A path dynamic programming method and system for extreme weather scenarios, a terminal and a storage medium

Abstract

The application relates to the technical field of path planning, and discloses a path dynamic planning method and system for an extreme weather scene, a terminal and a storage medium, the method comprising the following steps: acquiring a target road network of a target city, collecting a street view image set under an extreme weather scene according to a road sampling interval; acquiring a road center line of the target road network, performing road node binding on each street view image according to a road network diagram to obtain a target street view image set; performing identification analysis on all objects in the target street view image set to obtain a target analysis result, performing risk assessment to obtain an image risk result, and obtaining a corresponding road risk result according to the image risk result; performing path planning according to the road risk result to obtain a candidate path set, and performing path confirmation on the candidate path set to obtain target planning path information. The application dynamically provides a planning path with higher safety by comprehensively assessing the risks of objects around roads under an extreme weather scene.

Description

Technical Field

[0001] This invention relates to the field of route planning technology, and in particular to a method, system, terminal, and computer-readable storage medium for dynamic route planning in extreme weather scenarios. Background Technology

[0002] Currently, achieving fine-grained road risk perception and dynamic route optimization for extreme weather scenarios (such as heavy rain and blizzards) remains a key challenge in the field of urban safe travel. Existing route planning technologies are mainly based on macro-level data such as traffic flow, road structure, and estimated travel time, employing shortest path or fastest path algorithms such as Dijkstra's algorithm (a greedy algorithm for finding the single-source shortest path in a weighted graph) and incorporating real-time traffic event information. These technologies can provide limited risk avoidance in situations such as construction, accidents, or road closures.

[0003] However, existing technologies generally rely on macroscopic information such as traffic flow and road attributes, lacking the ability to identify microscopic road environments (e.g., lateral obstacles, temporarily fallen objects, and loose billboards), and are unable to effectively perceive short-term, sudden dangers arising under extreme weather conditions. Therefore, existing technologies still have significant shortcomings in safe route planning for extreme weather scenarios: First, traditional navigation lacks a refined understanding of road visual environment risks and cannot identify potential hazards such as trees and billboards; second, methods based on macroscopic event pushes or point-based monitoring are insufficient for responding to localized, sudden dangers, making it difficult to form timely and comprehensive risk assessments; third, existing route planning models cannot incorporate visual risk quantification information into the path cost function, resulting in recommended routes potentially passing through high-risk areas, thus failing to guarantee user driving safety.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a method, system, terminal, and storage medium for dynamic path planning in extreme weather scenarios. This invention aims to address the problem that existing technologies lack the ability to identify the micro-environment of roads, cannot effectively perceive sudden dangers arising under extreme weather conditions, and are unable to form timely and comprehensive risk assessments, resulting in low safety levels in the recommended planned paths.

[0006] To achieve the above objectives, the present invention provides a path dynamic planning method for extreme weather scenarios, the method comprising the following steps: Obtain the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and collect street view image sets of the target city under extreme weather scenarios according to the road sampling intervals; Obtain the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set; All objects in the target street view image set are identified and analyzed to obtain target analysis results. Based on the target analysis results, the target street view image set is risk-assessed to obtain image risk results. Based on the image risk results, the corresponding road risk results are obtained. Based on the road risk results, path planning is performed to obtain a set of candidate paths. Then, the candidate path set is confirmed according to user requirements to obtain the target planned path information.

[0007] Optionally, in the aforementioned path dynamic planning method for extreme weather scenarios, the step of collecting street view image sets of the target city under extreme weather scenarios at intervals according to the road specifically includes: ; The step of binding road nodes to each street view image in the street view image set according to the road network map specifically involves: ; in, A collection of street view images, Street view image, Number the street views. This is an image acquisition function. The longitude coordinates of the data collection point. The latitude coordinates of the collection point. The orientation angle of the street view image. For street view image acquisition time, The number of street view images. This represents the distance from road nodes within a road segment to street view sampling points. The coordinates of the points on the road segment. This is the nearest road segment.

[0008] Optionally, the path dynamic planning method for extreme weather scenarios, wherein the step of identifying and analyzing all objects in the target street view image set to obtain target analysis results specifically includes: Define a target hazardous object category, and perform identification processing on all objects in the target street view image set according to the target hazardous object category to obtain object identification results, wherein the object identification results include object category, object bounding box, object confidence score and object coordinate position; Depth maps are generated from the target street view image set to obtain object depth maps, and the relative distance of each object in the object depth map is calculated to obtain the object relative distance; The object scale is calculated based on the relative distance of the object to obtain the object scale information, and the target analysis result is obtained based on the object scale information, the relative distance of the object, and the object recognition result. Specifically, the calculation of the relative distance to each target object in the object depth map is as follows: ; The calculation of object scale based on the relative distance of the objects specifically involves: ; in, The relative distance between objects. The bounding box of the object. is the x-coordinate of a pixel in the object's depth map. y = ... This is a depth map of the object. Number the street views. For object categories, For the number of road sections, This refers to the object's scale information.

