Multi-traffic-flow interactive path optimization method and system based on AI decision

Abstract

The invention discloses a multi-traffic-flow interactive path optimization method and system based on AI decision, and relates to the technical field related to path optimization, and the method comprises the steps: obtaining a real-time position and an input end point of a target vehicle through an interactive vehicle navigation application; and taking the real-time position as a driving starting point, performing local driving path planning in combination with the input ending point, and outputting K standby feasible paths. And sending the K standby feasible paths to the AI decision cloud. And carrying out traffic sensing node local activation to obtain K groups of return path sensing information, analyzing the information, and screening and positioning a target driving path from K standby feasible paths. And constructing a multi-traffic-flow interaction condition based on the real-time position and the input end point. And sharing the target driving path to the associated vehicle through a V2X communication technology based on a multi-traffic-flow interaction condition. The technical problem that a traditional navigation system only depends on static map data and simple real-time traffic information to carry out path planning, so that the accuracy and efficiency of path planning are difficult to improve is solved.

Description

Technical Field

[0001] This application relates to the field of path optimization technologies, and specifically to a multi-traffic-flow interactive path optimization method and system based on AI decision-making. Background Art

[0002] With the acceleration of the urbanization process, the problem of traffic congestion has become increasingly serious. Traditional navigation systems often rely only on static map data and simple real-time traffic information for path planning, making it difficult to comprehensively consider multi-traffic-flow interaction and dynamic traffic perception data, resulting in inaccurate and inefficient path planning. In addition, traditional systems lack an intelligent deployment and activation mechanism for traffic perception nodes and cannot effectively use real-time traffic data for path optimization. Therefore, there is an urgent need for a path optimization method that can combine real-time traffic perception data, multi-traffic-flow interaction information, and intelligent decision-making to improve traffic efficiency and driving experience.

[0003] Therefore, in the prior art, traditional navigation systems rely only on static map data and simple real-time traffic information for path planning, resulting in the technical problems that it is difficult to improve the accuracy and efficiency of path planning. Summary of the Invention

[0004] This application provides a multi-traffic-flow interactive path optimization method and system based on AI decision-making, which solves the technical problems that in the prior art, traditional navigation systems rely only on static map data and simple real-time traffic information for path planning, resulting in difficult improvement of the accuracy and efficiency of path planning. By combining real-time traffic perception data and multi-traffic-flow interaction information, as well as intelligently deploying and activating traffic perception nodes, real-time traffic data is effectively used for path optimization, improving traffic efficiency and driving experience, and achieving more accurate and efficient path planning.

[0005] This application provides a multi-traffic-flow interactive path optimization method based on AI decision-making. The method includes: interacting with an in-vehicle navigation application to obtain the real-time position of a target vehicle and an input destination; using the real-time position as the driving starting point, combining with the input destination to perform local driving path planning, and outputting K alternative feasible paths; sending the K alternative feasible paths to an AI decision-making cloud; the AI decision-making cloud locally activates traffic perception nodes according to the K alternative feasible paths to obtain K sets of backhaul path perception information; the AI decision-making cloud screens and locates a target driving path from the K alternative feasible paths by analyzing the K sets of backhaul path perception information; constructing multi-traffic-flow interaction conditions based on the real-time position and the input destination; and the target vehicle shares the target driving path with associated vehicles through V2X communication technology based on the multi-traffic-flow interaction conditions.

[0006] In the implementation manner, taking the real-time position as the driving starting point, combining with the input end point to perform local driving path planning, and outputting K alternative feasible paths, the method includes: loading a local road network according to the real-time position and the input end point; traversing the local road network to obtain an initial driving path; obtaining M exploratory driving paths by randomly perturbing the initial driving path; and screening the K alternative feasible paths from the M exploratory driving paths according to driving characteristics.

[0007] In the implementation manner, before the AI decision cloud end locally activates traffic perception nodes according to the K alternative feasible paths to obtain K groups of backhaul path perception information, the method further includes: abstracting the target traffic road network of the target city to obtain a traffic road network topology; positioning W traffic perception sections and H traffic perception convergence points in the traffic road network topology according to the traffic flow of the target traffic road network; configuring sensors in the target city according to the W traffic perception sections and the H traffic perception convergence points to obtain W + H traffic perception nodes; and completing the local deployment of the AI decision cloud end by communicatively connecting the W + H traffic perception nodes with the AI decision cloud end.

