Dynamic navigation method for roadblock, electronic equipment and storage medium

The dynamic navigation method uses vehicle images to identify roadblocks and update navigation maps, generating detour plans to address the challenge of invisible roadblocks, enhancing traffic safety and efficiency.

CN119085678APending Publication Date: 2024-12-06CHONGQING SELIS PHOENIX INTELLIGENT INNOVATION TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411220999.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

In complex and ever-changing road environments, especially with large-scale invisible roadblocks like accident scenes or natural obstacles obscured by fog and smoke, vehicles struggle to accurately judge the situation, leading to potential safety risks and inefficient traffic management.

Method used

A dynamic navigation method that utilizes initial images from basic vehicles to determine roadblock areas, updates navigation maps, and generates detour plans for affected vehicles, leveraging image stitching, roadblock recognition models, and real-time traffic analysis to ensure timely and safe route adjustments.

Benefits of technology

Enables quick and accurate evaluation of roadblock scope, providing real-time information for emergency management and clear detour instructions, reducing traffic jams and ensuring safety by dynamically re-planning routes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119085678A_ABST
    Figure CN119085678A_ABST
Patent Text Reader

Abstract

The invention provides a dynamic navigation method for a roadblock, electronic equipment and a storage medium, the method is applied to a cloud, and the method comprises the following steps: determining a roadblock area according to initial images uploaded by each basic vehicle and position coordinates corresponding to each initial image; according to the roadblock area, updating the driving cost of each road section corresponding to the roadblock area in a navigation map; determining target vehicles influenced by the roadblock area, for each target vehicle, generating at least two detour schemes corresponding to the target vehicle based on the updated navigation map, and sending each detour scheme to the target vehicle. According to the technical scheme of the invention, the method achieves the timely recognition of a large-scale sudden roadblock, the updating of the navigation map, and the re-dynamic planning of the detouring scheme of each vehicle, and improves the traffic efficiency and safety of a road.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of cloud navigation technology, and in particular to a dynamic navigation method, electronic device and storage medium for roadblocks. Background Art

[0002] In complex and ever-changing road environments, especially when faced with large-scale invisible roadblocks, such as accident scenes or natural obstacles obscured by heavy fog and smoke, it is difficult for a single vehicle to accurately judge the situation of the roadblock, and there may be problems with being unable to avoid it in time, and even safety risks may arise.

[0003] At present, most emergency management departments arrive at the scene to assess the specific scope and impact of each roadblock and upload it to the cloud, which is then sent to each vehicle. This process consumes manpower and material resources and takes a long time, which will directly affect the route selection and traffic smoothness of subsequent vehicles. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, the present application aims to provide a dynamic navigation method, electronic device and storage medium for roadblocks, so as to realize the timely identification of sudden large-scale roadblocks, update the navigation map, and dynamically re-plan the detour plan of each vehicle, thereby improving the traffic efficiency and safety of the road.

[0005] The present application embodiment provides a dynamic navigation method for roadblocks, which is applied to the cloud. The method includes:

[0006] Determine the roadblock area based on the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image;

[0007] According to the roadblock area, updating the travel cost of each road section corresponding to the roadblock area in the navigation map;

[0008] Target vehicles affected by the roadblock area are determined, and for each target vehicle, at least two detour solutions corresponding to the target vehicle are generated based on the updated navigation map, and each detour solution is sent to the target vehicle.

[0009] According to the technical solution provided in the embodiment of the present application, optionally, determining the roadblock area according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image includes:

[0010] Based on the pre-trained roadblock recognition model, the initial images uploaded by each basic vehicle are recognized to determine whether there is a roadblock in each initial image, and the initial image with the roadblock is determined as the target image;

[0011] Perform image stitching according to the target images and the position coordinates corresponding to the target images to obtain at least one stitched image;

[0012] For each spliced ​​image, the spliced ​​image is identified based on the roadblock identification model to determine the roadblock area corresponding to the spliced ​​image.

[0013] According to the technical solution provided in the embodiment of the present application, optionally, performing image stitching according to each target image and the position coordinates corresponding to each target image to obtain at least one stitched image includes:

[0014] Determine an image set according to each target image, determine a first image from the image set, and delete the first image from the image set;

[0015] Determine each second image to be spliced ​​with the first image according to the position coordinates corresponding to the first image, the position coordinates corresponding to each target image in the image set after deleting the first image, and a preset splicing distance, and delete each second image from the image set;

[0016] Performing image stitching on the first image and each second image to obtain a stitched image;

[0017] Determine whether the image set after deleting each second image is empty;

[0018] If not, the process returns to the step of determining a first image from the image set and deleting the first image from the image set until the image set is empty.

