A dynamic traffic prediction method and system for autonomous driving vehicle path planning
Through real-time traffic analysis and monitoring analysis, the path planning method of autonomous driving vehicles solves the problem of information delay, optimizes the path in real time to avoid road congestion, and improves user experience and commuting efficiency.
Patent Information
- Application Number
- CN202411237773.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-05
AI Technical Summary
The existing autonomous vehicle path planning methods have information delays in the path planning scenario, and it is difficult to predict in advance or judge road congestion based on actual conditions, resulting in insufficient replacement of congested roads in the planned path.
By obtaining the real-time location, target location and user needs of the vehicle, combining traffic analysis methods to analyze and optimize target paths in real time, using traffic prediction methods to predict congestion, and monitoring roads in real time through monitoring and analysis methods to generate replacement paths to avoid congestion.
It reduces the delay in congestion information, improves the user's riding experience and the commuting efficiency of autonomous driving, and ensures that the path planning is more in line with the actual needs of users.
Smart Images

Figure CN118776584B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of path planning, and in particular to a dynamic traffic prediction method and system for path planning of an autonomous driving vehicle. Background Art
[0002] Path planning refers to calculating the best path from the starting point to the destination under given input conditions. This process is widely used in many fields, including robot navigation, autonomous driving, logistics and transportation. Path planning of autonomous vehicles is one of its core functions, involving the selection of the best path from the starting point to the destination. The goal of path planning is to ensure that the vehicle drives safely and efficiently under various environmental conditions.
[0003] The patent publication number is CN111829545A. The method and system for dynamic planning of an autonomous driving vehicle and its motion trajectory include: obtaining the current position of the autonomous driving vehicle; perceiving environmental data of the environment surrounding the autonomous driving vehicle; extracting lane information about lanes from the environmental data; obtaining the first drivable area of the autonomous driving vehicle based on the current position of the autonomous driving vehicle, high-definition maps and lane information; extracting static information about static objects from the environmental data; extracting dynamic information about dynamic objects from the environmental data, and predicting the motion trajectory of the dynamic objects based on the dynamic information; planning a second drivable area based on the first drivable area, static information, motion trajectory, and lane information. In addition, the present invention also provides a dynamic planning system for an autonomous driving vehicle and its motion trajectory. The technical solution of the present invention effectively solves the problem of how to reasonably plan the motion trajectory of an autonomous driving vehicle.
[0004] The above method can achieve the effect of reasonably planning the motion trajectory of the autonomous driving vehicle according to its motion trajectory planning method. However, the above method relies more on navigation software in the scenario of path planning, that is, from the starting point to the end point. However, when using the navigation software in this scenario, there is a disadvantage that the information contains delays. At present, when the navigation software prompts that there is congestion on the road, the congestion on the road has already occurred, which is not convenient for predicting in advance or judging the congestion of the road according to the actual situation and replacing the congested road in the planned path. Therefore, the present invention is proposed. Summary of the invention
[0005] The purpose of the present invention is to provide a dynamic traffic prediction method and system for path planning of an autonomous driving vehicle to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above object, the present invention provides the following technical solution: a dynamic traffic prediction method for path planning of an autonomous driving vehicle, the method comprising:
[0007] Data acquisition: obtain the real-time location of the vehicle, target location, user needs and real-time time;
[0008] Path planning: Based on the real-time location and the target location, several initial paths and auxiliary information corresponding to the initial paths are obtained through path planning. The auxiliary information includes cost information and time information. Based on user needs and the auxiliary information, the path determination method is used to select the optimal initial path from several initial paths to obtain the target path and target priority;
[0009] Path traffic situation analysis and optimization: Based on the target path, the traffic analysis method is used to analyze in real time whether the road sections in the target path will be congested, and the congested road sections are optimized to obtain result information. The result information includes the predicted congested road sections and their predicted optimized road sections and the actual congested road sections and the actual optimized road sections. Based on the result information, the corresponding road sections in the target path are replaced to obtain the optimized target path. Based on the optimized target path, the traffic analysis method is used to obtain subsequent result information. Based on the subsequent result information, the optimized target path is further optimized, and the vehicle travels based on the continuously optimized target path;
[0010] The traffic analysis method includes: extracting road information in a target path before or after optimization to obtain a number of target roads, predicting congestion conditions on the target roads by a traffic prediction method to obtain prediction results, the prediction results including predicted congestion information and predicted non-congestion information, the predicted congestion information including predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments, monitoring the target roads in real time by a monitoring and analysis method to obtain analysis results, the analysis results including actual congestion information and actual non-congestion information, the actual congestion information including actual congested road segments and their estimated congestion durations, generating replacement paths by a path optimization method based on the predicted congestion information or the actual congestion information to obtain predicted optimized road segments or actual optimized road segments, combining the predicted congested road segments generated by the target path before optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain result information, combining the predicted congested road segments generated by the target path after optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain subsequent result information.
