Cooperative path planning method and system for intelligent connected vehicles

Abstract

The application provides a cooperative path planning method and system of an intelligent connected vehicle, relates to the technical field of traffic optimization, determines a plurality of groups of initial planning paths of a plurality of intelligent connected vehicles in a preset area, identifies a first connected vehicle set located at a first intersection at a first prediction time, performs a traditional vehicle shunting feature prediction at the first intersection, performs intersection connected vehicle shunting optimization in combination with a first shunting prediction information, a driving purpose certainty label, and an initial planning path corresponding to the first connected vehicle set, returns an optimization result to a corresponding intelligent connected vehicle control terminal, solves the problem that intelligent connected vehicles and traditional vehicles form a decision conflict, cause path conflict and low shunting efficiency, and cause traffic hazards and congestion, and achieves the effects of improving the traffic flow and overall traffic efficiency of the intersection, reducing traffic congestion and conflict, and improving traffic safety and efficiency.

Description

Technical Field

[0001] This invention relates to the field of traffic optimization technology, specifically to a collaborative path planning method and system for intelligent connected vehicles. Background Technology

[0002] With the continuous development of intelligent connected vehicle technology, information exchange between vehicles has become an important means to improve road traffic safety and efficiency. V2X (Vehicle-to-Everything) communication, as a key technology, enables real-time data interaction between intelligent connected vehicles and other vehicles, road infrastructure, pedestrians, and networks, thereby making the transportation system more intelligent and automated. Important applications of V2X communication technology include vehicle-to-vehicle (V2V) communication, vehicle-to-infrastructure (V2I) communication, vehicle-to-pedestrian (V2P) communication, and information exchange between vehicles and networks / cloud platforms (V2N, V2C). Among these, V2V communication, by sharing vehicle location information, speed, and driving intentions, can effectively reduce traffic accidents and improve traffic flow.

[0003] While the application of intelligent connected vehicles has improved traffic management and driving safety to some extent, current technologies primarily focus on optimizing the travel paths of individual vehicles or improving traffic flow through information sharing. However, for the collaborative path planning problem between traditional vehicles (i.e., non-connected vehicles and intelligent connected vehicles), existing technologies do not fully consider the behavior and reaction patterns of traditional vehicles, nor do they deeply integrate path optimization for connected vehicles in complex traffic scenarios such as intersections. This leads to decision-making and path conflicts between intelligent connected vehicles and traditional vehicles in scenarios with high traffic volume and frequent intersections, thus affecting the overall efficiency and safety of the traffic system. Furthermore, existing technologies mainly focus on shortest path planning or traffic flow optimization based on static data, lacking technical solutions for intersection diversion optimization and failing to effectively predict the diversion behavior of traditional vehicles at intersections and incorporate it into the path planning decisions of connected vehicles. Therefore, the reactions and path planning of traditional vehicles have not been fully coordinated with those of intelligent connected vehicles, resulting in the underutilization of the potential for traffic flow and safety.

[0004] In summary, existing technologies often lack collaborative traffic management between intelligent connected vehicles and traditional vehicles, which leads to decision-making conflicts between them, resulting in path conflicts and low diversion efficiency, causing traffic hazards and congestion. Summary of the Invention

[0005] This application provides a collaborative path planning method and system for intelligent connected vehicles, which addresses the technical problem of the lack of collaborative traffic management between intelligent connected vehicles and traditional vehicles in the prior art, which leads to decision conflicts between intelligent connected vehicles and traditional vehicles, resulting in path conflicts and low diversion efficiency, causing traffic hazards and congestion.

[0006] In view of the above problems, this application provides a collaborative path planning method and system for intelligent connected vehicles.

[0007] In a first aspect, this application provides a cooperative path planning method for intelligent connected vehicles, the method comprising:

[0008] Connecting to a V2X communication system, multiple initial planned paths for various intelligent connected vehicles within a preset area are determined, wherein each intelligent connected vehicle carries a corresponding driving purpose deterministic label; the multiple initial planned paths are parsed to identify a first set of connected vehicles located at a first intersection at a first prediction time; based on the first prediction time at the first intersection, the diversion characteristics of traditional vehicles are predicted to generate first diversion prediction information; combining the first diversion prediction information, the driving purpose deterministic label, and the initial planned paths corresponding to the first set of connected vehicles, intersection connected vehicle diversion optimization is performed to generate a first path optimization result set corresponding to the first set of connected vehicles; the first path optimization result set is returned to the control terminal of the corresponding intelligent connected vehicle.

[0009] Secondly, this application provides a cooperative path planning system for intelligent connected vehicles, the system comprising:

[0010] An initial planning path acquisition module is used to connect to a V2X communication system and determine multiple initial planning paths for multiple intelligent connected vehicles within a preset area, wherein each intelligent connected vehicle carries a corresponding driving purpose deterministic label; a path parsing module is used to parse the multiple initial planning paths and identify a first set of connected vehicles located at a first intersection at a first prediction time; a diversion feature prediction module is used to predict the diversion features of traditional vehicles at the first intersection based on the first prediction time and generate first diversion prediction information; a diversion optimization module is used to combine the first diversion prediction information, the driving purpose deterministic label, and the initial planning paths corresponding to the first set of connected vehicles to perform intersection diversion optimization for connected vehicles and generate a first path optimization result set corresponding to the first set of connected vehicles; and an optimization control module is used to return the first path optimization result set to the control terminal of the corresponding intelligent connected vehicle.