[0009] Optionally, the path dynamic planning method for extreme weather scenarios, wherein the step of performing a risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtaining corresponding road risk results based on the image risk results, specifically includes: Risk weighting is applied to each type of target object in the target analysis results to obtain risk weighting results. Object risk assessment is performed on the risk weighting results to obtain object risk coefficients. Based on the object risk coefficients, risk assessment is performed on each street scene image in the target street scene image set to obtain image risk results. All the street view images are mapped to their corresponding road segments to obtain a mapping table. The image risk results are then aggregated with the mapping table to obtain the road risk results.

[0010] Optionally, in the path dynamic planning method for extreme weather scenarios, the step of performing object risk assessment on the risk-weighted results specifically involves: ; The risk assessment of each street view image in the target street view image set based on the object risk coefficient is specifically as follows: ; The aggregation process based on the image risk results and the mapping table specifically involves: ; in, The risk factor of the object. Risk-weighted results The combined risk of the object's size and distance. For risk confidence level, To mitigate the additional risk of nearby dangerous objects, This refers to object scale information. These are the object's relative coordinates. , and All are risk coefficients. The relative distance between objects. This represents the maximum effective distance from which the risk of identifiable objects can be detected in a street view image. The risk index of street view images, The risk index of the road, It is an average function. Street view image, This is the nearest road segment.

[0011] Optionally, the path dynamic planning method for extreme weather scenarios, wherein the step of planning a path based on the road risk results to obtain a candidate path set, and confirming the candidate path set according to user requirements to obtain target planned path information, specifically includes: Based on the road risk results, a comprehensive risk cost is calculated to obtain a comprehensive risk result. The road network map is then updated based on the road risk results to obtain a target road network map. The target road network map is used to solve for the path, and the path solution information is obtained. Then, multi-path matching is performed based on the path solution information to obtain a set of candidate paths. The system obtains user requirements, customizes the candidate path set according to the user requirements to obtain the target planned path, marks the target planned path with risk according to the road risk results to obtain a road risk level map, and generates target planned path information based on the road risk level map and the target planned path.

[0012] Optionally, in the aforementioned path dynamic planning method for extreme weather scenarios, the step of calculating the comprehensive risk cost based on the road risk results specifically includes: ; The specific steps for solving the path in the target road network map are as follows: ; The step of customizing the candidate path set according to the user's needs specifically involves: ; in, To consider the overall risk outcome, The length of the road segment Speed ​​limits are based on road segments. The risk index of the road, , and These are all hyperparameters representing the relative weights of the factors. For heuristic functions, Road nodes To the target road node The straight-line distance Based on the maximum speed of the road segment, As a risk penalty coefficient, For nodes Local average risk of surrounding road sections, Plan a path to the goal. For the first Planned route, For the candidate path set, For the first The total travel time for the planned route For the first The total distance of the planned path, For the first The cumulative risk value of each planned path.

[0013] Optionally, the path dynamic planning method for extreme weather scenarios, wherein the path dynamic planning system for extreme weather scenarios includes: The image acquisition module is used to acquire the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and acquire a set of street view images of the target city under extreme weather scenarios according to the road sampling intervals. The image processing module is used to acquire the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set. The risk assessment module is used to identify and analyze all objects in the target street view image set to obtain target analysis results, perform risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtain corresponding road risk results based on the image risk results. The route confirmation module is used to plan a route based on the road risk results, obtain a set of candidate routes, and confirm the route of the candidate route set according to user requirements to obtain the target planned route information.

[0014] In addition, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a path dynamic planning program for extreme weather scenarios stored in the memory and executable on the processor, wherein when the path dynamic planning program for extreme weather scenarios is executed by the processor, it implements the steps of the path dynamic planning method for extreme weather scenarios as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a path dynamic planning program for extreme weather scenarios, and the path dynamic planning program for extreme weather scenarios, when executed by a processor, implements the steps of the path dynamic planning method for extreme weather scenarios as described above.

