Open type unmanned vehicle group dynamic evolution method based on perception contribution

Abstract

The invention relates to the field of unmanned driving, and provides an open type unmanned vehicle group dynamic evolution method based on perception contribution. Interference caused by interference factors around the unmanned vehicle to the perception contribution of the unmanned vehicle is defined; the influence of external interference factors on perception contributions among the unmanned vehicles and vehicle group dynamic evolution reasons caused by the external interference factors are analyzed; based on perception contribution, dynamic evolution behaviors such as amplification, reduction, combination, decomposition and extinction of the unmanned vehicle group are defined, and a corresponding evolution behavior processing method is provided. According to the method, a brand new thought is provided for realizing efficient perception cooperation of the unmanned vehicle group in a complex environment and improving the overall operation stability, and a theoretical foundation is laid for technical research and development and practical application in related fields. The method has important theoretical significance and practical value for promoting the breakthrough of an unmanned vehicle group perception cooperation technology and promoting unmanned large-scale application in an urban scene.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving, and more specifically to an open method for dynamic evolution of autonomous vehicle swarms based on perception contributions. Background Technology

[0002] The dynamic evolution of autonomous vehicle swarms refers to changes in the relationships between swarm members, requiring corresponding interventions. Researchers have conducted relevant studies on the dynamic evolution of autonomous vehicle swarms, including traditional methods and swarm intelligence-based methods.

[0003] Traditional dynamic evolution methods for autonomous vehicle swarms adjust platooning by centrally controlling vehicle acceleration, deceleration, and lane-changing behavior, leveraging close-range cooperative driving to reduce air resistance and improve traffic efficiency. Researchers have proposed several traditional dynamic evolution methods, including: an autonomous vehicle flow grouping method that indirectly guides human-driven vehicle behavior using autonomous vehicles, reducing stop-and-go phenomena in merging areas and improving traffic flow through grouped cooperative scheduling; a vehicle platooning protocol based on stochastic time series analysis to identify platoon formation and predict platoon splitting, reducing false alarm rates caused by road curvature through a two-step splitting prediction method; a collaborative governance-based autonomous vehicle dynamic evolution method that improves overall swarm mobility by shifting from isolated to cooperative autonomous driving paradigms, achieving comprehensive benefits such as reduced congestion, increased travel options and fairness, and reduced pollutant emissions; and a novel framework for determining appropriate platoon merging, lane-changing, and space reservation operations. However, such methods rely heavily on preset control logic and idealized environmental assumptions, and do not fully consider complex dynamic factors such as communication delays, traffic light regulation, and interference from mixed pedestrian and vehicle traffic in open scenarios. This makes the evolution of vehicle groups susceptible to external disturbances and difficult to maintain robustness and adaptability.

[0004] Swarm intelligence-based dynamic evolution methods for autonomous vehicle swarms generally rely on emergent behavior in biological systems. Researchers have proposed a modeling and analysis method for the dynamic evolution of autonomous vehicle swarms in a highway scenario. This method defines five vehicle states and their transition rules, constructs a state machine-based dynamic evolution framework for the swarm, manages continuous changes in the swarm topology through time slicing, and develops dynamic evolution detection and prediction methods, utilizing historical state sequences to identify evolutionary events such as swarm splitting and merging. However, this method only predicts dynamic evolution events without providing specific processing methods. Subsequent researchers, addressing the challenges of handling dynamic evolution events in open scenarios, have fully considered complex interference factors in urban environments, analyzed the causes of swarm dynamic evolution, abstracted dynamic evolution events, and designed corresponding dynamic evolution processing methods.

[0005] Swarm intelligence-based dynamic evolution methods endow vehicles with autonomous decision-making capabilities through multi-agent collaborative mechanisms, supporting self-organization and adjustment of vehicle groups during evolutionary events such as splitting and merging, significantly improving dynamic adaptability in complex environments. However, existing dynamic evolution methods in open scenarios do not incorporate differences in perception capabilities into the evolutionary rule design. Specifically, existing swarm intelligence-based dynamic evolution methods still suffer from the following problems: Existing research focuses solely on the impact of external disturbances on vehicle-group communication links, neglecting the role of environmental disturbances in the perceptual interaction relationships between autonomous vehicles. Inherent disturbances in open scenarios alter the nature of perceptual contribution relationships among vehicle group members, triggering dynamic evolutionary behavior within the group. Current dynamic evolution methods fail to establish a mapping relationship between "disturbance factors - perceptual interaction - vehicle group evolution," resulting in theoretical blind spots in these methods. Summary of the Invention

[0006] To address the issue that current dynamic evolution methods for autonomous vehicle swarms do not consider the impact of inherent interference and blind spots in open scenarios on the perception and interaction among swarm members, this invention analyzes the influence of external interference factors on the perception contributions among autonomous vehicles and the resulting reasons for the dynamic evolution of the swarm. Based on perception contributions, dynamic evolution behaviors such as expansion, reduction, merger, decomposition, and extinction of autonomous vehicle swarms are defined, and their impact on the contribution degree, persistence, accessibility, and real-time nature of perception contributions among members is analyzed. Corresponding dynamic evolution behavior processing methods are proposed, providing effective guarantees for the stable and orderly dynamic evolution behavior of autonomous vehicle swarms in open scenarios.

[0007] The technical solution of this invention: The open-ended dynamic evolution method for autonomous vehicle swarms based on perception contributions includes the following steps: Step 1. Define the interference caused by ambient disturbances around the autonomous vehicle to its perception. Step 2. Analysis of the reasons for the dynamic evolution of the autonomous vehicle swarm; Step 3. Based on perception contributions, define the dynamic evolutionary behavior of the autonomous vehicle swarm, including expansion, contraction, merger, decomposition, and extinction. Analyze the impact of dynamic evolutionary behavior on the contribution degree, persistence, accessibility, and real-time performance of perception contributions among members, and propose corresponding methods for handling dynamic evolutionary behavior, including the following steps: Step 3.1 Design and handling method for augmented events in autonomous vehicle swarms; Step 3.2: Event reduction design and handling methods for autonomous vehicle swarms; Step 3.3 Design and handling method for parallel events of autonomous vehicle swarms; Step 3.4 Decomposition event design and processing method for autonomous vehicle swarm; Step 3.5 Design of the event of the demise of the autonomous vehicle swarm.

[0008] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses a dynamic evolution method for autonomous vehicle swarms based on perception contributions in open scenarios, providing a novel approach for achieving efficient perception collaboration and improving overall operational stability in complex environments. It lays a theoretical foundation for technological development and practical applications in related fields. This invention has significant theoretical and practical value in promoting breakthroughs in perception collaboration technology for autonomous vehicle swarms and facilitating the large-scale application of autonomous driving in urban scenarios. Attached Figure Description

[0009] Figure 1 This is a flowchart of an embodiment of the present invention.

[0010] Figure 2 An event diagram is added to this invention.

[0011] Figure 3 This is a schematic diagram of the emergent events of the present invention.

[0012] Figure 4 This is a schematic diagram illustrating the event reduction method of the present invention.

[0013] Figure 5 This is a schematic diagram illustrating the event detachment of the present invention.

