System and method for determining mixed autonomous vehicle traffic situations on basis of multi-source data fusion analysis

Abstract

The present invention relates to a traffic situation determination system based on multi-source data fusion analysis of a road in which autonomous vehicles and non-autonomous vehicles are mixed. The system comprises: a data collection unit for receiving traffic situation data from a roadside device, an intelligent traffic system, and autonomous vehicles; a data operation unit for determining a traffic state for each link by calculating a service level and a congestion index for each road from the data collected by the data collection unit; and a data storage unit for storing the traffic state information for each link and providing the traffic state information determined by the data operation unit to the autonomous vehicles, wherein the data collection unit comprises: a map matching unit for visually outputting driving information of the autonomous vehicles by displaying location information of the autonomous vehicles on a map; and a basic information generation unit for performing traffic management indexing on each of the data provided from the roadside device, the intelligent traffic system, and the autonomous vehicles using an AI model algorithm.

Description

System and method for determining mixed traffic situations involving autonomous vehicles based on multi-collection data fusion analysis

[0001] The present invention relates to a system and method for determining traffic conditions involving mixed autonomous vehicles based on multi-collected data fusion analysis. In particular, it relates to a system for determining traffic conditions by fusion-combining data collected in real-time from multiple collection devices, such as driving information of autonomous vehicles, Edge RSU collection information, and Vehicle Detection System (VDS) information of an Intelligent Transportation System (ITS), in a traffic environment where autonomous vehicles and non-autonomous vehicles are mixed. The system extracts data necessary to determine traffic conditions from data that differs in form and collection purpose by each collection device to generate basic analysis information, and thereby supports the operation of autonomous vehicles.

[0002] Autonomous driving technology consists of perception technology that receives information and signals necessary for driving, judgment technology that processes them, control technology such as steering, braking, and acceleration of the vehicle, and networks. This autonomous driving technology is a technology that performs driving on its own without driver intervention, and according to the J3016 classification criteria of the Society of Automotive Engineers (SAE), it is classified into six levels, ranging from Level 0, which has no autonomous driving function at all, to Level 5, which enables fully autonomous driving without a driver in all situations.

[0003] Vehicles equipped with such autonomous driving technology have reached the stage of paid commercialization abroad, such as providing Level 4 taxi services. In Korea, a temporary operation permit system has been in operation since 2016 to commercialize autonomous vehicles, and support for the designation of autonomous vehicle test zones is being provided to offer opportunities to experience autonomous vehicles.

[0004] Even after autonomous vehicles are commercialized in the future, it is predicted that for a considerable period, traffic conditions will continue to be mixed, with non-autonomous vehicles without autonomous driving capabilities, autonomous vehicles where the driver drives the vehicle and the vehicle system provides limited support to the driver (Level 1, Level 2), and autonomous vehicles where the system drives (Level 3 to Level 5).

[0005] Therefore, to ensure traffic safety and efficient traffic flow in situations where conventional and autonomous vehicles coexist on the road, it is necessary to monitor the operating status of individual autonomous vehicles and assess the risk level of road conditions to proactively manage factors causing traffic accidents and congestion.

[0006] In particular, in the case of autonomous vehicles, abnormal situations such as sudden changes in driving speed or sudden braking may occur if they deviate from the Operational Design Domain (ODD), which defines the basic range for the autonomous driving system to operate. In such cases, this affects not only the autonomous vehicle but also the driving environment of conventional vehicles, potentially causing adverse effects on overall traffic flow. Accordingly, to minimize traffic problems and resulting adverse effects that may arise in situations where conventional and autonomous vehicles coexist, a driving support system for autonomous vehicles based on traffic operation management is required.

[0007] An example of a technology for determining traffic conditions in a mixed situation of general vehicles and autonomous vehicles is disclosed in Patent Documents 1 to 3, etc.