[0016] In this invention, a target road network of a target city is obtained. Different road sampling intervals are set based on the road level information of the target road network, and street view image sets of the target city under extreme weather scenarios are collected according to the road sampling intervals. The center lines of the target road network are obtained, a road network map is constructed based on the road center lines, and road nodes are bound to each street view image in the street view image set according to the road network map to obtain a target street view image set. All objects in the target street view image set are identified and analyzed to obtain target analysis results. A risk assessment is performed on the target street view image set based on the target analysis results to obtain image risk results, and corresponding road risk results are obtained based on the image risk results. Path planning is performed based on the road risk results to obtain a candidate path set, and the candidate path set is confirmed according to user needs to obtain target planned path information. This invention provides a highly safe planned path by comprehensively assessing the risks of objects around roads under extreme weather scenarios. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the path dynamic planning method for extreme weather scenarios of the present invention; Figure 2 This is a structural diagram of a preferred embodiment of the path dynamic planning system for extreme weather scenarios of the present invention; Figure 3 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0020] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0021] The preferred embodiment of the path dynamic planning method for extreme weather scenarios described in this invention, such as... Figure 1 As shown, the path dynamic planning method for extreme weather scenarios includes the following steps: Step S10: Obtain the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and collect street view image sets of the target city under extreme weather scenarios according to the road sampling intervals.

[0022] Specifically, addressing the shortcomings of existing technologies in recognizing the micro-environment of roads, effectively perceiving sudden dangers arising under extreme weather conditions, and struggling to form timely and comprehensive risk assessments, resulting in low safety levels in recommended planned routes, this invention proposes a dynamic route planning method for extreme weather scenarios. By integrating multi-source data from street view, meteorology, and road topology at the urban scale, a road risk assessment mechanism is constructed. The corresponding risk assessment results are used as dynamic weights for route search, forming a closed loop of "risk perception—risk quantification—multi-objective route optimization—online adaptation." Ultimately, this provides optimal planned routes that prioritize safety or balance safety and efficiency for different user scenarios (e.g., motor vehicle scenarios, non-motor vehicle scenarios, and pedestrian scenarios).

[0023] The specific processing procedure involves first constructing a multi-source street view data framework for road safety analysis in extreme weather scenarios, enabling automated acquisition, spatiotemporal alignment, and structured management of street view images. To obtain high-quality street view images covering the urban road network, this embodiment employs a scripted and parallel-running street view data crawling strategy for data collection, specifically acquiring the target road network of the target city. ,in, For the first Each road section, To determine the number of road segments, different road sampling intervals are set based on the road classification information of the target road network (e.g., a sampling interval of 20 m for main roads, 30 m for secondary roads, and 40 m for branch roads) to ensure the spatial continuity of the image distribution. Street view image sets of the target city under extreme weather scenarios are then collected according to these road intervals. The representation of the collected street view image sets is as follows: ; in, A collection of street view images, Street view image, Number the street views. This is an image acquisition function. The longitude coordinates of the data collection point. The latitude coordinates of the collection point. The orientation angle of the street view image. For street view image acquisition time, The number of street view images is limited. To enhance the comprehensiveness of visual information, each acquisition point can automatically generate multiple shooting directions (e.g., 90 degrees, 180 degrees, etc.) based on the road shape, thereby ensuring the visibility of potential hazards such as buildings, roadside trees, and water bodies in extreme weather conditions. Furthermore, considering the complex sources of street view images, diverse shooting times, and the existence of blurry, obstructed, or underexposed images, this invention performs effectiveness screening on the street view images. For example, an image quality evaluation function is set to effectively eliminate low-quality street view images with raindrop obstruction, insufficient light, or excessive darkness at night, ensuring that subsequent risk identification is not affected.

[0024] Each street view image is accompanied by complete metadata before being stored in the database, including but not limited to: spatial coordinates, street view image orientation angle, acquisition platform, resolution, and timestamp. To support rapid retrieval and parallel processing of large-scale images, this invention constructs a dual-index structure: a spatial index and a temporal index. The spatial index is based on a BallTree (a tree-like data structure for efficiently processing high-dimensional spatial data) to quickly locate street view image sets within a specific area, supporting queries for various spatial ranges such as road segments, buffer zones, and pedestrian areas. The temporal index uses timestamps as keys, supporting historical comparisons of street views of the same road segment across different months and years, enabling "multi-temporal" risk monitoring. This dual-index structure ensures that local image queries can be completed quickly, significantly improving the overall system efficiency.

[0025] Step S20: Obtain the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street view image in the street view image set based on the road network map to obtain the target street view image set.