[0008] In the implementation manner, positioning W traffic perception sections and H traffic perception convergence points in the traffic road network topology according to the traffic flow of the target traffic road network, the method includes: interactively obtaining multiple traffic flow records of multiple local sections in the traffic road network topology; analyzing the multiple traffic flow records to obtain multiple traffic flow fluctuation thresholds and multiple traffic flow stable values; presetting a traffic flow fluctuation scale and a traffic flow threshold; using the traffic flow fluctuation scale and the traffic flow threshold to map and traverse the multiple traffic flow fluctuation thresholds and the multiple traffic flow stable values to position W traffic perception sections; and positioning H traffic perception convergence points in the traffic road network topology according to the section intersection situation.

[0009] In an implementation manner, the AI decision cloud locates the target driving path by screening from K alternative feasible paths through analyzing the K groups of feedback path perception information. The method includes: interactively obtaining multiple groups of sample path perception information of multiple sample driving paths and multiple sample quantified passing efficiencies; constructing a passing efficiency quantification model with the multiple groups of sample path perception information and the multiple sample quantified passing efficiencies as training data; extracting Q pieces of real-time path perception information from the first group of feedback path perception information, where the real-time path perception information includes real-time traffic flow density, real-time traffic flow speed, and real-time road condition information; synchronizing the Q pieces of real-time path perception information to the passing efficiency quantification model and analyzing to output a first quantified passing efficiency; and so on, analyzing the K groups of feedback path perception information to obtain K quantified passing efficiencies; serializing the K quantified passing efficiencies and screening and locating the target driving path from the K alternative feasible paths according to the sorting result.

[0010] In an implementation manner, the K alternative feasible paths are screened from the M exploratory driving paths according to driving characteristics. The method includes: making an online data call for the first exploratory driving path to obtain a first driving characteristic, where the first driving characteristic includes a first driving distance and multiple first driving duration records; outputting a first driving feasibility by quantitatively evaluating the first driving characteristic; and so on, calculating M driving feasibilities of the M exploratory driving paths; presetting a feasibility threshold, comparing the M driving feasibilities with the feasibility threshold, and screening the K alternative feasible paths from the M exploratory driving paths.

[0011] In an implementation manner, the method for outputting a first driving feasibility by quantitatively evaluating the first driving characteristic includes: calculating a first average driving time based on the multiple first driving duration records; counting the multiple first driving duration records and outputting a first path application frequency; presetting a path application frequency threshold; judging whether the first path application frequency meets the path application frequency threshold; if the first path application frequency meets the path application frequency threshold, adding up the first driving distance, the first average driving time, and the first path application frequency based on a preset weighting rule, and outputting the first driving feasibility.

[0012] The present application also provides a multi-traffic-flow interactive path optimization system based on AI decision-making, including: an interaction module for interacting with in-vehicle navigation applications to obtain the real-time position of the target vehicle and the input destination; a feasible path acquisition module for performing local driving path planning with the real-time position as the driving starting point and combining the input destination to output K alternative feasible paths; a cloud interaction module for sending the K alternative feasible paths to the AI decision-making cloud; a perception information acquisition module for the AI decision-making cloud to locally activate traffic perception nodes based on the K alternative feasible paths to obtain K sets of backhaul path perception information; a feasible path screening module for the AI decision-making cloud to screen and locate the target driving path from the K alternative feasible paths by analyzing the K sets of backhaul path perception information; an interaction condition construction module for constructing multi-traffic-flow interaction conditions based on the real-time position and the input destination; and a sharing association module for the target vehicle to share the target driving path with associated vehicles based on the multi-traffic-flow interaction conditions through V2X communication technology.

[0013] It is intended to solve the technical problem in the prior art that traditional navigation systems only rely on static map data and simple real-time traffic information for path planning, resulting in difficulties in improving the accuracy and efficiency of path planning. By combining real-time traffic perception data and multi-traffic-flow interaction information, as well as intelligently deploying and activating traffic perception nodes, real-time traffic data is effectively utilized for path optimization, improving traffic efficiency and driving experience, and achieving more accurate and efficient path planning. The proposed multi-traffic-flow interactive path optimization method and system based on AI decision-making in the present application obtain the real-time position of the target vehicle and the input destination by interacting with in-vehicle navigation applications; perform local driving path planning with the real-time position as the driving starting point and combining the input destination to output K alternative feasible paths; send the K alternative feasible paths to the AI decision-making cloud; the AI decision-making cloud locally activates traffic perception nodes based on the K alternative feasible paths to obtain K sets of backhaul path perception information; the AI decision-making cloud screens and locates the target driving path from the K alternative feasible paths by analyzing the K sets of backhaul path perception information; constructs multi-traffic-flow interaction conditions based on the real-time position and the input destination; and the target vehicle shares the target driving path with associated vehicles based on the multi-traffic-flow interaction conditions through V2X communication technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the operations above or below do not necessarily need to be executed precisely in order. On the contrary, according to the needs, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0015] Figure 1 Schematic flowchart of the multi-traffic-flow interactive path optimization method based on AI decision-making provided by the embodiments of the present application;

[0016] Figure 2 Schematic structural diagram of the multi-traffic-flow interactive path optimization system based on AI decision-making provided by the embodiments of the present application.

