A method and system for intelligent traffic vehicle behavior modeling based on collaborative learning

Abstract

The present invention provides a method and system for modeling intelligent traffic vehicle behavior based on collaborative learning, which relates to the field of unmanned driving technology, including: obtaining traffic data in a target area; performing data calibration processing based on the traffic data to obtain a collaborative perception data set; generating a vehicle spatiotemporal trajectory function based on the collaborative perception data set to obtain a conflict domain spatiotemporal distribution map; extracting the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution map to obtain a multi-dimensional interaction feature vector; performing behavior modeling based on the multi-dimensional interaction feature vector to obtain a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles. The present invention effectively solves the problems of single-vehicle sensor occlusion and cross-vehicle data heterogeneity through the calibration of multi-vehicle collaborative perception data and the dynamic fusion of road geometric constraints, significantly improving the global consistency of vehicle position and motion state estimation.

Description

Technical Field

[0001] The present invention relates to the field of unmanned driving technology, and in particular to a method and system for intelligent traffic vehicle behavior modeling based on collaborative learning. Background Art

[0002] In the field of autonomous driving and intelligent transportation, existing technologies are mostly based on single-vehicle perception systems (such as lidar and cameras) and independent decision-making frameworks, making them difficult to handle the complex interactions of multiple vehicles in intersecting paths. Traditional approaches face significant bottlenecks, particularly in dynamic environments such as unsignaled intersections. Single-vehicle perception is susceptible to occlusion and sensor noise, leading to cumulative trajectory prediction errors and threatening the safety of autonomous driving systems. Intelligent traffic management solutions based on fixed rules or centralized scheduling lack the ability to dynamically model multi-vehicle interactions, making it difficult to generate globally optimal strategies in real time. Existing technologies attempt to achieve vehicle-road collaboration through V2X communication, such as by employing preset priority levels or time slot allocation mechanisms. However, these approaches still have drawbacks: First, they rely heavily on high-precision infrastructure and cannot adapt to dynamic traffic flows; second, they lack collaborative learning mechanisms between vehicles, resulting in low conflict resolution success rates and reduced traffic efficiency. Furthermore, existing approaches often rely on isolated learning models, failing to fully exploit the potential correlations in multi-vehicle interaction data. This significantly reduces the robustness of behavior prediction in the face of communication delays or perception failures of some vehicles.

[0003] Based on the above shortcomings of the existing technology, there is an urgent need for an intelligent traffic vehicle behavior modeling method and system based on collaborative learning. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for modeling intelligent traffic vehicle behavior based on collaborative learning to improve the above-mentioned problems. To achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for modeling intelligent traffic vehicle behavior based on collaborative learning, comprising:

[0006] Acquire traffic data within the target area, including real-time status data of the current vehicle, road structure data, and neighboring vehicle interaction data;

[0007] Performing data calibration processing on the traffic data and eliminating abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative perception data set;

[0008] Generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation;

[0009] Based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles are extracted by quantifying the temporal behavior pattern association and the interactive game relationship, thereby obtaining a multi-dimensional interactive feature vector;

[0010] Behavior modeling is performed based on the multi-dimensional interaction feature vector, and a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles is obtained by constructing a layered graph network model based on node dynamic feature encoding and edge relationship weight update.

[0011] In a second aspect, the present application also provides an intelligent traffic vehicle behavior modeling system based on collaborative learning, comprising:

[0012] An acquisition module is used to acquire traffic data within the target area, wherein the traffic data includes real-time status data of the current vehicle, road structure data, and neighboring vehicle interaction data;

[0013] a calibration module, configured to perform data calibration processing based on the traffic data and eliminate abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative perception data set;

[0014] A prediction module, configured to generate a spatiotemporal trajectory function of a vehicle based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation;

[0015] An extraction module is configured to extract the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, by associating temporal behavior patterns and quantifying interactive game relationships, to obtain a multi-dimensional interactive feature vector;

[0016] A modeling module is used to perform behavior modeling based on the multi-dimensional interaction feature vector, and obtain a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles by constructing a hierarchical graph network model based on node dynamic feature encoding and edge relationship weight update.

[0017] The beneficial effects of the present invention are:

[0018] The present invention effectively solves the problems of single-vehicle sensor occlusion and cross-vehicle data heterogeneity through multi-vehicle collaborative perception data calibration and utilizes dynamic fusion of road geometric constraints, significantly improving the global consistency of vehicle position and motion state estimation; through interactive feature extraction and hierarchical graph network optimization, the driving intention probability, game relationship matrix and traffic rules are jointly encoded into node and edge attributes, realizing the joint optimization of single-vehicle strategy and multi-vehicle collaboration, and improving the success rate of conflict resolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flowchart of an intelligent traffic vehicle behavior modeling method based on collaborative learning according to an embodiment of the present invention;

[0021] Figure 2 Schematic diagram of the structure of an intelligent transportation vehicle behavior modeling system based on collaborative learning according to an embodiment of the present invention;

[0022] Figure 3 Schematic diagram of the structure of an intelligent traffic vehicle behavior modeling device based on collaborative learning described in an embodiment of the present invention.

[0023] Markings in the figure: 800, an intelligent traffic vehicle behavior modeling device based on collaborative learning; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, calibration module; 903, prediction module; 904, extraction module; 905, modeling module. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0025] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0026] Example

[0027] This embodiment provides an intelligent transportation vehicle behavior modeling method based on collaborative learning.

[0028] See also Figure 1 , the figure shows that the method includes steps S100 to S500.

