Intelligent vehicle and artificial vehicle mixed lane changing scene representation method and system based on space-time network

Abstract

This invention discloses a method and system for representing lane-changing scenarios involving intelligent vehicles and human vehicles in mixed traffic based on a spatiotemporal network. The method includes: acquiring motion state data and environmental interaction data of all vehicles in the mixed traffic flow within a target road segment; constructing a basic architecture of a discretized spatiotemporal network covering the target spatiotemporal domain based on the motion state data and environmental interaction data; modeling the lane-changing behavior of intelligent vehicles as deterministic trajectories in the discretized spatiotemporal network, and modeling the uncertain lane-changing behavior of human vehicles as probability clouds in the discretized spatiotemporal network, while simultaneously constructing a risk coefficient field; establishing interaction constraints between the deterministic trajectories and the probability clouds, and transforming the interaction constraints into mathematical constraints; constructing an objective function for the discretized spatiotemporal network, and obtaining the final discretized spatiotemporal network based on the risk coefficient field; and obtaining the optimal spatiotemporal trajectories of intelligent vehicles and human vehicles based on the final discretized spatiotemporal network.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation and autonomous driving technology, specifically relating to a method and system for representing a mixed lane-changing scenario of intelligent vehicles and human vehicles based on a spatiotemporal network. Background Technology

[0002] With the rapid development of intelligent connected vehicle technology, a mixed traffic situation involving both connected and automated vehicles (CAVs) and human-driven vehicles (HDVs) will persist on future roads for a long time. Lane changing, as a fundamental operation of vehicles during road travel, is a key factor affecting traffic flow stability and road capacity. In a mixed traffic environment, the lane changing behaviors of CAVs and HDVs differ fundamentally: the trajectory of a CAV can be precisely planned and controlled by the system, exhibiting determinism and controllability; while the lane changing behavior of an HDV is influenced by individual driver decisions, exhibiting significant uncertainty and randomness. How to accurately describe and represent the lane changing behaviors of these two types of vehicles in a mixed traffic environment, especially the spatiotemporal interaction between the uncertain behavior of HDVs and the deterministic behavior of CAVs, is a fundamental problem that urgently needs to be solved in the field of intelligent connected vehicle trajectory planning and control.

[0003] Currently, the main technical solutions for representing lane-changing scenarios in mixed-traffic environments fall into the following categories: (1) Microscopic simulation model-based methods. Car-following models and lane-changing models (such as the Gipps model, MOBIL model, etc.) are used to simulate vehicle behavior, and the uncertainty of HDV is simulated by adding random terms. These methods can well characterize the microscopic behavior of vehicles, but they have problems such as high computational complexity, difficulty in direct integration with optimization control algorithms, and the randomness of simulation results leads to a lack of consistency in scene representation.

[0004] (2) Motion planning-based methods. These methods represent lane-changing scenarios as the vehicle's state space and generate the CAV's trajectory through sampling or optimization methods. These methods typically treat HDVs as dynamic obstacles or uncertain disturbances, employing deterministic boundaries or probabilistic occupancy for obstacle avoidance. However, these methods lack a structured description of the spatiotemporal evolution of HDV behavior, making it difficult to handle complex scenarios such as continuous lane changes by HDVs.

[0005] (3) Graph Neural Network-Based Methods. The interaction relationships between vehicles are modeled as a graph structure, and the interaction features are learned using graph neural networks. This type of method can capture the spatial interaction between vehicles, but the modeling of the time dimension is relatively coarse and it is difficult to accurately express the continuous occupancy relationship of spatiotemporal resources during lane changing.

[0006] Existing technologies have the following shortcomings in representing mixed-traffic lane-changing scenarios: First, there is a lack of a unified modeling framework for the differences in behavior between the two types of vehicles. Existing methods typically treat CAV and HDV as homogeneous objects, using the same representation, failing to fully utilize the deterministic advantages brought by the controllability of CAV, and also failing to effectively express the uncertain characteristics of HDV behavior.

[0007] Second, it is difficult to accurately represent the spatiotemporal resource occupancy during lane changing. Lane changing is a continuous process, during which vehicles need to occupy the spatiotemporal resources of both the current lane and the target lane simultaneously. Existing methods mostly adopt the assumption of instantaneous lane changing or simplify the process, which cannot accurately describe this key physical characteristic.

[0008] Third, there is a lack of characterization of the cumulative probability effect of continuous lane changing in HDV. HDV may engage in continuous lane changing behavior across multiple time steps, and the probability of its occupancy of spatiotemporal resources needs to be calculated by superposition. Existing methods typically assume that lane changing behavior at each time step is independent, neglecting the cumulative probability effect of continuous lane changing.

