Vehicle-cloud hierarchical decoupling and collaborative control method and system based on gravity-repulsion model

Abstract

The application belongs to the technical field of intelligent networked vehicle information physical system, and discloses a vehicle-cloud layered decoupling and collaborative control method and system based on a gravity-repulsion model. The application constructs a multi-layer collaborative architecture of "path decision-speed guidance-vehicle control-execution feedback", and abstracts multi-source heterogeneous traffic information into standardized digital objects. The upper layer decides the path based on the gravity-repulsion model, the gravity model represents the spatial benefit to the terminal point, and the repulsion model avoids the bottleneck by combining the instant congestion penalty and long-term recursive prediction. The lower layer CAV controller collects environmental data in real time, completes speed control and dynamic lane changing, and realizes the collaborative optimization of global reachability and local efficiency. Through layered decoupling and cross-domain integration, the application solves the problems of IVCPS function coupling and information heterogeneity. According to the SUMO simulation verification, the application can significantly shorten the driving time, improve the smoothness and lane changing safety, and is suitable for intelligent networked vehicles and vehicle-road cooperative systems in dynamic traffic scenarios.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent connected vehicles and cyber-physical systems (IVCPS), specifically relating to a vehicle-cloud hierarchical decoupling and cooperative control method and system based on a gravity-repulsion model. Background Technology

[0002] As intelligent connected vehicles, vehicle-road cooperative systems, and cyber-physical systems (IVCPS) continue to develop towards greater complexity and dynamism, IVCPS is no longer a simple superposition of single perception, single decision-making, or single control functions. Instead, it is gradually evolving into a complex functional system involving the coupling of multiple links such as environmental perception, state representation, path planning, speed guidance, behavior control, and execution feedback.

[0003] In dynamic traffic scenarios, vehicle operation is simultaneously affected by multiple factors such as road network topology, traffic signals, surrounding vehicles, road capacity, and task objectives. There are significant differences in time scale, information expression, and control granularity among different functional modules, which poses a great challenge to the organization, integration, and coordination of complex IVCPS functions.

[0004] Most existing research focuses on a single problem in path planning, longitudinal control, lateral lane changing, or signal coordination, lacking a unified approach that systematically analyzes the distribution, hierarchical relationships, and cross-domain integration patterns of complex functional modules from the perspective of the IVCPS multidimensional architecture. On the one hand, traffic signal status, preceding vehicle information, adjacent lane information, vehicle status, and road network information in dynamic scenarios are often heterogeneous in origin and dispersed in expression, making it difficult to form a standardized perception information state. This leads to inconsistent interfaces between upper-level decision-making and lower-level control, and poor module reusability. On the other hand, path decision-making, vehicle control, and speed guidance often adopt a serial splicing approach, lacking a layered decoupling mechanism oriented towards dynamic demand characteristics. This makes it difficult to reflect the collaborative patterns of IVCPS at different functional layers and to support the cross-domain integration design of basic functions. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a vehicle-cloud hierarchical decoupling and collaborative control method and system based on a gravity-repulsion model, which aims to solve the problems of high module coupling, inconsistent information expression, and difficulty in cross-domain integration in the existing complex function collaboration of IVCPS.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A vehicle-cloud hierarchical decoupling and cooperative control method based on a gravity-repulsion model includes the following steps:

[0008] S1. Standardization of Perception Information: Traffic signal status, distance to the vehicle in front, information of vehicles in adjacent lanes, vehicle operating status, road network topology and control output are uniformly abstracted into standardized digital objects to establish a unified information expression mechanism for IVCPS;

[0009] S2. Multi-layer collaborative architecture construction: Construct a multi-layer collaborative architecture of "path decision-speed guidance-vehicle control-execution feedback", with functions divided into "perception information state layer-path decision layer-speed guidance layer-lane change control layer-execution feedback layer". The upper-layer task decision and the lower-layer control execution are coupled collaboratively through standard interfaces.