[0029] Step S100: Acquire traffic data within the target area, where the traffic data includes real-time status data of the current vehicle, road structure data, and neighboring vehicle interaction data;

[0030] This step utilizes a multi-source, heterogeneous data synchronization mechanism to acquire three types of data within the target area: real-time vehicle status data, including position, speed, acceleration, and heading angle collected by onboard sensors, as well as vehicle identification and communication status acquired through V2X communication; road structure data, including intersection lane topology, lane curvature, and traffic rule constraints analyzed through high-precision maps; and neighboring vehicle interaction data, including historical behavior sequences, intent labels, and interaction event records of surrounding vehicles acquired through V2V communication. In complex scenarios such as unsignaled intersections, single-vehicle sensors can easily experience blind spots due to occlusion or environmental interference (such as rain, fog, and backlight). While V2X / V2V communication can supplement neighboring vehicle data, it suffers from heterogeneous issues such as asynchronous timestamps and inconsistent coordinate systems. This step fuses multimodal data to provide global input for subsequent collaborative perception and modeling.

[0031] Step S200: performing data calibration processing based on traffic data and eliminating abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative sensing data set;

[0032] This step addresses the spatiotemporal heterogeneity of multi-source data through a four-layer progressive calibration process: First, by combining the clock deviation observations of multiple vehicles with the periodic clock calibration signals from V2V communications, timestamp offsets are dynamically corrected to eliminate temporal asynchrony in cross-vehicle data. Subsequently, the positions of multiple vehicles are mapped to the global coordinate system defined by the high-precision map, and spatial benchmark consistency is verified by combining lane topology connectivity (such as steering logic relationships). Furthermore, a vehicle observation topology map is constructed based on the relative positions and motion trends of neighboring vehicles, predicting the acceleration and heading angle of obscured vehicles and filling sensor blind spots. Finally, a quadratic programming algorithm is used to force vehicle positions to within the lane boundaries of the high-precision map, eliminating abnormal drift caused by sensor noise or communication delays. In unsignaled intersection scenarios, this calibration mechanism specifically addresses temporal asynchrony, spatial misalignment, and occlusion, outputting a collaborative perception dataset that is spatiotemporally aligned and conforms to lane geometry. This provides high-precision input for trajectory modeling, ensures the accuracy of subsequent collision domain risk quantification, and prevents the generation of illegal trajectories.

[0033] Step S300: Generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation;

[0034] This step utilizes dynamic trajectory modeling and conflict risk quantification mechanisms to generate a continuous spatiotemporal trajectory function for vehicles based on a calibrated collaborative perception dataset. Combining the real-time motion states (speed and acceleration) of the vehicles and sensor performance parameters (such as detection error and system response delay), it predicts the spatiotemporal intersection of multiple vehicle trajectories at the intersection and their time window confidence intervals. It also dynamically calculates an adaptive safety radius based on the maximum vehicle deceleration and environmental conditions (such as the road adhesion coefficient) to quantify the risk level of the conflict domain. In unsignaled intersection scenarios, traditional fixed safety distance models are unable to cope with sudden lane changes or inclement weather. However, this step uses dynamic safety radius adjustments (such as increasing redundancy in rainy and foggy weather) and time window overlap probability analysis to accurately identify high-risk conflict areas and their timeliness. It then outputs a spatiotemporal distribution map of the conflict domain, including location, time, and risk intensity. This provides a quantitative basis for subsequent interactive game analysis, significantly improving the reliability and real-time performance of multi-vehicle collaborative decision-making in complex scenarios.

[0035] Step S400: Based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles are extracted by associating temporal behavior patterns with the quantification of interactive game relationships to obtain a multi-dimensional interactive feature vector.

[0036] This step combines driving intention inference with multi-vehicle game relationship modeling. Based on the historical motion sequences in the vehicle's spatiotemporal trajectory function and the risk level of the conflict domain's spatiotemporal distribution map, it uses a long-short-term memory network to analyze temporal behavior patterns. Combined with the steering logic constraints of the intersection lane topology, it outputs a probability distribution of driving intentions. Simultaneously, based on the overlap of conflict time windows, relative speeds, and historical interaction data, it constructs a competition and cooperation relationship matrix to quantify vehicle strategy benefits. Through feature concatenation and normalization, it integrates intention probabilities, game relationships, and vehicle identity labels into a multidimensional interaction feature vector. In dynamic game scenarios, traditional rule engines struggle to cope with the uncertainty of neighboring vehicles' intentions. However, this step, through a data-driven feature extraction mechanism, explicitly encodes implicit interaction relationships into structured features, supporting layered graph network collaborative optimization and achieving intention-interpretable and strategy-adaptive behavior modeling in multi-vehicle conflict scenarios.

[0037] Step S500 : Behavior modeling is performed based on the multi-dimensional interaction feature vector. By constructing a layered graph network model based on node dynamic feature coding and edge relationship weight update, a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles is obtained.

[0038] This step uses a layered graph network architecture and a multi-agent collaborative optimization mechanism to map multi-dimensional interaction feature vectors into a spatiotemporal graph structure, where nodes represent the dynamic state of vehicles and edge weights quantify the intensity of interactions between vehicles. Based on the encoding of node dynamic features and iterative updates of edge relationship weights, a multi-agent reinforcement learning framework is used to jointly optimize single-vehicle strategies and multi-vehicle collaborative relationships, using conflict resolution success rate and traffic efficiency as global reward functions. At the same time, illegal strategy action spaces are pruned using traffic rule constraints. This model breaks through the limitations of traditional centralized scheduling or independent decision-making. Through the divide-and-conquer optimization of a layered graph network (nodes focus on single-vehicle behavior generation, while edges focus on multi-vehicle interaction and coordination), it directly supports highly robust collaborative decision-making of autonomous driving systems in complex scenarios.

