ，（15） in, For the final parameter vector, For symbolic functions, for The dimensional parameters in the positive axis direction are the dimensional parameters on the driver's right side. for The dimensional parameters in the negative axis direction are the same as the dimensional parameters on the driver's left side. for The dimensional parameters in the positive axis direction are the dimensional parameters in front of the driver. for The dimensional parameters in the negative axis direction are the dimensional parameters behind the driver. for The speed parameter in the positive direction of the axis is the speed parameter on the driver's right side. for The speed parameter in the negative direction of the axis is the speed parameter on the driver's left side. for The speed parameter in the positive direction of the axis is the speed parameter in front of the driver. for The speed parameter in the negative direction of the axis is the speed parameter behind the driver.

5. The method according to claim 3, characterized in that, Collect historical vehicle trajectory datasets and optimize the parameters of the approach resistance model, including: Collect historical vehicle trajectory datasets, obtain the required preset number of valid vehicle pairs, construct a local coordinate system based on the target vehicle in each vehicle pair, and construct the mapping coordinates of the surrounding vehicles corresponding to the target vehicle. Formula (16) is used to construct the overall likelihood function based on maximizing the probability of the presence of surrounding vehicles as the first objective function. At the same time, based on the modification of the approach resistance model, the first objective function is transformed and modified, and the log-likelihood function is obtained using formula (17) as the stable objective function. ，（16） ，（17） in, Let be the overall likelihood function, which is also the first objective function. The multiplication symbol is used. For the preset quantity, Surrounding vehicles based on parameter vectors Relative to the target vehicle Approaching the resistance value, Let be the log-likelihood function, which is also the stable objective function. To correct the parameters, It is the natural logarithm function; The stable objective function is iterated by adjusting the parameter vector using a numerical optimization algorithm to obtain the maximum value of the stable objective function; Determine whether at least one of the following conditions is met: the maximum value of the current stable objective function tends to be stable, the parameter vector tends to converge, and the number of iterations reaches the preset maximum value. Under the given conditions, output the parameter vector and the current optimized approximate resistance model.

6. The method according to claim 5, characterized in that, The interaction index is determined and quantified using the optimized approach resistance model, generating a vehicle interaction characterization map, including: Obtain historical vehicle trajectory datasets and construct structured interactive analysis scenarios; Obtain the parameter vector of the approach resistance model based on each structured interaction analysis scenario, determine the approach resistance threshold and interaction space area corresponding to each structured interaction analysis scenario, and thus determine the interaction index; Based on the discrete points generated by the interaction indicators, curve fitting is performed to generate a vehicle interaction representation map.

7. The method according to claim 6, characterized in that, Obtain historical vehicle trajectory datasets and construct structured interactive analysis scenarios, including: Obtain historical vehicle trajectory datasets, construct vehicle interaction datasets, and classify them by interaction type, identifying two interaction types: longitudinal interaction and lateral interaction. Based on the relative speed of the vehicles, they are continuously grouped into a preset number of interval groups, and the number of vehicles with effective interaction in each interval group is greater than or equal to a preset first effective interaction threshold. For each interaction type and each interval group, the corresponding speed-interaction type analysis scenario is obtained by combining them. For each analysis scenario, the corresponding vehicle pair data of effective interaction is added to form a structured interaction analysis scenario.

8. The method according to claim 6, characterized in that, Obtain the parameter vector of the approach resistance model based on each structured interaction analysis scenario, determine the corresponding approach resistance threshold and interaction space area for each structured interaction analysis scenario, and thus determine the interaction indicators, including: For each structured interaction analysis scenario, the parameter vector is optimized using the proximity resistance model to obtain the optimal parameter vector for each structured interaction analysis scenario. Based on the scale parameters of the proximity resistance model, a reasonable proximity resistance threshold is set. Formulas (18)-(19) are used to determine the interaction space area for each structured interaction analysis scenario based on the proximity resistance threshold and the optimal parameter vector for each scenario. ，（18） ，（19） in, For each structured interaction analysis scenario, the corresponding proximity resistance threshold is defined. for The critical distance in the positive axis direction is the critical distance to the driver's right side. for The critical distance in the negative direction of the axle is the critical distance to the driver's left. for The critical distance in the positive axis direction is the critical distance in front of the driver. for The critical clearance in the negative axle direction is the critical clearance behind the driver. The area of ​​the interaction space under effective interaction conditions; Using the interaction space area in each structured interaction analysis scenario, formula (20) is used to construct the quasi-density and quasi-flow interaction indicators for each structured interaction analysis scenario, and the structural state points for each structured interaction analysis scenario are obtained. ，（20） in, This refers to the total area of ​​interaction space equivalent to that occupied by vehicle interactions per unit road area at relative speed in a structured interaction analysis scenario. The effective interaction rate per unit time through a unit cross section at relative velocity. The relative speed for structured interactive analysis scenarios.

9. The method according to claim 6, characterized in that, Based on the discrete points generated by the interaction metrics, curve fitting is performed to generate a vehicle interaction representation map, including: Obtain the structural state points in all structured interaction analysis scenarios, and perform curve fitting on the discrete structural state points according to different interaction types; Based on the fitted curves, corresponding quasi-flow-relative velocity interaction plots and quasi-flow-quasi-density interaction plots are generated. Feature annotation is performed on the two interaction graphs to generate the final vehicle interaction representation graph.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 9.

Images