[0011] Furthermore, the traffic prediction method includes: extracting the starting point of the target road to obtain a node, obtaining the time of arrival at the node based on the real-time time to obtain the instantaneous time, numbering and sorting the nodes based on the order in which vehicles travel on the target path to obtain numbered nodes, extracting time information of adjacent numbered nodes based on related relationships to obtain a predicted time period for the target road, presetting a deviation value, adding the deviation value to the predicted time period for the travel to obtain a travel time period, obtaining date information and weather information, integrating the travel time period, date information and weather information to obtain target features, and predicting congestion conditions of the target road in segments based on the target features and previous surveillance videos through a segmented processing method to obtain predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments.
[0012] Furthermore, the segmentation processing method includes: obtaining previous surveillance videos of the target road, presetting a segmentation standard and a sample number, segmenting and sorting the target road based on the segmentation standard to obtain a number of sub-road segments, mining video segments similar to the target features in the previous surveillance videos based on the sample number and the sub-road segments to obtain a target video set, the target video set including a number of target videos, judging through visual recognition whether there is congestion in the target video to obtain a congestion result, the congestion result including the presence of congestion and the absence of congestion, obtaining the probability of congestion based on the number of existing congestion and the number of samples to obtain a predicted congestion probability, extracting the target video with congestion, judging through visual recognition and recording the time interval from the presence of congestion to the absence of congestion on the target road to obtain the target time, and averaging all target times to obtain the predicted congestion time.
[0013] Furthermore, the monitoring and analysis method includes: presetting a reaction time, a driving speed and an image acquisition frequency, obtaining a reaction mileage based on the reaction time and the driving speed, adding a reaction mileage to the real-time position of the vehicle to obtain a monitoring point, obtaining a real-time monitoring video of the monitoring point to obtain a point video, obtaining an image of the point video in real time based on the image acquisition frequency to obtain a monitoring image, analyzing whether there is a congestion behavior in the image based on visual recognition to obtain a congestion result, when the congestion result is that there is a congestion behavior, presetting a range value, adding the range value to the monitoring point to obtain a monitoring range, obtaining a real-time monitoring video of the monitoring range to obtain a range video, mining the section of the target road with congestion behavior through visual recognition based on the range video to obtain the actual congested road section, mining the source of the congestion behavior in the target road through visual recognition based on the range video to obtain the cause of the congestion, recording the cause of the congestion and extracting the road information of the target road to obtain congestion characteristics, and retrieving the congestion duration of similar situations online based on the congestion characteristics to obtain an estimated congestion duration.
[0014] Furthermore, the path optimization method includes: when processing the predicted congested road segment, extracting the starting point and the end point of the predicted congested road segment to obtain the predicted optimized starting point and the predicted optimized end point, performing path planning based on the predicted optimized starting point and the predicted optimized end point to obtain the predicted replacement road segment, selecting the optimal section in the predicted replacement road segment based on the target priority to obtain the predicted optimized road segment, and when processing the actual congested road segment, extracting the starting point and the end point of the actual congested road segment to obtain the actual optimized starting point and the actual optimized end point, performing path planning based on the actual optimized starting point and the actual optimized end point to obtain the actual replacement road segment, and selecting the optimal section in the actual replacement road segment based on the target priority to obtain the actual optimized road segment.