[0039] Furthermore, step S200 includes steps S210 to S240.

[0040] Step S210: Perform time synchronization processing based on the traffic data, correct the timestamp offset between vehicles by combining the clock deviation observation data of multiple vehicles, and use the periodic clock calibration signal communicated between vehicles as the time reference to obtain a multi-vehicle status data set;

[0041] This step addresses the temporal asynchrony of multi-vehicle data by establishing a unified time base through periodic inter-vehicle clock calibration signals (such as the time synchronization messages embedded in V2V communications). This process combines the clock deviation observations from multiple vehicles to dynamically correct the timestamp offsets in each vehicle's sensor data. The specific process includes: periodically broadcasting calibration signals to align all vehicles' local times with the clock of a reference node (such as the host vehicle or roadside unit); calculating timestamp offsets based on the combined observations of multi-vehicle clock deviations; and performing time axis resampling of the original state data to eliminate time misalignment across vehicle data.

[0042] Step S220: performing unified spatial coordinate processing on the multi-vehicle state data set, mapping the multi-vehicle position data to a global map coordinate system, and obtaining a vehicle position data set;

[0043] This step uses the global coordinate system defined by the high-precision map to uniformly convert the position data collected by each vehicle based on its own local coordinate system (such as the center of the vehicle body as the origin) into global coordinates under the map coordinate system. In practice, due to the different driving directions of multiple vehicles, local coordinate system deviations may occur. Directly using the original position data will cause trajectory intersection prediction errors. Through coordinate transformation and lane topology logic verification, all vehicle position data are ensured to be aligned under a unified spatial reference, eliminating relative position calculation errors caused by coordinate system differences.

[0044] Step S230: performing occlusion compensation processing based on the vehicle position dataset, constructing a vehicle observation topology map and predicting the acceleration and heading angle of the occluded vehicle based on the relative positions and motion trends of neighboring vehicles to obtain a vehicle status dataset;

[0045] This step addresses the issue of missing vehicle states due to sensor occlusion. By constructing a vehicle observation topology graph (with nodes representing vehicles and edges representing the relative positions and motion trends between vehicles), the acceleration and heading angle of the obscured vehicle are inferred using the real-time positions, speeds, and historical behavior sequences of neighboring vehicles. By extrapolating motion trends and correlating interaction histories, the dynamic state of the obscured vehicle is predicted, ensuring the integrity of multi-vehicle data.

[0046] Step S240: Perform road geometry constraint correction processing based on the vehicle state dataset, and use a quadratic programming algorithm to constrain the vehicle position to the lane boundary defined by the map to obtain a collaborative perception dataset.

[0047] This step addresses the issue of abnormal vehicle position drift caused by sensor noise or communication delays. Using the lane boundaries defined by the high-precision map as hard constraints, a quadratic programming optimization objective is constructed: the vehicle position is forced to be within the lane boundaries while minimizing the deviation between the original and corrected vehicle positions. For curved lanes (such as roundabouts), the lane boundary constraints are converted into lateral displacement limits in the Frenet coordinate system, and the correction range is dynamically adjusted to accommodate changes in lane geometry. Traditional rigid boundary corrections can easily lead to trajectory distortion (e.g., a vehicle in a curved lane being incorrectly corrected to an adjacent lane). This step, through dynamic coordinate system conversion and smoothing optimization, ensures that the corrected vehicle position both conforms to physical rules and maintains motion continuity.

[0048] Furthermore, step S300 includes steps S310 to S340.

[0049] Step S310: Generate a vehicle spatiotemporal trajectory function based on the collaborative sensing dataset and combining the vehicle's historical position and real-time motion status;

[0050] Specifically, data preprocessing is first performed. In the data preprocessing stage, single-point data errors are eliminated by spatial weighted averaging. The weight calculation uses the Gaussian kernel function, and the formula is: ;

[0051] ; ;

[0052] in, Indicates the The original coordinates of the cooperative vehicles; Indicates the serial number of the collaborative sensing vehicle; represents the number of collaborative sensing vehicles; , represents the coordinates of the vehicle; represents the spatial correlation attenuation coefficient; Indicates the communication radius; Represents the estimated vehicle position after spatiotemporal fusion; Indicates the The spatial credibility weight of the coordinated vehicles; represents the Hadamard product.

[0053] Then, a prediction model is established to integrate historical trends and real-time data. The formula is: ; ;

[0054] in, express The predicted space-time coordinates of the moment; represents the time-varying fusion acceleration; represents the position vector calibrated by collaborative data; represents the fused velocity vector; represents the historical average acceleration; Indicates real-time acceleration measurement value; Indicates the weight of historical data; Indicates the time sequence number.

[0055] The final space-time trajectory function is expressed as: ;in, represents the space-time trajectory function; represents the prediction time window; represents the confidence interval; Indicates the time synchronization reference point.

[0056] Step S320: performing trajectory intersection prediction processing based on the vehicle space-time trajectory function, and calculating the space-time intersection coordinates of multiple vehicle trajectories and the confidence intervals of the intersection time windows to obtain a space-time distribution table of trajectory intersections;

[0057] It can be understood that this step integrates the predicted driving routes of each vehicle and uses a spatiotemporal overlay analysis method to identify the intersection areas of different vehicle trajectories in time and space dimensions. First, based on the motion trend prediction model of each vehicle, the sequence of location points that the vehicle may pass through in the future time period is calculated. Then, a cross-validation algorithm is used to find the intersection positions and corresponding time nodes between different vehicle paths. For each predicted intersection point, combined with factors such as the positioning accuracy and communication delay of each vehicle, the time fluctuation range and spatial position deviation range of the intersection event are calculated, and finally these prediction results are integrated into a structured data table. This table records in detail the most likely occurrence time, location coordinates and corresponding credibility assessment indicators of each potential conflict point.