[0014] Figure 6 This is a schematic diagram of the migration event of the present invention.

[0015] Figure 7 This is a schematic diagram of the merger and acquisition event of the present invention.

[0016] Figure 8 This is a schematic diagram illustrating the decomposition of events in this invention.

[0017] Figure 9 This is a schematic diagram of the simulation scenario of the present invention.

[0018] Figure 10 The diagram shows the relationship between ACTS and maximum vehicle speed when n=200 and n=300.

[0019] Figure 11 The diagram shows the relationship between ACTS and maximum vehicle speed when n=400 and n=500.

[0020] Figure 12 This is a schematic diagram showing the relationship between ACTS and maximum vehicle speed when n=600.

[0021] Figure 13 The diagram shows the ACS distribution of the CPDE, UDE, and SCDE algorithms in each experiment during a simulation period of 50 to 250 seconds.

[0022] Figure 14 The diagram shows the ACS distribution of the CPDE, UDE, and SCDE algorithms during a simulation period of 50 to 250 seconds.

[0023] Figure 15 This is a schematic diagram showing the relationship between APTS and maximum vehicle speed when n=200 and n=300.

[0024] Figure 16 This is a schematic diagram showing the relationship between APTS and maximum vehicle speed when n=400 and n=500.

[0025] Figure 17 This is a schematic diagram showing the relationship between APTS and maximum vehicle speed when n=600.

[0026] Figure 18 The diagram shows the APS distribution of the CPDE, UDE, and SCDE algorithms in each experiment during a simulation period of 50 to 250 seconds.

[0027] Figure 19 The diagram shows the APS distribution of the CPDE, UDE, and SCDE algorithms during a simulation period of 50 to 250 seconds.

[0028] Figure 20 The diagram shows the relationship between AATS and maximum vehicle speed when n=200 and n=300.

[0029] Figure 21 The diagram shows the relationship between AATS and maximum vehicle speed when n=400 and n=500.

[0030] Figure 22 This is a schematic diagram showing the relationship between AATS and the maximum vehicle speed when n=600.

[0031] Figure 23 The diagram shows the AAS distribution of the CPDE, UDE, and SCDE algorithms in each experiment during a simulation period of 50 to 250 seconds.

[0032] Figure 24 This is a schematic diagram showing the AAS distribution of the CPDE, UDE, and SCDE algorithms during a simulation period of 50 to 250 seconds.

[0033] Figure 25 This is a schematic diagram showing the relationship between ATTS and maximum vehicle speed when n=200 and n=300.

[0034] Figure 26 This is a schematic diagram showing the relationship between ATTS and maximum vehicle speed when n=400 and n=500.

[0035] Figure 27This is a schematic diagram showing the relationship between ATTS and maximum vehicle speed when n=600.

[0036] Figure 28 The diagram shows the AAS distribution of CPDE, UDE, and SCDE in each experiment during a simulation period of 50 to 250 seconds.

[0037] Figure 29 The diagram shows the ATS distribution of the CPDE, UDE, and SCDE algorithms during a simulation period of 50 to 250 seconds.

[0038] Figure 30 The relationship between APB and maximum vehicle speed when n=200 and n=300.

[0039] Figure 31 The relationship between APB and maximum vehicle speed when n=400 and n=500.

[0040] Figure 32 The relationship between APB and maximum vehicle speed when n=600. Detailed Implementation

[0041] The technical solutions provided in this application will be further described below with reference to specific embodiments and accompanying drawings. The advantages and features of this application will become clearer from the following description.

[0042] like Figure 1 As shown, the embodiments of the present invention specifically include the following four steps: Step 1: Relevant definitions; Step 2: Analysis of the reasons for the dynamic evolution of the autonomous vehicle swarm, including: Step 2.1: Analysis of the impact of external interference factors on the contribution of perception; Step 2.2: Analysis of the reasons for the dynamic evolution of the vehicle group; Step 3: Construct a dynamic evolution method for autonomous vehicle swarms based on perception contributions, including: Step 3.1: Design and handling methods for expanded events in autonomous vehicle swarms; Step 3.2: Event reduction design and handling methods for autonomous vehicle swarms; Step 3.3: Design and handling methods for concurrent events in autonomous vehicle swarms; Step 3.4: Decomposition event design and handling method for autonomous vehicle swarms; Step 3.5: Design of the event of the demise of the autonomous vehicle swarm; Step 3.6: Dynamic evolution algorithm for autonomous vehicle swarm based on perception contribution.

[0043] Step 4: Simulation experiment verification.

[0044] Details are as follows: Step 1: Relevant Definitions driverless vehicles and The interference caused by surrounding disturbances (human-driven vehicles, roadside obstacles, traffic lights, and pedestrians, etc.) to its perception is as follows: (1) In the formula, Indicates driverless vehicles and The interference from surrounding factors on the device itself is calculated as follows: (2) in, and They represent and Driverless vehicles at all times and The number of invalid sensing squares caused by interference, which can be represented as: (3) Indicates driverless vehicles and The interference of surrounding disturbance factors on the perceived contribution relationship between them is calculated as follows: (4) in, (5) in, , , , , and Representing time respectively driverless vehicles and The contribution of the perceived contribution includes the probability of interruption, the transmission throughput, the reliability, the latency, the efficiency, and the maximum contribution time.

[0045] Step 2: Analysis of the Reasons for the Dynamic Evolution of the Autonomous Vehicle Swarm Step 2.1 Analysis of the impact of external interference factors on the contribution of perception In open scenarios, autonomous vehicle swarms undergo dynamic evolution due to interference from human-driven vehicles, roadside obstacles, traffic lights, and pedestrians. The perception contributions of the autonomous vehicle swarm change due to these external disturbances, thus altering the topology of collaborative perception within the swarm. To analyze the dynamic evolution of autonomous vehicle swarms, it is necessary to examine the impact of these external disturbances on the contribution, persistence, accessibility, and real-time nature of the perception contribution relationships among swarm members, which can be represented as: 1) ,in Indicates driverless vehicles and Contribution degree of perceived contribution relationship between them It decreases over time.

[0046] 2) ,in , , These represent driverless vehicles. and The persistence of perceived contribution relationship between them Accessibility Real-time It decreases over time.

[0047] Step 2.2 Reasons for the dynamic evolution of the vehicle group Due to changes in the external environment surrounding the autonomous vehicle swarm, the contribution, persistence, reachability, and real-time nature of the perception contributions between vehicle nodes are not zero, resulting in perception contribution relationships and causing the swarm to evolve dynamically. When autonomous vehicles... and When subjected to external disturbances, the contribution of the perception contribution relationship between them decreases, and at least one of the indicators such as persistence, accessibility, or real-time performance decreases. That is, when the autonomous vehicle swarm is affected by external disturbances in an open scenario, the contribution, persistence, accessibility, and real-time performance of the perception contribution between its member nodes decrease. When one or more of them drop to zero, the perception contribution relationship between the vehicle nodes will be broken, leading to dynamic evolution of the vehicle swarm.