[0011] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0012] The collaborative path planning method for intelligent connected vehicles provided in this application determines multiple initial planned paths for multiple intelligent connected vehicles within a preset area by connecting to a V2X communication system. Each intelligent connected vehicle carries a corresponding driving purpose deterministic label. The method parses the multiple initial planned paths to identify a first set of connected vehicles located at a first intersection at a first prediction time. Based on the diversion characteristics of traditional vehicles at the first intersection at the first prediction time, it generates first diversion prediction information. Finally, it combines the first diversion prediction information, the driving purpose deterministic label, and the initial planned paths corresponding to the first set of connected vehicles to execute traffic flow prediction. The intersection traffic diversion optimization generates a first path optimization result set corresponding to the first set of connected vehicles. This first path optimization result set is then returned to the control terminal of the corresponding intelligent connected vehicle. This solves the technical problem in existing technologies where there is a lack of collaborative traffic management between intelligent connected vehicles and traditional vehicles, leading to decision conflicts, path conflicts, low diversion efficiency, traffic hazards, and congestion. The solution achieves collaborative path planning between intelligent connected vehicles and traditional vehicles, thereby improving traffic flow and overall traffic efficiency at intersections, reducing traffic congestion and conflicts, and enhancing traffic safety and efficiency. Attached Figure Description

[0013] Figure 1 This application provides a flowchart illustrating the collaborative path planning method for intelligent connected vehicles.

[0014] Figure 2 This application provides a schematic diagram of the collaborative path planning system structure for intelligent connected vehicles.

[0015] Figure labeling: Initial planning path acquisition module 11, path parsing module 12, diversion feature prediction module 13, diversion optimization module 14, optimization control module 15. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0017] Example 1, as Figure 1 As shown, this application provides a cooperative path planning method for intelligent connected vehicles, the method comprising:

[0018] A V2X communication system is connected to determine multiple initial planned paths for various intelligent connected vehicles within a preset area. Each intelligent connected vehicle carries a corresponding driving purpose deterministic tag. Further, the driving purpose deterministic tag includes a destination-deterministic tag and a destination-uncertain tag. The destination-deterministic tag is used when the vehicle has a definite driving task and destination, while the destination-uncertain tag is used when the vehicle does not have a definite driving task and destination.

[0019] Specifically, V2X (Vehicle-to-Everything) communication systems refer to communication systems between vehicles, road infrastructure, pedestrians, networks, and other traffic-related entities. V2X technology enables vehicles to perceive their surrounding traffic environment through real-time information exchange, thereby optimizing driving decisions and improving traffic safety, smoothness, and intelligence. V2X communication systems can collect and share real-time data such as the location, speed, and direction of travel of multiple vehicles, generating multiple feasible initial planned paths for each intelligent connected vehicle. These paths are generated based on factors including the vehicle's current traffic conditions, driving objectives, and the system's predictions of road conditions. Notably, each intelligent connected vehicle's initial planned path is not singular but rather a set of multiple feasible route candidates formed by analyzing different possible paths. Further optimization analysis is then used to select the optimal path for execution.

[0020] To ensure the accuracy and relevance of route planning, each intelligent connected vehicle carries a driving purpose deterministic tag. This tag clarifies the vehicle's current driving task and destination, serving as a crucial basis for route planning. Depending on different task requirements, driving purpose deterministic tags are divided into two types: "destination-determined tags" and "destination-uncertain tags." Specifically, a destination-determined tag indicates the vehicle's state when it has a definite driving task and target destination. For example, a vehicle might be performing a passenger-carrying task with a specific address as its destination; the system will optimize the route based on this destination information. In this case, the vehicle's route planning will focus on reaching the destination quickly and directly, while avoiding potential traffic congestion and delays. Conversely, a destination-uncertain tag indicates that the vehicle does not have a clear destination or task. This typically occurs in autonomous driving scenarios, such as a driverless taxi in an "empty" state when there are no fixed passengers. For example, in a ride-sharing platform, an autonomous vehicle might be searching for passengers; at this time, the vehicle's task and destination are not yet determined and may be dynamically adjusted based on real-time traffic flow, surrounding demand, and other information. Therefore, in route planning, vehicles with uncertain destination labels will consider more variables, such as the behavior of other vehicles and changes in traffic signals, in order to flexibly adapt to changing traffic environments.

[0021] By assigning appropriate driving purpose deterministic tags to each intelligent connected vehicle, the system can design personalized route planning schemes for each vehicle, ensuring that in different traffic scenarios, whether the vehicle has a definite destination or is in a destinationless state, it can obtain the optimal driving route. The core of this process lies in enabling information exchange between vehicles through V2X communication technology, thereby performing intelligent route adjustment and optimization in a multi-vehicle collaborative environment. Through the combination of multiple key steps such as the V2X communication system and driving purpose tags, the efficiency and accuracy of route planning are ensured, enabling real-time adjustments and optimizations in dynamically changing traffic environments to improve the traffic efficiency and safety of intelligent connected vehicles.

[0022] The multiple initial planned paths are analyzed to identify the first set of connected vehicles located at the first intersection at the first prediction time.

[0023] Optionally, the initial planned routes are generated by the V2X communication system based on the current location, speed, destination, and other information of each intelligent connected vehicle. These routes take into account the vehicle's current condition and the surrounding traffic environment. By evaluating different routes, multiple preliminary route planning schemes are generated. However, these routes are not necessarily optimal, but rather a set of feasible routes after preliminary screening. Next, the system needs to further analyze and optimize these routes to ensure that the route chosen by each vehicle can meet the requirements of traffic flow and safety during actual driving.

[0024] Based on this, the first set of connected vehicles located at the first intersection at the first predicted time is identified. This step aims to provide crucial traffic dynamic information for subsequent route optimization. Specifically, the first predicted time refers to the future time calculated by the system based on real-time traffic flow predictions and vehicle positions during the route planning process. For example, near intersections, intelligent connected vehicles and conventional vehicles may be affected by traffic light control or other traffic conditions; therefore, these factors must be considered when determining the route. Through the V2X communication system, vehicles can share current location, speed, and other data. Based on this information, the system calculates which vehicles will arrive at the first intersection at the first predicted time. The first set of connected vehicles refers to the set of all intelligent connected vehicles near the intersection at that time, including all vehicles that may intersect with the intersection. The movement trajectories of these vehicles at the intersection are one of the key factors in route planning and must be given special attention. For example, suppose there are three intelligent connected vehicles A, B, and C, which come from different road segments and are expected to approach the same intersection at some future time. At the initial prediction moment, the system collects data such as the location, speed, and driving intentions of these vehicles through the V2X communication system, and predicts their specific arrival time and location at the intersection. These vehicles are thus identified as the "first connected vehicle set" and become the targets for subsequent traffic diversion optimization.