[0010] S3. Gravity-Repulsion Path Decision: At each decision point, consider all candidate next path segments. Calculate the overall score The road segment with the highest score is selected as the driving direction. The formula for calculating the overall score is:

[0011]

[0012] In the formula, The attractive force moving towards the endpoint, calculated for the gravity model; The congestion resistance of the road segment calculated using the repulsion model;

[0013] S4. Local Control Coordination: During the path replanning interval, the lower-level CAV controller collects environmental data in real time, integrates traffic light status, preceding vehicle constraints, adjacent lane gaps and lane change benefits, completes target acceleration / speed calculation and dynamic lane change decision, and achieves coordinated optimization of global reachability and local operating efficiency.

[0014] S5. Cross-domain integrated verification: Integrate road network information, traffic facility information, vehicle operating status, control logic and simulation execution environment into the same digital closed loop, and establish mapping relationships and collaborative mechanisms between the perception domain, decision domain, control domain and execution domain.

[0015] Furthermore, the gravity model The calculation formula is:

[0016]

[0017] In the formula, For the current road section The shortest path distance to the destination. As a candidate for the next road segment The shortest path distance to the destination. and All were obtained through preprocessing using the inverse Dijkstra algorithm; This is expressed as the safe average speed for the next road segment;

[0018]

[0019] In the formula, This represents the current real-time average speed. The maximum speed limit for the road.

[0020] Furthermore, the repulsive force model includes instantaneous repulsive force and long-term repulsive force;

[0021] Instantaneous repulsive force:

[0022]

[0023] In the formula, The length of the road segment; This refers to the weighting coefficient for parking penalties. Number of parking spaces on the road section;

[0024] Long-term repulsive force:

[0025]

[0026] Where the boundary condition is d=0 or there is no downstream road segment, return. Ultimately adopted As the total repulsive force Substitute into the comprehensive scoring formula; To preset the maximum search depth; This is the discount factor.

[0027] This invention also provides a vehicle-cloud hierarchical decoupling and cooperative control system based on a gravity-repulsion model, comprising:

[0028] Standardized perception module: used to abstract multi-source heterogeneous traffic information into standardized digital objects and establish a unified information expression mechanism;

[0029] Multi-layer collaborative decision-making module: used to build the "path decision-vehicle speed guidance-vehicle control-execution feedback" architecture, execute the above gravity-repulsion path decision, and output driving direction commands;

[0030] Lower-level CAV control module: used to collect environmental data in real time, complete target acceleration / velocity calculation and dynamic lane change decision, and realize micro-level smooth control;

[0031] Cross-domain integration module: used to integrate the perception domain, decision domain, control domain and execution domain into the same digital closed loop, and establish inter-domain mapping and collaboration mechanisms.

[0032] Furthermore, the multi-layer collaborative decision-making module includes a preprocessing unit, which is used to execute the reverse Dijkstra algorithm with the destination as the source point, and pre-calculate the shortest path distance matrix from all road segments to the destination. .

[0033] Beneficial effects:

[0034] 1. Significantly reduced travel time: By using long-term repulsive force to detect static congestion or queues of vehicles ahead, vehicles can proactively detour around potential bottleneck sections, effectively reducing overall travel time.

[0035] 2. Significantly improved driving smoothness: The algorithm directly penalizes vehicles in queues on road sections, preventing vehicles from entering poor road conditions with frequent stops and starts. This can significantly reduce the number of stops along the way, reducing energy consumption and mechanical wear.

[0036] 3. Improved lane change safety and coordination efficiency: By working in conjunction with the lower-level CAV controller, vehicles can perform dynamic lane changes based on the safe distance between adjacent lanes when traveling at high speeds, thereby improving the overall vehicle-road coordination efficiency and driving safety.

[0037] 4. Enhanced Module Reusability and Interface Standardization: By standardizing the perception information state and unifying the interface model, the problem of heterogeneous information being expressed in a scattered manner is solved, and the reusability of upper-level decision-making and lower-level control modules is improved.

[0038] 5. Cross-domain integration and collaboration capability optimization: Through a layered decoupling mechanism and cross-domain integration design, efficient collaboration of multiple domain functions such as path planning, speed control, and lane change control is achieved, providing support for the flexible expansion and optimization of complex IVCPS functions.