[0017] Explanation of reference numerals: Interaction module 11, Feasible path acquisition module 12, Cloud interaction module 13, Perception information acquisition module 14, Feasible path screening module 15, Interaction condition construction module 16, Shared association module 17. Detailed implementation manners

[0018] The above description is only an overview of the technical solutions of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically gives the detailed implementation manners of the present application.

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "first" and "second" are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that comprises a series of steps or units need not be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0021] The embodiments of this application provide a multi-traffic-flow interactive path optimization method and system based on AI decision-making, as Figure 1 shown, the method includes:

[0022] Interact with the in-vehicle navigation application to obtain the real-time position of the target vehicle and the input destination; use the real-time position as the driving starting point, combine with the input destination to perform local driving path planning, and output K alternative feasible paths; send the K alternative feasible paths to the AI decision cloud.

[0023] By interacting with the navigation application installed on the vehicle, obtain the real-time position of the target vehicle and the input destination. The target vehicle is the vehicle that needs to perform path planning. The real-time position is the current position information obtained through the vehicle's GPS or other positioning systems. The input destination is the destination position input by the user through the in-vehicle navigation application. Subsequently, use the real-time position as the driving starting point, combine with the destination input by the user, call the local map data, and generate multiple possible paths through the path planning algorithm, and output K alternative feasible paths. Further, send the K alternative feasible paths to the AI decision cloud through the network, and the AI decision cloud is used to further analyze and process the feasible paths.

[0024] The method provided by the embodiments of this application further includes: loading the local road network according to the real-time position and the input destination; traversing the local road network to obtain the initial driving path; obtaining M exploratory driving paths by randomly perturbing the initial driving path; and screening the K alternative feasible paths from the M exploratory driving paths according to the driving characteristics.

[0025] Taking the real-time position as the driving starting point, combining with the input end point to perform local driving route planning, and outputting K alternative feasible routes. The method includes: loading a local road network, i.e., the road network information along the way, from the local map database according to the real-time position of the vehicle and the input end point. Exemplarily, when the vehicle departs from point A to point B, the system loads all possible roads between point A and point B. Subsequently, traverse the local road network and obtain an initial driving route by random selection. Further, generate new routes by randomly perturbing the initial driving route, such as randomly changing a certain node or section in the route, to obtain M exploratory driving routes. After obtaining the M exploratory driving routes, screen and obtain the K alternative feasible routes according to driving characteristics.

[0026] The AI decision cloud activates the traffic perception nodes locally according to the K alternative feasible routes to obtain K sets of backhaul route perception information; the AI decision cloud analyzes the K sets of backhaul route perception information, screens and locates the target driving route from the K alternative feasible routes; constructs multi-traffic flow interaction conditions based on the real-time position and the input end point; the target vehicle shares the target driving route with associated vehicles through V2X communication technology based on the multi-traffic flow interaction conditions.

[0027] After receiving the alternative routes, the AI decision cloud analyzes the areas covered by the routes and activates the traffic perception nodes in these areas. Based on the traffic perception nodes, perception information is sent back to the cloud. The perception information includes: traffic flow density, speed, road condition information, etc. For example, for K alternative routes, the cloud activates the sensor nodes along the routes. These nodes collect traffic flow density, speed, and road condition information in real time and send the data back to the cloud. Each route corresponds to a set of perception information, obtaining K sets of backhaul route perception information. The AI decision cloud analyzes the K sets of backhaul route perception information, sorts and screens the K alternative feasible routes according to the traffic efficiency to locate the target driving route. Finally, constructs multi-traffic flow interaction conditions based on the real-time position and the input end point, that is, when there are vehicles with the same real-time position and input end point, set traffic flow interaction conditions for them. The target vehicle shares the target driving route screened by the AI decision cloud with other associated vehicles through V2X communication technology. For example, the target vehicle sends the route information to other vehicles going to the same destination, and these vehicles can adjust their driving routes according to the shared route information to achieve collaborative optimization. This solves the technical problem in the prior art that traditional navigation systems only rely on static map data and simple real-time traffic information for route planning, resulting in difficulties in improving the accuracy and efficiency of route planning. By combining real-time traffic perception data and multi-traffic flow interaction information, as well as intelligently deploying and activating traffic perception nodes, effectively utilize real-time traffic data for route optimization, improve traffic efficiency and driving experience, and achieve more accurate and efficient route planning.