[0009] To address the shortcomings of existing technologies, this invention provides a method for representing lane-changing scenarios involving both intelligent vehicles (CAVs) and human-driven vehicles (HDVs) based on a spatiotemporal network. This method constructs a multi-lane spatiotemporal network, representing the lane-changing behavior of CAVs as deterministic trajectories within the network and the lane-changing behavior of HDVs as probabilistic clouds. It also establishes interactive constraints between the two types of vehicles regarding spatiotemporal resource occupancy, ultimately generating a standardized scenario representation model that can be used for CAV trajectory planning and decision-making. Summary of the Invention

[0010] This invention aims to address the shortcomings of existing technologies and provides the following solutions: A method for representing a lane-changing scenario where intelligent vehicles and human vehicles share the same lanes based on a spatiotemporal network includes the following steps: Acquire motion status data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment; Based on the motion state data and the environmental interaction data, a basic architecture for a discretized spatiotemporal network covering the target spatiotemporal domain is constructed. The lane-changing behavior of intelligent vehicles is modeled as a deterministic trajectory in the discretized spatiotemporal network, and the uncertain lane-changing behavior of manual vehicles is modeled as a probability cloud in the discretized spatiotemporal network. At the same time, a risk coefficient field is constructed. Establish the interaction constraints between the deterministic trajectory and the probability cloud, and transform the interaction constraints into mathematical constraints. Construct the objective function of the discretized spatiotemporal network, and obtain the final discretized spatiotemporal network based on the risk coefficient field; The optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle are obtained based on the final discretized spatiotemporal network.

[0011] Preferably, the motion state data includes: the vehicle's position, speed, and acceleration; The methods for obtaining the motion state data include: for intelligent vehicles, acquiring it through onboard sensors or vehicle-to-infrastructure (V2I) systems; for manual vehicles, acquiring it through roadside sensing devices. The environmental interaction data includes: lane line information, road boundaries, exit / forced lane change point locations, relative distances to surrounding vehicles, and relative speeds of surrounding vehicles.

[0012] Preferably, the method for constructing the infrastructure of the discretized spatiotemporal network includes: Define lane set L For a three-lane scenario L ={0,1,2}, representing the three lanes from the inside out; Each lane l ∈ L Divided into a continuous set of spatial nodes Sl ,node s ∈ Sl lane l The first s There are 1 spatial location unit, and the actual length of each spatial node is Δ. s ; The planning period is divided into a continuous set of time nodes. T The time node index is t ∈ T The time interval between adjacent time nodes is Δ t ; Define spatiotemporal resource nodes ( l , s , t ), indicating lane l Spatial nodes on s In time t Spatiotemporal resource units on the surface; Define spacetime arc ( l , s , t )→( l’ , s’ , t +1) indicates that the vehicle started from time t arrive t +1, from the lane l spatial nodes s Move to the lane l’ spatial nodes s’ The transfer process; Define the spatiotemporal node network as V ,in V ( l , s, t ) indicates the relationship between nodes ( l , s , t ) are adjacent, and from node ( l , s , t The spacetime nodes that can be reached from the starting point; According to the lane set L The set of spatial nodes Sl and the set of time nodes T The basic architecture for constructing the discretized spatiotemporal network ( L , Sl , T ).

[0013] Preferably, the method for obtaining the deterministic trajectory includes: For each intelligent vehicle, define the initial lane. Initial spatial nodes Target lane and the spatial node where the forced lane change point is located ; Introducing 0-1 deterministic trajectory variables xc ( l , s , t , l’ , s’ , t +1), indicating whether the intelligent vehicle is from time t arrive t +1 occupied from the lane l spatial nodes s to the lane l′ spatial nodes s′ The spacetime arc; Introducing 0-1 deterministic trajectory variables yc ( l , s , t This indicates whether the intelligent vehicle is occupying the lane. l Spatiotemporal resource nodes on l , s , t ); When an intelligent vehicle changes lanes, it simultaneously occupies the spatiotemporal resources of both the current lane and the target lane during the lane change process, which is represented as: in, lcurrent Indicates the lane the intelligent vehicle is currently in. ltragetIndicates the lane the intelligent vehicle is in after changing lanes. tstart Indicates the start time of lane change. tend Indicates the end time of the lane change.

[0014] Preferably, the method for obtaining the probability cloud includes: For each manual vehicle, its lane-changing behavior at each decision moment is described by a probabilistic model: in, Pstay This represents the probability of going straight. Pleft This represents the probability of changing lanes to the left. Pright This indicates the probability of changing lanes to the right. exist t At any time, manual vehicles occupy the lane. l Spatial nodes, in t At time +1, with probability Pstay Continue to occupy the lane l The subsequent spatial nodes, in probability Plef Start moving to the left lane l -1 lane change, with probability Pright Start moving to the right lane l +1 lane change; When a manually operated vehicle changes lanes consecutively across multiple time steps, the spatiotemporal resource occupancy probabilities of subsequent lanes are superimposed to calculate the probability cloud. Ph ( l , s , t ).

