A multi-unmanned vehicle path coordination intelligent planning system

Abstract

The present application relates to the technical field of multi-vehicle cooperative scheduling, and particularly relates to a multi-unmanned vehicle path cooperative intelligent planning system, which comprises the following steps: constructing a local dynamic map with the vehicle as the center; forming a local dynamic passable area network; identifying key nodes in the local dynamic passable area network; calculating the dynamic conflict strength of the key nodes; marking the corresponding key nodes as high-conflict cooperative nodes when the dynamic conflict strength exceeds an adaptive threshold; and generating a flexible cooperative avoidance strategy. The present application accurately identifies the space-time traffic conflicts of multiple unmanned vehicles in complex scenarios, and realizes cooperative avoidance by using a distributed negotiation and flexible traffic control mechanism, thereby effectively improving the overall efficiency and running smoothness of multi-vehicle concurrent traffic while ensuring driving safety.

Description

Technical Field

[0001] This invention relates to the field of multi-vehicle collaborative scheduling technology, and in particular to a multi-unmanned vehicle path collaborative intelligent planning system. Background Technology

[0002] With the development of autonomous driving technology, collaborative passage of multiple autonomous vehicles in urban roads, industrial parks, and complex traffic scenarios has become an important application. Existing autonomous driving systems are mostly based on single-vehicle autonomous planning, using onboard sensors and local environment modeling to generate paths and avoid obstacles, which can meet the needs of independent vehicle driving to a certain extent. With the introduction of vehicle-to-everything (V2X) and collaborative perception technologies, some solutions are beginning to utilize information sharing between vehicles to extend perception of the status of nearby vehicles, thereby improving safety and traffic efficiency. However, when multiple vehicles simultaneously approach high-conflict areas such as intersections, merging points, and narrow road sections, the mutual influence between vehicles in both spatial and temporal dimensions is significantly enhanced. Relying solely on single-vehicle path planning or simple priority rules is insufficient to accurately depict the spatiotemporal competition between multiple vehicles, easily leading to traffic conflicts, reduced efficiency, and even safety risks.

[0003] Existing technologies for handling multi-vehicle conflicts often employ fixed rules, such as first-come-first-served, main road priority, centralized signal control, or simple time-based avoidance strategies. These methods generally suffer from insufficient responsiveness to dynamic environmental changes and imprecise characterization of simultaneous multi-vehicle competition scenarios. On one hand, there is a lack of unified modeling of the evolution of future traversable areas over time, making it difficult to identify potential high-risk conflict nodes in a timely manner. On the other hand, existing methods often fail to simultaneously integrate the inherent conflict risks of the road structure with the real-time movement status of multiple vehicles, resulting in inaccurate conflict assessments. Furthermore, in the collaborative control phase, some solutions rely on centralized node scheduling or rigid time control, lacking distributed negotiation capabilities and flexible adjustment mechanisms for travel time, making it difficult to balance traffic efficiency, system robustness, and vehicle ride smoothness. Therefore, a multi-autonomous vehicle path collaborative intelligent planning system is urgently needed to effectively solve the challenges of conflict determination and collaborative control in multi-vehicle concurrent traffic, thereby improving overall traffic safety and efficiency. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides a multi-unmanned vehicle path cooperative intelligent planning system.

[0005] A multi-autonomous vehicle path cooperative intelligent planning system, the intelligent planning system comprising performing the following: For each autonomous vehicle within the collaborative planning scope, a local dynamic map centered on the vehicle is constructed; static and dynamic obstacle information is integrated to generate a dynamic passable area that changes over time; a local dynamic passable area network is formed with the vehicle and its associated vehicles as nodes and the overlapping and connection relationships between dynamic passable areas as edges. In a local dynamic traversable area network, key nodes are identified where the dynamic traversable areas of multiple vehicles overlap and their expected arrival times are the same. Based on the real-time motion state of each associated vehicle and the time-varying characteristics of its dynamic traversable area, the spatial overlap and temporal competition trend of key nodes within a predetermined time window are predicted. Combined with the current static conflict metric, the dynamic conflict intensity of key nodes is calculated. When the dynamic conflict intensity exceeds an adaptive threshold, the corresponding key node is marked as a high-conflict cooperative node. For each high-conflict cooperative node, the vehicles involved negotiate through communication to generate a flexible cooperative avoidance strategy with the overall passage time optimization objective. The flexible cooperative avoidance strategy includes vehicle passage order, reference time window, flexible passage time margin, and recommended micro-paths planned within the local dynamic passable area network. Based on the flexible cooperative avoidance strategy, the vehicles involved select a path to pass through the high-conflict cooperative node within their corresponding flexible passage time margin.

[0006] Optionally, the local dynamic map includes static road structure, traffic signs, and obstacle location information.