[0008] For example, Patent Document 1 (Korean Published Patent Application No. 10-2024-0078714 (published June 4, 2024)) relates to a road risk derivation system using the mixing ratio of autonomous vehicles and general vehicles. By deriving the road risk based on behavioral information of general vehicles collected through roadside infrastructure such as radar, and FCWS information and BCA information of autonomous vehicles, the reliability of the road risk can be fundamentally improved, and thus, a technology is disclosed that can predict dangerous situations in advance and prevent traffic accidents.

[0009] Patent Document 2 (Korean Registered Patent Publication No. 10-2588414 (Registered on Oct. 06, 2023)) discloses a road traffic network system for on-site control of unexpected situations and disasters on roads where autonomous vehicles are mixed, by including an integrated control server that receives information on the detection of unexpected situations and disasters from multiple facilities around the road and roads connected to a communication network, collecting information from road facilities, field terminals, and autonomous vehicles, and the integrated control server assigns a set grade according to the elements of the disaster and unexpected grade included in the received information, and transmits a response manual for each set grade to multiple registered terminals.

[0010] The following patent document 3 (U.S. Registered Patent Publication No. 11922805 (registered on March 5, 2024)) relates to a system and method for intelligent traffic control, and discloses a technology configured to generate a 2D or 3D map based on data of an autonomous vehicle and traffic data in a traffic environment where autonomous vehicles and non-autonomous vehicles are mixed, make navigation decisions based on the map, and the traffic control commands include stop, move, yield, and cross commands, and output unmanned / manned mode commands for the autonomous vehicle.

[0011] The technology disclosed in the aforementioned patent documents did not disclose a technology that fuses and analyzes speed and traffic volume information collected from autonomous vehicles, Edge RSU equipment, ITS centers, autonomous vehicles, and smart intersection equipment, and supports the safe driving of autonomous vehicles by determining traffic situations where autonomous and non-autonomous vehicles are mixed.

[0012] The objective of the present invention is to solve the problems described above by providing a system and method for determining mixed traffic situations involving autonomous vehicles based on multi-collected data fusion analysis, which configures data collected from Edge RSU equipment, ITS centers, autonomous vehicles, and smart intersection equipment to correspond to traffic conditions and matches this with spatiotemporal data to establish an AI-based big data automatic classification system.

[0013] Another objective of the present invention is to provide a system and method for determining mixed traffic conditions involving autonomous vehicles based on multi-collected data fusion analysis to support the operation of autonomous vehicles in a road environment where autonomous and non-autonomous vehicles are mixed, thereby preventing traffic problems that may adversely affect the overall traffic situation, such as abnormal situations like sudden changes in speed and sudden braking, when an autonomous vehicle deviates from the Operational Design Domain (ODD).

[0014] Another objective of the present invention is to provide a system and method for determining mixed traffic situations involving autonomous vehicles based on multi-collected data fusion analysis, which can improve the accuracy of data indexing by classifying data using a traffic management indexing method utilizing an AI model.

[0015] Another objective of the present invention is to provide a system and method for determining traffic conditions involving a mix of autonomous and non-autonomous vehicles based on multi-collected data fusion analysis, which provides real-time road congestion information by calculating the Level of Service (LOS) of a road by calculating a congestion index of a road where autonomous and non-autonomous vehicles are mixed using a constructed classification system.

[0016] Another objective of the present invention is to provide a system and method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis, which supports the safe driving of autonomous vehicles in road environments where autonomous and non-autonomous vehicles are mixed by generating a real-time road congestion index and providing it to the autonomous vehicle.