[0026] Specifically, in order to locate street view images within a road map structure, this invention constructs a road network map based on road centerlines, that is, by obtaining the road centerlines of the target road network and constructing a road network map based on the road centerlines. ,in, For the set of road nodes, This is a collection of road segments. For each street view image... By performing road node binding to obtain the target street view image set, specifically, the street view images are bound to the nearest road segments using the shortest vertical distance algorithm. The corresponding expression is: ; in, This represents the distance from road nodes within a road segment to street view sampling points. These are the coordinates of a point on the road segment. When... If so, it is considered that the street view image has been effectively bound to the nearest road segment, where, The purpose of binding road nodes is to define the maximum distance (e.g., 5m) from a road node to a street view sampling point on a road segment. This is to play a key role in subsequent risk aggregation, enabling "image risk" to be accurately mapped to "road risk" and establishing a clear visual-road coupling link.

[0027] Step S30: Identify and analyze all objects in the target street view image set to obtain target analysis results; perform risk assessment on the target street view image set based on the target analysis results to obtain image risk results; and obtain corresponding road risk results based on the image risk results.

[0028] Specifically, after obtaining the target street view image, a deep learning visual model is used to detect dangerous objects in the road. To identify objects in the target street view image set that may pose safety hazards under extreme weather conditions, this invention employs a target detection model (e.g., YOLO, You Only Look Once, a real-time target detection algorithm based on deep learning) to identify and analyze all objects in the target street view image set. Specifically, this involves defining target dangerous object categories. The corresponding expression is: ; in, Trees It is a column type. For billboards, It is a fence type. For buildings, the system identifies all objects in the target street view image set according to the target hazardous object category, obtaining object recognition results, including object category, object bounding box, object confidence score, and object coordinates. The object bounding box contains the pixel region of the target object (i.e., the hazardous object), and the object confidence score reflects the reliability of the detection result. The significance of this detection lies in automatically capturing hazardous objects that may pose a disaster risk through the target detection model, such as fallen trees caused by typhoons, loose billboards, and temporary obstacles, thereby providing basic input for subsequent risk analysis.

[0029] Since relying solely on the category of hazardous objects is insufficient to determine the degree of risk of an object, this invention generates depth maps from the target street scene image set using a monocular depth estimation model, thereby obtaining object depth maps. ,in, is the x-coordinate of a pixel in the object's depth map. Here, represents the ordinate of a pixel in the object depth map. This object depth map is used to calculate the relative distance between each object and the road center. Specifically, the relative distance of each object in the object depth map is calculated to obtain the object's relative distance, and the corresponding expression is: ; in, The relative distance between objects. The bounding box of the object. Number the street views. For object categories, This refers to the number of road segments. To improve the quantification of the hazard level of objects, this invention introduces the "effective scale" index to comprehensively reflect the relationship between object size and distance. For example, the larger the object and the closer it is, the higher its potential hazard. For instance, a large tree within a few meters of the roadside has a higher risk of collapsing in a typhoon than a small utility pole further away. Specifically, the object scale is calculated based on the relative distance to obtain the object scale information, and the corresponding expression is: ; in, The object scale information is then used; subsequently, the target analysis result is obtained based on the object scale information, the relative distance between the objects, and the object recognition result. This invention achieves a quantitative mapping from two-dimensional pixel space to three-dimensional spatial relationships through depth estimation and scale transformation, enabling objects to be used for rigorous risk modeling.

[0030] This invention comprehensively considers factors such as object category, size, confidence level, and distance, and also introduces an expert experience weighting system to weight the risk of objects. For example, trees or billboards pose a significantly higher risk than fences in a typhoon. Specifically, the risk of each type of target object in the target analysis results is weighted to obtain a risk weighting result. Based on this, this invention constructs a risk function for object risk assessment, that is, it uses the risk function to assess the object risk of the weighted results, obtaining the object risk coefficient, the corresponding expression of which is: ; in, The risk factor of the object. Risk-weighted results The combined risk of the object's size and distance. For risk confidence level, To mitigate the additional risk of nearby dangerous objects, This refers to object scale information. These are the object's relative coordinates. , and All are risk coefficients. The relative distance between objects. This represents the maximum effective distance for identifying object risk in the street view image. Based on this object risk coefficient, a risk assessment is performed on each street view image in the target street view image set to obtain the image risk result, which is expressed as: ; in, The risk index of the street view image is used; then, all the street view images are mapped to their corresponding road segments to obtain a mapping table, and the image risk results are aggregated with the mapping table to obtain the road risk results, the corresponding expression of which is: ; in, The risk index of the road, The average function is used. To improve the stability and spatial consistency of road risk results, this invention further utilizes road topology to smooth the risk of adjacent road segments, making road segments with continuous spatial relationships exhibit more consistent risk levels and reducing local abrupt changes caused by noise in single images.