[0007] Optionally, in the local dynamic map, areas occupied by static obstacles or permanently impassable are marked as static restricted areas; for dynamic obstacles, their trajectory over a future period is predicted based on their current motion state, and a safety envelope is extended along the trajectory to generate dynamically occupied areas that change over time; from the passable space of the local dynamic map, the static restricted areas and the dynamically occupied areas at each moment are removed to obtain a series of continuous polygonal regions that change over time, which serve as dynamically passable areas.

[0008] Optionally, the construction of a local dynamic passable area network includes defining the current position of the vehicle and all associated vehicles within the communication range as nodes of the network; calculating the spatial overlap area or the shortest connection distance of the dynamic passable areas of any two nodes within a predetermined future time window; if the spatial overlap area is greater than a preset overlap area threshold or the shortest connection distance is less than a preset minimum safety distance, then a network edge is established between the two nodes, thereby forming a local dynamic passable area network.

[0009] Optionally, the identification of the key node includes filtering out multiple candidate nodes for future spatiotemporal intersection points of vehicle dynamic traversable areas in the local dynamic traversable area network; for each candidate node, estimating its estimated arrival time based on the planned path length from each associated vehicle to the corresponding node and its current motion state; when the difference between the estimated arrival times of at least two associated vehicles is less than the time synchronization tolerance, the candidate node is determined to be the key node.

[0010] Optionally, for each key node, the real-time position, speed, and acceleration of each associated vehicle are obtained, and the time-varying characteristics of the shape and position of its dynamic passable area as it evolves over time are superimposed. The curves of the spatial overlap area ratio of each vehicle's dynamic passable area near the node as a function of time are simulated and predicted within a predetermined time window starting from the earliest expected arrival time, as well as the degree of dispersion of the actual possible arrival time distribution of each vehicle, are used as quantitative representations of the spatial overlap curve and the time competition trend, respectively.

[0011] Optionally, the static conflict metric of the key node at the current moment is calculated; simultaneously, based on the quantitative representation of the spatial overlap curve and the temporal competition trend, the dynamic conflict prediction value is calculated; and the static conflict metric and the dynamic conflict prediction value are weighted and fused to obtain the dynamic conflict intensity of the key node.

[0012] Optionally, the dynamic conflict intensity is compared with a preset adaptive threshold. When the dynamic conflict intensity exceeds the adaptive threshold, the corresponding key node is marked as a high-conflict cooperative node. The adaptive threshold is dynamically adjusted according to the vehicle density and average vehicle speed in the current network.

[0013] Optionally, for each high-conflict cooperative node, the objective function is to minimize the total or average time for all involved vehicles to pass through the node. Through negotiation or iterative calculation, a flexible cooperative avoidance strategy is generated, which includes vehicle passing order, reference time window, flexible passage time margin and recommended micropath.

[0014] Optionally, after receiving the agreed-upon flexible cooperative avoidance strategy, each vehicle involved can, based on its real-time location, local environmental perception information, and vehicle dynamics constraints, fine-tune the specific time of passing through the high-conflict cooperative node and its driving speed on the recommended micropath within the elastic passage time margin, so as to pass safely and smoothly.

[0015] The beneficial effects of this invention are: This invention, by introducing a spatiotemporal passable zone modeling and dynamic conflict detection mechanism, can accurately identify the location and time when multiple autonomous vehicles may have spatiotemporal conflicts on a specific road segment, overcoming the limitations of traditional methods based on fixed priorities or single-vehicle perspectives. Based on road structure constraints and vehicle operating status, a dynamic traffic conflict map is established, and vehicles are further guided to negotiate flexible passage opportunities within local collaborative areas, realizing a shift from conflict avoidance to resource coordination, effectively improving the safety and robustness of traffic.

[0016] This invention employs a distributed negotiation strategy combined with a dynamic adjustment mechanism for passage time margin during the passage control phase. This allows each vehicle to flexibly choose its passage time based on its own status and passage window, provided that the safe distance and trajectory feasibility are met. This avoids drastic fluctuations in acceleration and deceleration caused by rigid control, thereby achieving more stable lateral and longitudinal motion control. This mechanism not only optimizes the changes in vehicle acceleration and deceleration, but also reduces energy consumption and passenger discomfort. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram illustrating the execution process of the planning system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the functional modules of the planning system according to an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] It should be noted that the use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0021] Generally, terms can be understood at least partly from their use in context. For example, depending at least partly on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood not necessarily to convey an exclusive set of factors, but rather, alternatively, depending at least partly on the context, to allow for the presence of other factors that are not necessarily explicitly described.