[0058] Step S330: Conflict time window prediction processing is performed based on the spatiotemporal distribution table of trajectory intersections. The high conflict risk time window is obtained by calculating the expected time distribution of vehicles arriving at the intersection and the probability of overlap of the conflict time windows.

[0059] First, a Gaussian distribution model is constructed for each vehicle's estimated arrival time. The mean is calculated from the trajectory function, and the standard deviation integrates sensor noise (such as GPS error) and vehicle dynamic fluctuations (such as acceleration and deceleration randomness) to quantify temporal uncertainty. Subsequently, the overlap probability of multiple vehicle arrival time distributions is calculated. For two vehicles, the collision risk is determined by integrating the joint probability density function of their time differences within a safety threshold. For multi-vehicle scenarios, a Monte Carlo sampling method is used to simulate a large number of possible arrival time combinations and calculate the frequency of intersections of all vehicle time windows to efficiently estimate the collision probability. Furthermore, an environmental visibility factor is introduced to dynamically adjust the safety time threshold, such as extending the base threshold in rainy and foggy weather, enhancing the model's adaptability to complex conditions. Finally, high-risk time windows and their probability levels are output, providing a quantitative basis for subsequent behavioral modeling, ensuring that the autonomous driving system prioritizes urgent conflicts, ensuring safety while optimizing traffic efficiency. By integrating probability theory with dynamic environmental parameters, this step addresses the real-time and adaptability limitations of traditional fixed-threshold methods, significantly improving the accuracy of collision prediction in complex traffic scenarios.

[0060] Step S340: Dynamically calculate and process the safety radius according to the high-collision risk time window, dynamically adjust the safety radius threshold based on the vehicle's maximum deceleration, system response delay, and sensor detection error parameters, and obtain a spatiotemporal distribution map of the collision domain.

[0061] It should be noted that this step calculates the baseline safety distance based on the vehicle's maximum deceleration, system response delay (including sensor data processing and decision-making time), and real-time vehicle speed. This calculation covers the spatial margin required for the vehicle to come to a complete stop from sensing the vehicle. Secondly, sensor detection error parameters (such as LiDAR point cloud density decay and camera resolution degradation in low light) and environmental visibility factors (such as rain and fog) are introduced to dynamically expand the safety radius threshold to ensure increased redundancy in adverse conditions. Furthermore, the safety radius is associated with the conflict time window, and multiple vehicle safety zones are superimposed along the spatiotemporal distribution of vehicle trajectories. Overlapping areas are identified as high-conflict domains. Multi-dimensional risk quantification is performed based on the overlapping area, time window urgency (such as the remaining time percentage), and environmental risk level (such as the road slipperiness coefficient), forming a spatiotemporal distribution map of conflict domains that includes location, time, and risk intensity.

[0062] Furthermore, step S400 includes steps S410 to S440.

[0063] Step S410: Perform driving intention classification processing based on the vehicle's spatiotemporal trajectory function, using the acceleration change rate and heading angle deviation characteristics of the vehicle's historical trajectory and the steering logic constraints of the intersection lane topology as input values ​​of the long-short-term memory network to output the probability distribution characteristics of the driving intention;

[0064] Step S420: Modeling the interaction relationship based on the spatial-temporal distribution diagram of the conflict domains, and constructing a competition-cooperation relationship matrix by calculating the spatial-temporal proximity of the conflict domains between vehicles.

[0065] Step S430: Calculate the game strategy benefits based on the competition and cooperation relationship matrix, quantify the benefits of each vehicle adopting the acceleration, deceleration, and yielding strategies in the conflict domain, and generate a strategy benefit distribution table;

[0066] Step S440: Perform feature fusion processing based on the probability distribution characteristics and the strategy return distribution table to obtain a multi-dimensional interactive feature vector.

[0067] As can be understood, first, in step S410, based on the acceleration rate and heading angle deviation characteristics of the vehicle's historical trajectory, combined with the steering logic constraints of the intersection lane topology (e.g., no straight driving in the left-turn lane), a long-short-term memory network is used to analyze temporal behavior patterns and output a probability distribution of driving intentions (e.g., going straight, turning left, slowing down and yielding). Lane rules are also used to enforce exclusion of illegal intentions, ensuring that the prediction results comply with traffic regulations. Subsequently, in step S420, based on the spatiotemporal distribution map of the conflict domain, the spatiotemporal proximity between vehicles (e.g., time difference and relative distance to the conflict domain) is quantified. Combined with historical interaction data (e.g., frequency of yielding and following habits between vehicles), a competition and cooperation relationship matrix is ​​constructed to characterize the vehicles' game tendencies in the conflict domain (e.g., aggressive competition or conservative yielding). Furthermore, in step S430, the benefits of each vehicle's acceleration, deceleration, and yielding strategies are calculated based on a non-cooperative game model. The remaining time window of the conflict domain is introduced as a benefit decay factor to encourage vehicles to make decisions early while eliminating physically infeasible strategies (e.g., aggressive actions exceeding maximum acceleration). A strategy benefit distribution table is generated. Ultimately, through feature concatenation and normalization, the intent probability distribution, the game relationship matrix, and vehicle identity tags (such as emergency vehicle and truck priority) are integrated into a multidimensional interactive feature vector. Feature dimensionality reduction is then performed based on the spatial and temporal extent of the conflict domain to eliminate redundant information. This process transcends the static limitations of traditional rule-based engines. Through data-driven dynamic feature extraction and game quantification, it enables intention-interpretable and strategy-adaptive multi-vehicle collaborative modeling in complex scenarios. This provides high-information input for the subsequent hierarchical graph network, significantly improving intersection conflict resolution success rates and traffic efficiency.