[0028] The method provided by the embodiment of the present application further includes: abstracting the target traffic road network of the target city to obtain a traffic road network topology; positioning W traffic perception sections and H traffic perception sink points in the traffic road network topology according to the traffic flow of the target traffic road network; performing sensor configuration in the target city according to the W traffic perception sections and H traffic perception sink points to obtain W + H traffic perception nodes; and completing the local deployment of the AI decision cloud by communicatively connecting the W + H traffic perception nodes with the AI decision cloud.

[0029] Before the AI decision cloud performs local activation of traffic perception nodes according to the K alternative feasible paths to obtain K sets of feedback path perception information, the method further includes: abstracting the traffic network of the target city into a topology graph, where roads are represented as edges, and intersections and nodes are represented as vertices, to obtain a traffic road network topology. Further, according to the traffic flow of the target traffic road network, W traffic perception sections and H traffic perception sink points are positioned in the traffic road network topology, that is, the actual traffic perception sections and traffic perception sink points are mapped into the topology graph. The traffic perception section refers to a section selected on the traffic road network for installing sensors to monitor the traffic flow density and speed in real time. The traffic perception sink point refers to a node where multiple sections of the traffic road network meet, such as an intersection or a transportation hub. Further, sensor configuration, such as traffic flow detectors, cameras, etc., is performed in the target city according to the W traffic perception sections and H traffic perception sink points to obtain W + H traffic perception nodes. Finally, by connecting the W + H traffic perception nodes with the AI decision cloud through V2X communication technology, real-time data transmission and processing are ensured, and the local deployment of the AI decision cloud is completed.

[0030] The method provided by the embodiment of the present application further includes: interactively obtaining multiple traffic flow records of multiple local sections in the traffic road network topology; analyzing the multiple traffic flow records to obtain multiple traffic flow fluctuation thresholds and multiple traffic flow stable values; presetting a traffic flow fluctuation scale and a traffic flow threshold; using the traffic flow fluctuation scale and the traffic flow threshold to map and traverse the multiple traffic flow fluctuation thresholds and multiple traffic flow stable values to position W traffic perception sections; and positioning H traffic perception sink points in the traffic road network topology according to the section intersection situation.

[0031] According to the traffic flow of the target traffic road network, W traffic perception sections and H traffic perception convergence points are located in the traffic road network topology. The method includes: interactively obtaining multiple traffic flow records of multiple local sections in the traffic road network topology, where the traffic flow records are vehicle flow data at different time nodes collected on different sections. Analyze the multiple traffic flow records to obtain multiple traffic flow fluctuation thresholds and multiple traffic flow stable values. The multiple traffic flow fluctuation thresholds are the fluctuation ranges of the vehicle flow on each section, and the multiple traffic flow stable values are the average values of the traffic flow on each section. Further, a traffic flow fluctuation scale and a traffic flow threshold are preset. The traffic flow fluctuation scale is used to measure the severity of the traffic flow fluctuation. The traffic flow fluctuation scale is a preset fluctuation range. When it is greater than this traffic flow fluctuation scale, the corresponding traffic flow fluctuates greatly, and the corresponding section can be set as a traffic perception section. The traffic flow threshold is a set passing traffic flow threshold, which is used to determine whether a section is a main section. When it is greater than this value, the corresponding section is a main section with a high passing traffic flow, and the corresponding section can be set as a traffic perception section to obtain section information in a timely manner. Using the traffic flow fluctuation scale and the traffic flow threshold, map and traverse the multiple traffic flow fluctuation thresholds and the multiple traffic flow stable values, and determine whether the multiple traffic flow fluctuation thresholds are greater than or equal to the traffic flow fluctuation scale, and at the same time whether the multiple traffic flow stable values are greater than or equal to the traffic flow threshold. When both are satisfied, W traffic perception sections are located. Finally, in the traffic road network topology, according to the section intersection situation, locate the convergence points where the traffic flow is concentrated, and locate H traffic perception convergence points according to the intersection points in the traffic road network topology.