[0031] Step S40: Perform path planning based on the road risk results to obtain a candidate path set, and confirm the candidate path set according to user requirements to obtain the target planned path information.

[0032] Specifically, after obtaining the road risk results, safe route planning for extreme weather scenarios is required. This involves generating multiple selectable routes that meet actual needs, prioritizing the avoidance of high-risk road sections while ensuring traffic efficiency. Traditional route planning primarily optimizes based on road length or time, making it difficult to consider safety. However, in extreme weather scenarios, this invention incorporates a road risk index into a comprehensive cost function, giving road risk a controllable and adjustable influence on route selection. Specifically, a comprehensive risk cost is calculated based on the road risk results to obtain the comprehensive risk result, expressed as: ; in, To consider the overall risk outcome, The length of the road segment Speed ​​limits are based on road segments. The risk index of the road, , and These are all hyperparameters representing the relative weights of factors; by setting different weight combinations, various navigation strategies can be supported. For example, for safety priority, the weights can be increased. If distance is prioritized, then increase Prioritizing efficiency will increase If safety, distance, and efficiency are taken into account, then a balance can be achieved. , and This integrated cost-benefit mechanism provides a unified optimization objective for path search, enabling it to automatically favor less dangerous roads in extreme weather scenarios.

[0033] To further enhance safety, this invention employs a "hard constraint" screening mechanism to prohibit traffic on certain extremely high-risk road sections. Specifically, a risk threshold is set. When satisfied If a road segment is deemed unsafe for passage under extreme weather conditions, it should be removed from the searchable road segments. Then, the road network map is updated based on the road risk results to obtain the target road network map. ,in, The updated set of road segments corresponds to the following expression: ; in, These are prohibited road sections. Risk-based closures effectively prevent planned routes from traversing high-risk areas such as fallen trees, flying debris, and structural damage, thereby ensuring the safety and reliability of the discovered routes.

[0034] Subsequently, based on the comprehensive cost and prohibition handling, the existing A* search algorithm (A-Star Algorithm) is improved by introducing a heuristic function. The improved heuristic search algorithm is then used for path finding. To prioritize searching away from high-risk areas, the heuristic function also considers distance and local risk factors. Specifically, the improved heuristic search algorithm is used to find paths on the target road network map, obtaining path solution information, the corresponding expression of which is: ; in, For heuristic functions, Road nodes To the target road node The straight-line distance Based on the maximum speed of the road segment, This is a risk penalty coefficient used to adjust the degree to which the navigation process avoids high-risk areas. For nodes The local average risk of surrounding road sections; compared with the traditional heuristic search algorithm, the improved heuristic search algorithm of this invention can guide the search process to actively deviate from high-risk areas, and prioritize the exploration of lower-risk parts among candidate paths with similar costs. It also has strong adaptability and can adjust the risk penalty coefficient according to different extreme weather events.

[0035] To enhance the system's usability, this invention not only generates the optimal path but also constructs multiple alternative solutions using the K-shortest path algorithm, resulting in a candidate path set. Each candidate path in this set includes indicators such as total distance, total time, and cumulative risk value. Furthermore, to meet user preferences, users can customize the path. Specifically, user requirements are obtained, and the candidate path set is customized based on these requirements to obtain the target planned path, expressed as: ; in, Plan a path to the goal. For the first Planned route, For the candidate path set, For the first The total travel time for the planned route For the first The total distance of the planned path, For the first The cumulative risk value of a planned route is an integration of the risk indices of each road segment within the planned route, used to reflect the overall safety of the planned route. This customization mechanism allows the system to output different types of optimal planned routes (i.e., target planned routes) according to different needs. For example, a safety-first route avoids all high-risk areas; an efficiency-first route pursues the shortest time within a controllable risk range; and a balanced route considers risk, distance, and time simultaneously.

[0036] Finally, based on the road risk results, the target planned route is risk-marked (e.g., different risk levels are marked with colors or symbols to identify high-risk road sections under weather conditions such as typhoons and heavy rain), resulting in a road risk level map. Based on the road risk level map and the target planned route, target planned route information is generated, outputting information including route descriptions, risk levels for each road segment, and potential risk point alerts. This provides users with intuitive, interpretable, and highly customizable safe travel suggestions in extreme weather scenarios. Furthermore, this invention supports basic parameter adjustment functions, such as setting risk weights, selecting different weather scenarios, and viewing route distance, time, and cumulative risk values.