[0022] like Figures 1-2 As shown, a multi-autonomous vehicle path cooperative intelligent planning system includes the following modules: Local dynamic mapping module: used to build a local dynamic map centered on each autonomous vehicle within the collaborative planning range; it integrates static obstacle information and dynamic obstacle trajectories to generate a dynamic passable area that evolves over time; The passable area network construction module is used to represent the overlapping and connection relationships between the dynamic passable areas of each vehicle as edges of the graph, with the vehicle and its associated vehicles as nodes in the graph network, thus forming a dynamic passable area network that reflects the local environment's traffic relationships. Conflict Node Identification and Evaluation Module: In the dynamic traversable area network, it identifies overlapping areas where multiple vehicles are expected to arrive simultaneously as key nodes, and calculates the dynamic conflict intensity of key nodes based on vehicle motion state, spatial overlap degree within time window and conflict trend, combined with static conflict metric value, and determines whether it exceeds an adaptive threshold to mark it as a high-conflict cooperative node. The flexible collaborative avoidance strategy generation module is used to generate a flexible collaborative avoidance strategy for each high-conflict collaborative node through communication negotiation between the vehicles involved. This strategy includes the vehicle passage order, recommended reference time window, elastic passage time margin, and recommended micro-path within the local network. It is used to guide vehicles to complete the avoidance and passage tasks within the corresponding elastic passage time margin.

[0023] More specifically, the intelligent planning system performs the following steps to achieve intelligent path planning: S1: For each autonomous vehicle within the collaborative planning range, construct a local dynamic map centered on the vehicle itself; integrate static and dynamic obstacle information to generate a dynamic passable area that changes over time; using the vehicle itself and its associated vehicles as nodes, and the overlapping and connection relationships between dynamic passable areas as edges, form a local dynamic passable area network. S1 specifically includes: S11, Local Dynamic Map Construction: For each autonomous vehicle within the collaborative planning range, it collects its own perception information through its sensors and receives its own status and perception information shared by all related vehicles within the communication range through vehicle-to-everything (V2X) communication. It then fuses the perception information with the shared information between vehicles to build a local dynamic map with the vehicle as the coordinate reference, based on the vehicle's current position and heading. The vehicle's own perception information is data collected by various sensors onboard to the vehicle, such as LiDAR, millimeter-wave radar, cameras, IMU, and GPS, including lane lines, obstacles ahead, and traffic light positions. This information can cover areas directly detectable by the vehicle, but is usually spatially limited. The shared information between vehicles is obtained through V2X technology by communicating with other vehicles within the communication range to acquire their status, such as position, speed, heading, and their perceived local environment. This information expands the perception capabilities of a single vehicle, enabling the acquisition of environmental data over a wider range and greater distance. The fusion of perceived and shared information involves integrating the vehicle's own sensory information with that of other vehicles to form a comprehensive environmental information set. This fusion process includes deduplication, alignment, and consistency correction to ensure all information is referenced to the current time and the vehicle's own coordinates. Based on this fusion, a coordinate system is constructed with the vehicle's position as the origin and its orientation as the positive direction. Environmental elements are plotted within this coordinate system to form a local dynamic map. The map content includes: Static road structure: lanes, curbs, medians, etc., do not change over time; Traffic sign information: speed limit signs, traffic light location and status, stop lines, etc.; Obstacle information: This includes static obstacles, such as parked vehicles and fences, as well as dynamic obstacles, such as pedestrians, bicycles, and other vehicles.

[0024] Constructing a local dynamic map with reference to this vehicle includes the following steps: 1. Set the current position of the vehicle as the origin (0,0), and the current direction of the vehicle's head as the positive X-axis direction of the coordinate system, and establish a right-handed local vehicle coordinate system; 2. The data collected by sensors such as lidar and cameras are transformed into coordinates, and the transformed point cloud, obstacle outline, lane lines, etc. are projected onto a two-dimensional map at the center of the vehicle. 3. Based on the GPS coordinates, heading, and relative position of the perceived target of the other vehicle, calculate its position in the vehicle's coordinate system, and integrate this data into the vehicle's map for unified representation; 4. Use grid maps, vector maps, or polygon representations to store the spatial location and attributes of map elements, and label obstacle attributes, dynamic element identifiers, and passable area boundaries, etc. 5. The local dynamic map is updated periodically, removing outdated dynamic targets and integrating new sensor results to maintain timeliness. Figure 1 It typically only covers an area of ​​tens to hundreds of meters, with the aim of meeting the requirements of low latency, high frequency, and real-time updates.

[0025] S12 generates dynamically passable areas. In the local dynamic map, spatial occupancy analysis is performed on static obstacles, marking their occupied areas or permanently impassable closed areas as static restricted zones. First, obstacles that do not move are processed, such as parked cars, roadside buildings, guardrails, green belts, and road construction closures. For these types of obstacles, the areas they occupy are directly marked as permanently impassable areas on the local map, called static restricted zones.