[0039] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Detailed Implementation

[0040] To make the technical solutions, advantages, and objectives of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the protection scope of this application.

[0041] This invention provides a vehicle-cloud hierarchical decoupling and collaborative control method based on a gravity-repulsion model. The overall workflow is calculated in real time using a gravity-repulsion scoring formula, and specifically includes the following steps:

[0042] S1. Standardization of Perception Information: Traffic signal status, distance to the vehicle in front, information of vehicles in adjacent lanes, vehicle operating status, road network topology and control output are uniformly abstracted into standardized digital objects to establish a unified information expression mechanism for IVCPS;

[0043] S2. Multi-layer collaborative architecture construction: Construct a multi-layer collaborative architecture of "path decision-speed guidance-vehicle control-execution feedback", with functions divided into "perception information state layer-path decision layer-speed guidance layer-lane change control layer-execution feedback layer". The upper-layer task decision and the lower-layer control execution are coupled collaboratively through standard interfaces.

[0044] S3. Gravity-Repulsion Path Decision: At each decision point, consider all candidate next path segments. Calculate the overall score The road segment with the highest score is selected as the driving direction. The formula for calculating the overall score is:

[0045]

[0046] In the formula, The attractive force moving towards the endpoint, calculated for the gravity model; The congestion resistance of the road segment calculated using the repulsion model;

[0047] Gravity is used to measure the distance from the current road segment Pass through candidate road sections The gravitational pull towards the endpoint is essentially the spatial gain (an approximation of the time gain) obtainable per unit speed. (Gravity model) The calculation formula is:

[0048]

[0049] In the formula, For the current road section The shortest path distance to the destination. As a candidate for the next road segment The shortest path distance to the destination. and All were obtained through preprocessing using the inverse Dijkstra algorithm; This is expressed as the safe average speed for the next road segment;

[0050] The calculation of safe average speed incorporates the maximum speed limit of the road. To and current real-time average speed And a conservatism factor (e.g., 0.7) was set:

[0051]

[0052] In the formula, This represents the current real-time average speed. The maximum speed limit for the road.

[0053] The repulsive force model includes instantaneous repulsive force and long-term repulsive force;

[0054] Instantaneous repulsion: Calculated based on the travel time and the number of parked vehicles on the road segment, severely penalizing congested nodes;

[0055]

[0056] In the formula, The length of the road segment; The parking penalty weighting coefficient (recommended value is 2.0); Number of parking spaces on the road section;

[0057] Long-term repulsion: By recursively searching connected road segments within depth d, the minimum congestion resistance of future paths is obtained, and a discount factor is used. (Recommended value: 0.7) Attenuation;

[0058]

[0059] Where the boundary condition is d=0 or there is no downstream road segment, return. Ultimately adopted As the total repulsive force Substitute into the comprehensive scoring formula; This is the preset maximum search depth.

[0060] S4. Local Control Coordination: During the path replanning interval, the lower-level CAV controller collects environmental data in real time, integrates traffic light status, preceding vehicle constraints, adjacent lane gaps and lane change benefits, completes target acceleration / speed calculation and dynamic lane change decision, and achieves coordinated optimization of global reachability and local operating efficiency.

[0061] S5. Cross-domain integrated verification: Integrate road network information, traffic facility information, vehicle operating status, control logic and simulation execution environment into the same digital closed loop, and establish mapping relationships and collaborative mechanisms between the perception domain, decision domain, control domain and execution domain.

[0062] Collaborative control execution process:

[0063] Preprocessing stage: When the system starts, a directed graph of the road network is constructed, and the reverse Dijkstra algorithm is executed with the destination as the source point to calculate the static shortest distance matrix from all nodes to the destination.

[0064] Periodic path replanning (upper layer): During vehicle travel, at set time thresholds... (e.g., 10 seconds) triggers path replanning. The current vehicle location is obtained, the attraction and repulsion forces of all legal downstream candidate road segments are calculated, the highest-scoring road segment is determined, and the vehicle's route is updated.