[0019] According to the technical solution provided by the embodiment of the present application, optionally, updating the driving cost of each road section corresponding to the roadblock area in the navigation map according to the roadblock area includes:

[0020] For each roadblock area, determining the roadblock point cloud data corresponding to the roadblock area according to the roadblock area and the radar point cloud data corresponding to the initial image uploaded by each basic vehicle;

[0021] Determine the roadblock volume corresponding to the roadblock area according to the roadblock point cloud data;

[0022] According to the roadblock volume, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

[0023] According to the technical solution provided in the embodiment of the present application, optionally, after sending each detour solution to the target vehicle, the method further includes:

[0024] According to the received target solutions selected by each target vehicle from the detour solutions, the travel cost of each road section in the navigation map is updated in real time;

[0025] Based on the updated navigation map, the detour cost corresponding to each target solution is updated in real time.

[0026] According to the technical solution provided in the embodiment of the present application, optionally, updating the driving cost of each road section in the navigation map in real time according to the target solution selected by each target vehicle from each detour solution received includes:

[0027] Determine target road sections corresponding to the target solutions according to the target solutions selected by the target vehicles from the detour solutions received;

[0028] According to the movement information of each target vehicle and the target plan, the future traffic volume of each target road section is predicted;

[0029] The travel cost of each target road section in the navigation map is updated in real time according to the future traffic volume of each target road section.

[0030] According to the technical solution provided in the embodiment of the present application, optionally, the determining of the target vehicles affected by the roadblock area, generating at least two detour solutions corresponding to the target vehicle based on the updated navigation map for each target vehicle, and sending each detour solution to the target vehicle includes:

[0031] Determine the first vehicle according to the navigation path of each basic vehicle and the coordinate position of the roadblock area, and determine the second vehicle according to the current position of each basic vehicle and the coordinate position of the roadblock area;

[0032] determining the first vehicle and the second vehicle as target vehicles affected by the roadblock area;

[0033] For each target vehicle, a path search is performed based on the updated navigation map according to the current position and target position of the target vehicle, and at least two detour plans and the detour costs corresponding to each detour plan are obtained, and each detour plan and the detour costs corresponding to each detour plan are sent to the target vehicle.

[0034] According to the technical solution provided in the embodiment of the present application, optionally, after determining the roadblock area according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image, the method further includes:

[0035] For each roadblock area, judging whether there is a passable lane in the road section corresponding to the roadblock area according to the roadblock area and the preset safety distance;

[0036] If there is a passable lane, the base vehicle traveling in the passable lane is selected as the third vehicle, and roadblock warning information is generated according to the passable lane, and the roadblock warning information is sent to the third vehicle;

[0037] According to the driving time of the third vehicle in the roadblock area, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

[0038] In summary, the present application proposes a dynamic navigation method for roadblocks, which determines the roadblock area in the cloud based on the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image, so as to facilitate the self-analysis of the roadblock situation in the cloud, and then, according to the roadblock area, updates the driving cost of each road section corresponding to the roadblock area in the navigation map, and determines the target vehicles affected by the roadblock area. For each target vehicle, at least two detour plans corresponding to the target vehicle are generated based on the updated navigation map, and each detour plan is sent to the target vehicle, so as to achieve the effect of timely identification of sudden large-scale roadblocks, update the navigation map, and dynamically re-plan the detour plan of each vehicle. Compared with the existing technology, it can quickly and accurately evaluate the scope of the roadblock, provide real-time information for the emergency management department, so as to accelerate the rescue and cleanup operations, and provide clear detour instructions for the driver to avoid blindly approaching unknown risk areas, ensure the safety of people and vehicles, reduce traffic jams caused by roadblocks, and improve the overall traffic efficiency and safety of the road network. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of a dynamic navigation method for roadblocks provided in an embodiment of the present application;

[0040] Figure 2 is a flow chart of another dynamic navigation method for roadblocks provided in an embodiment of the present application;

[0041] Figure 3 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0043] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0044] Figure 1It is a flow chart of a dynamic navigation method for roadblocks provided in an embodiment of the present application.

[0045] See also Figure 1 , the dynamic navigation method for roadblocks is applied in the cloud, specifically including:

[0046] S110: Determine a roadblock area according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image.

[0047] Among them, the basic vehicles are the vehicles connected to the cloud. It is understandable that each basic vehicle can upload motion information (such as speed, heading angle, etc.) and collected surrounding environment information (such as images, radar point clouds, etc.) to the cloud for subsequent roadblock identification and detour planning. The initial image is the image information around the basic vehicle collected by each basic vehicle through the on-board camera. The position coordinates are the current position information determined by the positioning system of the basic vehicle, such as longitude and latitude information. The roadblock area is an area where there are obstacles to road traffic, such as areas covered by heavy fog and smoke, areas where the ground has collapsed, areas where fallen rocks have accumulated, etc.

[0048] Specifically, the cloud can receive the initial images uploaded by each basic vehicle, as well as the coordinates of the basic vehicle when the initial images were collected. Furthermore, these initial images can be spliced ​​to obtain panoramic images of each road section, and roadblocks can be identified in the panoramic images of each road section. For example, deep learning image analysis can be used to identify whether the panoramic image contains roadblocks, the coverage of the roadblocks, and the location information of the roadblocks, so as to determine the roadblock area.