[0015] Furthermore, basic feedback information is obtained, predicted congestion information of the target road before optimization is recorded to obtain predicted reference information, actual congestion information of the target road before optimization is recorded to obtain actual reference information, predicted congestion information of the target road after optimization is obtained to obtain current predicted reference information, actual congestion information of the target road after optimization is obtained to obtain current actual reference information, and it is determined whether there is a predicted optimized road segment in the current predicted reference information; when there is an optimized road segment, the predicted congestion time and predicted congestion probability of the predicted congested road segment are compared with those of the predicted optimized road segment to obtain predicted comparison information, and the comparison information is fed back to the user based on the predicted basic feedback information, and it is decided whether to replace the predicted congested road segment with the predicted optimized road segment according to the user's feedback, and it is determined whether there is an actual optimized road segment in the current actual reference information; when there is an actual optimized road segment, the estimated congestion time of the actual congested road segment is compared with that of the actual optimized road segment to obtain actual comparison information, and the actual comparison information is fed back to the user based on the basic feedback information, and it is decided whether to replace the actual congested road segment with the actual optimized road segment according to the user's feedback.
[0016] Furthermore, the path determination method includes: obtaining user preferences based on user needs, prioritizing auxiliary information based on user preferences to obtain target priorities, and when the target priority is that cost information is higher than time information, selecting the initial path with the lowest cost among several initial paths to obtain the target path; when the target priority is that time information is higher than cost information, selecting the initial path with the shortest cost and time among several initial paths to obtain the target path.
[0017] A dynamic traffic prediction system for path planning of an autonomous driving vehicle uses the above-mentioned dynamic traffic prediction method for path planning of an autonomous driving vehicle.
[0018] Compared with the prior art, the present invention has the following beneficial effects:
[0019] The dynamic traffic prediction method and system for path planning of an autonomous driving vehicle can, through a set traffic prediction method, combine historical data to determine whether a road section on a target path is congested, and extract the road section with congestion and related information to obtain predicted congestion information; through a set monitoring and analysis method, the target road is monitored in real time to determine whether a congestion exists, and the road section with congestion and related information are generated to obtain actual congestion information; through a monitoring and analysis method in conjunction with a traffic prediction method, real-time and historical data can be combined to determine whether a road section on a target path is congested, which is beneficial to reducing the interference of the delay of congestion information on user travel and improving the user's riding experience.
[0020] At the same time, in the path optimization stage, the result information and subsequent result information generated by the traffic analysis method are used to optimize and update the target path in real time, so that the vehicle driving path can be adjusted in real time according to the congestion situation, which is conducive to improving the commuting efficiency of autonomous driving. Through the set path optimization method, new road sections can be found to replace congested or uncongested paths based on the results generated by the monitoring and analysis method and the traffic prediction method, so as to improve the efficiency of path planning, and the replacement selection of road sections is combined with user needs, which is conducive to making the travel path more in line with the actual needs of users.
[0021] At the same time, by collecting subsequent result information generated after optimization and judging whether the optimized road section will still cause congestion, the road section can be further optimized. Specifically, the comparison information can be fed back to the user, who can decide the selection, or the user can customize the priority and make a direct selection. By feeding back to the user and letting the user choose, the user's travel experience can be improved, and the real-time replacement and optimization of the path can better meet the needs of the user. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a schematic diagram of the structure of the path planning stage of the present invention;
[0023] Figure 2 A schematic diagram of the path traffic situation analysis and optimization structure of the present invention;
[0024] Figure 3 It is a schematic diagram of the structure of the traffic prediction method of the present invention;
[0025] Figure 4 It is a schematic diagram of the structure of the monitoring and analysis method of the present invention. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0027] Dynamic traffic prediction is an important research direction in the field of traffic management and autonomous driving, which aims to predict traffic conditions in the future by analyzing real-time and historical data. This prediction can help improve traffic flow, increase safety, and optimize travel experience.