[0017] To achieve the above objective, the multi-collection data fusion analysis-based traffic situation determination system for autonomous vehicles mixed with autonomous vehicles according to the present invention comprises: a data collection unit that receives traffic situation data from a roadside device, an intelligent transportation system, and an autonomous vehicle; a data computation unit that calculates a service level and a congestion index for each road from the data collected by the data collection unit and determines the traffic status for each link; and a data storage unit that stores the traffic status information for each link and provides the traffic status information determined by the data computation unit to the autonomous vehicle, wherein the data collection unit includes: a map matching unit that displays the location information of the autonomous vehicle on a map and visually outputs the operation information of the autonomous vehicle; and a basic information generation unit that performs traffic management indexing using an AI model algorithm for each of the data received from the roadside device, the intelligent transportation system, and the autonomous vehicle.

[0018] In addition, according to the multi-collection data fusion analysis-based autonomous vehicle mixed traffic situation determination system of the present invention, the basic information generation unit is characterized by performing indexing of data received from the roadside device, the intelligent traffic system, and the autonomous vehicle according to the driving state of the autonomous vehicle, traffic flow situation, the time of operation of the autonomous vehicle, weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the autonomous vehicle.

[0019] In addition, according to the autonomous vehicle mixed traffic situation determination system based on multi-collected data fusion analysis according to the present invention, the driving state of the autonomous vehicle is classified into normal driving, slow driving, stopping, accident, and vehicle breakdown; the traffic flow situation is classified using the driving speed and location of the autonomous vehicle, and data collected from the roadside device and the intelligent traffic system; the operating time period of the autonomous vehicle is classified into peak hours (morning, afternoon) and off-peak hours (dawn, morning, afternoon, night); the weather environment is classified by reflecting weather sensor data from the roadside device and weather information provided by the Korea Meteorological Administration; the road scale is classified into multi-lane roads with 6 or more lanes, 4-lane roads, 2-lane roads, and mixed pedestrian and vehicle roads based on the number of lanes and road width according to the location information of the autonomous vehicle; the road type is classified into links and nodes based on the location of the autonomous vehicle displayed in the map matching unit, wherein the links are classified into main roads, connecting roads, junctions, interchanges, overpasses, underpasses, tunnels, and bridges; and the nodes Data is classified into signalized intersections, non-signaled intersections, and roundabouts; the traffic conditions are classified to include traffic volume by direction, traffic volume by lane, traffic signal display, and surrounding road environment information regarding the road type; and the operation distribution of the autonomous vehicles is characterized by including data classification based on the operation distribution rate of the autonomous vehicles among the total traffic volume of the links and nodes to reflect traffic situations where the autonomous vehicles and non-autonomous vehicles are mixed.

[0020] In addition, according to the autonomous vehicle mixed traffic situation determination system based on multi-collected data fusion analysis according to the present invention, the basic information generation unit is characterized by including an MLOps (Machine Learning Operations) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost is applied.

[0021] In addition, according to the autonomous vehicle mixed traffic situation determination system based on multi-collected data fusion analysis according to the present invention, the data processing unit is characterized by calculating the level of service (LOS) of the road based on vehicle density, lane occupancy rate, and traffic speed in the link, based on the state information of the autonomous vehicle, road function, and attribute information of the link.

[0022] In addition, according to the autonomous vehicle mixed traffic situation determination system based on multi-collection data fusion analysis according to the present invention, the data processing unit [calculates] the congestion index (C) according to the [mathematical formula] below. n The current traffic situation is determined using ), and is characterized by determining that if the congestion index value is 0.85 or higher, it is determined to be congested, and if the congestion index value is less than 0.85, it is determined to be smooth or slow.

[0023] [Mathematical Formula]

[0024]

[0025] ( )

[0026] In addition, according to the autonomous vehicle mixed traffic situation determination system based on multi-collection data fusion analysis according to the present invention, the roadside device is an Edge RSU, and the intelligent traffic system is an ITS.