[0049] It is understandable that the on-board sensor array of each base vehicle (including but not limited to cameras, radars, LiDAR, etc.) can upload the collected sensor data to the cloud. In this way, when a roadblock is detected by a certain base vehicle, the base vehicles on both sides of the base vehicle can collect images and depth information of the edge of the roadblock from different perspectives, and can use wide-angle lenses and 3D (Three-Dimensional) scanning technology to cover as much range as possible. At the same time, the base vehicle uploads the positioning information of the GPS (Global Positioning System) and the sensor data to the cloud to ensure the accuracy of the geographical coordinates of the data.

[0050] For example, the cloud receives the initial images uploaded by each basic vehicle, and can use advanced image stitching technology to stitch the multi-view images to form a panoramic view of the roadblock. Assume that a series of initial images I1, I2, ..., I n, these initial images cover different angles of the roadblock. In order to generate a panoramic view, it is necessary to solve the image registration problem, that is, to find each pair of initial images I i and I j The geometric transformation relationship between ij This can be achieved through feature matching and RANSAC (Random Sample Consensus) algorithm. Feature matching can use SIFT (Scale-invariant feature transform), SURF (Speeded-Up Robust Features) or ORB (Oriented FAST and Rotated BRIEF) algorithms. Let p i is the initial image I i A feature point on the initial image I j The corresponding point on is p′ j , then the geometric transformation relationship Tij can be estimated by minimizing the reprojection error:

[0051]

[0052] Once the geometric transformation relationship between all the initial images is determined, a fusion algorithm in multi-view geometry (such as optical flow-based fusion or radiance-based fusion) can be used to generate a panoramic view of the roadblock area. Furthermore, semantic segmentation and instance segmentation models can be applied to identify and distinguish roadblock areas from passable areas, and accurately measure the area covered by the roadblock.

[0053] S120: Update the driving cost of each road section corresponding to the roadblock area in the navigation map according to the roadblock area.

[0054] The navigation map is a map containing the road sections and the driving costs of each road section, which is used to plan the routes for each basic vehicle. The driving cost is the cost required to pass the corresponding road section, which is related to the estimated travel time, number of lanes, traffic volume, etc.

[0055] Specifically, the sections affected by each roadblock area can be determined, and then the impact of the roadblock area on these sections can be determined, that is, the driving cost can be determined. For each section affected by the roadblock area, the roadblock type in the section and the impact parameter of the roadblock on the section can be determined, and the updated driving cost of the section can be determined based on the roadblock type, impact parameter and driving cost of the section. For example, if the road collapses completely, the section will be unable to continue to be passed, and the driving cost can be considered to be infinite; if two of the four lanes are blocked by fallen rocks, the driving cost can be considered to be twice that before the roadblock; if there is heavy fog, the driving cost can be updated based on visibility. Therefore, the driving cost of each section corresponding to the roadblock area in the navigation map can be updated through the analysis of the roadblock area, so that the path can be replanned on this basis for each vehicle affected by the roadblock area in the future.

[0056] Based on the above example, the travel cost of each road section corresponding to the roadblock area in the navigation map can be updated according to the roadblock area in the following manner:

[0057] For each roadblock area, determine the roadblock point cloud data corresponding to the roadblock area based on the radar point cloud data corresponding to the roadblock area and the initial images uploaded by each basic vehicle;

[0058] According to the roadblock point cloud data, determine the roadblock volume corresponding to the roadblock area;

[0059] According to the volume of the roadblock, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

[0060] Among them, radar point cloud data is data collected by the radar sensor of the base vehicle, which is data obtained and uploaded synchronously with the initial image and is consistent with the initial image in time and shooting space. Roadblock point cloud data is point cloud data corresponding to the roadblock area, which can be point cloud data obtained by combining various radar point cloud data related to the roadblock area. Roadblock volume is the volume data of the roadblock in three-dimensional space.

[0061] Specifically, each roadblock area can be processed separately. According to the roadblock area, the radar point cloud data contained in the area can be obtained from the radar point cloud data corresponding to the initial image uploaded by each basic vehicle, and spliced ​​and combined to obtain the roadblock point cloud data corresponding to the roadblock area. Alternatively, the radar point cloud data corresponding to each initial image used to determine the roadblock area can be spliced ​​and combined accordingly to obtain the roadblock point cloud data corresponding to the roadblock area. Further, the volume of the roadblock point cloud data is calculated in three-dimensional space to obtain the roadblock volume corresponding to the roadblock area. Since different roadblock volumes have different effects on road sections, for each road section affected by the roadblock, the influence coefficient of the driving cost of the road section can be determined according to the roadblock volume in the road section. Then, the driving cost of the road section before being affected by the roadblock can be processed by the influence coefficient to obtain the updated driving cost of the road section. Then, the driving cost of each road section corresponding to the roadblock area in the navigation map can be updated.