[0048] Step 3: Construct a dynamic evolution method for autonomous vehicle swarms based on perception contributions. Step 3.1 Expansion of the autonomous vehicle fleet The expansion of an autonomous vehicle swarm refers to the emergence of new perceptual contributions among swarm members due to changes in the external environment surrounding the swarm. This could be due to the appearance of new nodes or new perceptual contributions among existing swarm members, which are respectively referred to as addition events and emergence events in the autonomous vehicle swarm.

[0049] (1) Amplification event Augmentation events include addition events and emergence events.

[0050] Add events: such as Figure 2 As shown, an "addition event" refers to the emergence of a detached node as a member of an autonomous vehicle swarm in an open scenario. These detached nodes can effectively interact with one or more members of the swarm through perceptual contributions; they may be newly emerging nodes or nodes that have left other autonomous vehicle swarms. An addition event is represented as: (6) in, This represents a swarm of driverless vehicles. Indicating a fleet of self-driving cars An additional event is in progress. It is an unmanned vehicle The neighbor's vehicles gathered. and They represent and The direction of travel, express and There was no interference from external factors. Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between them Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time performance of the perceived contribution relationship between nodes are all not equal to zero, if the free node... If the above conditions are met, the vehicle will be added to the vehicle group. middle.

[0051] Emergent events: such as Figure 3 As shown, an emergent event refers to the emergence of new perceptual contribution relationships among members of a vehicle corps in an open scenario. These vehicle corps members already have perceptual contribution relationships with one or more other members within the corps, and simultaneously develop perceptual contribution relationships with other members of the corps. An emergent event is represented as: (7) in, Indicating a fleet of self-driving cars An emergent event occurred. Indicating a fleet of self-driving cars The set of members, Indicates to self-driving vehicles Establish a set of vehicle group members who contribute to perception. express and The difference set.

[0052] (2) Amplification method The method for expanding the autonomous vehicle swarm is shown in Algorithm 1, and is described in detail below: (1) When When an additional event occurs: 1) When At that time, free nodes Broadcast join request JR (lines 21-24); if Received from car group members Sending message A will join the autonomous vehicle swarm. Then, an update message U is sent to all vehicle group members (this message includes updated vehicle group member information and perception contribution information between members), and all vehicle group members update node information and perception contribution relationships between vehicle group members (lines 17-21).

[0053] 2) When At that time, members of the vehicle group Received free node When sending a join request (JR), if the autonomous vehicle node... and If they travel in the same direction and there is zero external interference between them, then Will Add to candidate node set . For sets Calculate all elements , , , The optimal candidate vehicle set is obtained using multi-objective optimization theory. If the maximum number of vehicle group members has not been reached at this time, and , , , If not equal to zero, then Towards The mid-range node sends a message allowing it to join the vehicle group. A(Agree) (This message includes updated vehicle group node information and perception contribution information among members.) (Lines 3-16) (2) When Emergent events occur: 1) If a fleet of self-driving cars members Received from another member The establishment of perceived contribution requests (CPRs) involves requests where the degree of contribution, persistence, reachability, and real-time nature of these requests are not equal to zero. ， Towards Send A. All vehicle group members update node information and perception contribution relationships. (Lines 25-34) Step 3.2 Reduction of the autonomous vehicle fleet The reduction of an autonomous vehicle corps refers to the disruption of the perception contribution relationships among its members due to external interference. A reduction event occurs when one or more perception contribution relationships of a vehicle node are disrupted; a disengagement event occurs when a vehicle node loses its perception contribution relationships with all other corps members; and a migration event occurs when a vehicle node can only obtain valid perception information from other corps members but can no longer contribute valid perception information to the corps.

[0054] (1) Reduced events Reduction events include elimination events, detachment events, and migration events.

[0055] 1) Event reduction like Figure 4 As shown, a reduction event refers to an event in an open scenario where, due to external interference, at least one of the following factors—contribution, persistence, reachability, and real-time performance—of a vehicle group's perceived contribution decreases to zero, preventing some vehicle group members from maintaining their perceived contribution relationship. A reduction event is represented as: (8) in, Indicating a fleet of self-driving cars A reduction event is underway. express and The interaction was affected by external factors. Indicates driverless vehicles and At least one of the following factors—contribution, persistence, accessibility, and real-time performance—is zero. express and It was not affected by external factors. Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between them Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time nature of the perceived contribution relationship between them are all not equal to zero.

[0056] 2) Escape Event like Figure 5 As shown, a disengagement event refers to an event in an open scenario where an autonomous vehicle swarm is affected by external disturbances, causing at least one of the following factors—contribution, persistence, reachability, or real-time performance—of a vehicle node and any other node in the swarm to drop to zero. The vehicle node can no longer contribute effective perception information to other swarm members, nor can it obtain the effective perception information it needs from other swarm members. A disengagement event can be represented as: (9) in, Indicating a fleet of self-driving cars An detachment event is in progress.

[0057] 3) Migration events like Figure 6 As shown, a migration event refers to a situation in an open scenario where one or more vehicle nodes within a vehicle group can only obtain but not contribute effective perception information from other members of the group. If they can obtain and contribute effective perception information from other members of neighboring vehicle groups, they will actively leave the current vehicle group and attempt to join another. The migration event is represented as: (10) in, Indicating a fleet of self-driving cars A migration event is underway. Indicates driverless vehicles It is unable to provide effective perception information to other members of the vehicle group. Indicates driverless vehicles . contributions. Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between them Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time nature of the perceived contribution relationship between them are all not equal to zero.

[0058] (2) Reduction method The method for reducing the number of autonomous vehicles is shown in Algorithm 2, and is described in detail below: 1) When At that time, a reduction event occurs: Autonomous vehicle node and The perceived contribution relationship between them is broken. Send a reduction message V to all members of the vehicle group, and the perception contribution relationship between all vehicle group member nodes will be updated accordingly. (Lines 4-8, 10-11) 2) When At that time, an escape event occurs: Autonomous vehicle node After leaving the car group, it becomes a free node, broadcasts the message "JR" to its surroundings, and continues to attempt to rejoin other car groups. (Lines 13-18) 3) When When a migration event occurs: Autonomous vehicle node Upon discovering that its contribution to the rest of the vehicle group is 0, the vehicle group node broadcasts the message JR (lines 19-21) to its surroundings. Received free node When sending a JR, Algorithm 1 is used to determine whether to agree. If agreeing, the request is sent to the node. Send message A allowing entry into the vehicle group; if receive Message A sent, Will voluntarily withdraw from the vehicle group And join the car group and towards the group of vehicles Members send message L to migrate the vehicle group, vehicle group Members update node information and the relationship of perceived contribution between nodes. (Lines 22-28) Step 3.3 Merging of the autonomous vehicle swarm The merging of autonomous vehicle swarms refers to the process where two or more vehicle swarms that are geographically close and exhibit similar behavior establish new perception contribution relationships between their vehicle nodes, thus forming a new swarm. A merging event occurs when a vehicle node in one swarm receives a request from a vehicle node in another swarm to establish a new perception contribution relationship.