[0026] In autonomous driving or multi-vehicle cooperative systems, dynamic obstacles are not stationary but constantly in motion. Therefore, simply knowing their current location is insufficient; it is essential to predict their future positions to plan avoidance paths and prevent conflicts. Thus, for dynamic obstacles, it is necessary to predict their future trajectory sequence in the time domain. Specifically, assuming that the obstacle maintains its current speed and heading within a short time window (i.e., uniform linear motion), the position at each future moment can be calculated step-by-step over time, denoted as: ; For dynamic obstacles in time The predicted location; Indicates the time of the obstacle The horizontal position at that time Indicates the time of the obstacle The longitudinal position at that time; By calculating at multiple future time points, such as 0.5 seconds, 1 second, 1.5 seconds, and so on, up to a certain set time window, such as 3 seconds, a sequence of trajectory points can be obtained. These points, when strung together, constitute the predicted trajectory of the obstacle in the future, and thus obtain the corresponding dynamically occupied area.

[0027] In autonomous driving or multi-vehicle cooperative systems, it's insufficient to simply predict where a dynamic obstacle will appear in the future; its actual spatial occupancy must also be considered. Even if a vehicle's future position is predicted to be safe, a certain safety distance must be maintained to prevent collisions caused by sudden acceleration, steering, or sensor errors. Therefore, a redundant occupancy range, or safety envelope, is established for each predicted location. The radius of the safety envelope is calculated based on vehicle dimensions and a preset buffer coefficient, as follows: ; For the safety envelope radius; The radius of the obstacle body, For redundant safety spacing; By removing static restricted areas and dynamically occupied areas from the passable space of the local dynamic map, the actual passable area can be obtained in each time slice, forming a series of continuous polygonal areas, i.e., dynamic passable areas.

[0028] S13, Construct a local dynamic traversable area network: Using each vehicle as a node, construct an area connection network based on the spatiotemporal relationships between dynamically traversable areas; this includes marking the current location of the vehicle and all associated vehicles within its communication range, or their next reachable location in their planned paths, as network nodes; assuming any two vehicle nodes are respectively... and Determine their corresponding dynamic passable areas respectively. and ; In the future scheduled time window Within, calculate the area of ​​overlap between the two regions. minimum connection distance ; Represents a node With nodes In time The overlapping area of ​​the dynamically passable region; Represents a node With nodes Dynamic passable areas in time The minimum connection distance; If any of the following conditions are met: Condition 1, This means that the planned feasible areas of the two vehicles overlap spatially at a certain moment, indicating that they may both want to pass through the same location at the same time. Condition 2, This means that in some scenarios where vehicles pass each other or intersect, even if the paths do not intersect, there is still a risk if the distance between them is too close. Then at the node and Establish network edges between Ultimately, a locally dynamic traversable area network was constructed. ,in, This indicates a dynamically accessible area network. For a set of nodes, For the set of connecting edges, The overlap area threshold is used to determine whether there is overlap and avoid misjudgment interference due to slight edge overlap; the value can be estimated based on the vehicle size; it is set as a proportion of the effective projected area of ​​a certain vehicle body, for example, 1.5~2.5 square meters, or 10%~20% of the area of ​​the vehicle's envelope area. The minimum safe distance is used to determine whether the distance is too close, preventing lateral collisions or insufficient braking distance during high-speed travel. It is usually calculated based on the vehicle's dynamic characteristics, taking into account vehicle speed, braking capacity, and system response time. Common static settings are 1.0~2.5 meters (low-speed urban scenarios) or 3.0~5.0 meters (high-speed scenarios). Alternatively, a speed-related minimum distance can be used, i.e.: safe distance = vehicle speed × reaction time + minimum physical distance.

[0029] S2: In a local dynamic traversable area network, identify key nodes where the dynamic traversable areas of multiple vehicles overlap and their expected arrival times are the same; based on the real-time motion state of each associated vehicle and the time-varying characteristics of its dynamic traversable area, predict the spatial overlap and temporal competition trend of key nodes within a predetermined time window, and calculate the dynamic conflict intensity of key nodes by combining the current static conflict metric; when the dynamic conflict intensity exceeds the adaptive threshold, the corresponding key node is marked as a high-conflict cooperative node. S2 specifically includes: S21, Key Node Identification: In a local dynamic traversable area network, candidate nodes are selected from multiple dynamically traversable areas where there is a possibility of spatiotemporal intersection in the future. Candidate nodes refer to locations that may be reached by more than one vehicle in the future, representing a set of potential conflict points. The specific selection method is as follows: Each vehicle has its own path planning results, which consist of multiple discrete target points, such as intersections, lane change locations, U-turn points, etc. The system iterates through the next hop position in all vehicle paths, i.e., the next reachable position; If a target point or spatial location appears in the path of multiple vehicles, that is, multiple vehicles are expected to pass through the location, then this location is marked as a candidate node; or, by analyzing the dynamic passable area of ​​the vehicles in the future evolution trajectory, it can be detected whether the areas of multiple vehicles have spatial overlap in a certain area, and the geometric center or intersection center of the overlapping area can also be defined as a candidate node. The selection criteria are as follows: Criterion 1: At least two vehicles' predicted paths contain the node or the region. Standard 2: The node must be within the valid time window of the current planning cycle, such as within the next 10 seconds; Standard 3: Nodes must be identifiable and locatable within the system's allowable control accuracy range.