[0028] like Figure 1-Figure 4 As shown, the present invention provides a technical solution: a dynamic traffic prediction method and system for path planning of an autonomous driving vehicle, the method comprising:
[0029] Data acquisition: obtain the real-time location of the vehicle, target location, user needs and real-time time;
[0030] Path planning: Based on the real-time location and the target location, several initial paths and auxiliary information corresponding to the initial paths are obtained through path planning. The auxiliary information includes cost information and time information. Based on user needs and the auxiliary information, the path determination method is used to select the optimal initial path from several initial paths to obtain the target path and target priority;
[0031] Path traffic situation analysis and optimization: Based on the target path, the traffic analysis method is used to analyze in real time whether the road sections in the target path will be congested, and the congested road sections are optimized to obtain result information. The result information includes the predicted congested road sections and their predicted optimized road sections and the actual congested road sections and the actual optimized road sections. Based on the result information, the corresponding road sections in the target path are replaced to obtain the optimized target path. Based on the optimized target path, the traffic analysis method is used to obtain subsequent result information. Based on the subsequent result information, the optimized target path is further optimized, and the vehicle travels based on the continuously optimized target path;
[0032] It should be noted that the real-time position of the vehicle in the data acquisition stage can be obtained according to the vehicle positioning, the target position is the terminal position to be reached, which can be set according to the use requirements, and the user requirements are the user's use requirements, which can be specifically related instructions of passengers or staff. In the path planning stage, the path planning is carried out according to the real-time position and the target position to obtain several initial paths. In this process, the real-time position and the target position can be directly obtained by inputting the real-time position and the target position into the navigation system on the market. The cost information is the cost of driving to the target position through the initial path, which is usually the sum of the fuel cost of the mileage and the road toll. The time information is the time consumed by driving to the target position through the initial path. Through the path determination method, the initial path that meets the user's requirements is automatically selected according to the user's requirements to obtain the target path. In the path traffic situation analysis and optimization stage, the set traffic analysis method is used to perform congestion analysis on the road sections passed through the target path and optimize the congested road sections. The optimization results generated in real time by the traffic analysis method, that is, the result information and the subsequent result information, are used to optimize and update the target path in real time through the optimization results, so that the vehicle driving path can be adjusted in real time according to the congestion situation, which is conducive to improving the commuting efficiency of autonomous driving.
[0033] The traffic analysis method includes: extracting road information in a target path before or after optimization to obtain a number of target roads, predicting congestion conditions on the target roads by a traffic prediction method to obtain prediction results, the prediction results including predicted congestion information and predicted non-congestion information, the predicted congestion information including predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments, monitoring the target roads in real time by a monitoring and analysis method to obtain analysis results, the analysis results including actual congestion information and actual non-congestion information, the actual congestion information including actual congested road segments and their estimated congestion durations, generating replacement paths by a path optimization method based on the predicted congestion information or the actual congestion information to obtain predicted optimized road segments or actual optimized road segments, combining the predicted congested road segments generated by the target path before optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain result information, combining the predicted congested road segments generated by the target path after optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain subsequent result information.
[0034] It should be noted that the road information of the target path is the different roads existing in the target path, which are usually the names of the roads, that is, the target path is divided according to the different road names, and the starting point of each road on the target path is a node. The instant time is the real-time time to reach each node. The instant time can be obtained by selecting the node as the end point and the real-time position as the starting point, and combining the end point and the starting point through the navigation software. Through the set traffic prediction method, it can be combined with historical data to determine whether there is congestion on the section of the target path, and extract the section with congestion and related information to obtain predicted congestion information for subsequent use. Through the set monitoring and analysis method, the target road is monitored in real time to determine whether there is congestion, and the section with congestion and related information are generated to obtain actual congestion information. Through the monitoring and analysis method combined with the traffic prediction method, it can be combined with real-time and historical data to determine whether there is congestion on the section of the target path, and at the same time reduce the interference of the delay of congestion information on user travel, which is conducive to improving the user's riding experience. Through the set path optimization method, new sections can be found to replace congested or congested paths according to the results generated by the monitoring and analysis method and the traffic prediction method, so as to improve the efficiency of path planning.