[0065] High-frequency micro-control (lower layer): During the intervals between replanning cycles, the CAV controller collects environmental data in real time (including the status of traffic lights ahead, distance to the vehicle ahead, and the status of vehicles ahead in the left and right lanes). The speed controller calculates the target acceleration / velocity, and the lane change controller calculates the lane change intention. These commands are then fused and sent to the vehicle execution unit to ensure that the vehicle not only achieves optimal route selection in dynamic traffic flow but also exhibits smooth and conflict-free micro-behavior.

[0066] This embodiment also provides a vehicle-cloud hierarchical decoupling and cooperative control system based on a gravity-repulsion model, including:

[0067] Standardized perception module: used to abstract multi-source heterogeneous traffic information into standardized digital objects and establish a unified information expression mechanism;

[0068] Multi-layer collaborative decision-making module: This module constructs a "path decision-speed guidance-vehicle control-execution feedback" architecture, executes the aforementioned gravity-repulsion path decision-making, and outputs driving direction commands. The multi-layer collaborative decision-making module includes a preprocessing unit, which executes the reverse Dijkstra algorithm with the destination as the source point, pre-calculating the shortest path distance matrix from all road segments to the destination. ;

[0069] Lower-level CAV control module: used to collect environmental data in real time, complete target acceleration / velocity calculation and dynamic lane change decision, and realize micro-level smooth control;

[0070] Cross-domain integration module: used to integrate the perception domain, decision domain, control domain and execution domain into the same digital closed loop, and establish inter-domain mapping and collaboration mechanisms.

[0071] The innovative aspects of this application can be summarized as follows:

[0072] 1. A standardized decoupling method for perception information in dynamic traffic scenarios is proposed. The project abstracts traffic signal status, distance to the vehicle in front, information of vehicles in adjacent lanes, vehicle operating status, road network topology, and control output into standardized digital objects, establishes a unified information expression mechanism for IVCPS, solves the problems of difficulty in reusing multi-source heterogeneous information and inconsistent module interfaces, and provides a standardized foundation for the layered design of complex functions.

[0073] 2. A multi-layered collaborative architecture of "path decision-making, speed guidance, vehicle control, and execution feedback" is proposed. Unlike traditional methods that separate path planning and control execution, this project starts from the overall system collaboration, incorporating high-level global path decision-making, mid-level local speed guidance and behavior control, and low-level execution feedback into a unified framework. This enables collaborative operation at different time scales and functional granularities, reflecting the hierarchical organizational characteristics of IVCPS's complex functions.

[0074] 3. A layered decoupling method for complex functions based on a functional digital model is proposed. Instead of implementing system functions by stacking algorithms, the project divides functions into "perception information layer - path decision layer - speed guidance layer - lane change control layer - execution feedback layer". This allows upper-layer task decision-making and lower-layer control execution to be coupled collaboratively through standard interfaces, forming clear inter-layer relationships and decoupling logic, which can effectively support the expansion, replacement and reconstruction of complex functions.

[0075] 4. A collaborative mechanism for path selection and local control oriented towards dynamic demand characteristics is proposed. At the global path layer, the project adopts the gravity-repulsion coupling concept, incorporating destination approach benefits, real-time congestion status, parking penalties, and forward traffic conditions into path evaluation; at the local control layer, traffic light status, preceding vehicle constraints, adjacent lane gaps, and lane change benefits are comprehensively considered, achieving collaborative optimization between global accessibility and local operational efficiency.

[0076] 5. A cross-domain integration design concept for IVCPS basic functions is proposed. The project integrates road network information, traffic facility information, vehicle operating status, control logic, and simulation execution environment into the same digital closed loop, and establishes the mapping relationship and coordination mechanism between the perception domain, decision domain, control domain, and execution domain, providing an achievable and verifiable technical path for the cross-domain integration of IVCPS basic functions.