[0015] Preferably, the interaction constraints include: Safety distance constraints: in, This indicates that the manual vehicle occupies a spatiotemporal node ( l , s’ , t The probability of ) dfront This indicates the number of nodes corresponding to the safe distance ahead. dback This indicates the number of nodes corresponding to the rear safety distance; Exclusive constraints on spatiotemporal resources: in, This indicates the time and space nodes occupied by all intelligent connected vehicles. l , s , t The occupancy status of ).

[0016] Preferably, the mathematical constraints include: flow balance constraints, forced lane-changing constraints, space resource occupancy constraints, lane-changing process constraints, and dynamic constraints.

[0017] Preferably, the objective function includes: an overall benefit function and an overall risk function; The overall benefit function is: in, Cl lane l Includes all intelligent connected vehicles. Indicating intelligent connected vehicles c Index of the target lane lane l The index of the last spatiotemporal node, ω LC represents the loss item for lane changing by intelligent connected vehicles. Ll lane l The adjacent lane, zc ( l , s , t , l’ (This refers to intelligent connected vehicles) c Is it in time? t From the lane l of s Node to Lane l’ Change lanes. ωa The coefficient representing the acceleration loss in the objective function. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωjerk This represents the coefficient representing the jerk loss in the objective function. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωE This represents the loss coefficient representing the gap between the objective function and the maximum throughput capacity. This represents the energy at the speed corresponding to the maximum throughput capacity of the current road segment. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t Longitudinal dynamic energy on ) The overall risk function is: in, EThe set of spatiotemporal arcs representing a spatiotemporal network. r ( l , s , t , l’ , s’ , t +1) indicates that from the spatiotemporal node ( l , s , t ) to spacetime node ( l’ , s , t+1 On the spacetime arc of ), the risks posed by human drivers.

[0018] The present invention also provides a system for representing a scenario of intelligent vehicles and human vehicles sharing lanes based on a spatiotemporal network. The system applies the above-mentioned method and includes: a data acquisition module, a network infrastructure construction module, a trajectory and probability modeling module, a constraint construction module, a network construction module, and a trajectory optimization module. The data acquisition module is used to acquire motion status data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment; The network infrastructure construction module constructs a discrete spatiotemporal network infrastructure covering the target spatiotemporal domain based on the motion state data and the environmental interaction data. The trajectory and probability modeling module is used to model the lane-changing behavior of the intelligent vehicle as a deterministic trajectory in the discretized spatiotemporal network, and to model the uncertain lane-changing behavior of the manual vehicle as a probability cloud in the discretized spatiotemporal network, while constructing a risk coefficient field. The constraint construction module is used to establish the interaction constraints between the deterministic trajectory and the probability cloud, and to transform the interaction constraints into mathematical constraints. The network construction module is used to construct the objective function of the discretized spatiotemporal network and obtain the final discretized spatiotemporal network based on the risk coefficient field. The trajectory optimization module obtains the optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle based on the final discretized spatiotemporal network.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention introduces a probabilistic cloud to describe the uncertain behavior of manually driven vehicles, which can accurately express the complex interaction between intelligent vehicles and human vehicles in terms of spatiotemporal resource occupation in mixed traffic environments, and provides a standardized scenario input for lane-changing decisions and trajectory planning of intelligent connected vehicles. Attached Figure Description

[0020] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a simplified diagram of the spatiotemporal network according to an embodiment of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] The scenario modeled in this invention is a vehicle lane-changing interaction scenario under mixed traffic conditions of intelligent connected vehicles and manually driven vehicles in an urban expressway or highway environment.

[0025] Consider a one-way multi-lane road section (taking a three-lane road as an example), with lanes numbered from the inside out as lane 0, lane 1, and lane 2. Two types of vehicles operate on the road simultaneously: one type is intelligent vehicles (CAVs), whose motion is fully controllable and whose lane-changing behavior can be precisely planned by the system; the other type is human vehicles (HDVs), whose lane-changing behavior is uncertain, influenced by individual driver decisions, and difficult to predict accurately.

[0026] In the time dimension, continuous time is discretized into time slices of equal length. t , t +1, t +2, ...), each time slice corresponds to a vehicle's position update or decision moment. In the spatial dimension, each lane is divided into continuous spatial nodes (1, 2, 3, ...), each node is set to a fixed value, and each vehicle occupies a fixed number of nodes according to its actual vehicle length. The vehicle's driving process is manifested by occupying these spatial nodes in sequence.

[0027] Taking a three-lane scenario as an example, a spatiotemporal network is constructed for each lane, such as... Figure 2As shown, red represents the HDV's occupation of spatiotemporal resources, and blue represents the CAV's occupation of spatiotemporal resources under a certain decision. The following are descriptions of these aspects: Example 1 In this embodiment, as Figure 1 As shown, a method for representing a lane-changing scenario where intelligent vehicles and human vehicles share the same lanes based on a spatiotemporal network includes the following steps: S1. Obtain motion status data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment.