[0059] (1) Conjunction event like Figure 7 As shown, a concatenation event refers to an open scenario where vehicle nodes within a vehicle group interact with other members of the same vehicle group through perceptual contributions, resulting in the concatenation of the perceptual contribution relationships between these vehicle groups to form a new vehicle group. A concatenation event is represented as: (11) in, Indicating a fleet of self-driving cars A merging event is occurring. Indicating a fleet of self-driving cars , , It is a set of candidate vehicle groups to be merged.

[0060] (2) Co-construction method The method for constructing a swarm of autonomous vehicles is shown in Algorithm 3, and is described in detail below: 1) Autonomous vehicle swarm Members of the car group The broadcast vehicle group merge request (MR) message carries... Occlusion level of all members. (Lines 1-2) 2) When the swarm of driverless cars vehicle node Received from When performing MR, first determine and After merging the vehicle groups, determine if the group has reached its maximum size. If the merged group has not reached its maximum size, and the two nodes are traveling in the same direction, then the vehicle group will be... Join the vehicle consolidation group It also calculates and constructs the contribution of the perception contribution of the preceding and following vehicle groups, as well as the changes in persistence, accessibility, and real-time performance; otherwise, it sends a request to the node. Send a message rejecting the merger. (Lines 3-9) 3) Based on Pareto optimality theory The autonomous vehicle swarm within the set is sorted using a non-dominated method. Then proceed in sequence to The autonomous vehicle swarm sends an agreement and formation message (AM) (which includes information about the merged vehicle nodes and perception contributions among members) until the swarm reaches its maximum size. (Lines 10-12) 4) If Received from consent and constitute a message, All nodes in the [database name] have been added. A success message is sent to all members of the vehicle group, and all members update their vehicle group node information and perception contribution information. (Lines 12-17) Step 3.4 Decomposition of the autonomous vehicle swarm The decomposition of an autonomous vehicle swarm refers to the process by which a swarm of vehicles is broken down into two or more swarms due to external disturbances. A decomposition event occurs when there is no perception contribution relationship between the different parts of the swarm, but there is still a perception contribution relationship between the individual vehicle nodes within the swarm.

[0061] (1) Decompose the event like Figure 8 As shown, a decomposition event refers to a situation in an open scenario where, due to external interference, at least one of the contribution, persistence, reachability, and real-time performance of vehicle nodes within the vehicle group drops to zero, preventing the maintenance of the perception contribution relationship. The decomposition event is represented as: (12) in, Indicating a fleet of self-driving cars Decomposition is in progress. , Indicates from The set of autonomous vehicles decomposed from the data. This represents the set of remaining vehicles.

[0062] (2) Decomposition method The decomposition method for the autonomous vehicle swarm is shown in Algorithm 4, and is described in detail below: When self-driving vehicles and The perception contribution relationship between them is broken and is the set of remaining vehicles. and When the last pair of perception contribution relationships between members is established, the autonomous vehicle swarm decomposition event will occur (lines 1-6). Vehicle node Remove the set from the vehicle group members The vehicles in the middle, To become a new swarm of driverless vehicles (lines 7-8).

[0063] Step 3.5 The Demise of Autonomous Vehicle Swarms The demise of an autonomous vehicle swarm refers to the absence of any perception contribution relationships within the swarm. A demise event occurs when the last remaining perception contribution relationship within the swarm disappears.

[0064] A disappearance event refers to an event in an open scenario where external interference causes the perception contribution, persistence, reachability, and real-time performance of vehicles within a vehicle group to drop to zero, resulting in the disappearance of perception contribution relationships within the group. The disappearance event is represented as follows: (13) in, This indicates that the swarm of self-driving cars is experiencing a demise.

[0065] Step 3.6 Dynamic Evolution Algorithm for Autonomous Vehicle Swarm Based on Perception Contribution The overall process of the Contributed Perception Autonomous Vehicle Group Dynamic Evolution Method (CPDE) is shown in Algorithm 5, and is described in detail below: (1) When the group of driverless cars When an addition or emergence event occurs, execute Algorithm 1; (lines 1-3) (2) When the group of driverless cars When a reduction, detachment, or migration event occurs, execute Algorithm 2; (lines 4-6) (3) When the group of driverless vehicles When a concatenation event occurs, execute Algorithm 3; (lines 7-10) (4) When the group of driverless vehicles When a decomposition event occurs, Algorithm 4 is executed. (Lines 11-13) Step 4: Simulation experiment verification.

[0066] 4.1 Simulation Experiment Setup (1) Simulation Experiment Environment: The simulation experiment was conducted on a high-performance server. The specific parameters of the simulation experiment environment are shown in Table 1. Among them, the Simulation of Urban Mobility (SUMO) was used to generate road topology in the open environment, construct interference factors in the open environment, and extract the trajectory of unmanned vehicles; Network Simulator 3 (NS-3) was used to simulate the communication between unmanned vehicles in the open scene during the simulation.

[0067] Table 1 Simulation Experiment Environment Parameters (2) Simulation Experiment Scenario: First, a typical traffic area in Jiading District, Shanghai, was used as the experimental scenario. A simulated road network of 3.7 km × 7.8 km (geographic coordinate range: 31.25°N–31.29°N, 121.16°E–121.25°E) was constructed based on OpenStreetMap open-source geographic data. The specific road network structure is as follows: Figure 9 As shown, the experimental area comprises a multi-level road system, encompassing 203 road elements including expressways (Beijing-Shanghai Expressway and Shenyang-Haikou Expressway), national highways, provincial highways, and county roads, and includes 127 signalized intersections. The area boundary is formed by the Beijing-Shanghai Expressway, Shenyang-Haikou Expressway, Cao'an Road, and Moyu South Road, creating an experimental environment with typical urban-intercity mixed traffic characteristics. To simulate a realistic open-world scenario, 10% of the vehicles were set up as manned vehicles. In addition, 25 buses, 50 non-motorized vehicles, 100 pedestrians, and 50 randomly placed roadside obstacles were also present in the area. The randomtrip script in SUMO was used to generate the routes for all vehicles, and the Krauss and MOBIL models were used to model vehicle behavior, simulating vehicle following and lane-changing behaviors, respectively.

[0068] (3) Communication Simulation Setup: A high-fidelity vehicle-to-everything (V2X) communication system based on NS-3 was constructed for the simulation experiment. WifiNet network interface devices supporting the IEEE 802.11p protocol were deployed for all vehicle nodes. The multipath fading effect in an open environment was simulated using the YansWifiChannel channel model. A logarithmic distance path loss model was adopted.

[131] A path loss exponent n=2.0 and a random propagation delay model (delay range 50-200 ms, variance 50 ms) characterize the signal attenuation law and multipath delay features, respectively. Combined with a 20 dBm transmit power and a 10 MHz channel bandwidth, an effective communication range of 300 meters is achieved. The physical layer is set to a maximum sensing distance of 150 meters, constructing a 15×15 sensing range grid. The vehicle communication distance is set to 250m, and a 1-second periodic beacon broadcast and CSMA / CA contention mechanism are configured.