[0032] The method provided by the embodiment of the present application further includes: interactively obtaining multiple groups of sample path perception information and multiple sample quantified passing efficiencies of multiple sample driving paths; using the multiple groups of sample path perception information and multiple sample quantified passing efficiencies as training data to construct a passing efficiency quantification model; extracting Q real-time path perception information from the first group of backhaul path perception information, where the real-time path perception information includes real-time traffic density, real-time traffic speed, and real-time road condition information; synchronize the Q real-time path perception information to the passing efficiency quantification model, and analyze and output the first quantified passing efficiency; and so on, analyze the K groups of backhaul path perception information to obtain K quantified passing efficiencies; serialize the K quantified passing efficiencies, and screen and locate the target driving path from the K alternative feasible paths according to the sorting result.

[0033] The AI decision-making cloud locates the target driving path by screening and selecting from K alternative feasible paths through analyzing the K groups of backhaul path perception information. The method includes: interacting with historical traffic data to obtain multiple groups of sample path perception information corresponding to multiple sample driving paths and multiple sample quantified traffic efficiencies. The multiple sample quantified traffic efficiencies are traffic efficiency parameters marked manually, and the higher the traffic efficiency, the shorter the travel time of the corresponding road section. The sample path perception information includes perception information of parameters such as traffic flow density, traffic flow speed, and real-time road conditions. Using the multiple groups of sample path perception information and multiple sample quantified traffic efficiencies as training data to construct a traffic efficiency quantification model. The traffic efficiency quantification model is constructed based on a neural network model. Using multiple groups of sample path perception information as training data and the corresponding multiple sample quantified traffic efficiencies as supervision data to perform supervised training on the model until the accuracy of the traffic efficiency output by the model meets the requirements, and obtaining a trained traffic efficiency quantification model.

[0034] Further, extract Q real-time path perception information from the first group of backhaul path perception information, where Q is the number of traffic perception nodes in the first group of backhaul paths. Among them, the real-time path perception information includes real-time traffic flow density, real-time traffic flow speed, and real-time road conditions. The first group of backhaul paths is a random one among the K groups of backhaul paths. Synchronize the Q real-time path perception information to the traffic efficiency quantification model to obtain the first quantified traffic efficiency analyzed and output by the traffic efficiency quantification model. Use the same processing method to obtain and analyze the K groups of backhaul path perception information to obtain K quantified traffic efficiencies. Further, serialize the K quantified traffic efficiencies, and screen and locate the target driving path from the K alternative feasible paths according to the sorting result. The target driving path is the path with the highest traffic efficiency in the sorting result.

[0035] The method provided by the embodiment of this application further includes: making an online data call for the first exploratory driving path to obtain the first driving feature, where the first driving feature includes the first driving distance and multiple first driving duration records; outputting the first driving feasibility by quantitatively evaluating the first driving feature; and so on, calculating the M driving feasibilities of the M exploratory driving paths; presetting a feasibility threshold, comparing the M driving feasibilities with the feasibility threshold, and screening the K alternative feasible paths from the M exploratory driving paths.

[0036] The K alternative feasible paths are selected from the M exploratory driving paths according to driving characteristics. The method includes: making a network data call for the first exploratory driving path to obtain a first driving characteristic, where the first driving characteristic includes a first driving distance and multiple first driving duration records. By quantitatively evaluating the first driving characteristic, a first driving feasibility is output, and the driving feasibility is used to quantitatively measure the quality of the driving characteristic. By analogy, M driving feasibilities of the M exploratory driving paths are calculated. Here, both M and K are positive integers greater than 1. A feasible threshold is preset, and the M driving feasibilities are compared with the feasible threshold. The feasible threshold is a preset minimum standard of feasibility. When the corresponding path is less than the feasible threshold, its feasibility is poor, and vice versa. Paths greater than the feasible threshold are selected from the M exploratory driving paths to obtain the K alternative feasible paths.

[0037] The method provided in the embodiment of the present application further includes: calculating a first average driving time based on the multiple first driving duration records; counting the multiple first driving duration records and outputting a first path application frequency; presetting a path application frequency threshold; determining whether the first path application frequency meets the path application frequency threshold; if the first path application frequency meets the path application frequency threshold, then adding up the first driving distance, the first average driving time, and the first path application frequency based on a preset weighting rule, and outputting the first driving feasibility.