[0028] Motion state data includes: vehicle position, speed, and acceleration; methods for obtaining motion state data include: for intelligent vehicles, acquisition through onboard sensors or vehicle-to-infrastructure (V2I) systems; for manually operated vehicles, acquisition through roadside sensing devices (such as cameras and radar); environmental interaction data includes: lane line information, road boundaries, exit / forced lane change point locations, relative distances to surrounding vehicles, and relative speeds of surrounding vehicles. The continuous motion trajectory is discretized and aligned to a unified time reference to prepare for subsequent spatiotemporal network construction.

[0029] S2. Based on motion state data and environmental interaction data, construct the basic architecture of a discretized spatiotemporal network covering the target spatiotemporal domain.

[0030] In this embodiment, the method for constructing the infrastructure of a discretized spatiotemporal network includes: Define lane set L For a three-lane scenario L ={0,1,2} represents the three lanes from the inside out.

[0031] Each lane l ∈ L Divided into a continuous set of spatial nodes Sl ,node s ∈ Sl lane l The first s There are 1 spatial location unit, and the actual length of each spatial node is Δ. s (Each vehicle occupies a fixed number of nodes based on its actual length).

[0032] The planning period is divided into a continuous set of time nodes. T The time node index is t ∈ T The time interval between adjacent time nodes is Δ t .

[0033] Define spatiotemporal resource nodes ( l , s , t ), indicating lane l Spatial nodes ons In time t The spatiotemporal resource units on the road; the occupation of roads by vehicles is manifested as the occupation of these spatiotemporal nodes.

[0034] Define spacetime arc ( l , s , t )→( l’ , s’ , t +1) indicates that the vehicle started from time t arrive t +1, from the lane l spatial nodes s Move to the lane l’ spatial nodes s’ The transfer process.

[0035] Define the spatiotemporal node network as V ,in V ( l , s , t ) indicates the relationship between nodes ( l , s , t ) are adjacent, and from node ( l , s , t The spatiotemporal nodes that can be reached from the starting point; this representation must respect spatiotemporal transitions, such as the time of the next node must be greater than that of the current node, and since vehicles cannot travel in reverse, the spatial location index of the next node must also be greater than that of the current node.

[0036] According to lane set L Spatial Node Set Sl and time node set T The basic architecture for building discretized spatiotemporal networks L , Sl , T ).

[0037] S3. Model the lane-changing behavior of intelligent vehicles as deterministic trajectories in a discretized spatiotemporal network, and model the uncertain lane-changing behavior of manual vehicles as probability clouds in a discretized spatiotemporal network, while constructing a risk coefficient field.

[0038] In this embodiment, the method for obtaining a deterministic trajectory includes: For each intelligent vehicle, define the initial lane. Initial spatial nodes Target lane and the spatial node where the forced lane change point is located (e.g., the location that must be changed before export).

[0039] Introducing 0-1 deterministic trajectory variables xc ( l , s , t , l’ , s’ , t +1), indicating whether the intelligent vehicle is from time t arrive t +1 occupied from the lane l spatial nodes s to the lane l′ spatial nodes s′ The spacetime arc; if occupied, it is 1, otherwise it is 0.

[0040] Introducing 0-1 deterministic trajectory variables yc ( l , s , t This indicates whether the intelligent vehicle is occupying the lane. l Spatiotemporal resource nodes on l , s , t If occupied, the value is 1; otherwise, it is 0.

[0041] When the intelligent vehicle changes lanes, during the duration of the lane change (from tstart arrive tend This simultaneously occupies the spatiotemporal resources of both the current lane and the target lane, represented as: in, lcurrent Indicates the lane the intelligent vehicle is currently in. ltraget Indicates the lane the intelligent vehicle is in after changing lanes. tstart Indicates the start time of lane change. tend This indicates the end time of the lane change. This reflects the physical fact that a lane change cannot be completed instantly and requires a vehicle to occupy two lanes simultaneously.

[0042] In this embodiment, the method for obtaining the probability cloud includes: For each manual vehicle, its lane-changing behavior at each decision moment is described by a probabilistic model: in, Pstay This represents the probability of going straight. Pleft This represents the probability of changing lanes to the left. Pright This represents the probability of changing lanes to the right.

[0043] exist t At any time, manual vehicles occupy the lane. l spatial nodess , s +1, s +2 ( s , s +1, s +2 corresponds to vehicle length), in t At time +1, with probability Pstay Continue to occupy the lane l The subsequent spatial nodes, in probability Plef Start moving to the left lane l -1 lane change, with probability Pright Start moving to the right lane l +1 lane change; When a manually operated vehicle changes lanes consecutively across multiple time steps, the spatiotemporal resource occupancy probabilities of subsequent lanes are superimposed to calculate a probability cloud. Ph ( l , s , t For example, if a manual vehicle first changes lanes to the right (probability 0.1) and then to the left (probability 0.7), the final probability of occupying the spacetime resources of a specific lane is the product of the probabilities of the two events (0.1 × 0.7 = 0.07).