[0027] A method for determining traffic conditions involving a mix of autonomous vehicles based on multi-collected data fusion analysis according to the present invention for achieving the above objective comprises: i) receiving traffic condition data from a roadside device, an intelligent transportation system, and an autonomous vehicle from a data collection unit; ii) displaying location information of the autonomous vehicle on a map to visually output operation information of the autonomous vehicle; iii) performing traffic management indexing using an AI model algorithm for each of the data received from the roadside device, the intelligent transportation system, and the autonomous vehicle; iv) calculating a service level and a congestion index for each road based on the results of the traffic management indexing and determining the traffic condition for each link in a data processing unit; and v) storing the traffic condition information for each link and providing the traffic condition information determined by the data processing unit to the autonomous vehicle from a data storage unit.

[0028] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collection data fusion analysis according to the present invention, step iii) is characterized by performing indexing of data received from the roadside device, the intelligent traffic system, and the autonomous vehicle according to the driving state of the autonomous vehicle, traffic flow conditions, the time of operation of the autonomous vehicle, weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the autonomous vehicle.

[0029] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention, the driving state of the autonomous vehicle is classified into normal driving, slow driving, stopping, accident, and vehicle breakdown; the traffic flow situation is classified using the driving speed and location of the autonomous vehicle, and data collected from the roadside device and the intelligent transportation system; the operating time period of the autonomous vehicle is classified into peak hours (morning, afternoon) and off-peak hours (dawn, morning, afternoon, night); the weather environment is classified by reflecting weather sensor data from the roadside device and weather information provided by the Korea Meteorological Administration; the road scale is classified into multi-lane roads with 6 or more lanes, 4-lane roads, 2-lane roads, and mixed pedestrian and vehicle roads based on the number of lanes and road width according to the location information of the autonomous vehicle; the road type is classified into links and nodes based on the location of the autonomous vehicle displayed in the map matching unit, wherein the links are classified into main roads, connecting roads, junctions, interchanges, overpasses, underpasses, tunnels, and bridges; and the nodes Data is classified into signalized intersections, non-signaled intersections, and roundabouts; the traffic conditions are classified to include traffic volume by direction, traffic volume by lane, traffic signal display, and surrounding road environment information regarding the road type; and the operation distribution of the autonomous vehicles is characterized by including data classification based on the operation distribution rate of the autonomous vehicles among the total traffic volume of the links and nodes to reflect traffic situations where the autonomous vehicles and non-autonomous vehicles are mixed.

[0030] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention, the AI ​​model is characterized by including an MLOps (Machine Learning Operations) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost is applied.

[0031] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collection data fusion analysis according to the present invention, step iv) is characterized by calculating the level of service (LOS) of the road based on vehicle density, lane occupancy rate, and traffic speed in the link, based on the status information of the autonomous vehicle, road function, and attribute information of the link.

[0032] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collection data fusion analysis according to the present invention, the service level of the road comprises: a step of determining the driving speed, location, time information of the autonomous vehicle, and the type information of the road on which the autonomous vehicle is driving; a step of analyzing link attribute information including traffic volume, travel speed, and number of lanes when the autonomous vehicle is driving on an expressway or a dedicated automobile road, or on a multi-lane road with four or more lanes; and a step of determining the service level according to vehicle density (pcpkmpl) or traffic volume-to-capacity ratio (V / C), and is characterized by determining the service level according to the travel speed when the autonomous vehicle is driving on a road with fewer than four lanes.

[0033] In addition, according to the method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention, step iv) is a congestion index (C according to the [mathematical formula] below. nThe current traffic situation is determined using ), and if the congestion index value is 0.85 or higher, it is determined to be delayed, and if the congestion index value is less than 0.85, it is determined to be smooth or slow.

[0034] [Mathematical Formula]

[0035]

[0036] ( )

[0037] In addition, according to the method for determining mixed traffic situations of autonomous vehicles based on multi-collected data fusion analysis according to the present invention, the CS n , the above CF n and the above CO n Determine whether the value of is negative or positive; if the values ​​are all positive, determine that the congestion situation has worsened, and if the values ​​are all negative, determine that the congestion situation has eased, and CS n , CF n , CO n It is characterized by determining that the stagnation is maintained when the values ​​of are all negative and all positive.