[0035] like Figure 3 As shown, the traffic prediction method includes: extracting the starting point of the target road to obtain a node, obtaining the time of arriving at the node based on the real-time time to obtain the instant time, numbering and sorting the nodes based on the order in which vehicles travel on the target path to obtain numbered nodes, extracting the time information of adjacent numbered nodes based on the correlation to obtain the predicted travel time period of the target road, presetting a deviation value, adding the deviation value to the predicted travel time period to obtain the travel time period, obtaining date information and weather information, integrating the travel time period, date information and weather information to obtain target features, and predicting the congestion situation of the target road by segmentation based on the target features and previous monitoring videos through a segmentation processing method to obtain predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments.
[0036] It should be noted that numbering the nodes is conducive to the subsequent acquisition and passage time period, and it is also conducive to analyzing the sequence of nodes. The specific numbers can be: node 1, node 2, node 3, ..., node n, node n+1. By combining the instant time of node 1 and node 2, a time range can be obtained. This time range is the pre-passage time period. Due to the influence of vehicle speed and traffic lights, there are differences in the actual pre-passage time period. By customizing the deviation value and adding it to the pre-passage time period, the passage time period can be obtained, specifically the pre-passage time period of the first target road, making the data richer and more accurate. Date information and weather information can be directly retrieved on the information retrieval platform. The weather information must be consistent with the area where the vehicle's real-time location is located. By setting the passage time period, the processing volume of previous monitoring videos can be shortened, which is conducive to reducing system occupancy and improving operating speed.
[0037] The segmentation processing method includes: obtaining previous monitoring videos of the target road, presetting a segmentation standard and a sample quantity, segmenting and sorting the target road based on the segmentation standard to obtain a number of sub-road segments, mining video segments similar to target features in previous monitoring videos based on the sample quantity and the sub-road segments to obtain a target video set, the target video set including a number of target videos, judging through visual recognition whether there is congestion in the target video to obtain a congestion result, the congestion result including the presence of congestion and the absence of congestion, obtaining the probability of congestion based on the number of existing congestion and the number of samples to obtain a predicted congestion probability, extracting the target video with congestion, judging through visual recognition and recording the time interval from the presence of congestion to the absence of congestion on the target road to obtain the target time, and averaging all target times to obtain the predicted congestion time.
[0038] It should be noted that the classification standard and the number of samples are customized by the user. The classification standard can be a specific road length or a monitoring area of a camera on the road, etc. The number of samples is the number of videos collected that are similar to the target features. By further dividing the target road, the amount of data processed by the system at one time can be reduced, and the system response speed can be improved. The specific process can be to process the sub-road segments in sequence or to process multiple sub-road segments at the same time. Based on the number of samples and sub-road segments, the target video set is obtained by mining the video segments similar to the target features in the previous monitoring videos. The target video set can be obtained by obtaining the sub-videos in the previous monitoring videos at the same location, the same time period, the same date, and the same weather. The date has particularity. The specific analysis is to determine whether the date is a weekday or a holiday. When the current date is a holiday, when selecting a video, the date of the selected video should also be on the holiday, which is conducive to ensuring the accuracy of the sample.
[0039] like Figure 4As shown, the monitoring and analysis method includes: presetting a reaction time, a driving speed and an image acquisition frequency, obtaining a reaction mileage based on the reaction time and the driving speed, adding a reaction mileage to the real-time position of the vehicle to obtain a monitoring point, obtaining a real-time monitoring video of the monitoring point to obtain a point video, obtaining an image of the point video in real time based on the image acquisition frequency to obtain a monitoring image, analyzing whether there is a congestion behavior in the image based on visual recognition to obtain a congestion result, when the congestion result is that there is a congestion behavior, presetting a range value, adding the range value to the monitoring point to obtain a monitoring range, obtaining a real-time monitoring video of the monitoring range to obtain a range video, mining a section of the target road with congestion behavior through visual recognition based on the range video to obtain an actual congested road section, mining a source of congestion behavior in the target road through visual recognition based on the range video to obtain a congestion cause, recording the congestion cause and extracting road information of the target road to obtain a congestion feature, and retrieving the congestion duration of similar situations online based on the congestion feature to obtain an estimated congestion duration.