[0068] Furthermore, step S500 includes steps S510 to S540.

[0069] Step S510: constructing a spatiotemporal graph structure based on the multidimensional interactive feature vector, converting the vehicle identity labels, intention probability distribution, and conflict domain associations into node attributes and edge connection relationships of the spatiotemporal graph structure, and combining the connectivity constraints of the intersection lane topology to obtain a hierarchical spatiotemporal graph structure;

[0070] As can be understood, this step maps multidimensional interaction features into a hierarchical spatiotemporal graph structure to intuitively represent the dynamic interactions between vehicles. Nodes are defined as vehicle entities, whose attributes include real-time status (speed, acceleration), intent probability (e.g., 70% left turn, 30% straight), and identity labels (e.g., emergency vehicle, regular vehicle). Edges represent interactions between vehicles, with weights determined by the spatiotemporal proximity of the conflict domain (time difference, distance difference) and historical cooperative tendencies (e.g., frequent yielding). Secondly, this step automatically filters illegal connections (e.g., potentially conflicting edges crossing solid lines) by introducing lane topology connectivity constraints, such as prohibiting lane changes on solid lines or specific turning restrictions, ensuring strict compliance with traffic regulations. This hierarchical design (e.g., single-vehicle status at the bottom layer and multi-vehicle interactions at the top layer) supports divide-and-conquer optimization, reduces computational complexity, and provides a scalable topological framework for subsequent feature aggregation and policy optimization.

[0071] Step S520: Aggregate node features based on the hierarchical spatiotemporal graph structure, perform bias correction on the attention weights in combination with the vehicle priority labels, and obtain the node's encoded feature vector;

[0072] Preferably, this step is based on a layered spatiotemporal graph and employs a gated graph attention network (GGAT) to aggregate neighborhood information. Each node utilizes a multi-head attention mechanism to weightedly fuse its own features with those of neighboring vehicles. This weighting introduces a vehicle priority bias: for example, ambulance nodes are given a higher weight in the attention calculation, forcing the model to prioritize their dynamic intent. Furthermore, lane topology information (such as the turning authority of the vehicle's lane) is leveraged to constrain feature propagation paths and prevent the spread of illicit information (e.g., a vehicle in a straight lane receiving information about a neighbor turning left). This step outputs a node encoding feature vector that not only captures the state of the individual vehicle but also implicitly reflects the influence of neighborhood interactions, providing a high-information node representation for dynamic updates of edge weights.

[0073] Step S530: Iteratively update the edge relationship weights based on the encoded feature vectors, and modify the weights based on the real-time conflict domain risk changes and the yielding habits between vehicles to obtain a dynamic edge relationship matrix;

[0074] Specifically, edge weights in this step are dynamically adjusted based on the real-time scenario. These include: conflict timeliness attenuation. For example, when the remaining time in the conflict domain decreases from 10 seconds to 2 seconds, the corresponding edge weight increases exponentially, prioritizing high-risk interactions based on urgency. Historical interaction habit correction: If vehicle A has a historical yield probability of 80% for vehicle B, the competition weight is reduced in the current interaction, emphasizing collaboration. Sensor confidence compensation: For vehicles experiencing increased LiDAR detection error in low visibility, the reliability of their associated edge weights is reduced. Through an online iterative mechanism, the edge relationship matrix responds to environmental changes in real time, ensuring a strict match between interaction intensity and risk level, thus avoiding the decision lag inherent in static models.

[0075] Step S540: Based on the encoded feature vector and the dynamic edge relationship matrix, a vehicle behavior model is constructed through collaborative strategy optimization and traffic rule constraint strategy space pruning processing.

[0076] It should be noted that this step utilizes the Multi-Agent Proximal Policy Optimization (MAPPO) framework for joint training based on node encoding features and a dynamic edge matrix. The global reward function is designed as a weighted sum of conflict resolution success rate and traffic efficiency, incentivizing the model to optimize overall traffic flow while maintaining safety. Policy space pruning directly embeds traffic rules, such as prohibiting lane changes in solid-line areas and stopping and waiting at yield signs. During training, the node policy generator (controlling individual vehicle behavior) and the edge relationship optimizer (coordinating multi-vehicle interactions) are updated alternately, achieving hierarchical optimization. The final model output is a deployable policy network that supports real-time inference in the hundreds of milliseconds. It generates compliant, coordinated, and dynamically adaptable vehicle behaviors in complex scenarios such as unsignaled intersections. For example, this can trigger coordinated multi-vehicle deceleration in high-risk conflict zones while maintaining efficient traffic flow in low-risk scenarios.

[0077] Example 2:

[0078] like Figure 2 As shown, this embodiment provides an intelligent traffic vehicle behavior modeling system based on collaborative learning, the system including:

[0079] Acquisition module 901 is used to acquire traffic data in the target area, the traffic data including the real-time status data of the current vehicle, road structure data and neighboring vehicle interaction data;

[0080] A calibration module 902 is configured to perform data calibration processing based on traffic data and eliminate abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative sensing dataset;

[0081] Prediction module 903, used to generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation;

[0082] Extraction module 904 is used to extract the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles based on the vehicle spatiotemporal trajectory function and the spatiotemporal distribution diagram of the conflict domain, through the association of temporal behavior patterns and the quantification of interactive game relationships, to obtain a multi-dimensional interactive feature vector;

[0083] The modeling module 905 is used to perform behavior modeling based on the multi-dimensional interaction feature vector. By constructing a layered graph network model based on node dynamic feature encoding and edge relationship weight update, a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles is obtained.