[0044] Define the risk coefficient field This indicates that the artificial vehicle is in the spacetime arc ( l , s , t )→( l’ , s’ , t The risk coefficient brought about by +1) is positively correlated with the probability that the artificial vehicle occupies the spatiotemporal arc.

[0045] S4. Establish the interaction constraints between the deterministic trajectory and the probability cloud, and transform the interaction constraints into mathematical constraints.

[0046] Interaction constraints include: Intelligent vehicles need to maintain a safe distance from both the vehicle in front and behind. In a spatiotemporal network, this means that within a certain range before and after the node occupied by the intelligent vehicle, there cannot be any other vehicles (or the probability of other vehicles exceeding a safe threshold). This is known as a safe distance constraint. in, This indicates that the manual vehicle occupies a spatiotemporal node ( l , s’ , t The probability of ) dfront This indicates the number of nodes corresponding to the safe distance ahead. dback This indicates the number of nodes corresponding to the rear safety distance; A single spatiotemporal node cannot be occupied by two intelligent vehicles simultaneously, nor can the sum of deterministic occupation by an intelligent vehicle and probabilistic occupation by a human vehicle exceed 1 (i.e., they cannot coexist probabilistically). This is the spatiotemporal resource exclusivity constraint. in, This indicates the time and space nodes occupied by all intelligent connected vehicles. l , s , t The occupancy status of ).

[0047] The probability occupancy of HDV is handled conservatively here, and is regarded as the "occupancy ceiling in a probabilistic sense".

[0048] Mathematical constraints include: flow balance constraints, forced lane-changing constraints, space resource occupancy constraints, lane-changing process constraints, and dynamic constraints.

[0049] in: (1) Flow balance constraints include: To ensure the continuity of the intelligent vehicle's trajectory, the initial node has only one outflow arc, intermediate nodes have inflow equal to outflow, and the endpoint node is located at the end of the target lane.

[0050] ① For the initial node: any intelligent vehicle at the initial time... t 0 must start from the initial lane Starting from the initial spatial node, occupying a spacetime arc to enter the next time step: in, Indicates in t 0 o'clock from the lane spatial nodes Departure from the set of lanes and spatial nodes that are reachable in the next time step (including going straight and changing lanes to adjacent lanes). The variable is a 0-1 decision variable, indicating whether the intelligent vehicle occupies the spatiotemporal arc.

[0051] ② For intermediate nodes: any intermediate spatiotemporal node of any intelligent vehicle other than the initial and final times ( l , s , t The number of spacetime arcs flowing into this node is equal to the number of spacetime arcs flowing out of this node. in, lperv The lane index represents the preceding spatiotemporal node. sperv The spatial node index representing the preceding spatiotemporal node. lnext The lane index represents the subsequent spatiotemporal node. snextThe spatial node index represents the subsequent spatiotemporal node.

[0052] ③ For the destination node: For any intelligent vehicle, it must eventually reach the target lane. End space node (or exit node), and in the last time step tend There is no outflow arc at this point: And for the endpoint node have: .

[0053] (2) Mandatory lane-changing constraints include: Ensure that intelligent vehicles must complete lane changes before reaching a mandatory lane-changing point (such as an exit), and that they cannot change lanes again after reaching the mandatory lane-changing point.

[0054] ① Lane change constraints must be completed before a forced lane change is implemented. For any intelligent vehicle, if its initial lane Not equal to the target lane Then you must reach the spatial node where the forced lane change point is located. Previously, the lane change to the target lane was completed, among which... Indicates that in the lane l The index of the last spatial node: .

[0055] ② Lane changing is prohibited after reaching the mandatory lane change point: .

[0056] (3) Space resource occupancy constraints include: Establish spatiotemporal arc occupancy variables xc Spatiotemporal node occupancy variables yc The mapping relationship between them ensures that the intelligent vehicle occupies the corresponding start and end nodes while occupying the spatiotemporal arc.

[0057] ① Mapping relationship between spacetime arcs and spacetime nodes: in, l’’ Represents spatiotemporal nodes ( l, s, t The corresponding feasible successor lane index, s’’ Represents spatiotemporal nodes ( l, s, t The corresponding feasible successor space index.

[0058] ② For any intelligent vehicle, it needs to maintain a safe distance from the vehicle in front and also from the vehicle behind. Therefore, the occupation of spatial and temporal resources should consider both the area in front of and behind the vehicle. Lenc Hc represents the total length of the vehicle, Hc represents the distance the vehicle needs to maintain for safe driving, and Δ represents the following distance. s The length of a spatial unit.

[0059] Where M represents an extremely large integer.

[0060] ② No single node in a spatiotemporal network can be occupied by two intelligent vehicles simultaneously: .

[0061] (4) Lane changing process constraints include: ① Any CAV can only change lanes in one direction: .

[0062] ② The start and end times of lane changing must meet the minimum lane changing time threshold, where This is an integer decision variable, representing the time required for the lane change.