[0040] It should be noted that the previous monitoring videos and real-time monitoring videos can be obtained through the camera locations on the road. The reaction time is the time it takes for the autonomous driving vehicle to make a decision to change the road section in real time. The reaction time is proportional to safety and is determined based on actual conditions. The driving speed is the speed at which the vehicle is driving on the road. The image acquisition frequency is the speed frequency of acquiring images in the video information, which can be specifically one or more images acquired at intervals of n seconds. The reaction mileage is the mileage required to change the road section. The driving speed is generally km / h. The specific measurement can be obtained by converting the reaction time into hours and multiplying it by the driving speed. By adding the reaction mileage to the real-time position of the vehicle to obtain the monitoring point, it can be ensured that the vehicle changes lanes safely. The image is analyzed through visual recognition to determine whether there is congestion. Specifically, features can be extracted from videos or images through deep learning models such as convolutional neural networks. The model is trained to identify congestion, and the target detection algorithm is used to identify the status of different objects in the image in real time, thereby inferring whether there is congestion. The video or image is divided into multiple areas, and the flow and density changes in each area are monitored to identify traffic congestion and other existing visual recognition technologies to judge and analyze whether there is congestion in the image or video. When there is congestion, a preset range value is used to expand the monitoring range in combination with other cameras adjacent to the detection point. The actual congested road section can be obtained more comprehensively based on the range video in the monitoring range in combination with the visual recognition technology. Based on the range video, the source of the congestion behavior in the target road is mined through visual recognition to obtain the cause of the congestion. The specific operation can be to mine the behavioral information between the source of the congestion and the non-congestion to obtain the cause of the congestion. The specific cause of the congestion may be a traffic accident or road repair, etc. The congestion feature is the specific feature of the target road, such as a one-way one-lane road, a one-way two-lane road, or a two-way three-lane road, etc., based on Internet search, for example: How long will the traffic jam last if there is a rear-end collision on a one-way one-lane road in an urban road? The estimated congestion duration can be obtained based on the searched results.
[0041] like Figure 3 and Figure 4 As shown, the path optimization method includes: when processing the predicted congested road segment, extracting the starting point and the end point of the predicted congested road segment to obtain the predicted optimized starting point and the predicted optimized end point, performing path planning based on the predicted optimized starting point and the predicted optimized end point to obtain the predicted replacement road segment, selecting the optimal road segment in the predicted replacement road segment based on the target priority to obtain the predicted optimized road segment, and when processing the actual congested road segment, extracting the starting point and the end point of the actual congested road segment to obtain the actual optimized starting point and the actual optimized end point, performing path planning based on the actual optimized starting point and the actual optimized end point to obtain the actual replacement road segment, and selecting the optimal road segment in the actual replacement road segment based on the target priority to obtain the actual optimized road segment.
[0042] It should be noted that both the process of obtaining predicted replacement road segments through path planning based on the predicted optimization starting point and the predicted optimization end point and obtaining actual replacement road segments through path planning based on the actual optimization starting point and the actual optimization end point can be obtained using navigation software. After obtaining, the predicted optimization road segments and actual optimization road segments can be obtained by selecting the corresponding replacement road segments according to the user's target priority.
[0043] Obtain basic feedback information, record the predicted congestion information of the target road before optimization to obtain predicted reference information, record the actual congestion information of the target road before optimization to obtain actual reference information, obtain the predicted congestion information of the target road after optimization to obtain current predicted reference information, obtain the actual congestion information of the target road after optimization to obtain current actual reference information, determine whether there is a predicted optimized road segment in the current predicted reference information, when there is an optimized road segment, compare the predicted congestion time and predicted congestion probability of the predicted congested road segment with the predicted optimized road segment to obtain predicted comparison information, feed back the comparison information to the user based on the predicted basic feedback information, decide whether to replace the predicted congested road segment with the predicted optimized road segment according to the user's feedback, determine whether there is an actual optimized road segment in the current actual reference information, when there is an actual optimized road segment, compare the estimated congestion time of the actual congested road segment with the actual optimized road segment to obtain actual comparison information, feed back the actual comparison information to the user based on the basic feedback information, and decide whether to replace the actual congested road segment with the actual optimized road segment according to the user's feedback.