[0025] By analyzing the initial planned routes of vehicles and predicting traffic dynamics at intersections, it is possible to identify in real time which intelligent connected vehicles are likely to arrive at intersections at the same time, thus providing necessary information support for subsequent route optimization and diversion strategies. This process takes place in the early stages of route planning, laying the foundation for subsequent fine-tuning of routes.

[0026] Based on the first prediction time, the diversion characteristics of traditional vehicles at the first intersection are predicted to generate the first diversion prediction information.

[0027] For example, the prediction of traffic diversion characteristics for traditional vehicles refers to analyzing the traffic flow of traditional vehicles (i.e., non-connected vehicles) passing through the first intersection to predict their driving behavior and make route optimization decisions based on this. First, based on the first prediction time, the system predicts the possible traffic conditions at that moment by real-time monitoring of the intersection and its surrounding traffic environment. The first prediction time refers to the future time calculated by the system based on vehicle location information and traffic light status, etc. For example, if the system predicts that a large number of vehicles will pass through the intersection within the next 5 minutes, then this 5 minutes is the first prediction time. At this point, the system generates a prediction result based on historical data, traffic flow models, and real-time road information.

[0028] Next, the traffic flow characteristics of the first intersection are analyzed and identified, especially the distribution of conventional vehicles. Specifically, the diversion characteristics of conventional vehicles refer to their travel behavior and route selection at the intersection. Typically, the travel paths of conventional vehicles are relatively fixed and are greatly influenced by factors such as traffic lights and road conditions. The system uses historical traffic flow data, intersection traffic signal cycles, and traffic volume in each direction to predict the possible traffic flow distribution in different lanes. For example, if one lane has a higher traffic volume while another lane has a lower traffic volume, conventional vehicles may tend to choose the lane with the lower traffic volume. These behaviors and preferences will affect the route planning decisions of intelligent connected vehicles, therefore requiring detailed analysis through a diversion prediction model.

[0029] Furthermore, generating the first diversion prediction information refers to generating a set of data reflecting the traditional traffic flow conditions of different lanes and directions by predicting the traditional traffic flow characteristics of the first intersection. This prediction information typically includes the traffic density, vehicle transit time, and potential congestion for each lane. For example, assuming the northbound lane of the first intersection has a high traffic volume at the first prediction time, the system will predict that this lane may become congested in the future by analyzing historical and real-time data, generating high congestion probability information for that lane. In contrast, if the eastbound lane has a lower traffic volume, the system will predict that the lane has a higher traffic flow smoothness, generating low congestion probability information. By predicting these traditional vehicle diversion characteristics, necessary decision-making basis can be provided for subsequent route optimization. This diversion prediction information will be input into the intelligent connected vehicle's route planning system to help the intelligent connected vehicle determine which lanes are currently relatively smooth and which lanes are likely to be congested, thereby making reasonable route selection and adjustments.

[0030] By predicting and analyzing traffic flow, potential traffic congestion areas can be identified in advance, and this information can be integrated into the route planning of intelligent connected vehicles to ensure that vehicles can avoid congested areas, improve traffic efficiency, and enhance the overall safety and smoothness of traffic.

[0031] Combining the first diversion prediction information, the driving purpose deterministic label, and the initial planned path corresponding to the first set of connected vehicles, intersection connected vehicle diversion optimization is performed to generate a first path optimization result set corresponding to the first set of connected vehicles. The first path optimization result set is then returned to the control terminal of the corresponding intelligent connected vehicle.

[0032] Specifically, when optimizing the traffic flow of connected vehicles at intersections, the system needs to comprehensively consider traffic flow prediction information and driving destination labels, analyze the vehicle's initial path and destination state, and generate an optimized driving path. The process of optimizing the traffic flow of connected vehicles at intersections includes the following key steps: First, path selection and adjustment. Based on the initial traffic flow prediction information, the system analyzes the traffic volume and traffic conditions of each lane and selects the most suitable lane for each intelligent connected vehicle. For example, in lanes with high traffic volume, the system may suggest that vehicles choose other lanes to avoid congestion, while in lanes with smooth traffic flow, vehicles can pass smoothly. Second, combining destination and smoothness. For vehicles with a known destination, the system guides them to the shortest path or the lane with the highest smoothness based on their destination label. For vehicles with an uncertain destination, the system will adjust their driving path based on real-time traffic flow and prediction results to ensure that they can choose the optimal path during their journey and avoid congestion. Third, dynamic path updates. At each intersection, the system dynamically adjusts the path planning, optimizing the path selection based on the current traffic flow conditions and the real-time location of the vehicles to ensure smooth passage. For example, if an unexpected congestion occurs in a lane, the system will automatically replan the route, adjusting the vehicle's direction or choice. Through these steps, a first set of optimized route results is generated for the first set of connected vehicles. This means that an optimized driving route is generated for each intelligent connected vehicle, taking into full account the traffic flow conditions at intersections, the driving intentions of vehicles, and the behavior of other traffic participants, aiming to maximize traffic efficiency and reduce congestion. Finally, returning the first set of optimized route results to the control terminal of the corresponding intelligent connected vehicle is a crucial step in the entire route optimization process. The system transmits the optimized route results to the control terminal of each intelligent connected vehicle, and the vehicle makes real-time adjustments based on this information, thus smoothly executing the new route planning and driving along the optimized route. In this way, intelligent connected vehicles can collaborate with other vehicles and traffic facilities in real time, ensuring that the most suitable driving route is always selected in complex traffic environments, thereby improving overall traffic flow and safety.