[0069] 4.2 Evaluation Indicators (1) ACS / ACTS: Average Contribution per Second (ACS) is the sum of the contributions of vehicle swarm members per second divided by the total number of vehicles in the simulation scenario at the current moment; Average Contribution in Total Simulation (ACTS) is the sum of ACS over all simulation seconds and the ratio of the total simulation time. ACS and ACTS are used to measure the effectiveness of autonomous vehicle swarm members in providing effective perception information. The higher the ACS and ACTS values, the more valuable the perception information contributed by autonomous vehicle swarm members.

[0070] (2) APS / APTS: Average Persistence per Second (APS) is the sum of the persistence of vehicle swarm members per second divided by the total number of vehicles in the simulation scenario at the current moment; Average Persistence in Total Simulation (APTS) is the sum of APS over all simulation seconds and the ratio of the total simulation time. APS and APTS are used to measure the ability of autonomous vehicle swarm members to continuously contribute effective perception information. The higher the APS and APTS values, the stronger the ability of autonomous vehicle swarm members to continuously contribute perception information.

[0071] (3) AAS / AATS: Average Accessibility per Seconds (AAS) is the sum of the accessibility among vehicle group members per second divided by the total number of vehicles in the simulation scenario at the current time; Average Accessibility in Total Simulation (AATS) is the sum of AAS over all simulation seconds and the ratio of the total simulation time. AAS and AATS are used to measure the reliability of the perception information links contributed by autonomous vehicle group members. The higher the AAS and AATS values, the more reliable the perception information links contributed by autonomous vehicle group members.

[0072] (4) ATS / ATTS: Average Timeliness per Second (ATS) is the sum of the real-time performance of all vehicle swarm members per second divided by the total number of vehicles in the simulation scenario at the current moment; Average Timeliness in Total Simulation (ATTS) is the sum of the ATS over all simulation seconds and the ratio of the total simulation time. ATS and ATTS are used to measure the speed at which members of the autonomous vehicle swarm contribute perception information. The higher the ATS and ATTS values, the faster the members of the autonomous vehicle swarm contribute perception information.

[0073] (5) AP: Average Perception Boundary (AP) is the sum of the perception boundaries of the autonomous vehicle group in the simulation scene at the current moment divided by the total number of vehicles in the group at the current moment.

[0074] 4.3 Simulation Experiment Results and Analysis The most advanced methods in the dynamic evolution of autonomous vehicle swarms were selected as benchmarks for evaluating the perception-sharing-based dynamic evolution method (CPDE) of this invention: the sidechain consensus-based dynamic evolution method (SCDE) and the urban scenario-based dynamic evolution method (UDE). SCDE is the latest distributed dynamic evolution method for autonomous vehicle swarms in highway scenarios, while UDE is the latest method for the open scenario dynamic evolution of semi-centralized autonomous vehicle swarms.

[0075] (1) Ability to provide effective perception information: Multiple sets of comparative experiments were conducted throughout the simulation process. The maximum number of vehicles and the maximum vehicle speed in each set of simulation experiments were different. The maximum number of vehicles was set to 200, 300, 400, 500 and 600 respectively; the maximum vehicle speed was 40, 50, 60, 70 and 80 km / h. Figures 10-12 The relationship between the average vehicle swarm contribution and the maximum vehicle speed for CPDE, UDE, and SCDE methods under different maximum vehicle numbers is described. Experimental results show that the average vehicle swarm contribution increases with the maximum number of vehicles. This is mainly because the increase in the number of autonomous vehicles in the scenario leads to more vehicles being able to share effective perception information. In addition, the increase in the maximum number of vehicles increases the probability of dynamic evolution events such as addition, emergence, and merger, resulting in a larger autonomous vehicle swarm in the simulation experiments.

[0076] Except for the special cases of a maximum number of vehicles of 300 and a maximum vehicle speed of 50 and a maximum number of vehicles of 500 and a maximum vehicle speed of 60, the remaining results show that ACTS decreases with increasing maximum vehicle speed. This decreasing trend is attributed to the fact that increased vehicle mobility reduces the duration of effective perception information in vehicle interactions, thereby reducing the total amount of effective perception information contributing throughout the simulation. Notably, compared to UDE and SCDE methods, CPDE exhibits the highest ACTS, with average ACTS values ​​of 1.76, 2.10, 2.26, 2.54, and 2.47 for different maximum vehicle numbers and maximum vehicle speeds of 40–80, respectively; UDE's average ACTS values ​​for the same range are 1.33, 1.58, 1.87, 2.11, and 1.86; while SCDE's are 0.48, 0.65, 0.85, 1.14, and 1.42. Compared to UDE, CPDE improved ACTS by 24%, 25%, 18%, 17%, and 25% at different maximum vehicle counts; the improvement was even more significant compared to SCDE. This indicates that CPDE-based autonomous vehicle swarms contributed the most effective perception information to each other compared to UDE and SCDE-based swarms. This advantage can be attributed to CPDE considering the impact of perception information between autonomous vehicles on the perception contribution chain during dynamic evolution, while SCDE and UDE only focus on whether the perception contribution chain is smooth and efficient, but do not consider whether the chain can contribute valuable perception information.

[0077] Figure 13 Figures (a)-(c) show the distribution of ACS for each simulation experiment over a period of 50 to 250 seconds using scatter plots. The horizontal axis represents simulation time, and the vertical axis represents ACS. It can be seen that the CPDE method has higher maximum, minimum, and mean ACS values ​​compared to the SCDE and UDE methods. CPDE consistently exhibits higher ACS values ​​compared to SCDE and UDE, with its average value significantly higher. Furthermore, CPDE's lowest ACS value per simulation second in each simulation experiment is also higher than 1, while UDE's is approximately 0.6, and SCDE's is only slightly higher than 0.

[0078] Figure 14(a) shows the ACS distribution of the three methods in the form of a box plot. Figure 14 Figures (b) and (c) show the results of two comparative experiments, in which Indicates the maximum number of vehicles. The maximum vehicle speed is represented, revealing the relationship between ACS and simulation time during dynamic evolution. It can be seen that the ACS of the CPDE method remains stable at approximately 3.6 throughout the simulation time, while the ACS of the UDE and SCDE methods are lower than that of the CPDE method. These findings collectively demonstrate the robustness of CPDE to external disturbances, enabling the vehicle group to contribute effective perception information during dynamic evolution.

[0079] (2) The ability to continuously contribute effective perceived information: Figures 15 to 17 The relationship between APTS and maximum vehicle speed for CPDE, UDE, and SCDE under different maximum vehicle numbers is shown. Experimental results show that in all three methods, the APTS at a maximum vehicle speed of 40 is higher than that at a maximum vehicle speed of 80. This trend is because as the maximum vehicle speed increases, the mobility of autonomous vehicles in the simulation scenario also increases, potentially leading to a shorter duration for autonomous vehicles to contribute perception information, thus reducing APTS. Furthermore, it can be seen that APTS decreases with increasing maximum vehicle number. This is attributed to the fact that the total number of vehicles in the simulation scenario increases with the maximum vehicle number. The increased number of autonomous vehicles leads to more frequent dynamic evolution events, thereby shortening the time for autonomous vehicles to continuously contribute perception information. In all scenarios with different maximum vehicle numbers, CPDE consistently exhibits a higher APTS compared to the other two comparison methods. Compared to UDE and SCDE methods, the average APTS for maximum vehicle speeds of 40-80 km / h under different maximum vehicle numbers were 8.71, 6.01, 6.03, 4.33, and 4.22, respectively; the average APTS for UDE at maximum vehicle speeds of 40-80 km / h were 5.95, 4.49, 4.24, 3.68, and 3.67, respectively; while those for SCDE were 2.23, 2.28, 2.27, 2.44, and 2.57. CPDE improved APTS by 32%, 25%, 30%, 15%, and 13% compared to UDE under different maximum vehicle numbers; and the improvement compared to SCDE was even more significant. This phenomenon indicates that CPDE can maintain effective perception information interaction for a longer period of time.