[0062] For example, based on the roadblock point cloud data, a three-dimensional model of the roadblock can be constructed, its spatial structure can be further refined, the volume of the roadblock can be calculated, and the actual degree of obstruction (impact coefficient) of the roadblock to road traffic can be evaluated. n is a collection of roadblock point cloud data, where each point p i With three-dimensional coordinates (x i ,y i ,z i ). To simplify the calculation, the set of roadblock point cloud data is first divided into multiple clusters by a clustering algorithm, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), each cluster representing a part of the roadblock. Then, for each cluster C, a plane or surface model F can be fitted by the least squares method. C (x,y,z), and calculate its volume V C .

[0063] If the cluster C is approximately a plane, its volume can be calculated by integration:

[0064]

[0065] Where z(x,y) is the plane equation F C (x,y,z) = 0 is the z coordinate at point (x,y). Finally, the total volume of the obstacle V total is the sum of all cluster volumes:

[0066] S130: Determine target vehicles affected by the roadblock area, generate at least two detour plans corresponding to each target vehicle based on the updated navigation map, and send each detour plan to the target vehicle.

[0067] The target vehicles are basic vehicles affected by the roadblock area, such as basic vehicles that are planned to enter the roadblock area or travel near the roadblock area. The detour plan is a plan to avoid driving in the roadblock area.

[0068] Specifically, each basic vehicle can be analyzed to determine whether it will be affected by the roadblock area. If it is affected by the roadblock area, the basic vehicle is used as the target vehicle. For each target vehicle, the current position and destination of the target vehicle can be determined, and a path search can be performed based on the updated navigation map, such as using the Dijkstra algorithm, to obtain at least two detour solutions corresponding to the target vehicle, and each detour solution is sent to the target vehicle for the driver of the target vehicle to choose to detour and avoid the roadblock area, thereby improving driving safety.

[0069] For example, the traffic network can be abstracted as a weighted directed graph G = (V, E), where V is a vertex set, representing the starting and ending points of road intersections or road sections; E is an edge set, representing road sections, and each edge e∈E has a weight w(e), i.e., the travel cost, which represents the cost required to travel this distance. For the path search problem, for a given starting point s (current location) and end point t (destination), the goal of dynamic route planning is to find at least two paths P with detour solutions from s to t. * .

[0070] Based on the above example, after sending each detour plan to the target vehicle, the target plan selected by each target vehicle from the detour plans can also be received, and the change of future traffic flow of each road section is estimated according to the selection of the target plan. The impact on each target plan can be specifically:

[0071] According to the target solution selected by each target vehicle from various detour solutions, the driving cost of each road section in the navigation map is updated in real time;

[0072] Based on the updated navigation map, the detour cost corresponding to each target solution is updated in real time.

[0073] The target solution is a detour solution selected by the target vehicle from various detour solutions provided by the cloud. The detour cost is the cost required to drive according to the target solution, such as distance, time, etc.

[0074] Specifically, the cloud can receive the target solution selected by each target vehicle from various detour solutions, and predict the traffic flow changes of each road section involved in the target solution based on the received target solutions, and then update the driving cost of each road section related to the target solution in the navigation map in real time. Since the driving cost of each road section related to the target solution changes, the detour cost corresponding to each target solution can be updated in real time based on the navigation map after the driving cost is updated, and sent to the corresponding target vehicle.

[0075] It is understandable that the result of some target vehicles selecting the target solution may have a dynamic impact on the detour cost of the detour solution of the target vehicles that have not completed the selection. For example, target vehicle A receives detour solution 1 and detour solution 2, and target vehicle B receives detour solution 1 and detour solution 3. It is known that some sections of detour solution 3 overlap with detour solution 1. If target vehicle A selects detour solution 1, then the choice of target vehicle A will have an impact on the detour cost of detour solution 1, specifically, the driving cost of each section in detour solution 1. Therefore, the detour costs corresponding to detour solution 1 and detour solution 3 received by target vehicle B also change in real time, and target vehicle B can make a choice based on the detour solutions after the detour cost changes. After target vehicle A selects detour solution 1, the detour cost of detour solution 1 will also change in real time.

[0076] Based on the above example, the travel cost of each road section in the navigation map can be updated in real time according to the target solution selected by each target vehicle from various detour solutions in the following manner:

[0077] Determine target road sections corresponding to the target solutions according to the target solutions selected by the target vehicles from the detour solutions received;

[0078] According to the movement information of each target vehicle and the target plan, the future traffic volume of each target road section is predicted;

[0079] According to the future traffic volume of each target road section, the driving cost of each target road section in the navigation map is updated in real time.

[0080] The target road section is the road section involved in each target solution. The motion information is the motion speed and heading angle of the target vehicle. The future traffic volume is the predicted traffic volume at each time point in the future.

[0081] Specifically, the road sections involved in each target scheme are taken as target sections. Furthermore, for each target section, the future flow of the target section can be predicted according to the motion information of each target vehicle that will pass through the target section and the distance between the target section and the target vehicle in the target scheme. The future flow of each target section at different future moments will affect the driving cost at the corresponding future moment. Therefore, the driving cost corresponding to each future moment can be updated according to the future flow of each target section at each future moment, and accordingly, the driving cost of each target section in the navigation map can be updated in real time.