[0033] Furthermore, by combining the first diversion prediction information, the driving purpose deterministic label, and the initial planned path corresponding to the first set of connected vehicles, intersection connected vehicle diversion optimization is performed to generate a first path optimization result set corresponding to the first set of connected vehicles, including:

[0034] The system invokes a traffic flow prediction model to analyze traditional traffic flow indicators in each direction of the first intersection based on the first traffic flow prediction information. The traffic flow prediction model is constructed by collecting historical traffic flow information and corresponding vehicle communication flow samples in each direction. The system then analyzes the lane-changing cost of the first intersection by combining the driving purpose deterministic label with the initial planned path corresponding to the first set of connected vehicles, generating a first lane-changing cost set. Finally, it optimizes traffic flow by combining the traditional traffic flow indicators in each direction with the first lane-changing cost set, generating a first path optimization result set.

[0035] In one specific embodiment, a traffic flow prediction model is invoked to analyze the traditional traffic flow indicators in each direction of the first intersection based on the first traffic flow prediction information. This aims to optimize the path planning of intelligent connected vehicles by predicting the traffic flow and driving status of traditional vehicles. This process involves dynamically analyzing the traffic flow at the intersection and combining historical data and real-time information to predict the traffic flow in each direction. The traffic flow prediction model is built by training on historical traffic flow information of the intersection and vehicle flow samples in each direction. The core of this model lies in predicting future traffic flow status using historical traffic data (such as past traffic flow, flow, traffic signal cycles, etc.) and current traffic information (such as real-time traffic density, traffic signal status, etc.). Historical data provides the foundation for the model, helping it learn traffic flow patterns, while real-time flow samples provide immediate analytical basis based on current traffic conditions. Using this data, the traffic flow prediction model can calculate the traditional traffic flow indicators for each lane or direction, i.e., the smoothness of each lane at a specific time. This indicator reflects the ease with which vehicles can pass through a lane, typically measured by parameters such as traffic density, speed, and congestion level. For example, if traffic density is high and speed is slow in a particular direction, the smoothness indicator for that direction may be low, and vice versa. By calculating these smoothness indicators, the system identifies which lanes are relatively uncongested and which are likely to be congested, providing data support for subsequent route optimization.

[0036] Next, the lane-changing cost analysis for the first intersection is performed by combining the driving purpose definite label with the initial planned path corresponding to the first set of connected vehicles. The purpose of this step is to calculate the cost of a connected vehicle changing lanes at the first intersection. The calculation of the lane-changing cost is based on the vehicle's driving purpose label and its initial planned path. The driving purpose definite label can be a "destination-definite label" or a "destination-undefinite label," the former indicating that the vehicle has a definite destination, and the latter indicating that the vehicle does not have a fixed destination in the short term. For example, if a connected vehicle has a definite destination and its initial planned path points to a potentially congested lane, the system will consider the congestion situation of that lane and the vehicle's lane-changing cost to optimize its path selection. The lane-changing cost typically consists of the following factors: path complexity, increased travel time, and potential congestion risk. During this process, the calculation of lane-changing costs also needs to consider generating a first set of lane-changing costs. This set includes the cost incurred by each vehicle when changing lanes. Specifically, this may include increased path complexity (e.g., switching from the left lane to the right lane may require longer time or more complex operations), increased total route distance (e.g., the route after the lane change is longer than the initial route), and increased historical congestion data (e.g., lanes in a certain direction have historically been frequently congested). These costs will help the system determine whether certain lane changes are worthwhile, avoiding unnecessary lane-changing operations.

[0037] Based on this, traffic flow optimization is performed by combining the traditional traffic flow indicators for each direction with the first lane-changing cost set. Specifically, the system comprehensively considers the traditional traffic flow indicators and lane-changing cost sets for different directions to perform traffic flow optimization. The purpose of traffic flow optimization is to select the most suitable path for each connected vehicle based on traffic flow and lane-changing cost. For example, if the traffic flow in a certain direction is high and the lane-changing cost is low, the system may recommend that the vehicle switch to another, more efficient lane; conversely, if the traffic flow in a certain direction is low and the lane-changing cost is high, the system may choose to maintain the original path. In this way, the system can generate a first set of path optimization results, which includes the optimal path selection for each connected vehicle when passing through the first intersection. The goal of path optimization is to minimize traffic congestion, improve vehicle throughput, and reduce lane-changing costs.

[0038] Finally, the first set of optimized path results can be returned to the control terminal of the corresponding intelligent connected vehicle. This step ensures the execution of path optimization. The system transmits the optimized path information to the control system of each connected vehicle, and the vehicle adjusts its route according to these optimization results, thereby ensuring that the vehicle can pass smoothly through intersections and effectively avoid congestion. Through this path optimization mechanism, intelligent connected vehicles can not only improve traffic flow but also coordinate with other vehicles to achieve more efficient overall traffic operation.

[0039] Furthermore, by combining the driving purpose deterministic label with the initial planned path corresponding to the first connected vehicle set, a lane-changing cost analysis is performed at the first intersection to generate a first lane-changing cost set, including:

[0040] The initial planned paths of each connected vehicle in the first connected vehicle set are analyzed, and path replanning for intersection changes is performed. The route complexity growth index, total route distance growth index, and historical congestion probability growth index of the replanned path compared to the initial planned path are calculated and weighted to generate each first initial lane change cost. Based on the driving purpose determinism label of each connected vehicle in the first connected vehicle set, the lane change cost is weighted and optimized. The first initial lane change cost is then generated and added to the first lane change cost set.

[0041] Furthermore, the initial planned paths of each connected vehicle within the first connected vehicle set are analyzed, and path replanning for intersection direction changes is performed. First, the initial path of each connected vehicle is analyzed in detail to determine whether a direction change operation is necessary at the intersection. The goal of path replanning is to dynamically adjust the travel route of each vehicle based on the specific traffic conditions at the intersection and the driving needs of the vehicles, in order to optimize traffic flow and avoid congestion as much as possible.