[0084] In some embodiments disclosed herein, the calibration module 902 includes:

[0085] The first calibration unit is used to perform time synchronization processing based on traffic data, correct the timestamp offset between vehicles by combining the clock deviation observation data of multiple vehicles, and use the periodic clock calibration signal communicated between vehicles as the time reference to obtain a multi-vehicle status data set;

[0086] The second calibration unit is used to perform unified spatial coordinate processing based on the multi-vehicle state data set, map the multi-vehicle position data to the global map coordinate system, and obtain a vehicle position data set;

[0087] A third calibration unit is configured to perform occlusion compensation processing based on the vehicle position dataset, obtain a vehicle status dataset by constructing a vehicle observation topology map and predicting the acceleration and heading angle of the occluded vehicle based on the relative positions and motion trends of neighboring vehicles;

[0088] The fourth calibration unit is used to perform road geometry constraint correction processing based on the vehicle state data set, and use a quadratic programming algorithm to constrain the vehicle position to the lane boundary defined by the map to obtain a collaborative perception data set.

[0089] In some embodiments disclosed in this application, the prediction module 903 includes:

[0090] The first prediction unit is configured to generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset and in combination with the vehicle's historical position and real-time motion state;

[0091] The second prediction unit is used to perform trajectory intersection prediction processing based on the vehicle space-time trajectory function, and obtain a space-time distribution table of trajectory intersections by calculating the space-time intersection coordinates of multiple vehicle trajectories and the confidence intervals of the intersection time windows;

[0092] The third prediction unit is used to perform conflict time window prediction processing based on the spatiotemporal distribution table of trajectory intersections, and obtain a high conflict risk time window by calculating the expected time distribution of vehicles arriving at the intersection and the overlap probability of the conflict time window;

[0093] The fourth prediction unit is used to calculate and process the dynamic safety radius according to the high-collision risk time window, dynamically adjust the safety radius threshold based on the vehicle's maximum deceleration, system response delay, and sensor detection error parameters, and obtain a spatiotemporal distribution map of the collision domain.

[0094] In some embodiments disclosed in this application, the extraction module 904 includes:

[0095] The first extraction unit is used to classify the driving intention according to the vehicle's spatiotemporal trajectory function, and outputs the probability distribution characteristics of the driving intention by using the acceleration change rate and heading angle deviation characteristics of the vehicle's historical trajectory and the steering logic constraints of the intersection lane topology as input values ​​of the long short-term memory network;

[0096] The second extraction unit is used to model the interaction relationship based on the spatiotemporal distribution diagram of the conflict domain and construct a competition and cooperation relationship matrix by calculating the spatiotemporal proximity of the conflict domains between vehicles;

[0097] The third extraction unit is used to calculate the game strategy benefits based on the competition and cooperation relationship matrix, quantify the benefits of each vehicle adopting the acceleration, deceleration and yielding strategies in the conflict domain, and generate a strategy benefit distribution table;

[0098] The fourth extraction unit is used to perform feature fusion processing based on the probability distribution characteristics and the strategy return distribution table to obtain a multi-dimensional interactive feature vector.

[0099] In some embodiments disclosed herein, the modeling module 905 includes:

[0100] The first modeling unit is used to construct a spatiotemporal graph structure based on the multi-dimensional interactive feature vector. It converts the vehicle identity labels, intention probability distribution, and conflict domain associations into node attributes and edge connection relationships of the spatiotemporal graph structure, and combines the connectivity constraints of the intersection lane topology to obtain a hierarchical spatiotemporal graph structure.

[0101] The second modeling unit is used to aggregate node features based on the hierarchical spatiotemporal graph structure, and to perform bias correction on the attention weights based on the vehicle priority labels to obtain the node's encoded feature vector;

[0102] The third modeling unit is used to iteratively update the edge relationship weights based on the encoded feature vectors, and to modify the weights based on the real-time conflict domain risk changes and the yielding habits between vehicles to obtain a dynamic edge relationship matrix;

[0103] The fourth modeling unit is used to construct a vehicle behavior model based on the encoded feature vector and the dynamic edge relationship matrix through collaborative strategy optimization and strategy space pruning processing constrained by traffic rules.

[0104] Example 3:

[0105] Corresponding to the above method embodiment, this embodiment also provides an intelligent traffic vehicle behavior modeling device based on collaborative learning. The intelligent traffic vehicle behavior modeling device based on collaborative learning described below and the intelligent traffic vehicle behavior modeling method based on collaborative learning described above can refer to each other.

[0106] Figure 3 FIG. 8 is a block diagram of an intelligent traffic vehicle behavior modeling device 800 based on collaborative learning according to an exemplary embodiment. Figure 3 As shown, the intelligent traffic vehicle behavior modeling device 800 based on collaborative learning may include: a processor 801, a memory 802. The intelligent traffic vehicle behavior modeling device 800 based on collaborative learning may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0107] The processor 801 is used to control the overall operation of the collaborative learning-based intelligent traffic vehicle behavior modeling device 800 to complete all or part of the steps of the collaborative learning-based intelligent traffic vehicle behavior modeling method described above. The memory 802 is used to store various types of data to support the operation of the collaborative learning-based intelligent traffic vehicle behavior modeling device 800. This data may include, for example, instructions for any application or method operating on the collaborative learning-based intelligent traffic vehicle behavior modeling device 800, as well as application-related data such as contact information, sent and received messages, images, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the intelligent traffic vehicle behavior modeling device 800 based on collaborative learning and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more thereof, so the corresponding communication component 805 may include: a Wi-Fi module, a Bluetooth module, an NFC module.