[0037] Furthermore, such as Figure 2 As shown, based on the above-mentioned path dynamic planning method for extreme weather scenarios, the present invention also provides a path dynamic planning system for extreme weather scenarios, wherein the path dynamic planning system for extreme weather scenarios includes: Image acquisition module 51 is used to acquire the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and acquire street view image sets of the target city under extreme weather scenarios according to the road sampling intervals. Image processing module 52 is used to obtain the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set; The risk assessment module 52 is used to identify and analyze all objects in the target street view image set to obtain target analysis results, perform risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtain corresponding road risk results based on the image risk results. The path confirmation module 53 is used to perform path planning based on the road risk results, obtain a set of candidate paths, and confirm the path of the candidate path set according to user requirements to obtain the target planned path information.

[0038] Furthermore, such as Figure 3 As shown, based on the above-mentioned path dynamic planning method for extreme weather scenarios, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 3 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0039] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a path dynamic planning program 40 for extreme weather scenarios, which can be executed by the processor 10 to implement the path dynamic planning method for extreme weather scenarios in this application.

[0040] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the path dynamic planning method for extreme weather scenarios.

[0041] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface.

[0042] In one embodiment, when the processor 10 executes the path dynamic planning program 40 for extreme weather scenarios stored in the memory 20, the following steps are performed: Obtain the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and collect street view image sets of the target city under extreme weather scenarios according to the road sampling intervals; Obtain the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set; All objects in the target street view image set are identified and analyzed to obtain target analysis results. Based on the target analysis results, the target street view image set is risk-assessed to obtain image risk results. Based on the image risk results, the corresponding road risk results are obtained. Based on the road risk results, path planning is performed to obtain a set of candidate paths. Then, the candidate path set is confirmed according to user requirements to obtain the target planned path information.

[0043] Specifically, the step of collecting street view image sets of the target city under extreme weather scenarios at intervals according to the road is as follows: ; The step of binding road nodes to each street view image in the street view image set according to the road network map specifically involves: ; in, A collection of street view images, Street view image, Number the street views. This is an image acquisition function. The longitude coordinates of the data collection point. The latitude coordinates of the collection point. The orientation angle of the street view image. For street view image acquisition time, The number of street view images. This represents the distance from road nodes within a road segment to street view sampling points. The coordinates of the points on the road segment. This is the nearest road segment.

[0044] Specifically, the step of identifying and analyzing all objects in the target street view image set to obtain target analysis results includes: Define a target hazardous object category, and perform identification processing on all objects in the target street view image set according to the target hazardous object category to obtain object identification results, wherein the object identification results include object category, object bounding box, object confidence score and object coordinate position; Depth maps are generated from the target street view image set to obtain object depth maps, and the relative distance of each object in the object depth map is calculated to obtain the object relative distance; The object scale is calculated based on the relative distance of the object to obtain the object scale information, and the target analysis result is obtained based on the object scale information, the relative distance of the object, and the object recognition result. Specifically, the calculation of the relative distance to each target object in the object depth map is as follows: ; The calculation of object scale based on the relative distance of the objects specifically involves: ; in, The relative distance between objects. The bounding box of the object. is the x-coordinate of a pixel in the object's depth map. y = ... This is a depth map of the object. Number the street views. For object categories, For the number of road sections, This refers to the object's scale information.

[0045] Specifically, the step of performing a risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtaining corresponding road risk results based on the image risk results, includes: Risk weighting is applied to each type of target object in the target analysis results to obtain risk weighting results. Object risk assessment is performed on the risk weighting results to obtain object risk coefficients. Based on the object risk coefficients, risk assessment is performed on each street scene image in the target street scene image set to obtain image risk results. All the street view images are mapped to their corresponding road segments to obtain a mapping table. The image risk results are then aggregated with the mapping table to obtain the road risk results.

[0046] Specifically, the object risk assessment based on the risk weighting result includes: ; The risk assessment of each street view image in the target street view image set based on the object risk coefficient is specifically as follows: ; The aggregation process based on the image risk results and the mapping table specifically involves: ; in, The risk factor of the object. Risk-weighted results The combined risk of the object's size and distance. For risk confidence level, To mitigate the additional risk of nearby dangerous objects, This refers to object scale information. These are the object's relative coordinates. , and All are risk coefficients. The relative distance between objects. This represents the maximum effective distance from which the risk of identifiable objects can be detected in a street view image. The risk index of street view images, The risk index of the road, It is an average function. Street view image, This is the nearest road segment.