[0042] After analyzing the initial path, the system calculates a route complexity growth metric for the replanned path compared to the initial planned path. This metric measures the increase in route complexity during the replanning. Specifically, a higher complexity growth occurs if the path crosses more intersections or involves more lane changes, and vice versa. This metric reflects the operational complexity of the replanned path and the control resources required by the vehicle. For example, if a vehicle changes from a straight lane to a left-turn lane, the operational complexity is higher than staying in the original lane, thus increasing the route complexity growth metric. Additionally, the system calculates a total route distance growth metric, which measures the change in total travel distance during the replanned path. This metric increases if the replanned path is longer than the initial path. This typically occurs when detours are chosen to avoid congestion. Calculating the total distance growth metric helps the system evaluate the efficiency of the replanned path, ensuring that the optimized path does not result in unnecessary time waste due to excessive detours. Simultaneously, a historical congestion probability growth metric is also calculated to assess the increased probability of congestion occurring on the replanned path over a certain period. The purpose of this indicator is to avoid selecting routes that have a history of frequent congestion. Specifically, the system predicts the congestion risk of the current route based on historical traffic data, such as traffic flow and the frequency of congestion over a past period. If a route has a history of frequent congestion, the replanned route may avoid selecting that route, thereby reducing the probability of traffic congestion.

[0043] Based on the steps described above, a weighted average is applied to generate the initial first-order switching costs. The purpose of this weighting step is to comprehensively consider factors such as route complexity, total distance increase, and historical congestion probability increase, and to assign weights based on the relative importance of these factors. For example, if avoiding congestion is more important than shortening the distance, the system might assign a higher weight to the congestion probability increase indicator. Through this weighting, the final generated initial first-order switching costs can comprehensively reflect the optimization effect of each path, helping the decision-making system assess which path is best suited for the current traffic conditions.

[0044] Next, based on the driving purpose determinism labels of each connected vehicle within the first connected vehicle set, a weight reduction assignment is performed on the lane-changing cost. The driving purpose determinism label reflects whether the vehicle's destination is clear. If the vehicle's destination is clear and unchangeable, the system considers the vehicle's route adjustment need to be low, thus reducing the lane-changing cost weight for that vehicle. This makes the optimization process more inclined to select more flexible route planning for vehicles with unclear goals, in order to better cope with traffic changes. After the weight reduction assignment, the system will optimize each of the first initial lane-changing costs. Specifically, the weight reduction assignment reduces the impact of certain cost factors on route selection, allowing the system to pay more attention to the vehicle's driving purpose and current road conditions during the optimization process. For example, for vehicles with a clear destination, route optimization focuses more on the shortest and least risky route; while for vehicles without a fixed destination, route optimization can more flexibly select routes with greater variability.

[0045] Finally, the first lane-change costs are generated and added to the first lane-change cost set. In this step, the system generates a final lane-change cost for each vehicle using the weighted and optimized lane-change costs described above. Each vehicle's lane-change cost represents the cost required for that vehicle to perform a lane-change operation under the current road conditions and destination. The lane-change costs of all vehicles are collected and formed into a set, which serves as the basis for path optimization, helping the system determine which vehicles need to change lanes and which vehicles should continue on their original routes.

[0046] By following the steps above, the routes of connected vehicles at intersections can be optimized, ensuring that each vehicle can pass through intersections in the best way in complex traffic environments, thereby improving overall traffic flow and avoiding traffic congestion.

[0047] Furthermore, based on the deterministic labels of the driving purposes of each connected vehicle within the first set of connected vehicles, a weighted reduction assignment is performed on the lane-changing cost. The weighted reduction assignment results are then used to optimize each initial lane-changing cost, generating each initial lane-changing cost, including:

[0048] If the driving purpose deterministic label of any connected vehicle is a destination deterministic label, the corresponding weight reduction value is assigned to 0; if the driving purpose deterministic label of any connected vehicle is a destination uncertain label, with the goal of minimizing the lane-changing cost in each connected vehicle, the corresponding weight reduction value is determined, the first lane-changing cost is adjusted, and the respective first lane-changing costs are generated.

[0049] For example, if the driving purpose determination label of any connected vehicle is a destination determination label, the corresponding reduction weight will be assigned a value of 0. The purpose of this step is to ensure that for connected vehicles with clear driving tasks and determined destinations, their route optimization process is not overly affected by lane-changing costs. Specifically, a destination determination label means that the connected vehicle's driving task is clear and the destination is fixed, such as having booked a fixed destination through a ride-hailing platform system, and the vehicle's driving route does not require much flexibility adjustment. Therefore, the system will assign a reduction weight of zero for lane-changing costs for these vehicles, meaning the originally calculated cost will not be changed. The system will assume that the route optimization for these vehicles only focuses on the shortest path or the most direct route, without considering the complexity or potential cost of changing direction at intersections.

[0050] If any connected vehicle has a definite driving purpose label but an uncertain destination label, the system determines the corresponding weight reduction value with the goal of minimizing the route-changing cost among all connected vehicles. In this case, the vehicle's driving purpose is unclear and may be in a flexible state, such as the autonomous vehicles in a ride-sharing platform, where the destination is undetermined or diverse. For these vehicles, minimizing the route-changing cost becomes a crucial objective in path optimization. This is because vehicles with uncertain destination labels are more likely to make multiple route adjustments on the road to adapt to constantly changing traffic environments and goals. Therefore, the system adjusts the route-changing costs of these vehicles based on current road conditions and other factors, reducing the weight assignment to make route selection more flexible and reduce the complexity of route adjustments.

[0051] Specifically, the system will adjust the lane-changing costs based on the individual lane-changing costs of each connected vehicle to generate the aforementioned first lane-changing costs. For vehicles with uncertain destinations, the system will dynamically adjust their lane-changing costs based on their location and the real-time traffic conditions around them. The adjustment process involves comparing the complexity of different paths, prioritizing paths with lower lane-changing costs. For example, if a path has high traffic volume and the vehicle is in a congested area, the system will assign a higher lane-changing cost to that path, while assigning lower costs to paths with smoother traffic or better road conditions. Therefore, the system will adjust the lane-changing costs in real time during the vehicle's journey to ensure that the goal is to minimize the lane-changing costs across all connected vehicles.