[0080] Figure 18 Figures (a)-(c) show the APS distribution of these three methods over a period of 50 to 250 seconds, respectively. Figure 19 (a) uses a box plot to represent these results. It can be seen that the CPDE method has higher maximum and minimum APS and mean APS compared to the SCDE and UDE methods. Figure 19 Figures (b) and (c) show the results of two separate simulation runs, in which Indicates the maximum number of vehicles. This represents the maximum speed of the vehicle. Compared to the other two comparison methods, CPDE exhibits a significantly higher APS. CPDE's minimum APS consistently exceeds 2.5 throughout the simulation, while the minimum APS of the other two methods is around 1. Furthermore, compared to UDE and SCDE, the APS value of the autonomous vehicle swarm based on the CPDE method fluctuates the least within the simulation timeframe of 50-250 seconds. This is because the CPDE method demonstrates high stability in the duration of the perception contribution transmission link during dynamic evolution. These results collectively indicate that, compared to UDE and SCDE, the CPDE-based autonomous vehicle swarm can maintain perception contribution transmission among its members for a longer period even during dynamic evolution processes involving swarm growth, reduction, merging, or splitting.

[0081] (3) The ability to maintain reliable and effective perceived information contribution: Figure 20 to Figure 22 The graphs show the relationship between the AATTS (Advanced Achievement Tests) and maximum vehicle speed for the CPDE, UDE, and SCDE methods under different maximum vehicle numbers. It is clear from these graphs that, except for the cases where the maximum vehicle number is 200 and the maximum vehicle speed is 40, and the cases where the maximum vehicle number is 500 and the maximum vehicle speed is 70, the AATTS of CPDE is consistently higher than that of its counterparts. Compared to the UDE and SCDE methods, the average AATTS for maximum vehicle speeds of 40–80 under different maximum vehicle numbers are 19.47, 18.64, 15.83, 14.77, and 11.76, respectively; the average AATTS for UDE are 18.11, 16.21, 12.58, 11.87, and 10.22, respectively; while those for SCDE are 13.86, 13.83, 12.79, 12.44, and 8.97. Compared to UDE, CPDE improved AATS by 7%, 13%, 21%, 20%, and 13% at different maximum vehicle numbers, respectively; the improvement was even more significant compared to UDE. This indicates that during simulation, CPDE can better maintain the perception contributions among vehicle group members, thereby improving the reliability of perception information. As the maximum vehicle speed increases, a gradual decrease in AATS is observed. This is because as vehicle speed increases, vehicle mobility increases, leading to more frequent changes in relative positions between vehicles, thus increasing the contribution links between autonomous vehicles. Conversely, as the maximum number of autonomous vehicles increases, AATS also decreases. This is because as the number of vehicles increases, dynamic evolution events become more frequent, reducing the time for vehicles to continuously contribute perception information.

[0082] Figure 23The distribution of AAS during the simulation was described over a period of 50 to 250 seconds. Figure 24 (a) uses a box plot to represent the distribution of these results. The experimental results show that the CPDE method has higher maximum and minimum AAS values ​​and a higher mean AAS compared to the SCDE and UDE methods. Figure 24 Figures (b) and (c) show the results of two separate simulation experiments, in which Indicates the maximum number of vehicles. This represents the vehicle's maximum speed. It can be observed that CPDE significantly outperforms its comparison methods in all cases. Results from two separate simulations also confirm this. This is because the CPDE method considers proactive dynamic evolution events in which vehicles actively seek out neighboring vehicle groups to obtain effective perception information, thereby enhancing the reliability of the perception information. In contrast, CPDE enables vehicle groups to maintain a higher level of reliability in perceptual information interaction during dynamic evolution.

[0083] (4) Efficiency in contributing effective perceived information: Figures 25-27 The relationship between ATTS and maximum vehicle speed is shown under five different maximum vehicle number conditions. It can be observed that, regardless of the maximum vehicle speed or maximum vehicle number conditions, the ATTS of CPDE is consistently higher than that of the comparison method. This indicates that CPDE-based AVGs can maintain more efficient perception information transmission during dynamic evolution. This is because CPDE can select vehicles or vehicle groups with higher contribution efficiency to join or merge into their respective groups. Furthermore, as the maximum vehicle speed increases from 60 to 80, ATTS gradually decreases. This is because increased vehicle mobility increases the likelihood of vehicle group shrinkage and splitting, leading to an increase in vehicle group size or a decrease in perception contribution, ultimately resulting in a decrease in ATTS. On the other hand, ATTS increases with the increase in the maximum vehicle number. This is because, with increased vehicle density, CPDE can select vehicles with higher efficiency and lower latency to form perception contributions.

[0084] Figure 28 shows the ATS distribution for each simulation experiment over a period of 50 to 250 seconds using a scatter plot. The horizontal axis represents simulation time, and the vertical axis represents ATS. Figure 29 (a) uses a box plot to illustrate the ATS distribution of the three methods, revealing the dynamic evolution of ATS with the autonomous vehicle swarm. The experimental results show that the CPDE method has higher maximum and minimum ATS values ​​and a higher mean ATS compared to the SCDE and UDE methods. The figure also clearly shows a trend indicating that CPDE consistently exhibits a higher ATS value compared to the SCDE and UDE methods, with its mean value being significantly higher. Figure 29Figures (b) and (c) show the results of two comparative experiments, in which Indicates the maximum number of vehicles. The figure represents the maximum vehicle speed, revealing the changes in ATS (Automatic Sensor Speed) as the autonomous vehicle swarm dynamically evolves. As can be seen from the figure, the average ATS of the CPDE method is above 2 throughout the simulation time, while the ACS (Advanced Perception Scale) of the UDE and SCDE methods is lower than that of the CPDE method. These findings collectively highlight that the CPDE-based autonomous vehicle swarm contributes more efficiently and effectively to perception information.

[0085] (5) The ability to fully perceive the surrounding environment: Figures 30 to 32 The relationship between AVG's APB and simulation time was demonstrated, covering different maximum vehicle speed ranges from 40 km / h to 80 km / h.

[0086] As can be observed from the graph, CPDE maintains the highest APB, followed closely by SCDE, while UDE lags slightly behind. Performance remains relatively stable over time. Compared to its peers, CPDE consistently maintains a higher level of environmental understanding. This means that as the vehicle group expands, contracts, splits, and merges, it can still acquire valuable perception information through the perceptual contributions of its members. Therefore, it can filter out interference from invalid perception areas, achieving superior understanding of the surrounding environment.