[0038] By quantitatively evaluating the first driving characteristic, a first driving feasibility is output. The method includes: calculating a first average driving time based on the multiple first driving duration records, where the multiple first driving duration records are driving duration records when driving on this path at the same historical time node. Subsequently, the multiple first driving duration records are counted, and a first path application frequency is output. The first path application frequency is the number of applications of this path with the same starting point and ending point in every 100 passing records. Further, a path application frequency threshold is preset. The preset path application frequency threshold is a preset passing frequency threshold. When it is greater than this threshold, the passing selection frequency of the corresponding passing path is higher. Determine whether the first path application frequency meets, that is, is greater than or equal to, the path application frequency threshold. If the first path application frequency meets the path application frequency threshold, then parameter normalization is performed on the first driving distance, the first average driving time, and the first path application frequency, and the weights of each parameter are configured according to actual preferences. For example, if the requirement for time is higher, a higher weight is configured for the first average driving time. The preset weighting rule is a weighting rule preset according to actual preferences. Weighted summation is performed according to the normalized first driving distance, the first average driving time, and the first path application frequency and the corresponding weight parameters, and the first driving feasibility is output.

[0039] In the above text, reference is made to Figure 1 a detailed description of the AI - decision - based multi - traffic - flow interactive path optimization method according to an embodiment of the present invention. Next, reference will be made to Figure 2 describe an AI - decision - based multi - traffic - flow interactive path optimization system according to an embodiment of the present invention.

[0040] The AI - decision - based multi - traffic - flow interactive path optimization system according to an embodiment of the present invention solves the technical problem in the prior art that traditional navigation systems only rely on static map data and simple real - time traffic information for path planning, resulting in difficulties in improving the accuracy and efficiency of path planning. By combining real - time traffic perception data and multi - traffic - flow interaction information, and intelligently deploying and activating traffic perception nodes, real - time traffic data is effectively utilized for path optimization, improving traffic efficiency and driving experience, and achieving more accurate and efficient path planning. The AI - decision - based multi - traffic - flow interactive path optimization system includes: an interaction module 11, a feasible path acquisition module 12, a cloud interaction module 13, a perception information acquisition module 14, a feasible path screening module 15, an interaction condition construction module 16, and a sharing and association module 17.

[0041] The interaction module 11 is used to interact with in - vehicle navigation applications to obtain the real - time position and input destination of the target vehicle.

[0042] The feasible path acquisition module 12 is used to take the real - time position as the driving starting point, combine with the input destination to perform local driving path planning, and output K alternative feasible paths.

[0043] The cloud interaction module 13 is used to send the K alternative feasible paths to the AI - decision cloud.

[0044] The perception information acquisition module 14 is used for the AI - decision cloud to locally activate traffic perception nodes according to the K alternative feasible paths to obtain K sets of back - transmitted path perception information.

[0045] The feasible path screening module 15 is used for the AI - decision cloud to analyze the K sets of back - transmitted path perception information and screen and locate the target driving path from the K alternative feasible paths.

[0046] The interaction condition construction module 16 is used to construct multi - traffic - flow interaction conditions based on the real - time position and input destination.

[0047] The sharing and association module 17 is used for the target vehicle to share the target driving path to associated vehicles through V2X communication technology based on the multi - traffic - flow interaction conditions.

[0048] Next, the specific configuration of the feasible path acquisition module 12 will be described in detail. The feasible path acquisition module 12 may further include: taking the real-time position as the driving starting point, combining with the input end point to perform local driving path planning, and outputting K alternative feasible paths. The method includes: loading a local road network according to the real-time position and the input end point; traversing the local road network to obtain an initial driving path; obtaining M exploratory driving paths by randomly perturbing the initial driving path; and screening the K alternative feasible paths from the M exploratory driving paths according to driving characteristics.

[0049] Next, the specific configuration of the perception information acquisition module 14 will be further described in detail. The perception information acquisition module 14 further includes: the AI decision cloud locally activates traffic perception nodes according to the K alternative feasible paths to obtain K groups of backhaul path perception information. Before that, the method further includes: abstracting the target traffic road network of the target city to obtain a traffic road network topology; positioning W traffic perception sections and H traffic perception convergence points in the traffic road network topology according to the traffic flow of the target traffic road network; configuring sensors in the target city according to the W traffic perception sections and H traffic perception convergence points to obtain W + H traffic perception nodes; and completing the local deployment of the AI decision cloud by communicatively connecting the W + H traffic perception nodes with the AI decision cloud.

[0050] Next, the specific configuration of the perception information acquisition module 14 will be described in detail. The perception information acquisition module 14 further includes: positioning W traffic perception sections and H traffic perception convergence points in the traffic road network topology according to the traffic flow of the target traffic road network. The method includes: interactively obtaining multiple traffic flow records of multiple local sections in the traffic road network topology; analyzing the multiple traffic flow records to obtain multiple traffic flow fluctuation thresholds and multiple traffic flow stability values; presetting a traffic flow fluctuation scale and a traffic flow threshold; using the traffic flow fluctuation scale and the traffic flow threshold to map and traverse the multiple traffic flow fluctuation thresholds and multiple traffic flow stability values to locate W traffic perception sections; and positioning H traffic perception convergence points in the traffic road network topology according to the section intersection situation.