[0063] in, Indicates the minimum duration of a single lane change. This indicates the longest duration of a single lane change.

[0064] ③ The time interval between two lane changes must be greater than the threshold: in, t’ Indicates the time index of the second lane change, Δ LC This indicates the minimum time interval between two lane changes.

[0065] ④ You can only change lanes to the adjacent lane: .

[0066] ⑤ During lane changing, both the current lane and the target lane must be occupied simultaneously in terms of time and space resources: For any intelligent vehicle, if we consider the time step... t Start from the lane l Towards the lane l’ Change lanes, and t + Tc (l , s , t , l’ Once the lane change is complete, you need to be in the lane. l Up, occupying the following lane l’ The spatiotemporal resources, and this event needs to be at the time step of starting the lane change. t Afterwards, and at the end of the lane change period t + Tc ( l , s , t , l’ Before that, there was: However, the following conditions must be met: firstly, the vehicle must actually need to change lanes, that is... zc ( l , s , t , l’ ) = 1, and the time requirement must also be met, that is: The constraint was found to be conditional and needed to be linearized. Therefore, 0-1 auxiliary decision variables were introduced. sc ( t , t’ (This indicates a smart car) c At time step t Time to step t’ Whether a lane-changing operation is currently in progress; if so, the value is 0; otherwise, it is 1. This leads to a linearized expression: .

[0067] (5) The dynamic constraints are: The speed, acceleration, and jerk of an intelligent vehicle between adjacent time steps must meet the upper and lower bounds of vehicle dynamics to ensure the physical feasibility of the trajectory.

[0068] ① For any intelligent vehicle, the transitions between nodes in the spatiotemporal network must satisfy the upper and lower bounds of the dynamics, where, , , The decision variables are integers, representing the vehicle's jerk, acceleration, energy, etc. at its current spatiotemporal location.

[0069] .

[0070] ② Regarding longitudinal dynamics, for any intelligent vehicle, a mapping from velocity to spatiotemporal resource arcs needs to be constructed. For any intelligent vehicle, if it is at time step... t From the lane l spatial nodes i Turning at time step t +1 lane l’ spatial nodes i’ Therefore, its speed needs to meet the requirements for reaching that space node. i’ If the requirement is met and the range of the spatial node cannot be exceeded, then there are constraints: .

[0071] S5. Construct the objective function of the discretized spatiotemporal network, and obtain the final discretized spatiotemporal network based on the risk coefficient field.

[0072] In this embodiment, the objective function of the model is divided into two parts. The first part represents the overall benefit of the motion decision scheme for traffic, which includes efficiency, number of lane changes, energy loss, and the speed at which the speed passing through the planned road segment is close to the maximum capacity speed. The objective function includes: an overall benefit function and an overall risk function; The overall benefit function is: in, Cl lane l Includes all intelligent connected vehicles. Indicating intelligent connected vehicles c Index of the target lane lane l The index of the last spatiotemporal node, ω LC represents the loss item for lane changing by intelligent connected vehicles. Ll lane l The adjacent lane, zc ( l , s , t , l’ (This refers to intelligent connected vehicles) c Is it in time? t From the lane l of s Node to Lane l’ A lane change is initiated; if yes, the value is 1; otherwise, it is 0. ωa The coefficient representing the acceleration loss in the objective function. Indicating intelligent connected vehiclesc At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωjerk This represents the coefficient representing the jerk loss in the objective function. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωE This represents the loss coefficient representing the gap between the objective function and the maximum throughput capacity. This represents the energy at the speed corresponding to the maximum throughput capacity of the current road segment. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t Longitudinal dynamic energy on ) The overall risk function is: in, E The set of spatiotemporal arcs representing a spatiotemporal network. r ( l , s , t , l’ , s’ , t +1) indicates that from the spatiotemporal node ( l , s , t ) to spacetime node ( l’ , s , t+1 On the spacetime arc of ), the risks posed by human drivers.

[0073] Ultimately, the discretized spatiotemporal network includes: spatiotemporal network structure ( L , Sl , T Deterministic trajectory variables of intelligent vehicles xc , yc ; Probability cloud distribution of artificial vehicles Ph ( l , s , t ); Mathematical expressions for all constraints; Risk coefficient field .

[0074] S6. The optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle are obtained based on the final discretized spatiotemporal network.

[0075] In this embodiment, the final discretized spatiotemporal network can serve as a standardized input for subsequent intelligent vehicle trajectory planning and decision-making. Since a lane change may occur at any moment for a human-powered vehicle, a probability cloud of spatiotemporal resource occupancy will be formed in the lane spatiotemporal network. The deterministic trajectory of the intelligent vehicle will inevitably come into contact with this probability cloud. Therefore, the planning problem of the intelligent vehicle is transformed into: finding a spatiotemporal trajectory that satisfies all constraints and has the optimal comprehensive objective function (highest traffic efficiency, fewest lane changes, minimum energy loss, and minimum risk) in the risk field composed of the probability cloud of human-powered vehicles.