[0087] UDE and SCDE primarily focus on ensuring smoother and more reliable communication links during dynamic evolution, even allowing vehicles that might not effectively contribute to the vehicle group's perception information to join. However, the CPDE method prioritizes the effectiveness of perception information provided by candidate joining vehicles or merging vehicle groups. CPDE still holds an advantage as the maximum vehicle speed increases, but the APB fluctuations of all algorithms become more pronounced, indicating a greater difficulty in maintaining a high level of comprehensive environmental perception at high speeds. These results demonstrate that, compared to other comparative methods, CPDE enables vehicle groups to maintain a stronger ability to comprehensively perceive their surroundings even when subjected to external disturbances.

[0088] In summary, during the dynamic evolution of vehicle groups, CPDE has a stronger ability to contribute effective perception information, a longer information transmission time, a more reliable perception information transmission link, a faster perception information transmission speed, and a higher comprehensive perception of the surrounding environment compared to the comparative methods.

[0089] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. An open-ended dynamic evolution method for autonomous vehicle swarms based on perception contribution, characterized in that, Includes the following steps: Step 1. Define the interference caused by ambient disturbances around the autonomous vehicle to its perception. Step 2. Analysis of the reasons for the dynamic evolution of the autonomous vehicle swarm; Step 3. Based on perception contributions, define the dynamic evolutionary behavior of the autonomous vehicle swarm, including expansion, contraction, merger, decomposition, and extinction. Analyze the impact of dynamic evolutionary behavior on the contribution degree, persistence, accessibility, and real-time performance of perception contributions among members, and propose corresponding methods for handling dynamic evolutionary behavior, including the following steps: Step 3.1 Design and handling method for augmented events in autonomous vehicle swarms; Step 3.2: Event reduction design and handling methods for autonomous vehicle swarms; Step 3.3 Design and handling method for parallel events of autonomous vehicle swarms; Step 3.4 Decomposition event design and processing method for autonomous vehicle swarm; Step 3.5 Design of the event of the demise of the autonomous vehicle swarm.

2. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, In step 1, the interference factors around the autonomous vehicle include, but are not limited to, manned vehicles, roadside obstacles, traffic lights, and pedestrians; the autonomous vehicle and The interference caused by the contribution of surrounding disturbance factors to its perception is as follows: (1) In the formula, Indicates driverless vehicles and The interference from surrounding factors on the device itself is calculated as follows: (2) in, and They represent and Driverless vehicles at all times and The number of invalid sensing squares caused by interference, represented as: (3) Indicates driverless vehicles and The interference of surrounding disturbance factors on the perceived contribution relationship between them is calculated as follows: (4) in, (5) in, , , , , and Representing time respectively driverless vehicles and The contribution of the perceived contribution includes the probability of interruption, the transmission throughput, the reliability, the latency, the efficiency, and the maximum contribution time.

3. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 2 includes the following steps: Step 2.1, Analysis of the impact of external interference factors on the contribution of perception, specifically: In open scenarios, the perception contribution of an autonomous vehicle swarm changes due to external interference factors, thereby altering the topology of collaborative perception within the swarm. The impact of external interference factors on the contribution, persistence, accessibility, and real-time performance of perception contribution relationships among autonomous vehicle swarm members is expressed as follows: 1) ,in Indicates driverless vehicles and Contribution degree of perceived contribution relationship between them It decreases over time; 2) ,in , , These represent driverless vehicles. and Persistence of perceived contribution relationship between them Accessibility Real-time It decreases over time; Step 2.2, Analysis of the reasons for the dynamic evolution of the vehicle group, specifically: Due to changes in the external environment surrounding the autonomous vehicle swarm, the contribution, persistence, accessibility, and real-time nature of the perception contributions between vehicle nodes are not zero, resulting in perception contribution relationships and causing the vehicle swarm to undergo dynamic evolution. When self-driving vehicles and When subjected to external disturbances, the contribution of the perception contribution relationship between them decreases, and at least one of the persistence, accessibility, or real-time indicators decreases. When the autonomous vehicle swarm is affected by external disturbances in an open scenario, the contribution, persistence, accessibility, and real-time of the perception contribution between its member nodes decrease. When one or more of them drop to zero, the perception contribution relationship between the vehicle nodes will be broken, leading to dynamic evolution of the vehicle swarm.

4. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 3.1 specifically includes: The expansion of autonomous vehicle swarms refers to the emergence of new perceptual contributions among swarm members due to changes in the external environment surrounding the swarm. The expansion events of autonomous vehicle swarms include addition events and emergent events; The term "addition event" refers to the occurrence of detached nodes becoming members of an autonomous vehicle swarm in an open scenario. These detached nodes interact effectively with one or more members of the swarm through perceptual contributions. They may be newly emerging nodes or nodes that have left other autonomous vehicle swarms. The addition event is represented as follows: (6) in, This represents a swarm of driverless vehicles. Indicating a fleet of self-driving cars An additional event is in progress. It is an unmanned vehicle The neighbor's vehicles gathered. and They represent and The direction of travel, express and There was no interference from external factors. Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time performance of the perceived contribution relationship between nodes are all not equal to zero, if the free node... If the above conditions are met, the vehicle will be added to the vehicle group. middle; Emergent events refer to the emergence of new perceptual contribution relationships among vehicle fodder members in an open scenario. These vehicle fodder members already have perceptual contribution relationships with one or more other members within the fodder, and simultaneously develop perceptual contribution relationships with other fodder members. Emergent events are represented as follows: (7) in, Indicating a fleet of self-driving cars An emergent event occurred. Indicating a fleet of self-driving cars The set of members, Indicates to self-driving vehicles Establish a set of vehicle group members who contribute to perception. express and The difference set; The method for expanding an autonomous vehicle swarm is described in detail below: (1) When When an additional event occurs: 1) When At that time, free nodes Broadcast join request JR; if Received from car group members Sending message A will join the autonomous vehicle swarm. Then, an update message U is sent to all vehicle group members (this message includes updated vehicle group member information and perception contribution information between members), and all vehicle group members update node information and perception contribution relationships between vehicle group members; 2) When At that time, members of the vehicle group Received free node When sending a join request (JR), if the autonomous vehicle node... and If they travel in the same direction and there is zero external interference between them, then Will Add to candidate node set ; For sets Calculate all elements , , , The optimal candidate vehicle set is obtained using multi-objective optimization theory. If the maximum number of vehicle group members has not been reached, and , , , If not equal to zero, then Towards The mid-range node sends a message allowing it to join the vehicle group. A The message A This includes updated vehicle group node information and perception contribution information among members; (2) When Emergent events occur: If self-driving car swarm members Received from another member The establishment of perceived contribution requests (CPRs) involves requests where the degree of contribution, persistence, reachability, and real-time nature of these requests are not equal to zero. ， Towards Send A; all vehicle group members update node information and perception contribution relationships.

5. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 3.2 specifically includes: The shrinkage of an autonomous vehicle swarm refers to the disruption of the perception contribution relationship between the members of the swarm due to the influence of external disturbance factors. The reduction events of an autonomous vehicle swarm include reduction events, disengagement events, and migration events: a reduction event occurs when one or more perception contribution relationships of a vehicle node are broken; a disengagement event occurs when a vehicle node's perception contribution relationship with all other swarm members is broken. When a vehicle node can only obtain effective perception information from other members of the vehicle group but can no longer contribute effective perception information to the vehicle group, the node leaves the vehicle group and tries to join another vehicle group, and a migration event occurs. Specifically, the reduction event refers to an event in an open scenario where, due to external interference, at least one of the following factors—contribution, persistence, reachability, and real-time performance—of a vehicle group member's perceived contribution drops to zero, preventing the group members from maintaining their perceived contribution relationship. The reduction event is represented as follows: (8) in, Indicating a fleet of self-driving cars A reduction event is underway. express and The interaction was affected by external factors. Indicates driverless vehicles and At least one of the following factors—contribution, persistence, accessibility, and real-time performance—is zero. express and It was not affected by external factors. Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time nature of the perceived contribution relationship between them are all not equal to zero; Specifically, the disengagement event refers to an event in an open scenario where an autonomous vehicle swarm is affected by external interference, causing at least one of the contribution, persistence, reachability, or real-time performance of a vehicle node and any other node in the swarm to drop to zero. The vehicle node can no longer contribute effective perception information to other swarm members, nor can it obtain the effective perception information it needs from other swarm members. The disengagement event is represented as: (9) in, Indicating a fleet of self-driving cars An detachment event is occurring; Specifically, the migration event refers to a situation in an open scenario where one or more vehicle nodes within a vehicle group can only obtain but cannot contribute effective perception information from other members of the group. If they obtain such information from other members of neighboring vehicle groups and can contribute effective perception information, they will proactively leave that vehicle group and attempt to join another. The migration event is represented as follows: (10) in, Indicating a fleet of self-driving cars A migration event is underway. Indicates driverless vehicles Unable to provide effective perception information to other members of the vehicle group; Indicates driverless vehicles Contributions; Indicates driverless vehicles and The degree of contribution between them Indicates driverless vehicles and The continuity between Indicates driverless vehicles and Accessibility between them Indicates driverless vehicles and Real-time performance between them Indicates driverless vehicles and The contribution degree, persistence, reachability, and real-time nature of the perceived contribution relationship between them are all not equal to zero; The methods for reducing the number of autonomous vehicles are described in detail below: 1) When At that time, a reduction event occurs: Autonomous vehicle node and The perceived contribution relationship between them is broken. Send a reduction message V to all members of the vehicle group, and the perception contribution relationship between all members of the vehicle group will be updated accordingly. 2) When At that time, an escape event occurs: Autonomous vehicle node After leaving the car group, it becomes a free node, broadcasts the message JR to the surrounding area, and continues to try to join other car groups; 3) When When a migration event occurs: Autonomous vehicle node Upon discovering that its contribution to the other members of the car swarm is 0, it broadcasts a message to its surroundings: JR, the car swarm node. Received free node When sending a JR, Algorithm 1 is used to determine whether to agree. If agreeing, the request is sent to the node. Send message A allowing entry into the vehicle group; if receive Message A sent, Will voluntarily withdraw from the vehicle group And join the car group and towards the group of vehicles Members send message L to migrate the vehicle group, vehicle group Members update node information and the relationship between nodes' perceived contributions.

6. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 3.3 specifically includes: The merging of autonomous vehicle swarms refers to the process of establishing new perception contribution relationships between the vehicle nodes of two or more vehicle swarms that are geographically close and have similar motion behaviors, and then merging the perception contribution relationships of the vehicle swarms to which the vehicle nodes belong to form a new vehicle swarm. The aforementioned concatenation event refers to an interaction in an open scenario where vehicle nodes within a vehicle group interact with other members of the same vehicle group through perception contributions, resulting in the concatenation of perception contribution relationships between these vehicle groups to form a new vehicle group. The concatenation event is represented as follows: (11) in, Indicating a fleet of self-driving cars A concatenation event is occurring; Indicating a fleet of self-driving cars , , It is a set of candidate vehicle groups to be merged; The method for constructing a swarm of autonomous vehicles is described in detail below: 1) Autonomous vehicle swarm Members of the car group The broadcast vehicle group requests MR, and the message carries... Occlusion level of all members; 2) When the swarm of driverless cars vehicle node Received from When performing MR, first determine and After merging the vehicle groups, determine if the group has reached its maximum size. If the merged group has not reached its maximum size, and the two nodes are traveling in the same direction, then the vehicle group will be... Join the vehicle consolidation group It also calculates and constructs the contribution of the perception contribution of the preceding and following vehicle groups, as well as the changes in persistence, accessibility, and real-time performance; otherwise, it sends a request to the node. Send a message rejecting the merger; 3) Based on Pareto optimality theory The autonomous vehicle swarm within the set is sorted using a non-dominated method. Then proceed in sequence to The autonomous vehicle swarm sends an agreement and assembly message AM until the swarm reaches its maximum size. The assembly message AM includes the swarm node information after assembly and the perception contribution information between members. 4) If Received from consent and constitute a message, All nodes in the [database name] have been added. The system then sends a success message to all vehicle group members, and all vehicle group members update the vehicle group node information and the perception contribution information among members.

7. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 3.4 specifically includes: Decomposition of an autonomous vehicle swarm refers to the process by which a vehicle swarm is split into two or more autonomous vehicle swarms due to the influence of external disturbance factors. A decomposition event occurs when there is no perception contribution relationship between different parts of the vehicle swarm, but there is still a perception contribution relationship between the internal vehicle nodes. Specifically, the decomposition event refers to a situation in an open scenario where, due to external interference, at least one of the contribution, persistence, reachability, and real-time performance of vehicle nodes within the vehicle group drops to zero, preventing the maintenance of the perception contribution relationship. The decomposition event is represented as: (12) in, Indicating a fleet of self-driving cars Decomposition is in progress; , Indicates from The set of autonomous vehicles decomposed from the data. Indicates the set of remaining vehicles; The method for decomposing an autonomous vehicle swarm is described in detail below: When self-driving vehicles and The perception contribution relationship between them is broken and is the set of remaining vehicles. and When the last pair of perception contribution relationships among members is established, the autonomous vehicle swarm decomposition event will occur; vehicle node Remove the set from the vehicle group members The vehicles in the middle, To become a new fleet of driverless vehicles.

8. The open-ended autonomous vehicle swarm dynamic evolution method based on perception contribution as described in claim 1, characterized in that, Step 3.5 specifically includes: The extinction of an autonomous vehicle swarm refers to the absence of any perception contribution relationship within the swarm; the extinction event occurs when the last perception contribution relationship in the swarm disappears. Specifically, the disappearance event refers to the situation in an open scenario where external interference factors affect the perception contributions of vehicle ensemble members, causing the contribution, persistence, reachability, and real-time performance of the perception contributions within the vehicle ensemble to all drop to zero, resulting in the disappearance of perception contribution relationships within the vehicle ensemble. The disappearance event is represented as follows: (13) in, This indicates that the swarm of self-driving cars is experiencing a demise.

Images