[0052] In this way, the various first lane-changing costs are generated. Adjusting and optimizing these costs ensures that vehicles with uncertain destinations can flexibly choose the optimal route in a dynamic traffic environment. Lane-changing costs consider not only road length and traffic flow, but also road segment complexity, traffic light status, and other factors. Ultimately, by adjusting the lane-changing cost for each connected vehicle, the system can generate a more accurate set of first lane-changing costs that reflects actual traffic needs, enabling each vehicle to choose the most suitable driving route at intersections and in other complex road conditions.

[0053] Furthermore, by combining the traditional traffic flow indicators in each direction with the first lane-changing cost set for traffic diversion optimization, the first path optimization result set is generated, including:

[0054] Based on the first lane-changing cost set, a target vehicle set with a lane-changing cost less than a preset cost threshold is extracted, and the corresponding lane-changing direction is determined. Using the traditional traffic flow indicators in each direction, as well as the target vehicle set and the corresponding lane-changing direction, traffic balance optimization is performed in each direction to generate traffic diversion optimization results in each direction. The initial planned paths of each connected vehicle in the first connected vehicle set are updated using the traffic diversion optimization results in each direction to generate the first path optimization result set.

[0055] Optionally, the first path optimization result set is generated by combining traditional traffic flow indicators in each direction with the first lane-changing cost set for diversion optimization. The core purpose of this step is to optimize the driving path of intelligent connected vehicles at intersections by combining traffic flow (traditional traffic flow indicators) with lane-changing cost information, thereby achieving more efficient traffic flow allocation. By combining these two types of information, the system can make more accurate path optimization decisions. Specifically, based on the first lane-changing cost set, a target vehicle set with lane-changing costs less than a preset cost threshold is extracted, and the corresponding lane-changing direction is determined. The purpose of this step is to screen out vehicles with relatively low lane-changing costs from all connected vehicles to be optimized. These vehicles are more adaptable to lane-changing adjustments and have less impact on traffic flow. The preset cost threshold is a standard set according to traffic conditions, and factors such as traffic congestion and road segment complexity are usually considered during system design. When the lane-changing cost of a connected vehicle is lower than this threshold, the system considers these vehicles to be a target vehicle set for path optimization and further determines their lane-changing direction. This direction determines where these vehicles will change lanes at the intersection, thereby achieving diversion.

[0056] Using the traditional traffic flow indicators for each direction, the target vehicle set, and the corresponding lane-changing direction, the system performs traffic balance optimization in each direction, generating diversion optimization results for each direction. In this step, the system not only considers the traffic flow of traditional vehicles in each direction but also incorporates the lane-changing direction of the target vehicles to further optimize the distribution of traffic flow. The goal of traffic balance optimization is to ensure that traffic flow in each direction is reasonably distributed, avoiding excessive congestion in some directions while other directions are too sluggish. This optimization process generates the optimal diversion scheme adapted to the current road conditions by comprehensively evaluating the traffic flow in each direction and the lane-changing strategies of the target vehicles. Then, the initial planned paths of each connected vehicle in the first connected vehicle set are updated using the diversion optimization results for each direction, generating the first path optimization result set. At this point, based on the previously calculated diversion optimization results for each direction, the system adjusts the initial planned path of each vehicle in the first connected vehicle set. The optimized path may include lane-changing suggestions, path length reduction, or traffic signal adjustments to ensure smoother and more efficient vehicle travel at and around intersections. Each connected vehicle's path is meticulously optimized to achieve the best traffic flow, ultimately generating a first set of optimized path results, which contains all the optimized paths.

[0057] By following the steps above, we can effectively optimize the traffic diversion problem of intelligent connected vehicles at intersections, ensure smoother traffic flow at intersections, reduce traffic congestion, and improve overall road utilization efficiency.

[0058] Furthermore, based on the prediction of the diversion characteristics of traditional vehicles at the first intersection at the first prediction time, first diversion prediction information is generated, including:

[0059] Historical traditional vehicle diversion data of the first intersection is collected, wherein the historical traditional vehicle diversion data carries historical time tags and includes historical data with periodic continuity; the historical time tags and the historical traditional vehicle diversion data are used as sample data to train and learn the relationship between diversion features and time, and generate a time-series diversion prediction model; the time-series diversion prediction model is used to predict the diversion at the first prediction time to generate the first diversion prediction information.

[0060] Specifically, historical traditional vehicle diversion data is collected at the first intersection. This historical traditional vehicle diversion data carries historical time-stamped labels and includes historical data with periodic continuity. This step aims to obtain historical traffic flow data for the first intersection to provide training samples for subsequent diversion prediction models. The historical traditional vehicle diversion data refers to the diversion and traffic conditions of traditional vehicles at a specific time, date, and road segment. For example, the data may include indicators such as traffic flow and traffic density for specific time periods (e.g., weekdays, holidays) or hourly. Simultaneously, this data also needs to carry historical time-stamped labels to ensure that each data point corresponds to a specific time and date for time-series processing during analysis.

[0061] The historical traffic diversion data includes historical data with periodic continuity. This means that the collected historical data should be continuous and periodic to ensure that the predictive model can identify and learn the regular changes in traffic flow. For example, when collecting data, traffic data for each day and each time point within two years can be selected to ensure sufficient time intervals and regularity between data points. This periodic data helps the model learn the traffic flow trends in different time periods (such as morning rush hour and evening rush hour), thereby improving the accuracy of the prediction results.

[0062] Using the historical time stamps and historical traditional vehicle diversion data as sample data, the system trains and learns the relationship between diversion characteristics and time, generating a time-series diversion prediction model. In this step, through the analysis of historical data, the system uses machine learning methods to train the model and learn the time-series characteristics of traffic flow. Specifically, the historical time stamps represent the time information of each data point. By combining them with the corresponding traditional vehicle diversion data, the model can learn the diversion patterns at different time points (e.g., different hours, different dates, or different seasons). This time-series diversion prediction model can capture the patterns of traffic flow changes over time, providing an accurate basis for future traffic flow prediction.