[0044] It should be noted that by collecting subsequent result information generated after optimization and determining whether the optimized road segment will still cause congestion, the road segment can be further optimized. Specifically, the comparison information can be fed back to the user, and the user can decide to select, or the priority can be customized for direct selection. By feeding back to the user and allowing the user to select, the user's travel experience can be improved, and the real-time replacement and optimization of the path can be more in line with the user's needs. When the user's feedback decides to replace the actual congested road segment with the actual optimized road segment, the optimized target path will not be changed. When the user's feedback decides not to replace the actual congested road segment with the actual optimized road segment and not to replace the predicted congested road segment with the predicted optimized road segment, the optimized target path will be further changed, and the actual optimized road segment or the predicted optimized road segment in the optimized target path will be replaced back to the original actual congested road segment and the predicted congested road segment.
[0045] The path determination method includes: obtaining user preferences based on user needs, prioritizing auxiliary information based on user preferences to obtain target priorities, when the target priority is that cost information is higher than time information, selecting the initial path with the lowest cost among several initial paths to obtain the target path, and when the target priority is that time information is higher than cost information, selecting the initial path with the shortest cost and time among several initial paths to obtain the target path.
[0046] It should be noted that the user's preference is determined based on the user's travel needs, specifically whether the user prefers a path with short travel time or a path with low cost. The target priority is determined based on the user's preference, and the target priority can be that cost information is higher than time information or time information is higher than cost information.
[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is limited by the attached embodiments and their equivalents.
Claims
1. A dynamic traffic prediction method for path planning of an autonomous driving vehicle, characterized by: The method comprises: Data acquisition: obtain the real-time location of the vehicle, target location, user needs and real-time time; Path planning: Based on the real-time location and the target location, several initial paths and auxiliary information corresponding to the initial paths are obtained through path planning. The auxiliary information includes cost information and time information. Based on user needs and the auxiliary information, the path determination method is used to select the optimal initial path from several initial paths to obtain the target path and target priority; Path traffic situation analysis and optimization: Based on the target path, the traffic analysis method is used to analyze in real time whether the road sections in the target path will be congested, and the congested road sections are optimized to obtain result information. The result information includes the predicted congested road sections and their predicted optimized road sections and the actual congested road sections and the actual optimized road sections. Based on the result information, the corresponding road sections in the target path are replaced to obtain the optimized target path. Based on the optimized target path, the traffic analysis method is used to obtain subsequent result information. Based on the subsequent result information, the optimized target path is further optimized, and the vehicle travels based on the continuously optimized target path; The traffic analysis method includes: extracting road information in a target path before or after optimization to obtain a plurality of target roads, predicting congestion conditions of the target roads by a traffic prediction method to obtain prediction results, the prediction results including predicted congestion information and predicted non-congestion information, the predicted congestion information including predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments, monitoring the target roads in real time by a monitoring and analysis method to obtain analysis results, the analysis results including actual congestion information and actual non-congestion information, the actual congestion information including actual congested road segments and their estimated congestion durations, generating replacement paths by a path optimization method based on the predicted congestion information or the actual congestion information to obtain predicted optimized road segments or actual optimized road segments, combining the predicted congested road segments generated by the target path before optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain result information, combining the predicted congested road segments generated by the target path after optimization, the predicted optimized road segments, the actual congested road segments, and the actual optimized road segments to obtain subsequent result information; The traffic prediction method includes: extracting the starting point of the target road to obtain a node, obtaining the time to reach the node based on the real-time time to obtain the instant time, numbering and sorting the nodes based on the order in which vehicles travel on the target path to obtain numbered nodes, extracting the time information of adjacent numbered nodes based on the relationship to obtain the predicted travel time period of the target road, presetting a deviation value, adding the deviation value to the predicted travel time period to obtain the travel time period, obtaining date information and weather information, integrating the travel time period, date information and weather information to obtain target features, and predicting the congestion situation of the target road by segmentation based on the target features and previous monitoring videos through a segmentation processing method to obtain predicted congested road segments and predicted congestion times and predicted congestion probabilities corresponding to the predicted congested road