[0076] The network can output standardized data structures or matrix forms, which facilitates interfaceing with optimization solvers, simulation systems, or real vehicle controllers.

[0077] Example 2 In this embodiment, a system for representing a mixed lane-changing scenario of intelligent vehicles and human vehicles based on a spatiotemporal network includes: a data acquisition module, a network infrastructure construction module, a trajectory and probability modeling module, a constraint construction module, a network construction module, and a trajectory optimization module.

[0078] The data acquisition module acquires motion state data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment. The network infrastructure construction module builds the infrastructure of a discretized spatiotemporal network covering the target spatiotemporal domain based on the motion state data and environmental interaction data. The trajectory and probability modeling module models the lane-changing behavior of intelligent vehicles as deterministic trajectories in the discretized spatiotemporal network and the uncertain lane-changing behavior of manual vehicles as probability clouds in the discretized spatiotemporal network, while also constructing a risk coefficient field. The constraint construction module establishes the interaction constraints between the deterministic trajectory and the probability cloud, and transforms these interaction constraints into mathematical constraints. The network construction module constructs the objective function of the discretized spatiotemporal network and obtains the final discretized spatiotemporal network based on the risk coefficient field. The trajectory optimization module obtains the optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle based on the final discretized spatiotemporal network.

[0079] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for representing a lane-changing scenario involving intelligent vehicles and human vehicles in mixed traffic based on a spatiotemporal network, characterized in that, Includes the following steps: Acquire motion status data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment; Based on the motion state data and the environmental interaction data, a basic architecture for a discretized spatiotemporal network covering the target spatiotemporal domain is constructed. The lane-changing behavior of intelligent vehicles is modeled as a deterministic trajectory in the discretized spatiotemporal network, and the uncertain lane-changing behavior of manual vehicles is modeled as a probability cloud in the discretized spatiotemporal network. At the same time, a risk coefficient field is constructed. Establish the interaction constraints between the deterministic trajectory and the probability cloud, and transform the interaction constraints into mathematical constraints. Construct the objective function of the discretized spatiotemporal network, and obtain the final discretized spatiotemporal network based on the risk coefficient field; The optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle are obtained based on the final discretized spatiotemporal network.

2. The method for representing a mixed lane-changing scenario of intelligent vehicles and manual vehicles based on spatiotemporal networks according to claim 1, characterized in that, The motion state data includes: the vehicle's position, speed, and acceleration; The methods for obtaining the motion state data include: for intelligent vehicles, acquiring it through onboard sensors or vehicle-to-infrastructure (V2I) systems; for manual vehicles, acquiring it through roadside sensing devices. The environmental interaction data includes: lane line information, road boundaries, exit / forced lane change point locations, relative distances to surrounding vehicles, and relative speeds of surrounding vehicles.

3. The method for representing a scenario of intelligent vehicles and manual vehicles sharing lanes based on spatiotemporal networks according to claim 1, characterized in that, The methods for constructing the infrastructure of the discretized spatiotemporal network include: Define lane set L For a three-lane scenario L ={0,1,2}, representing the three lanes from the inside out; Each lane l ∈ L Divided into a continuous set of spatial nodes Sl ,node s ∈ Sl lane l The first s There are 1 spatial location unit, and the actual length of each spatial node is Δ. s ; The planning period is divided into a continuous set of time nodes. T The time node index is t ∈ T The time interval between adjacent time nodes is Δ t ; Define spatiotemporal resource nodes ( l , s , t ), indicating lane l Spatial nodes on s In time t Spatiotemporal resource units on the surface; Define spacetime arc ( l , s , t )→( l’ , s’ , t +1) indicates that the vehicle started from time t arrive t +1, from the lane l spatial nodes s Move to the lane l’ spatial nodes s’ The transfer process; Define the spatiotemporal node network as V ,in V ( l , s , t ) indicates the relationship between nodes ( l , s , t ) are adjacent, and from node ( l , s , t The spacetime nodes that can be reached from the starting point; According to the lane set L The set of spatial nodes Sl and the set of time nodes T The basic architecture for constructing the discretized spatiotemporal network ( L , Sl , T ).

4. The method for representing a mixed lane-changing scenario of intelligent vehicles and manual vehicles based on spatiotemporal networks according to claim 3, characterized in that, The methods for obtaining the deterministic trajectory include: For each intelligent vehicle, define the initial lane. Initial spatial nodes Target lane and the spatial node where the forced lane change point is located ; Introducing 0-1 deterministic trajectory variables xc ( l , s , t , l’ , s’ , t +1), indicating whether the intelligent vehicle is from time t arrive t +1 occupied from the lane l spatial nodes s to the lane l′ spatial nodes s′ The spacetime arc; Introducing 0-1 deterministic trajectory variables yc ( l , s , t This indicates whether the intelligent vehicle is occupying the lane. l Spatiotemporal resource nodes on l , s , t ); When an intelligent vehicle changes lanes, it simultaneously occupies the spatiotemporal resources of both the current lane and the target lane during the lane change process, which is represented as: in, lcurrent Indicates the lane the intelligent vehicle is currently in. ltraget Indicates the lane the intelligent vehicle is in after changing lanes. tstart Indicates the start time of lane change. tend Indicates the end time of the lane change.