[0063] The time-series diversion prediction model is used to predict diversion at the first prediction time, generating the first diversion prediction information. At this point, the trained time-series diversion prediction model is applied to future traffic flow prediction. Specifically, the model predicts the traditional vehicle diversion situation at the first prediction time based on current historical data and time-series characteristics. This prediction result is the first diversion prediction information, which contains predicted vehicle flow data in each direction at the intersection at the first prediction time. This prediction information is crucial for subsequent route optimization and traffic scheduling because it provides the system with future traffic flow trends, helping intelligent connected vehicles make more reasonable route adjustments.

[0064] In summary, by collecting historical traditional vehicle diversion data with periodic continuity and utilizing a time-series diversion prediction model, accurate predictions of future traffic flow can be made. This method effectively improves the accuracy of traffic flow prediction and provides data support for route optimization for intelligent connected vehicles and diversion optimization at intersections.

[0065] Through the technical solutions of the above embodiments, the collaborative path planning method for intelligent connected vehicles provided in this application solves the technical problem of the lack of collaborative traffic management between intelligent connected vehicles and traditional vehicles in the prior art, which leads to decision conflicts between intelligent connected vehicles and traditional vehicles, resulting in path conflicts and low diversion efficiency, causing traffic hazards and congestion. It achieves collaborative path planning between intelligent connected vehicles and traditional vehicles, thereby improving traffic flow and overall traffic efficiency at intersections, reducing traffic congestion and conflicts, and improving traffic safety and efficiency.

[0066] Example 2, based on the same inventive concept as the cooperative path planning method for intelligent connected vehicles in the previous examples, such as... Figure 2 As shown, this application provides a cooperative path planning system for intelligent connected vehicles, the system comprising:

[0067] The initial planning path acquisition module 11 is used to connect to the V2X communication system and determine multiple initial planning paths for multiple intelligent connected vehicles in a preset area, wherein each intelligent connected vehicle carries a corresponding driving purpose deterministic label.

[0068] The path parsing module 12 is used to parse the multiple sets of initial planned paths and identify the first set of connected vehicles located at the first intersection at the first prediction time.

[0069] The diversion feature prediction module 13 is used to predict the diversion features of traditional vehicles at the first intersection based on the first prediction time, and generate first diversion prediction information.

[0070] The diversion optimization module 14 is used to combine the first diversion prediction information, the driving purpose deterministic label and the initial planning path corresponding to the first connected vehicle set to perform intersection connected vehicle diversion optimization, and generate a first path optimization result set corresponding to the first connected vehicle set.

[0071] The optimization control module 15 is used to return the first path optimization result set to the control terminal of the corresponding intelligent connected vehicle.

[0072] Furthermore, the traffic splitting optimization module 14 is also used to perform the following steps:

[0073] The system invokes a traffic flow prediction model to analyze traditional traffic flow indicators in each direction of the first intersection based on the first traffic flow prediction information. The traffic flow prediction model is constructed by collecting historical traffic flow information and corresponding vehicle communication flow samples in each direction. The system then analyzes the lane-changing cost of the first intersection by combining the driving purpose deterministic label with the initial planned path corresponding to the first set of connected vehicles, generating a first lane-changing cost set. Finally, it optimizes traffic flow by combining the traditional traffic flow indicators in each direction with the first lane-changing cost set, generating a first path optimization result set.

[0074] Furthermore, the traffic splitting optimization module 14 is also used to perform the following steps:

[0075] The initial planned paths of each connected vehicle in the first connected vehicle set are analyzed, and path replanning for intersection changes is performed. The route complexity growth index, total route distance growth index, and historical congestion probability growth index of the replanned path compared to the initial planned path are calculated and weighted to generate each first initial lane change cost. Based on the driving purpose determinism label of each connected vehicle in the first connected vehicle set, the lane change cost is weighted and optimized. The first initial lane change cost is then generated and added to the first lane change cost set.

[0076] Furthermore, the traffic splitting optimization module 14 is also used to perform the following steps:

[0077] If the driving purpose deterministic label of any connected vehicle is a destination deterministic label, the corresponding weight reduction value is assigned to 0; if the driving purpose deterministic label of any connected vehicle is a destination uncertain label, with the goal of minimizing the lane-changing cost in each connected vehicle, the corresponding weight reduction value is determined, the first lane-changing cost is adjusted, and the respective first lane-changing costs are generated.

[0078] Furthermore, the traffic splitting optimization module 14 is also used to perform the following steps:

[0079] Based on the first lane-changing cost set, a target vehicle set with a lane-changing cost less than a preset cost threshold is extracted, and the corresponding lane-changing direction is determined. Using the traditional traffic flow indicators in each direction, as well as the target vehicle set and the corresponding lane-changing direction, traffic balance optimization is performed in each direction to generate traffic diversion optimization results in each direction. The initial planned paths of each connected vehicle in the first connected vehicle set are updated using the traffic diversion optimization results in each direction to generate the first path optimization result set.

[0080] Furthermore, the initial planned path acquisition module 11 also includes:

[0081] The driving purpose determination label includes a destination determination label and a destination uncertainty label, wherein the destination determination label is the label when the vehicle has a definite driving task and a definite task destination, and the destination uncertainty label is the label when the vehicle does not have a definite driving task and a definite task destination.

[0082] Furthermore, the diversion feature prediction module 13 is also used to perform the following steps:

[0083] Historical traditional vehicle diversion data of the first intersection is collected, wherein the historical traditional vehicle diversion data carries historical time tags and includes historical data with periodic continuity; the historical time tags and the historical traditional vehicle diversion data are used as sample data to train and learn the relationship between diversion features and time, and generate a time-series diversion prediction model; the time-series diversion prediction model is used to predict the diversion at the first prediction time to generate the first diversion prediction information.