segments; The monitoring and analysis method comprises: presetting a reaction time, a driving speed and an image acquisition frequency, obtaining a reaction mileage based on the reaction time and the driving speed, adding a reaction mileage to the real-time position of the vehicle to obtain a monitoring point, obtaining a real-time monitoring video of the monitoring point to obtain a point video, obtaining an image of the point video in real time based on the image acquisition frequency to obtain a monitoring image, analyzing whether there is a congestion behavior in the image based on visual recognition to obtain a congestion result, when the congestion result is that there is a congestion behavior, presetting a range value, adding the range value to the monitoring point to obtain a monitoring range, obtaining a real-time monitoring video of the monitoring range to obtain a range video, mining a road section with congestion behavior in a target road through visual recognition based on the range video to obtain an actual congested road section, mining a source of congestion behavior in a target road through visual recognition based on the range video to obtain a congestion cause, recording the congestion cause and extracting road information of the target road to obtain a congestion feature, and retrieving the congestion duration of similar situations online based on the congestion feature to obtain an estimated congestion duration; The path optimization method includes: when processing the predicted congested road segment, extracting the starting point and the end point of the predicted congested road segment to obtain the predicted optimized starting point and the predicted optimized end point, performing path planning based on the predicted optimized starting point and the predicted optimized end point to obtain the predicted replacement road segment, selecting the optimal road segment in the predicted replacement road segment based on the target priority to obtain the predicted optimized road segment, and when processing the actual congested road segment, extracting the starting point and the end point of the actual congested road segment to obtain the actual optimized starting point and the actual optimized end point, performing path planning based on the actual optimized starting point and the actual optimized end point to obtain the actual replacement road segment, and selecting the optimal road segment in the actual replacement road segment based on the target priority to obtain the actual optimized road segment; The path determination method includes: obtaining user preferences based on user needs, prioritizing auxiliary information based on user preferences to obtain a target priority, when the target priority is that cost information is higher than time information, selecting an initial path with the lowest cost among several initial paths to obtain the target path, and when the target priority is that time information is higher than cost information, selecting an initial path with the shortest cost and time among several initial paths to obtain the target path.
2. The method for dynamic traffic prediction of path planning for an autonomous driving vehicle according to claim 1, characterized in that: The segmentation processing method includes: obtaining previous monitoring videos of a target road, presetting a segmentation standard and a sample quantity, segmenting and sorting the target road based on the segmentation standard to obtain a plurality of sub-road segments, mining video segments similar to target features in previous monitoring videos based on the sample quantity and the sub-road segments to obtain a target video set, the target video set including a plurality of target videos, judging through visual recognition whether there is congestion in the target video to obtain a congestion result, the congestion result including the presence of congestion and the absence of congestion, obtaining the probability of congestion based on the number of existing congestion and the number of samples to obtain a predicted congestion probability, extracting target videos with congestion, judging through visual recognition and recording the time interval from the presence of congestion to the absence of congestion on the target road to obtain a target time, and averaging all target times to obtain a predicted congestion time.
3. The method for dynamic traffic prediction of path planning for an autonomous driving vehicle according to claim 1, characterized in that: Obtain basic feedback information, record the predicted congestion information of the target road before optimization to obtain predicted reference information, record the actual congestion information of the target road before optimization to obtain actual reference information, obtain the predicted congestion information of the target road after optimization to obtain current predicted reference information, obtain the actual congestion information of the target road after optimization to obtain current actual reference information, determine whether there is a predicted optimized road segment in the current predicted reference information, when there is an optimized road segment, compare the predicted congestion time and predicted congestion probability of the predicted congested road segment with the predicted optimized road segment to obtain predicted comparison information, feed back the comparison information to the user based on the predicted basic feedback information, decide whether to replace the predicted congested road segment with the predicted optimized road segment according to the user's feedback, determine whether there is an actual optimized road segment in the current actual reference information, when there is an actual optimized road segment, compare the estimated congestion time of the actual congested road segment with the actual optimized road segment to obtain actual comparison information, feed back the actual comparison information to the user based on the basic feedback information, and decide whether to replace the actual congested road segment with the actual optimized road segment according to the user's feedback.
4. A dynamic traffic prediction system for autonomous vehicle path planning, characterized by: A dynamic traffic prediction method for path planning of an autonomous driving vehicle is used as described in any one of claims 1-3.