[0082] For example, a dynamic traffic flow model can be introduced for processing. In the real world, traffic flow changes dynamically, so a dynamic traffic flow model can be established. Assume f e (t) is the vehicle flow entering edge e at time t, g e (t) is the flow of vehicles leaving edge e at time t, q e (t) is the queue length of edge e. The edge can be understood as a road segment, so the dynamic traffic model can be expressed as:

[0083]

[0084] g e (t) = min(μ e ,f e (t)+q e (t))

[0085] Among them, ρ e (t) is the vehicle generation rate arriving at edge e, μ e is the maximum outflow rate of edge e.

[0086] Dynamic route planning needs to consider the traffic status at future moments, so a prediction model is introduced to estimate future traffic flow. is the predicted flow entering edge e at time (t+τ), then the dynamic route planning problem can be expressed as an optimization problem:

[0087]

[0088] Where P is the target solution, Δt is the travel time of the target road segment, and w e (t+τ) is the travel cost of edge e at time t+τ, which depends on the predicted traffic

[0089] For each target vehicle, a personalized detour solution can be generated based on the current location, destination (target location), preferences (such as shortest time, shortest distance or least cost, etc.), real-time traffic information and movement information. Let U be the set of all users, P uis the detour solution of user u, then the personalized detour suggestion problem can be expressed as:

[0090]

[0091] Among them, w e (t+τ) is the travel cost of edge e at time t+τ, which is affected by the target path selected by all users.

[0092] The dynamic navigation method for roadblocks provided in the embodiment of the present application determines the roadblock area in the cloud based on the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image, so as to facilitate the self-analysis of the roadblock situation in the cloud. Then, according to the roadblock area, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated, and the target vehicles affected by the roadblock area are determined. For each target vehicle, at least two detour plans corresponding to the target vehicle are generated based on the updated navigation map, and each detour plan is sent to the target vehicle, thereby realizing the effect of timely identifying sudden large-scale roadblocks, updating the navigation map, and dynamically re-planning the detour plan for each vehicle. Compared with the existing technology, it can quickly and accurately evaluate the scope of the roadblock, provide real-time information for the emergency management department, so as to accelerate the rescue and cleanup operations, and provide clear detour instructions for the driver to avoid blindly approaching unknown risk areas, ensure the safety of people and vehicles, reduce traffic jams caused by roadblocks, and improve the traffic efficiency and safety of the overall road network.

[0093] Figure 2 FIG. 1 is a flow chart of another dynamic navigation method for roadblocks provided in an embodiment of the present application. Based on the above embodiments, the process of determining the roadblock area and the process of generating a detour plan are exemplarily described. Figure 2 , the dynamic navigation method for roadblocks specifically includes:

[0094] S210: Based on the pre-trained roadblock recognition model, the initial images uploaded by the basic vehicles are recognized to determine whether there is a roadblock in each initial image, and the initial image with the roadblock is determined as the target image.

[0095] Among them, the roadblock recognition model is a model used to distinguish between roadblock areas and passable areas. It can be a deep learning model and needs to be trained in advance based on sample original images and sample segmented images. The sample original image is a directly collected image corresponding to the initial image, and the sample segmented image is an annotated image of the sample original image with roadblock areas and passable areas annotated. The target image is the initial image in which the roadblock is identified, that is, the initial image in which the roadblock area exists.

[0096] Specifically, the initial images uploaded by each basic vehicle are respectively input into the pre-trained roadblock recognition model to obtain the recognition results corresponding to each initial image. According to the recognition results corresponding to each initial image, it can be judged whether there is a roadblock in each initial image, that is, whether the roadblock area is recognized. If there is a roadblock, the corresponding initial image is used as the target image for subsequent stitching.

[0097] For example, a fully convolutional network (FCN) or other deep learning models is used to construct a roadblock recognition model. Let X be the input sample original image, Y be the output image of the roadblock recognition model, that is, the predicted sample predicted image, and Y true Segment the image for the sample. Each pixel y in Y represents the category to which the pixel belongs (roadblock or passable).

[0098] In the process of selecting the roadblock recognition model, a loss function L can be defined to measure the difference between the prediction result and the true label, that is, the difference between the sample prediction image and the sample segmentation image. For example, the cross entropy loss can be used:

[0099]

[0100] Among them, Y true is the true label, and Y(y) is the probability distribution of the sample prediction image predicted by the model.

[0101] S220 , performing image stitching according to the target images and the position coordinates corresponding to the target images to obtain at least one stitched image.

[0102] The stitched image is an image obtained by stitching target images that have intersections with each other.

[0103] Specifically, each target image is clustered according to the position coordinates to obtain multiple groups of target images, and it can be understood that each group of target images corresponds to the same roadblock area. Then, each group of target images is spliced ​​to obtain at least one spliced ​​image.