[0108] In an exemplary embodiment, a collaborative learning-based intelligent traffic vehicle behavior modeling device 800 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the above-mentioned collaborative learning-based intelligent traffic vehicle behavior modeling method.

[0109] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, the program instructions implement the steps of the aforementioned method for modeling intelligent transportation vehicle behavior based on collaborative learning. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the device 800 for modeling intelligent transportation vehicle behavior based on collaborative learning to implement the aforementioned method for modeling intelligent transportation vehicle behavior based on collaborative learning.

[0110] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A method for modeling intelligent traffic vehicle behavior based on collaborative learning, characterized in that: include: Acquire traffic data within the target area, including real-time status data of the current vehicle, road structure data, and neighboring vehicle interaction data; Performing data calibration processing on the traffic data and eliminating abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative perception data set; Generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation; Based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles are extracted by quantifying the temporal behavior pattern association and the interactive game relationship, thereby obtaining a multi-dimensional interactive feature vector; Behavior modeling is performed based on the multi-dimensional interaction feature vector, and a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles is obtained by constructing a layered graph network model based on node dynamic feature encoding and edge relationship weight update; Among them, based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, the probability distribution characteristics of vehicle driving intentions and the multi-vehicle competition and cooperation relationship matrix are extracted by quantifying the temporal behavior pattern association and interactive game relationship, and a multi-dimensional interaction feature vector is obtained, including: Performing driving intention classification processing based on the vehicle spatiotemporal trajectory function, using the acceleration change rate and heading angle deviation characteristics of the vehicle's historical trajectory and the steering logic constraints of the intersection lane topology as input values ​​of the long short-term memory network, and outputting the probability distribution characteristics of the driving intention; Modeling the interaction relationship based on the spatiotemporal distribution diagram of the conflict domains, and constructing a competition-cooperation relationship matrix by calculating the spatiotemporal proximity of the conflict domains between vehicles; Calculate the game strategy benefits based on the competition and cooperation relationship matrix, quantify the benefits of each vehicle's acceleration, deceleration, and yielding strategies in the conflict domain, and generate a strategy benefit distribution table; Perform feature fusion processing based on the probability distribution characteristics and the strategy return distribution table to obtain a multi-dimensional interactive feature vector; The behavior modeling is performed based on the multi-dimensional interaction feature vector. By constructing a layered graph network model based on node dynamic feature encoding and edge relationship weight update, a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles is obtained, including: A spatiotemporal graph structure is constructed based on the multidimensional interactive feature vector, and the vehicle identity labels, intention probability distribution, and conflict domain association relationships are converted into node attributes and edge connection relationships of the spatiotemporal graph structure. In combination with the connectivity constraints of the intersection lane topology, a hierarchical spatiotemporal graph structure is obtained; Node features are aggregated based on the hierarchical spatiotemporal graph structure, and attention weights are biased and corrected in combination with vehicle priority labels to obtain encoding feature vectors of the nodes; Iteratively updating edge relationship weights according to the encoded feature vectors, and modifying the weights based on real-time conflict domain risk changes and vehicle yielding habits to obtain a dynamic edge relationship matrix; According to the encoded feature vector and the dynamic edge relationship matrix, a vehicle behavior model is constructed through collaborative strategy optimization and traffic rule-constrained strategy space pruning processing.

2. The intelligent traffic vehicle behavior modeling method based on collaborative learning according to claim 1 is characterized in that: Data calibration is performed based on the traffic data, and abnormal location points outside lane boundaries are eliminated based on road geometric constraints to obtain a collaborative perception dataset, including: Performing time synchronization processing on the traffic data, correcting timestamp offsets between vehicles by combining clock deviation observation data from multiple vehicles, and utilizing periodic clock calibration signals communicated between vehicles as a time reference to obtain a multi-vehicle state dataset; Performing unified spatial coordinate processing on the multi-vehicle state data set, mapping the multi-vehicle position data to a global map coordinate system, and obtaining a vehicle position data set; Performing occlusion compensation processing based on the vehicle position dataset, obtaining a vehicle state dataset by constructing a vehicle observation topology map and predicting the acceleration and heading angle of the occluded vehicle based on the relative positions and motion trends of neighboring vehicles; Road geometry constraint correction processing is performed based on the vehicle state dataset, and a quadratic programming algorithm is used to constrain the vehicle position to within the lane boundary defined by the map, thereby obtaining a collaborative perception dataset.

3. The intelligent traffic vehicle behavior modeling method based on collaborative learning according to claim 1 is characterized in that: A vehicle spatiotemporal trajectory function is generated based on the collaborative perception dataset, and a spatiotemporal distribution map of the collision domain is obtained based on trajectory intersection time window prediction and dynamic safety radius calculation, including: Generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset and combining the vehicle's historical position and real-time motion status; Performing trajectory intersection prediction processing according to the vehicle space-time trajectory function, and obtaining a space-time distribution table of trajectory intersections by calculating the space-time intersection coordinates of multiple vehicle trajectories and the confidence intervals of the intersection time windows; Conflict time window prediction processing is performed based on the temporal and spatial distribution table of trajectory intersections, and a high conflict risk time window is obtained by calculating the expected time distribution of vehicles arriving at the intersection and the probability of overlap of the conflict time windows; Dynamic safety radius calculation is performed according to the high-collision risk time window, and the safety radius threshold is dynamically adjusted based on the vehicle's maximum deceleration, system response delay, and sensor detection error parameters to obtain a spatiotemporal distribution map of the collision domain.