5. The method for representing a scenario of intelligent vehicles and manual vehicles sharing lanes based on spatiotemporal networks according to claim 1, characterized in that, The methods for obtaining the probability cloud include: For each manual vehicle, its lane-changing behavior at each decision moment is described by a probabilistic model: in, Pstay This represents the probability of going straight. Pleft This represents the probability of changing lanes to the left. Pright This indicates the probability of changing lanes to the right. exist t At any time, manual vehicles occupy the lane. l Spatial nodes, in t At time +1, with probability Pstay Continue to occupy the lane l The subsequent spatial nodes, in probability Pleef Start moving to the left lane l -1 lane change, with probability Pright Start moving to the right lane l +1 lane change; When a manually operated vehicle changes lanes consecutively across multiple time steps, the spatiotemporal resource occupancy probabilities of subsequent lanes are superimposed to calculate the probability cloud. Ph ( l , s , t ).

6. The method for representing a mixed lane-changing scenario of intelligent vehicles and manual vehicles based on spatiotemporal networks according to claim 1, characterized in that, The interaction constraints include: Safety distance constraints: in, This indicates that the manual vehicle occupies a spatiotemporal node ( l , s’ , t The probability of ) dfront This indicates the number of nodes corresponding to the safe distance ahead. dback This indicates the number of nodes corresponding to the rear safety distance; Exclusive constraints on spatiotemporal resources: in, This indicates the time and space nodes occupied by all intelligent connected vehicles. l , s , t The occupancy status of ).

7. The method for representing a scenario of intelligent vehicles and manual vehicles sharing lanes based on spatiotemporal networks according to claim 1, characterized in that, The mathematical constraints include: flow balance constraints, forced lane-changing constraints, space resource occupancy constraints, lane-changing process constraints, and dynamic constraints.

8. The method for representing a mixed lane-changing scenario of intelligent vehicles and manual vehicles based on spatiotemporal networks according to claim 4, characterized in that, The objective function includes: the overall benefit function and the overall risk function; The overall benefit function is: in, Cl lane l Includes all intelligent connected vehicles. Indicating intelligent connected vehicles c Index of the target lane lane l The index of the last spatiotemporal node, ω LC represents the loss item for lane changing by intelligent connected vehicles. Ll lane l The adjacent lane, zc ( l , s , t , l’ (This refers to intelligent connected vehicles) c Is it in time? t From the lane l of s Node to Lane l’ Change lanes. ωa The coefficient representing the acceleration loss in the objective function. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωjerk This represents the coefficient representing the jerk loss in the objective function. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t longitudinal acceleration on ) ωE This represents the loss coefficient representing the gap between the objective function and the maximum throughput capacity. This represents the energy at the speed corresponding to the maximum throughput capacity of the current road segment. Indicating intelligent connected vehicles c At the spatiotemporal node ( l , s , t Longitudinal dynamic energy on ) The overall risk function is: in, E The set of spatiotemporal arcs representing a spatiotemporal network. r ( l , s , t , l’ , s’ , t +1) indicates that from the spatiotemporal node ( l , s , t ) to spacetime node ( l’ , s , t+1 On the spacetime arc of ), the risks posed by human drivers.

9. A system for representing lane-changing scenarios involving intelligent vehicles and human-operated vehicles based on spatiotemporal networks, wherein the system applies the method described in any one of claims 1-8, characterized in that... include: The module includes a data acquisition module, a network infrastructure construction module, a trajectory and probability modeling module, a constraint construction module, a network construction module, and a trajectory optimization module. The data acquisition module is used to acquire motion status data and environmental interaction data of all vehicles in the mixed traffic flow within the target road segment; The network infrastructure construction module constructs a discrete spatiotemporal network infrastructure covering the target spatiotemporal domain based on the motion state data and the environmental interaction data. The trajectory and probability modeling module is used to model the lane-changing behavior of the intelligent vehicle as a deterministic trajectory in the discretized spatiotemporal network, and to model the uncertain lane-changing behavior of the manual vehicle as a probability cloud in the discretized spatiotemporal network, while constructing a risk coefficient field. The constraint construction module is used to establish the interaction constraints between the deterministic trajectory and the probability cloud, and to transform the interaction constraints into mathematical constraints. The network construction module is used to construct the objective function of the discretized spatiotemporal network and obtain the final discretized spatiotemporal network based on the risk coefficient field. The trajectory optimization module obtains the optimal spatiotemporal trajectories of the intelligent vehicle and the manual vehicle based on the final discretized spatiotemporal network.

Images