[0104] Based on the above example, images may be stitched according to the target images and the position coordinates corresponding to the target images to obtain at least one stitched image:

[0105] Determine an image set according to each target image, determine a first image from the image set, and delete the first image from the image set;

[0106] Determine each second image to be spliced ​​with the first image according to the position coordinates corresponding to the first image, the position coordinates corresponding to each target image in the image set after deleting the first image, and the preset splicing distance, and delete each second image from the image set;

[0107] Performing image stitching on the first image and each second image to obtain a stitched image;

[0108] Determine whether the image set after deleting each second image is empty;

[0109] If not, the process returns to the step of determining a first image from the image set and deleting the first image from the image set until the image set is empty.

[0110] The image set is a set of target images. The first image is a basic image used to form a stitched image, and can be any target image in the image set. The preset stitching distance is the maximum distance between any two target images when they can be stitched together. The second image is another target image used to stitch into a stitched image in addition to the first image.

[0111] Specifically, each target image constitutes an image set, and any target image is determined from the image set as the first image, and the first image is deleted from the image set. According to the coordinate position of the first image, according to the position coordinates corresponding to each target image in the image set after deleting the first image, each target image with overlap between images is selected, that is, the image distance between the first image and each target image in the image set after deleting the first image is determined, and if the image distance is within the preset stitching distance, it is determined to be the second image. Then, the second image is spliced ​​with the first image to obtain a process image, and each target image with overlap between the process image is determined as the second image, until there is no longer any target image that can overlap with the process image, and the process image at this time is taken as a stitched image. It can be understood that the first image and the second images determined each time are stitched to obtain a stitched image. Delete each second image from the image set, and determine whether the image set is empty after deleting each second image. If it is empty, it means that the image stitching is completed and at least one stitched image is obtained for subsequent identification of the roadblock area; if it is not empty, it means that the stitching processing of each target image about the roadblock has not been completed. Therefore, return to the step of determining a first image from the image set and deleting the first image from the image set until the image set is empty.

[0112] S230 : For each stitched image, based on a roadblock recognition model, identify the stitched image and determine a roadblock area corresponding to the stitched image.

[0113] Specifically, for each spliced ​​image, a roadblock recognition model can be used for recognition, and a result image with the roadblock area and the passable area marked can be output, thereby knowing the roadblock area corresponding to the spliced ​​image.

[0114] S240: Update the travel cost of each road section corresponding to the roadblock area in the navigation map according to the roadblock area.

[0115] S250, determining a first vehicle according to the navigation paths of the basic vehicles and the coordinate positions of the roadblock area, and determining a second vehicle according to the current positions of the basic vehicles and the coordinate positions of the roadblock area; determining the first vehicle and the second vehicle as target vehicles affected by the roadblock area.

[0116] The first vehicle is each basic vehicle whose navigation path will pass through the roadblock area. The second vehicle is a basic vehicle that travels into or near the roadblock area.

[0117] Specifically, the navigation path of each basic vehicle is analyzed to determine whether it passes through the roadblock area, that is, whether the coordinates of each path point to be passed in the future in the navigation path are within the coordinate position of the roadblock area. If so, the basic vehicle is determined as the first vehicle. The current position of each basic vehicle is analyzed to determine whether it has traveled into the roadblock area or near the roadblock area, that is, whether the current position is within the coordinate position of the roadblock area. If so, the basic vehicle is determined as the second vehicle. Further, the first vehicle and the second vehicle are determined as target vehicles affected by the roadblock area.

[0118] S260. For each target vehicle, according to the current position and the target position of the target vehicle, a path search is performed based on the updated navigation map to obtain at least two detour plans and the detour costs corresponding to each detour plan, and each detour plan and the detour costs corresponding to each detour plan are sent to the target vehicle.

[0119] Specifically, for each target vehicle, a path search is performed using the current position of the target vehicle and the target position on the updated navigation map to obtain at least two detour plans, and the detour cost corresponding to each detour plan is calculated respectively. In order to facilitate the driver of the target vehicle to quickly select a plan, each detour plan and the detour cost corresponding to each detour plan are sent to the target vehicle for the driver to view and select.

[0120] Based on the above example, after the roadblock area is determined according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image, each basic vehicle that is still determined to be traveling in the passable lane within the road section corresponding to the roadblock area may be pre-reminded:

[0121] For each roadblock area, determine whether there is a passable lane in the road section corresponding to the roadblock area according to the roadblock area and the preset safety distance;

[0122] If there is a passable lane, the base vehicle traveling in the passable lane is selected as the third vehicle, and roadblock warning information is generated according to the passable lane, and the roadblock warning information is sent to the third vehicle;

[0123] According to the driving time of the third vehicle in the roadblock area, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

[0124] Among them, the preset safety distance is the minimum distance between the vehicle and the edge of the roadblock area when driving. The passable lanes are the lanes in the road section of the roadblock area that are not in the roadblock area and the preset safety distance of the roadblock area. The third vehicle is each basic vehicle that is determined to still be driven in the passable lane in the road section corresponding to the roadblock area. The roadblock warning information is information used to remind the number of passable lanes, lane positions, maximum driving speeds and other safe driving reminders. The driving time is the time consumed by the third vehicle to drive through the roadblock area in the passable lane.