[0030] After filtering, the time required for each vehicle to reach the candidate node is estimated based on its current status and the path length to that node; for each candidate vehicle node... Based on each associated vehicle From current position to node Length of planned path By combining the current speed and acceleration, the estimated arrival time is estimated; if the vehicle is accelerating or decelerating, an acceleration model is used; if the vehicle is traveling at a constant speed, a constant speed model is used, thus obtaining the estimated arrival time for each vehicle, as expressed in the following expression: Acceleration model: ; Uniform velocity model: ;in, For vehicles To node Path length; For vehicles Current speed; For vehicles Current acceleration; For vehicles Expected arrival node The time.

[0031] After obtaining the estimated arrival time, compare the estimated arrival time difference between any two vehicles participating in the candidate node. If the time difference between any pair of vehicles is less than a preset time synchronization tolerance, they are considered to have a time race. Once a time race occurs, the candidate node can be officially marked as a critical node. Specifically, if any two vehicles... satisfy: Then select the candidate node. Determined to be a critical node. Time synchronization tolerance is a threshold used to determine whether two or more vehicles are approaching the same spatial node at the same time. In layman's terms, it defines a range of time overlap that is allowed. As long as the expected arrival time difference of multiple vehicles is within this range, the system considers that they are competing for the right of way at the node, which may constitute a potential conflict.

[0032] S22, for each critical node It acquires the real-time status of all associated vehicles, including their location, speed, and acceleration, and combines this with the spatial characteristics of their dynamically passable areas evolving over time to predict a future time window. The spatial relationships at each moment within the timeframe are simulated and predicted, and are divided into two dimensions: From a spatial perspective, the degree of spatial overlap refers to whether and to what extent the dynamically passable areas of vehicles overlap at different future times. Looking at the time competition trend from a time perspective: that is, when are these cars expected to arrive, and are their arrival times highly concentrated; To achieve such a simulation, the following input elements must first be mastered: The current status of each vehicle, including its current position, speed, and acceleration; The dynamic passable area of ​​each vehicle at a future time is not a point, but a spatial region, usually represented as a polygon; Time range, i.e., the scheduled time window This refers to the time from the departure time of the earliest vehicle expected to arrive at the critical node to the end time when the latest vehicle may pass through. The steps for conducting simulation predictions are as follows: Time window Divide into several discrete time points, denoted as This yields a continuous time series output. For each associated vehicle participating in the critical node, based on its current state, predict its time using uniform acceleration. The location at that time is determined, and a dynamically accessible area is constructed centered on that location. All vehicles at the same time The dynamic traversable regions are subjected to overlap analysis to calculate their geometric intersection; the area of ​​this intersection is called the overlap area. Simultaneously, the sum of the smallest reference areas within the regions of these vehicles is calculated; this sum is called the reference area. Next, calculate the overlap ratio at that moment, using the following formula: , For a moment Next node The proportion of overlapping areas in the vicinity; finally, the overlap proportions at all time points are connected to form key nodes. The spatial overlap variation curve.

[0033] The time competition trend index measures the distribution of the expected arrival times of multiple related vehicles at a key node. Specifically, it is characterized by calculating the dispersion of these arrival times, i.e., the standard deviation. In multi-vehicle path coordination, the predicted arrival times of each vehicle at a key node are grouped together: By calculating the standard deviation of this set of times, the result is the critical node. Time Competition Trend Indicator The time competition trend indicator quantifies the proximity of the arrival times of multiple vehicles at key nodes, providing the system with the ability to predict conflicts in the time dimension; it allows the system not only to know which locations may have conflicts, but also when conflicts may occur.