4. An intelligent traffic vehicle behavior modeling system based on collaborative learning, characterized in that: include: An acquisition module is used to acquire traffic data within the target area, wherein the traffic data includes real-time status data of the current vehicle, road structure data, and neighboring vehicle interaction data; a calibration module, configured to perform data calibration processing based on the traffic data and eliminate abnormal location points outside lane boundaries based on road geometric constraints to obtain a collaborative perception data set; A prediction module, configured to generate a spatiotemporal trajectory function of a vehicle based on the collaborative perception dataset, and obtain a spatiotemporal distribution map of the collision domain based on trajectory intersection time window prediction and dynamic safety radius calculation; An extraction module is configured to extract the probability distribution characteristics of vehicle driving intentions and the competition and cooperation relationship matrix of multiple vehicles based on the vehicle spatiotemporal trajectory function and the conflict domain spatiotemporal distribution diagram, by associating temporal behavior patterns and quantifying interactive game relationships, to obtain a multi-dimensional interactive feature vector; a modeling module for performing behavior modeling based on the multi-dimensional interaction feature vector, and obtaining a vehicle behavior model that characterizes the interaction behavior rules and decision logic between vehicles by constructing a hierarchical graph network model based on node dynamic feature encoding and edge relationship weight update; Wherein, the extraction module includes: a first extraction unit, configured to perform driving intention classification processing based on the vehicle spatiotemporal trajectory function, and output a probability distribution feature of the driving intention by using the acceleration change rate and heading angle deviation characteristics of the vehicle's historical trajectory and the steering logic constraints of the intersection lane topology as input values ​​of a long short-term memory network; a second extraction unit, configured to perform interaction relationship modeling based on the spatiotemporal distribution graph of the conflict domains, and construct a competition and cooperation relationship matrix by calculating the spatiotemporal proximity of the conflict domains between vehicles; a third extraction unit, configured to calculate the game strategy benefits according to the competition-cooperation relationship matrix, quantify the benefits of each vehicle adopting the acceleration, deceleration, and yielding strategies in the conflict domain, and generate a strategy benefit distribution table; a fourth extraction unit, configured to perform feature fusion processing based on the probability distribution feature and the strategy return distribution table to obtain a multi-dimensional interactive feature vector; Wherein, the modeling module includes: A first modeling unit is configured to construct a spatiotemporal graph structure based on the multidimensional interactive feature vector, convert vehicle identity labels, intention probability distribution, and conflict domain associations into node attributes and edge connection relationships of the spatiotemporal graph structure, and combine the connectivity constraints of the intersection lane topology to obtain a hierarchical spatiotemporal graph structure; A second modeling unit is configured to aggregate node features based on the hierarchical spatiotemporal graph structure, perform bias correction on attention weights in combination with vehicle priority labels, and obtain an encoded feature vector of the node; a third modeling unit, configured to iteratively update edge relationship weights according to the encoded feature vectors, and modify the weights based on real-time conflict domain risk changes and the yielding habits between vehicles to obtain a dynamic edge relationship matrix; The fourth modeling unit is used to construct a vehicle behavior model based on the encoded feature vector and the dynamic edge relationship matrix through collaborative strategy optimization and traffic rule constraint strategy space pruning processing.

5. The intelligent traffic vehicle behavior modeling system based on collaborative learning according to claim 4 is characterized in that: The calibration module includes: a first calibration unit, configured to perform time synchronization processing based on the traffic data, correct the timestamp offset between vehicles by combining clock deviation observation data of multiple vehicles, and use the periodic clock calibration signal communicated between vehicles as a time reference to obtain a multi-vehicle state data set; a second calibration unit, configured to perform unified spatial coordinate processing on the multi-vehicle state data set, map the multi-vehicle position data to a global map coordinate system, and obtain a vehicle position data set; a third calibration unit, configured to perform occlusion compensation processing based on the vehicle position dataset, obtain a vehicle state dataset by constructing a vehicle observation topology map and predicting the acceleration and heading angle of the occluded vehicle based on the relative positions and motion trends of neighboring vehicles; The fourth calibration unit is used to perform road geometry constraint correction processing based on the vehicle state data set, and use a quadratic programming algorithm to constrain the vehicle position to the lane boundary defined by the map to obtain a collaborative perception data set.

6. The intelligent traffic vehicle behavior modeling system based on collaborative learning according to claim 4 is characterized in that: The prediction module includes: A first prediction unit is configured to generate a vehicle spatiotemporal trajectory function based on the collaborative perception dataset and in combination with the vehicle's historical position and real-time motion state; a second prediction unit, configured to perform trajectory intersection prediction processing according to the vehicle spatiotemporal trajectory function, and obtain a spatiotemporal distribution table of trajectory intersections by calculating the spatiotemporal intersection coordinates of multiple vehicle trajectories and confidence intervals of intersection time windows; The third prediction unit is used to perform conflict time window prediction processing based on the spatiotemporal distribution table of the trajectory intersection points, and obtain a high conflict risk time window by calculating the expected time distribution of the vehicle arriving at the intersection and the overlap probability of the conflict time window; the fourth prediction unit is used to perform dynamic safety radius calculation processing based on the high conflict risk time window, and dynamically adjust the safety radius threshold based on the vehicle's maximum deceleration, system response delay, and sensor detection error parameters to obtain a conflict domain spatiotemporal distribution map.

Images