[0125] Specifically, for each roadblock area, the roadblock area is expanded according to the preset safety distance to obtain an impassable area, and it is determined whether there is a passable lane outside the impassable area in the road section corresponding to the roadblock area. If there is no passable lane, the vehicle is directly reminded that it cannot pass and the detour plan can be changed. If there is a passable lane, the driver can choose whether to choose a detour plan or choose to drive in the passable lane to pass the road section corresponding to the roadblock area. Therefore, the basic vehicle that chooses to drive in the passable lane can be used as the third vehicle. According to each determined passable lane, a roadblock warning information including the number of each passable lane, lane position, and maximum driving speed is clearly generated, and the roadblock warning information is sent to the third vehicle to provide a safety reminder when the third vehicle drives through each road section corresponding to the roadblock area based on the passable lane, so as to avoid safety threats to the third vehicle caused by the roadblock area or other vehicles. Furthermore, after the third vehicle passes through the sections corresponding to the roadblock area based on the passable lane, the driving costs of the sections corresponding to the roadblock area in the navigation map can be updated according to the driving time of the third vehicle in the roadblock area, so as to provide prompts for the third vehicles that choose to pass in the future.

[0126] Optionally, in collaborative obstacle avoidance, the third vehicles can be combined to perform comprehensive obstacle avoidance, with the goal of ensuring that all third vehicles can safely adjust their routes when facing the roadblock area. Let O be the set of edge points in the roadblock area, d(v k ,o) is the third vehicle v k The distance from the edge point o, vk (t) is the third vehicle v k At time t, n is the total number of third vehicles, then the obstacle avoidance problem can be expressed as:

[0127]

[0128] Among them, r o is the preset safety distance. The goal of the above problem is to minimize the shortest distance between all vehicles and obstacles.

[0129] Optionally, when transmitting the initial image and detour plan, information broadcasting can be used for information exchange. The goal of information broadcasting is to quickly disseminate key information (such as the initial image, detour plan, roadblock location, traffic conditions, etc.) to all relevant vehicles and infrastructure. Let M be the message to be broadcast, T(v k , M) is the base vehicle v k The time to receive message M, n is the total number of basic vehicles, then the information broadcast problem can be expressed as minimizing the sum of the time for all basic vehicles to receive the message:

[0130]

[0131] The broadcast strategy may include direct broadcast, multi-hop broadcast or infrastructure-based broadcast.

[0132] The dynamic navigation method for roadblocks provided in the embodiment of the present application recognizes the initial images uploaded by each basic vehicle based on a pre-trained roadblock recognition model, determines whether there is a roadblock in each initial image, and determines the initial image with the roadblock as the target image. According to each target image and the position coordinates corresponding to each target image, image stitching is performed to obtain at least one stitched image, so as to effectively utilize the initial images taken and uploaded by each basic vehicle. Then, for each stitched image, based on the roadblock recognition model, the stitched image is identified, and the roadblock area corresponding to the stitched image is determined, so as to achieve the purpose of timely failure of the roadblock area, and further , determine the first vehicle according to the navigation path of each basic vehicle and the coordinate position of the roadblock area, and determine the second vehicle according to the current position of each basic vehicle and the coordinate position of the roadblock area; determine the first vehicle and the second vehicle as target vehicles affected by the roadblock area, and for each target vehicle, perform path search based on the updated navigation map according to the current position and target position of the target vehicle, obtain at least two detour plans and the detour costs corresponding to each detour plan, and send each detour plan and the detour costs corresponding to each detour plan to the target vehicle, so as to provide the driver with clear detour plan guidance and improve driving safety.

[0133] Figure 3 Schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 3 As shown, the electronic device 300 includes one or more processors 301 and a memory 302 .

[0134] The processor 301 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 300 to perform desired functions.

[0135] The memory 302 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 301 may run the program instructions to implement the dynamic navigation method for roadblocks and / or other desired functions of any embodiment of the present application described above. Various contents such as initial external parameters, thresholds, etc. may also be stored in the computer-readable storage medium.

[0136] In one example, the electronic device 300 may further include: an input device 303 and an output device 304, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown). The input device 303 may include, for example, a keyboard, a mouse, etc. The output device 304 may output various information to the outside, including early warning information, braking force, etc. The output device 304 may include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, etc.

[0137] Of course, to simplify, Figure 3 Only some of the components related to the present application in the electronic device 300 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 300 may also include any other appropriate components.

[0138] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the dynamic navigation method for roadblocks provided in any embodiment of the present application.

[0139] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0140] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the processor executes the steps of the dynamic navigation method for roadblocks provided in any embodiment of the present application.

[0141] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0142] It should be noted that the terms used in this application are only for describing specific embodiments, rather than limiting the scope of this application. As shown in the specification and claims of this application, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method or device including the elements.