[0034] S23, Dynamic Conflict Intensity Calculation: The above steps have yielded the spatial conflict trend (i.e., the spatial overlap change curve) and the temporal competition trend (i.e., the temporal competition trend index) of a key node within a future time window. However, to determine whether a key node truly constitutes a serious conflict point, the structural risk of the node itself must also be considered. That is, even if the vehicle conflict trend is minor, if the node is located at a highly sensitive structural location such as an intersection, ramp exit, or blind spot, the system should increase its alertness and prioritize handling it. Therefore, the core objective is to integrate static structural risk and dynamic conflict trend into a quantitative indicator, namely, dynamic conflict intensity, to provide a basis for determining whether to initiate cooperative avoidance. The specific steps are as follows: First, obtain the key nodes. Static conflict metric This value is a pre-set value derived from the road structure database, which is based on the inherent conflict risk score of the road structure where the node is located, such as whether it is an intersection, merging point, or turning area. Table 1 is a preset source from the road structure database. Road structure type Example Static conflict scoring recommendations Simple line segments Two-lane straight road 0.1～0.3 Turning point or roundabout Crossroads or T-junctions, small roundabouts 0.4～0.6 Intersection + Non-signal control There are no traffic lights at intersections in the urban area. 0.6～0.8 Multi-vehicle merging area ramp merging, highway exit 0.8～0.95 High-risk structural areas Blind spots on the road, lane changes at slopes 0.95～1.0 Then, based on the spatial overlap curve The trend of competing with time Calculate the dynamic conflict prediction value of the node. The formula is: ; Finally, the static conflict metric and the dynamic conflict prediction are weighted and fused to obtain the dynamic conflict intensity of the key nodes. Its expression is: , This is the weighting coefficient, and its value ranges from... This is used to balance the influence of static and dynamic factors. The principle for determining its value is that the more reliable the static information, the less stable the dynamic prediction. The larger the value, the more reliable the dynamic prediction and the weaker the structural differences. The smaller the value, the better. Specifically, in open roads with simple structures, dynamic judgment is the primary factor, and the value is 0.3 to 0.5. In urban core areas with dense intersections, static structure is the primary factor for conflict risk, and the value is 0.6 to 0.8. In highways and areas with ramps, both factors are important, and the value is 0.5 to 0.7.

[0035] S24, High-Conflict Collaborative Node Identification: In the previous steps, through a series of spatiotemporal predictions and fusion calculations, a final conflict intensity value for each key node was obtained. It has comprehensively considered the inherent risks of the road structure where the node is located; the degree of future path intersection and arrival time overlap of multiple vehicles in the area; however, the system cannot immediately handle all conflicts, and must first determine which nodes have a sufficiently high conflict risk and deserve priority scheduling; therefore, a threshold needs to be set: only when the conflict intensity of a key node exceeds this threshold is the system considered to be a truly high-conflict collaborative node. Specifically, a comparison method is used for judgment, and each key node has a dynamic conflict intensity value; an adaptive conflict threshold is pre-set. ;when If the node is identified as a high-conflict collaborative node, it is considered a normal critical node and may not be processed immediately or the response may be delayed.

[0036] Adaptive conflict threshold The change is based on the vehicle density and average speed within a unit area of ​​the current vehicle network, and the formula is: ; Vehicle density per unit area represents the degree of congestion in a given area. The denser the vehicle density, the greater the risk of conflict. The system should lower the judgment threshold, and the threshold should increase as the density increases. Let be the average speed of vehicles in the current network, representing the overall traffic flow. A slower speed indicates a higher likelihood of localized congestion and potential for more severe conflicts. Since speed is the denominator, when... The smaller, The larger the value, the higher the threshold will be; This is an empirical coefficient used to control the strength of the influence of the two types of factors on the final threshold, and can be set to... ; The base threshold offset is a constant term used to define the minimum response threshold, ensuring that the system can identify a small number of significant conflicts, even in low-density, high-speed scenarios.

[0037] S3: For each high-conflict cooperative node, the vehicles involved negotiate through communication to generate a flexible cooperative avoidance strategy with the overall passage time optimization objective. The flexible cooperative avoidance strategy includes vehicle passage order, reference time window, flexible passage time margin, and recommended micro-paths planned within the local dynamic passable area network. The vehicles involved select a path to pass through the high-conflict cooperative node according to the flexible cooperative avoidance strategy within their corresponding flexible passage time margin. S3 specifically includes: S31, Flexible Cooperative Avoidance Strategy Generation: When multiple vehicles are about to enter a high-conflict cooperative node simultaneously, a personalized traffic strategy is generated for each vehicle through communication and coordination between vehicles. This strategy avoids conflict without affecting overall traffic efficiency. Specifically, it includes the following steps: S311. Each vehicle involved in the incident broadcasts its current status, including its location, speed, acceleration, and current path information; S312. Simultaneously broadcast the proposed passage information, including the suggested order of passage priority. With expected time Among them, by order priority This refers to a passage priority level assigned to each vehicle based on system or rule-based judgment before it enters a high-conflict coordination node. This level determines which vehicle's passage request should be responded to earlier and more worthy of priority during negotiation. The assignment method can include setting it as a numerical level, such as 1, 2, 3...; or generating weights according to a scoring mechanism, then sorting and numbering them. Expected time Each vehicle autonomously calculates an ideal time to cross the high-conflict node based on its own status and planned route. The calculation formula is as follows: If the acceleration is constant: ; If the vehicle is traveling at a constant speed: ;in, This represents the path length between the current location and the node. The vehicle's current speed. This represents the current acceleration. If there is no significant acceleration, it can be set to 0.