[0051] Next, the specific configuration of the perception information acquisition module 14 will be described in detail. The perception information acquisition module 14 may further include: the AI decision cloud locates the target driving path by screening from K alternative feasible paths through analyzing the K groups of backhaul path perception information. The method includes: interactively obtaining multiple groups of sample path perception information of multiple sample driving paths and multiple sample quantified traffic efficiencies; constructing a traffic efficiency quantification model with the multiple groups of sample path perception information and multiple sample quantified traffic efficiencies as training data; extracting Q real-time path perception information from the first group of backhaul path perception information, where the real-time path perception information includes real-time traffic density, real-time traffic speed, and real-time road condition information; synchronizing the Q real-time path perception information to the traffic efficiency quantification model, and analyzing and outputting the first quantified traffic efficiency; and so on, analyzing the K groups of backhaul path perception information to obtain K quantified traffic efficiencies; serializing the K quantified traffic efficiencies, and screening and locating the target driving path from the K alternative feasible paths according to the sorting result.

[0052] Next, the specific configuration of the feasible path acquisition module 12 will be further described in detail. The feasible path acquisition module 12 further includes: screening the K alternative feasible paths from the M exploratory driving paths according to driving characteristics. The method includes: making an online data call for the first exploratory driving path to obtain the first driving characteristics, where the first driving characteristics include the first driving distance and multiple first driving duration records; outputting the first driving feasibility by quantitatively evaluating the first driving characteristics; and so on, calculating the M driving feasibilities of the M exploratory driving paths; presetting a feasibility threshold, comparing the M driving feasibilities with the feasibility threshold, and screening the K alternative feasible paths from the M exploratory driving paths.

[0053] Next, the specific configuration of the feasible path acquisition module 12 will be described in detail. The feasible path acquisition module 12 further includes: outputting the first driving feasibility by quantitatively evaluating the first driving characteristics. The method includes: calculating the first average driving time based on the multiple first driving duration records; counting the multiple first driving duration records, and outputting the first path application frequency; presetting a path application frequency threshold; determining whether the first path application frequency meets the path application frequency threshold; if the first path application frequency meets the path application frequency threshold, then summing the first driving distance, the first average driving time, and the first path application frequency based on a preset weighting rule, and outputting the first driving feasibility.

[0054] The multi-traffic-flow interactive path optimization system based on AI decision provided by the embodiment of the present invention can execute the multi-traffic-flow interactive path optimization method based on AI decision provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0055] Although various references are made in this application to certain modules in the system according to embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or the server. The various units and modules included are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be achieved; in addition, the specific names of the various functional units are only for the convenience of mutual distinction and are not used to limit the protection scope of the present invention.

[0056] The above specific implementation manners do not constitute a limitation on the protection scope of this application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims

1. A multi-vehicle interactive path optimization method based on AI decision-making, characterized in that: The method comprises: Interactive in-vehicle navigation application to obtain the real-time location of the target vehicle and input the destination; Taking the real-time location as the driving starting point, performing local driving path planning in combination with the input destination, and outputting K backup feasible paths; Send the K backup feasible paths to the AI ​​decision cloud; The AI ​​decision cloud performs local activation of traffic sensing nodes according to the K backup feasible paths to obtain K groups of return path sensing information; The AI ​​decision cloud analyzes the K groups of returned path perception information and selects and locates the target driving path from the K backup feasible paths; Establishing multi-vehicle flow interaction conditions based on the real-time location and the input destination; Based on the multi-vehicle flow interaction condition, the target vehicle shares the target driving path with associated vehicles through V2X communication technology.

2. The multi-vehicle interactive path optimization method based on AI decision-making as claimed in claim 1, characterized in that: Taking the real-time location as the driving starting point and combining the input destination to perform local driving path planning, and outputting K backup feasible paths, the method includes: loading a local road network based on the real-time location and the input destination; Traversing the local road network to obtain an initial driving path; By randomly perturbing the initial driving path, M exploration driving paths are obtained; The K backup feasible paths are obtained by screening the M exploratory driving paths according to driving characteristics.