[0143] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0144] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and its core ideas of this application. The above is only the preferred implementation method of this application. It should be pointed out that due to the limitations of textual expression and the objective existence of infinite specific structures, ordinary technicians in this technical field can make several improvements, modifications or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, should be regarded as the scope of protection of this application.

Claims

1. A dynamic navigation method for roadblocks, characterized in that: Applied in the cloud, including: Determine the roadblock area based on the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image; According to the roadblock area, updating the travel cost of each road section corresponding to the roadblock area in the navigation map; Target vehicles affected by the roadblock area are determined, and for each target vehicle, at least two detour solutions corresponding to the target vehicle are generated based on the updated navigation map, and each detour solution is sent to the target vehicle.

2. The method according to claim 1, characterized in that The determining of the roadblock area according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image includes: Based on the pre-trained roadblock recognition model, the initial images uploaded by each basic vehicle are recognized to determine whether there is a roadblock in each initial image, and the initial image with the roadblock is determined as the target image; Perform image stitching according to the target images and the position coordinates corresponding to the target images to obtain at least one stitched image; For each spliced ​​image, the spliced ​​image is identified based on the roadblock identification model to determine the roadblock area corresponding to the spliced ​​image.

3. The method according to claim 2, characterized in that The step of performing image stitching according to the target images and the position coordinates corresponding to the target images to obtain at least one stitched image includes: Determine an image set according to each target image, determine a first image from the image set, and delete the first image from the image set; Determine each second image to be spliced ​​with the first image according to the position coordinates corresponding to the first image, the position coordinates corresponding to each target image in the image set after deleting the first image, and a preset splicing distance, and delete each second image from the image set; Performing image stitching on the first image and each second image to obtain a stitched image; Determine whether the image set after deleting each second image is empty; If not, the process returns to the step of determining a first image from the image set and deleting the first image from the image set until the image set is empty.

4. The method according to claim 1, characterized in that The updating of the travel cost of each road section corresponding to the roadblock area in the navigation map according to the roadblock area includes: For each roadblock area, determining the roadblock point cloud data corresponding to the roadblock area according to the roadblock area and the radar point cloud data corresponding to the initial image uploaded by each basic vehicle; Determine the roadblock volume corresponding to the roadblock area according to the roadblock point cloud data; According to the roadblock volume, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

5. The method according to claim 1, characterized in that After sending each detour solution to the target vehicle, the method further includes: According to the received target solutions selected by each target vehicle from the detour solutions, the travel cost of each road section in the navigation map is updated in real time; Based on the updated navigation map, the detour cost corresponding to each target solution is updated in real time.

6. The method according to claim 5, characterized in that The method of updating the driving cost of each road section in the navigation map in real time according to the target solution selected by each target vehicle from each detour solution received includes: Determine target road sections corresponding to the target solutions according to the target solutions selected by the target vehicles from the detour solutions received; According to the movement information of each target vehicle and the target plan, the future traffic volume of each target road section is predicted; The travel cost of each target road section in the navigation map is updated in real time according to the future traffic volume of each target road section.

7. The method according to claim 1, characterized in that The determining of target vehicles affected by the roadblock area, generating at least two detour solutions corresponding to each target vehicle based on an updated navigation map, and sending each detour solution to the target vehicle includes: Determine the first vehicle according to the navigation path of each basic vehicle and the coordinate position of the roadblock area, and determine the second vehicle according to the current position of each basic vehicle and the coordinate position of the roadblock area; determining the first vehicle and the second vehicle as target vehicles affected by the roadblock area; For each target vehicle, a path search is performed based on the updated navigation map according to the current position and target position of the target vehicle, and at least two detour plans and the detour costs corresponding to each detour plan are obtained, and each detour plan and the detour costs corresponding to each detour plan are sent to the target vehicle.

8. The method according to claim 1, characterized in that After determining the roadblock area according to the initial images uploaded by each basic vehicle and the position coordinates corresponding to each initial image, the method further includes: For each roadblock area, judging whether there is a passable lane in the road section corresponding to the roadblock area according to the roadblock area and the preset safety distance; If there is a passable lane, the base vehicle traveling in the passable lane is selected as the third vehicle, and roadblock warning information is generated according to the passable lane, and the roadblock warning information is sent to the third vehicle; According to the driving time of the third vehicle in the roadblock area, the driving cost of each road section corresponding to the roadblock area in the navigation map is updated.

9. An electronic device, characterized in that: The electronic device comprises: Processor and memory; The processor is used to execute the steps of the dynamic navigation method for roadblocks as described in any one of claims 1 to 8 by calling the program or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program or instruction, which enables a computer to execute the steps of the dynamic navigation method for roadblocks as described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • Vehicle driving guiding method and device

    CN114373311A

  • Traffic event processing method, device and equipment

    CN115985090A

  • Traffic jam adjusting method and device, electronic equipment and medium

    CN116030654A

  • Navigation information updating method and device, electronic equipment and storage medium

    CN116989809A

  • Cross-intersection traffic management method and device, computer equipment and storage medium

    CN117351744A