[0047] Specifically, the step of performing route planning based on the road risk results to obtain a candidate route set, and then confirming the candidate route set according to user needs to obtain the target planned route information, includes: Based on the road risk results, a comprehensive risk cost is calculated to obtain a comprehensive risk result. The road network map is then updated based on the road risk results to obtain a target road network map. The target road network map is used to solve for the path, and the path solution information is obtained. Then, multi-path matching is performed based on the path solution information to obtain a set of candidate paths. The system obtains user requirements, customizes the candidate path set according to the user requirements to obtain the target planned path, marks the target planned path with risk according to the road risk results to obtain a road risk level map, and generates target planned path information based on the road risk level map and the target planned path.

[0048] Specifically, the calculation of comprehensive risk cost based on the road risk results includes: ; The specific steps for solving the path in the target road network map are as follows: ; The step of customizing the candidate path set according to the user's needs specifically involves: ; in, To consider the overall risk outcome, The length of the road segment Speed ​​limits are based on road segments. The risk index of the road, , and These are all hyperparameters representing the relative weights of the factors. For heuristic functions, Road nodes To the target road node The straight-line distance Based on the maximum speed of the road segment, As a risk penalty coefficient, For nodes Local average risk of surrounding road sections, Plan a path to the goal. For the first Planned route, For the candidate path set, For the first The total travel time for the planned route For the first The total distance of the planned path, For the first The cumulative risk value of each planned path.

[0049] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a path dynamic planning program for extreme weather scenarios, and the path dynamic planning program for extreme weather scenarios, when executed by a processor, implements the steps of the path dynamic planning method for extreme weather scenarios as described above.

[0050] In summary, this invention provides a method, system, terminal, and storage medium for dynamic path planning in extreme weather scenarios. The method includes: acquiring a target road network of a target city; setting different road sampling intervals based on the road level information of the target road network; and collecting a set of street view images of the target city under extreme weather scenarios according to the road sampling intervals; acquiring the road centerlines of the target road network; constructing a road network graph based on the road centerlines; and binding road nodes to each street view image in the street view image set according to the road network graph to obtain a target street view image set; identifying and analyzing all objects in the target street view image set to obtain target analysis results; performing risk assessment on the target street view image set based on the target analysis results to obtain image risk results; and obtaining corresponding road risk results based on the image risk results; performing path planning based on the road risk results to obtain a set of candidate paths; and confirming the candidate path set according to user needs to obtain target planned path information. This invention provides a highly safe planned path by comprehensively assessing the risks of objects around roads under extreme weather scenarios.

[0051] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0052] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0053] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A path dynamic programming method for extreme weather scenarios, characterized in that, The path dynamic programming method for extreme weather scenarios includes: Obtain the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and collect street view image sets of the target city under extreme weather scenarios according to the road sampling intervals; Obtain the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set; All objects in the target street view image set are identified and analyzed to obtain target analysis results. Based on the target analysis results, the target street view image set is risk-assessed to obtain image risk results. Based on the image risk results, the corresponding road risk results are obtained. Based on the road risk results, path planning is performed to obtain a set of candidate paths. Then, the candidate path set is confirmed according to user requirements to obtain the target planned path information.

2. The path dynamic planning method for extreme weather scenarios according to claim 1, characterized in that, The step of collecting street view image sets of the target city under extreme weather scenarios at intervals according to the road is specifically as follows: ； The step of binding road nodes to each street view image in the street view image set according to the road network map specifically involves: ； in, A collection of street view images, Street view image, Number the street views. This is an image acquisition function. The longitude coordinates of the data collection point. The latitude coordinates of the collection point. The orientation angle of the street view image. For street view image acquisition time, The number of street view images. This represents the distance from road nodes within a road segment to street view sampling points. The coordinates of the points on the road segment. This is the nearest road segment.

3. The path dynamic planning method for extreme weather scenarios according to claim 1, characterized in that, The process of identifying and analyzing all objects in the target street view image set to obtain target analysis results specifically includes: Define a target hazardous object category, and perform identification processing on all objects in the target street view image set according to the target hazardous object category to obtain object identification results, wherein the object identification results include object category, object bounding box, object confidence score and object coordinate position; Depth maps are generated from the target street view image set to obtain object depth maps, and the relative distance of each object in the object depth map is calculated to obtain the object relative distance; The object scale is calculated based on the relative distance of the object to obtain the object scale information, and the target analysis result is obtained based on the object scale information, the relative distance of the object, and the object recognition result. Specifically, the calculation of the relative distance to each target object in the object depth map is as follows: ； The calculation of object scale based on the relative distance of the objects specifically involves: ； in, The relative distance between objects. The bounding box of the object. is the x-coordinate of a pixel in the object's depth map. y = ... This is a depth map of the object. Number the street views. For object categories, For the number of road sections, This refers to the object's scale information.