3. The method for multi-vehicle interactive path optimization based on AI decision-making as claimed in claim 1, characterized in that: The AI ​​decision cloud performs local activation of traffic sensing nodes according to the K backup feasible paths to obtain K groups of return path sensing information. Before that, the method further includes: Abstract the target traffic network of the target city and obtain the traffic network topology; According to the traffic flow of the target traffic network, W traffic-sensing sections and H traffic-sensing sinks are located in the traffic network topology; According to the W traffic sensing sections and the H traffic sensing sinks, sensors are configured in the target city to obtain W+H traffic sensing nodes; By connecting the W+H traffic sensing nodes to the AI ​​decision cloud for communication, the localized deployment of the AI ​​decision cloud is completed.

4. The method for multi-vehicle interactive path optimization based on AI decision-making as claimed in claim 3, characterized in that: According to the traffic flow of the target traffic network, W traffic sensing sections and H traffic sensing sinks are located in the traffic network topology, and the method includes: Interactively obtaining a plurality of traffic flow records of a plurality of local road sections in the traffic network topology; Analyzing the plurality of traffic flow records to obtain a plurality of traffic flow fluctuation thresholds and a plurality of traffic flow stability values; Preset traffic flow fluctuation scale and traffic flow threshold; Using the traffic flow fluctuation scale and the traffic flow threshold, mapping and traversing the multiple traffic flow fluctuation thresholds and the multiple traffic flow stability values, and locating W traffic perception sections; According to the intersection of road sections, H traffic sensing points are located in the traffic network topology.

5. The method for multi-vehicle interactive path optimization based on AI decision-making as claimed in claim 4, characterized in that: The AI ​​decision cloud analyzes the K groups of returned path perception information to select and locate the target driving path from the K backup feasible paths. The method includes: Interactively obtain multiple groups of sample path perception information of multiple sample driving paths and multiple sample quantitative traffic efficiencies; Using the multiple groups of sample path perception information and the multiple sample quantified traffic efficiencies as training data to construct a traffic efficiency quantification model; Extracting Q pieces of real-time path perception information from the first set of returned path perception information, wherein the real-time path perception information includes real-time traffic density, real-time traffic speed, and real-time road condition information; Synchronizing the Q real-time path perception information to the traffic efficiency quantification model, analyzing and outputting a first quantified traffic efficiency; Similarly, the K groups of feedback path perception information are analyzed to obtain K quantified traffic efficiencies; The K quantified traffic efficiencies are sequenced, and the target driving path is screened and located from the K backup feasible paths according to the sequencing result.

6. The multi-vehicle interactive path optimization method based on AI decision-making as claimed in claim 2, characterized in that: The K backup feasible paths are obtained by screening the M exploratory driving paths according to the driving characteristics, and the method includes: Performing online data call on the first explored driving path to obtain a first driving feature, wherein the first driving feature includes a first driving distance and a plurality of first driving duration records; Outputting a first driving feasibility by quantitatively evaluating the first driving feature; By analogy, M driving feasibility of the M exploration driving paths are calculated; A feasibility threshold is preset, the M driving feasibility is compared with the feasibility threshold, and the K backup feasible paths are obtained by screening from the M explored driving paths.

7. The method for multi-vehicle interactive path optimization based on AI decision-making as claimed in claim 6, characterized in that: By quantitatively evaluating the first driving feature, outputting a first driving feasibility, the method includes: Calculate a first driving time average based on the multiple first driving time records; Counting the plurality of first driving duration records, and outputting a first path application frequency; Preset path application frequency threshold; Determining whether the first path application frequency meets the path application frequency threshold; If the first path application frequency meets the path application frequency threshold, the first driving distance, the first driving time average and the first path application frequency are added based on a preset weighted rule to output the first driving feasibility.

8. The multi-vehicle interactive path optimization system based on AI decision-making is characterized by: The system comprises: Interaction module, used for interactive vehicle navigation applications, obtaining the real-time position of the target vehicle and inputting the destination; A feasible path acquisition module is used to take the real-time position as the driving starting point, perform local driving path planning in combination with the input end point, and output K backup feasible paths; A cloud interaction module, used for sending the K backup feasible paths to the AI ​​decision cloud; A perception information acquisition module, used for the AI ​​decision cloud to locally activate traffic perception nodes according to the K backup feasible paths to obtain K groups of return path perception information; A feasible path screening module is used for the AI ​​decision cloud to screen and locate the target driving path from K backup feasible paths by analyzing the K groups of returned path perception information; An interaction condition building module, used for building multi-vehicle flow interaction conditions based on the real-time position and the input destination; A sharing association module is used for the target vehicle to share the target driving path with associated vehicles through V2X communication technology based on the multi-vehicle flow interaction condition.

Images