[0038] S313. Each vehicle receives corresponding information from all other vehicles and conducts distributed negotiation based on this information. Distributed negotiation involves each vehicle participating in the negotiation, broadcasting its own status and passage proposal to each vehicle without relying on a unified control node. By receiving information from other vehicles and through consensus negotiation, a consensus is finally reached. S314. After all vehicles have collected each other's information, they can work together to solve a collaborative optimization problem. The collaborative optimization problem includes minimizing the total travel time and minimizing the average travel time. An optimization model is constructed with the goal of minimizing the total or average travel time of all involved vehicles passing through the node. Minimize total travel time: , The final scheduled actual transit time; The time it takes for a vehicle to begin waiting or approaching a node; the goal of minimizing the total passage time is to minimize the cumulative time for all vehicles from approaching the node to successfully passing through it. Minimize average travel time: The goal of minimizing average travel time is to control traffic efficiency while avoiding excessively long waiting times for individual vehicles and improving collaborative fairness.

[0039] S315. After the negotiation is completed, the system generates a flexible cooperative avoidance strategy for each vehicle, which includes the following four components: The order in which vehicles pass is used to determine the priority of each vehicle's passage. Reference passage time window This is used to suggest the target time period for vehicles to pass through the node; Flexible passage time margin This refers to the acceptable range of advance or delay around the reference time window; Recommended micropaths represent the set of paths a vehicle takes from its current location or a passable area within a locally dynamic traversable area network to the next segment of the network after crossing a high-conflict node.

[0040] S32, Flexible Strategy Execution and Fine-tuning: When multiple vehicles reach a consensus on an avoidance strategy for the same high-conflict node, i.e., after negotiating a time window for each vehicle to pass, each involved vehicle must decide on the actual time to pass within the time period allowed by the strategy, and adjust its driving status accordingly; each involved vehicle will execute traffic control in the following manner: S321. Based on the vehicle's current real-time location, local environmental perception (such as obstacles and traffic signs), and its own vehicle dynamics constraints (such as maximum acceleration and turning radius), select a specific time to pass within the travel time margin. ,satisfy: ; S322. At the same time, depending on the timing of passage. Fine-tune the path segments in the recommended micropath, including controlling the lateral and longitudinal travel speeds and optimizing acceleration / deceleration, to avoid interference with new obstacles in the real environment; Longitudinal speed control refers to controlling the vehicle's forward speed along the recommended path to ensure it arrives at the target node position within the reference time window on time without violating its own dynamic limitations. Specifically, the vehicle calculates a suitable average speed based on its current position, current speed, and the remaining time and distance between the expected passage time. If the current speed is too high, it will appropriately decelerate and coast or plan intermediate waiting points; if it is too low, it will plan acceleration segments. The entire path segment will be adjusted according to the target speed to ensure passage within the time window.

[0041] Lateral speed control refers to adjusting the rate of change of lateral offset, i.e., lateral speed, based on road curvature and safety distance, during lane changes, avoidance maneuvers, or curved sections. Specifically, it includes: The obstacle avoidance target and path curvature are combined with environmental perception information to calculate the optimal lateral offset path within a locally passable area. Lateral control adopts a trajectory tracking method to control the steering wheel angle in real time so that the vehicle can smoothly transition laterally. Through the above-mentioned combined longitudinal and lateral control, the vehicle can meet the cooperative avoidance requirements while maintaining a feasible trajectory.

[0042] Optimization of acceleration and deceleration refers to controlling the vehicle's acceleration and deceleration within a given time window to achieve a smooth transition from the current position to a designated passage time and speed. The optimization objective is: Prioritizing stability, and under the premise of meeting the timing constraints, the trajectory that minimizes the rate of change of acceleration and deceleration, i.e. the minimum acceleration / deceleration amplitude, is selected. Energy consumption optimization avoids frequent or drastic acceleration and deceleration, improves traffic efficiency, and reduces energy consumption; Constraints: All acceleration changes must be within the permissible dynamic range of the vehicle.

[0043] S323. The vehicle passes through the high-conflict node in a safe, smooth, and low-interference manner and successfully joins the subsequent travel path.