[0084] Through the foregoing detailed description of the cooperative path planning method for intelligent connected vehicles, those skilled in the art can clearly understand the cooperative path planning system for intelligent connected vehicles in this embodiment. As the system disclosed in the embodiment corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0085] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cooperative path planning method for intelligent connected vehicles, characterized in that, include: Connect to the V2X communication system to determine multiple initial planned paths for multiple intelligent connected vehicles within a preset area, wherein each intelligent connected vehicle carries a corresponding driving purpose deterministic label; Analyze the multiple sets of initial planned paths to identify the first set of connected vehicles located at the first intersection at the first prediction time; Based on the first prediction time, the diversion characteristics of traditional vehicles at the first intersection are predicted to generate the first diversion prediction information. Combining the first diversion prediction information, the driving purpose determination label, and the initial planning path corresponding to the first connected vehicle set, perform intersection connected vehicle diversion optimization to generate a first path optimization result set corresponding to the first connected vehicle set. The first path optimization result set is returned to the control terminal of the corresponding intelligent connected vehicle; Specifically, by combining the first diversion prediction information, the driving purpose deterministic label, and the initial planned path corresponding to the first connected vehicle set, intersection connected vehicle diversion optimization is performed to generate a first path optimization result set corresponding to the first connected vehicle set, including: The diversion smoothness prediction model is invoked, and the traditional traffic flow indicators in each direction of the first intersection are analyzed based on the first diversion prediction information. The diversion smoothness prediction model is constructed by collecting historical diversion information and corresponding vehicle communication smoothness samples in each direction. By combining the driving purpose deterministic label with the initial planning path corresponding to the first connected vehicle set, the lane-changing cost analysis of the first intersection is performed to generate the first lane-changing cost set; By combining the traditional traffic flow indicators in each direction with the first lane change cost set, traffic diversion optimization is performed to generate the first path optimization result set. Specifically, by combining the driving purpose deterministic label with the initial planned path corresponding to the first connected vehicle set, a lane-changing cost analysis is performed at the first intersection to generate a first lane-changing cost set, including: The initial planned paths of each connected vehicle in the first set of connected vehicles are analyzed, and the path replanning for intersection reversal is performed. The route complexity growth index, total route distance growth index, and historical congestion probability growth index of the replanned path compared to the initial planned path are calculated and weighted to generate the initial lane change cost for each vehicle. Based on the driving purpose deterministic label of each connected vehicle in the first connected vehicle set, the lane change cost is reduced by weighting, and the reduced weighting result is used to optimize each first initial lane change cost to generate each first lane change cost and add it to the first lane change cost set. Specifically, based on the deterministic driving purpose labels of each connected vehicle within the first connected vehicle set, a weighted reduction assignment is performed on the lane-changing cost. The weighted reduction assignment results are then used to optimize each initial lane-changing cost, generating each initial lane-changing cost, including: If the driving purpose deterministic label of any connected vehicle is the destination deterministic label, the corresponding weight reduction value will be assigned to 0; If the driving purpose deterministic label of any connected vehicle is the destination uncertain label, with the goal of minimizing the lane-changing cost among all connected vehicles, the corresponding weight reduction assignment is determined, the first lane-changing cost is adjusted, and the various first lane-changing costs are generated.

2. The cooperative path planning method for intelligent connected vehicles as described in claim 1, characterized in that, By combining the traditional traffic flow indicators in each direction with the first lane-changing cost set for traffic diversion optimization, the first path optimization result set is generated, including: Based on the first lane-changing cost set, extract the target car set whose lane-changing cost is less than a preset cost threshold, and determine the corresponding lane-changing direction; Using the traditional traffic flow indicators in each direction, as well as the target vehicle set and the corresponding lane-changing direction, perform traffic balance optimization in each direction to generate diversion optimization results in each direction. The initial planned paths of each connected vehicle in the first connected vehicle set are updated based on the traffic diversion optimization results in each direction, thereby generating the first path optimization result set.

3. The cooperative path planning method for intelligent connected vehicles as described in claim 1, characterized in that, The driving purpose determination label includes a destination determination label and a destination uncertainty label, wherein the destination determination label is the label when the vehicle has a definite driving task and a definite task destination, and the destination uncertainty label is the label when the vehicle does not have a definite driving task and a definite task destination.

4. The cooperative path planning method for intelligent connected vehicles as described in claim 1, characterized in that, Based on the first prediction time, the diversion characteristics of traditional vehicles at the first intersection are predicted, and first diversion prediction information is generated, including: Collect historical traditional vehicle diversion data at the first intersection, wherein the historical traditional vehicle diversion data carries historical time tags and includes historical data with periodic continuity; Using the historical time tags and the historical traditional vehicle diversion data as sample data, the relationship between diversion features and time is trained to generate a time-series diversion prediction model; The time-series diversion prediction model is used to predict the diversion at the first prediction time to generate the first diversion prediction information.

5. A collaborative path planning system for intelligent connected vehicles, characterized in that, The system is used to implement the cooperative path planning method for intelligent connected vehicles according to any one of claims 1-4, the system comprising: The initial planning path acquisition module is used to connect to the V2X communication system and determine multiple initial planning paths for multiple intelligent connected vehicles in a preset area, wherein each intelligent connected vehicle carries a corresponding driving purpose deterministic label. The path parsing module is used to parse the multiple sets of initial planned paths and identify the first set of connected vehicles located at the first intersection at the first prediction time. The diversion feature prediction module is used to predict the diversion features of traditional vehicles at the first intersection based on the first prediction time, and generate the first diversion prediction information. The diversion optimization module is used to combine the first diversion prediction information, the driving purpose deterministic label and the initial planning path corresponding to the first connected vehicle set to perform intersection connected vehicle diversion optimization, and generate a first path optimization result set corresponding to the first connected vehicle set. An optimization control module is used to return the first path optimization result set to the control terminal of the corresponding intelligent connected vehicle.

Images