[0077] 6. Develop a collaborative verification prototype for complex dynamic scenarios. The project utilizes a simulation environment to achieve closed-loop operation of the entire process, including path replanning, speed adjustment, lane change execution, and effect evaluation. This not only verifies the effectiveness of the multi-layered collaborative mechanism but also provides experimental basis for subsequent design of complex IVCPS functional architecture, algorithm iteration, and system optimization.

[0078] It is hereby declared that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A vehicle-cloud hierarchical decoupling and cooperative control method based on a gravity-repulsion model, characterized in that, Includes the following steps: S1. Standardization of Perception Information: Traffic signal status, distance to the vehicle in front, information of vehicles in adjacent lanes, vehicle operating status, road network topology and control output are uniformly abstracted into standardized digital objects to establish a unified information expression mechanism for IVCPS; S2. Multi-layer collaborative architecture construction: Construct a multi-layer collaborative architecture of "path decision-speed guidance-vehicle control-execution feedback", with functions divided into "perception information state layer-path decision layer-speed guidance layer-lane change control layer-execution feedback layer". The upper-layer task decision and the lower-layer control execution are coupled collaboratively through standard interfaces. S3. Gravity-Repulsion Path Decision: At each decision point, consider all candidate next path segments. Calculate the overall score The road segment with the highest score is selected as the driving direction. The formula for calculating the overall score is: In the formula, The attractive force moving towards the endpoint, calculated for the gravity model; The congestion resistance of the road segment calculated using the repulsion model; S4. Local Control Coordination: During the path replanning interval, the lower-level CAV controller collects environmental data in real time, integrates traffic light status, preceding vehicle constraints, adjacent lane gaps and lane change benefits, completes target acceleration / speed calculation and dynamic lane change decision, and achieves coordinated optimization of global reachability and local operating efficiency. S5. Cross-domain integrated verification: Integrate road network information, traffic facility information, vehicle operating status, control logic and simulation execution environment into the same digital closed loop, and establish mapping relationships and collaborative mechanisms between the perception domain, decision domain, control domain and execution domain.

2. The vehicle-cloud hierarchical decoupling and cooperative control method based on a gravity-repulsion model according to claim 1, characterized in that, The gravity model The calculation formula is: In the formula, For the current road section The shortest path distance to the destination. As a candidate for the next road segment The shortest path distance to the destination. and All were obtained through preprocessing using the inverse Dijkstra algorithm; This is expressed as the safe average speed for the next road segment; In the formula, This represents the current real-time average speed. The maximum speed limit for the road.

3. The vehicle-cloud hierarchical decoupling and cooperative control method based on a gravity-repulsion model according to claim 2, characterized in that, The repulsive force model includes instantaneous repulsive force and long-term repulsive force; Instantaneous repulsive force: In the formula, The length of the road segment; This refers to the weighting coefficient for parking penalties. Number of parking spaces on the road section; Long-term repulsive force: Where the boundary condition is d=0 or there is no downstream road segment, return. Ultimately adopted As the total repulsive force Substitute into the comprehensive scoring formula; To preset the maximum search depth; This is the discount factor.

4. A vehicle-cloud hierarchical decoupling and cooperative control system based on a gravity-repulsion model, characterized in that, include: Standardized perception module: used to abstract multi-source heterogeneous traffic information into standardized digital objects and establish a unified information expression mechanism; Multi-layer collaborative decision-making module: used to construct the "path decision-vehicle speed guidance-vehicle control-execution feedback" architecture, execute the gravity-repulsion path decision as described in any one of claims 1-3, and output driving direction commands; Lower-level CAV control module: used to collect environmental data in real time, complete target acceleration / velocity calculation and dynamic lane change decision, and realize micro-level smooth control; Cross-domain integration module: used to integrate the perception domain, decision domain, control domain and execution domain into the same digital closed loop, and establish inter-domain mapping and collaboration mechanisms.

5. A vehicle-cloud hierarchical decoupling and cooperative control system based on a gravity-repulsion model according to claim 4, characterized in that, The multi-layer collaborative decision-making module includes a preprocessing unit, which is used to execute the reverse Dijkstra algorithm with the destination as the source point, and pre-calculate the shortest path distance matrix from all road segments to the destination. .

Images