4. The path dynamic planning method for extreme weather scenarios according to claim 1, characterized in that, The step of performing a risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtaining corresponding road risk results based on the image risk results, specifically includes: Risk weighting is applied to each type of target object in the target analysis results to obtain risk weighting results. Object risk assessment is performed on the risk weighting results to obtain object risk coefficients. Based on the object risk coefficients, risk assessment is performed on each street scene image in the target street scene image set to obtain image risk results. All the street view images are mapped to their corresponding road segments to obtain a mapping table. The image risk results are then aggregated with the mapping table to obtain the road risk results.

5. The path dynamic planning method for extreme weather scenarios according to claim 4, characterized in that, The object risk assessment based on the risk weighting result is specifically as follows: ； The risk assessment of each street view image in the target street view image set based on the object risk coefficient is specifically as follows: ； The aggregation process based on the image risk results and the mapping table specifically involves: ； in, The risk factor of the object. Risk-weighted results The combined risk of the object's size and distance. For risk confidence level, To mitigate the additional risk of nearby dangerous objects, This refers to object scale information. These are the object's relative coordinates. , and All are risk coefficients. The relative distance between objects. This represents the maximum effective distance from which the risk of identifiable objects can be detected in a street view image. The risk index of street view images, The risk index of the road, It is an average function. Street view image, This is the nearest road segment.

6. The path dynamic planning method for extreme weather scenarios according to claim 2, characterized in that, The process of planning a route based on the road risk results to obtain a set of candidate routes, and then confirming the candidate routes according to user requirements to obtain the target planned route information, specifically includes: Based on the road risk results, a comprehensive risk cost is calculated to obtain a comprehensive risk result. The road network map is then updated based on the road risk results to obtain a target road network map. The target road network map is used to solve for the path, and the path solution information is obtained. Then, multi-path matching is performed based on the path solution information to obtain a set of candidate paths. The system obtains user requirements, customizes the candidate path set according to the user requirements to obtain the target planned path, marks the target planned path with risk according to the road risk results to obtain a road risk level map, and generates target planned path information based on the road risk level map and the target planned path.

7. The path dynamic planning method for extreme weather scenarios according to claim 6, characterized in that, The calculation of comprehensive risk cost based on the road risk results is specifically as follows: ； The specific steps for solving the path in the target road network map are as follows: ； The step of customizing the candidate path set according to the user's needs specifically involves: ； in, To consider the overall risk outcome, The length of the road segment Speed ​​limits are based on road segments. The risk index of the road, , and These are all hyperparameters representing the relative weights of the factors. For heuristic functions, Road nodes To the target road node The straight-line distance Based on the maximum speed of the road segment, As a risk penalty coefficient, For nodes Local average risk of surrounding road sections, Plan a path to the goal. For the first Planned route, For the candidate path set, For the first The total travel time for the planned route For the first The total distance of the planned path, For the first The cumulative risk value of each planned path.

8. A path dynamic planning system for extreme weather scenarios, characterized in that, The path dynamic planning system for extreme weather scenarios includes: The image acquisition module is used to acquire the target road network of the target city, set different road sampling intervals according to the road level information of the target road network, and acquire a set of street view images of the target city under extreme weather scenarios according to the road sampling intervals. The image processing module is used to acquire the road centerline of the target road network, construct a road network map based on the road centerline, and bind road nodes to each street scene image in the street scene image set based on the road network map to obtain the target street scene image set. The risk assessment module is used to identify and analyze all objects in the target street view image set to obtain target analysis results, perform risk assessment on the target street view image set based on the target analysis results to obtain image risk results, and obtain corresponding road risk results based on the image risk results. The route confirmation module is used to plan a route based on the road risk results, obtain a set of candidate routes, and confirm the route of the candidate route set according to user requirements to obtain the target planned route information.

9. A terminal, characterized in that, The terminal includes a memory, a processor, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the steps of the path dynamic planning method for extreme weather scenarios as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program thereon, and the computer-readable storage medium stores a path dynamic planning program for extreme weather scenarios. When the path dynamic planning program for extreme weather scenarios is executed by a processor, it implements the steps of the path dynamic planning method for extreme weather scenarios as described in any one of claims 1-7.

Images