[0044] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0045] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-unmanned vehicle path cooperative intelligent planning system, characterized in that, The intelligent planning system includes the following: For each autonomous vehicle within the collaborative planning scope, a local dynamic map centered on the vehicle is constructed; static and dynamic obstacle information is integrated to generate a dynamic passable area that changes over time; a local dynamic passable area network is formed with the vehicle and its associated vehicles as nodes and the overlapping and connection relationships between dynamic passable areas as edges. In a local dynamic traversable area network, key nodes are identified where the dynamic traversable areas of multiple vehicles overlap and their expected arrival times are the same. Based on the real-time motion state of each associated vehicle and the time-varying characteristics of its dynamic traversable area, the spatial overlap and temporal competition trend of key nodes within a predetermined time window are predicted. Combined with the current static conflict metric, the dynamic conflict intensity of key nodes is calculated. When the dynamic conflict intensity exceeds the adaptive threshold, the corresponding key node is marked as a high-conflict collaborative node. For each high-conflict cooperative node, the vehicles involved negotiate through communication to generate a flexible cooperative avoidance strategy with the overall passage time optimization objective. The flexible cooperative avoidance strategy includes vehicle passage order, reference time window, flexible passage time margin, and recommended micro-paths planned within the local dynamic passable area network. Based on the flexible cooperative avoidance strategy, the vehicles involved select a path to pass through the high-conflict cooperative node within their corresponding flexible passage time margin.

2. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 1, characterized in that, The local dynamic map includes static road structure, traffic signs, and obstacle location information.

3. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 2, characterized in that, In the local dynamic map, areas occupied by static obstacles or permanently impassable are marked as static restricted areas. For dynamic obstacles, their trajectory over a future period is predicted based on their current motion state, and a safety envelope is extended along the trajectory to generate dynamically occupied areas that change over time. From the passable space of the local dynamic map, the static restricted areas and the dynamically occupied areas at each moment are removed to obtain a series of continuous polygonal regions that change over time, which serve as dynamically passable areas.

4. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 3, characterized in that, The construction of a local dynamic passable area network includes defining the current position of the vehicle and all associated vehicles within the communication range as nodes of the network; calculating the spatial overlap area or the shortest connection distance of the dynamic passable areas of any two nodes within a predetermined future time window; if the spatial overlap area is greater than a preset overlap area threshold or the shortest connection distance is less than a preset minimum safety distance, then a network edge is established between the two nodes, thereby forming a local dynamic passable area network.

5. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 1, characterized in that, The identification of the key node includes screening out candidate nodes for future spatiotemporal intersection points of multiple vehicle dynamic traversable areas in a local dynamic traversable area network; for each candidate node, estimating its estimated arrival time based on the planned path length of each associated vehicle to the corresponding node and its current motion state; when the difference between the estimated arrival times of at least two associated vehicles is less than the time synchronization tolerance, the candidate node is determined to be the key node.

6. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 5, characterized in that, For each key node, the real-time position, speed, and acceleration of each associated vehicle are obtained, and the time-varying characteristics of the shape and position of its dynamic passable area are superimposed over time. The simulation predicts the curve of the spatial overlap area ratio of each vehicle's dynamic passable area near the node as a function of time within a predetermined time window starting from the earliest expected arrival time, as well as the degree of dispersion of the actual possible arrival time distribution of each vehicle. These are used as quantitative representations of the spatial overlap curve and the time competition trend, respectively.

7. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 1, characterized in that, Calculate the static conflict metric value of the key node at the current moment; and calculate the dynamic conflict prediction value based on the quantitative representation of the spatial overlap curve and the temporal competition trend. The static conflict metric and the dynamic conflict prediction are then weighted and fused to obtain the dynamic conflict intensity of the key node.

8. The multi-unmanned vehicle path cooperative intelligent planning system according to claim 7, characterized in that, The dynamic conflict intensity is compared with a preset adaptive threshold. When the dynamic conflict intensity exceeds the adaptive threshold, the corresponding key node is marked as a high-conflict cooperative node. The adaptive threshold is dynamically adjusted according to the vehicle density and average vehicle speed in the current network.

9. A multi-unmanned vehicle path cooperative intelligent planning system according to claim 1, characterized in that, For each high-conflict collaborative node, the objective function is to minimize the total or average time for all involved vehicles to pass through the node. Through negotiation or iterative calculation, a flexible collaborative avoidance strategy is generated, which includes vehicle passing order, reference time window, flexible passage time margin, and recommended micropath.

10. A multi-unmanned vehicle path cooperative intelligent planning system according to claim 9, characterized in that, After receiving the agreed-upon flexible collaborative avoidance strategy, each vehicle involved in the incident, based on its real-time positioning, local environmental perception information, and vehicle dynamics constraints, fine-tunes the specific time of passing through the high-conflict collaborative node and its driving speed on the recommended micropath within the elastic passage time margin, so as to pass safely and smoothly.

Images