[0038] As described above, according to the system and method for determining mixed traffic conditions involving autonomous vehicles based on multi-collected data fusion analysis of the present invention, the effect of configuring data collected from Edge RSU equipment, ITS centers, autonomous vehicles, and smart intersection equipment to correspond to traffic conditions, and matching this with spatiotemporal data to establish an AI-based big data automatic classification system is obtained.

[0039] Furthermore, according to the system and method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis of the present invention, an effect is obtained in which the operation of an autonomous vehicle can be supported to prevent traffic problems that may adversely affect the overall traffic situation, such as abnormal situations like sudden changes in speed and sudden braking, which occur when an autonomous vehicle deviates from the operation design area in a road environment where autonomous vehicles and non-autonomous vehicles are mixed.

[0040] In addition, according to the system and method for determining mixed traffic situations involving autonomous vehicles based on multi-collected data fusion analysis of the present invention, the effect of improving the accuracy of data indexing is obtained by classifying data using a traffic management index method utilizing an AI model.

[0041] Furthermore, according to the system and method for determining traffic conditions involving mixed autonomous vehicles based on multi-collection data fusion analysis of the present invention, the effect of providing real-time road congestion information is obtained by calculating the level of service (LOS) of a road where autonomous vehicles and non-autonomous vehicles are mixed using a constructed classification system.

[0042] In addition, according to the system and method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis of the present invention, by generating a real-time road congestion index and providing it to the autonomous vehicle, the effect of supporting safe driving of the autonomous vehicle in a road environment where autonomous vehicles and non-autonomous vehicles are mixed is obtained.

[0043] FIGS. 1A and FIGS. 1B are drawings illustrating a system for collecting vehicle information at links and nodes where autonomous vehicles and non-autonomous vehicles are mixed, respectively, according to an embodiment of the present invention.

[0044] FIG. 2 is a block diagram illustrating a system for collecting vehicle information on a road where autonomous vehicles and non-autonomous vehicles are mixed, according to an embodiment of the present invention.

[0045] FIG. 3a is a diagram illustrating the configuration of a data operation unit according to an embodiment of the present invention.

[0046] FIG. 3b is a drawing for explaining the configuration of a data storage unit according to an embodiment of the present invention.

[0047] FIG. 4 is a diagram illustrating an example of a traffic management index based on an autonomous vehicle according to an embodiment of the present invention.

[0048] FIG. 5 is a flowchart of data classification according to traffic conditions in accordance with an embodiment of the present invention.

[0049] FIG. 6 is a flowchart showing an example of a traffic management indexing procedure according to an embodiment of the present invention.

[0050] FIG. 7 is a table showing an example of data collected according to an embodiment of the present invention being classified according to a traffic management indexing process and organized into a database.

[0051] FIG. 8 is a diagram showing the algorithm of an AI model for indexing real-time large-scale data for traffic management according to an embodiment of the present invention.

[0052] FIG. 9 is a data processing flowchart for determining a traffic situation in which autonomous vehicles are mixed, according to an embodiment of the present invention.

[0053] FIG. 10 is a diagram illustrating a detailed algorithm for determining LOS by road type according to an embodiment of the present invention.

[0054] FIG. 11 is a diagram illustrating a detailed algorithm for calculating a mixture index of a data operation unit according to an embodiment of the present invention.

[0055] FIG. 12 is a diagram showing the results of an AI modeling performance evaluation using test data according to an embodiment of the present invention.

[0056] The above and other objects and novel features of the present invention will become more apparent from the description in this specification and the accompanying drawings.

[0057] The size and thickness of each component shown in the description and drawings of the present invention are depicted arbitrarily for convenience of explanation, and therefore the present invention is not necessarily limited to what is illustrated. Additionally, thicknesses have been enlarged in the drawings to clearly represent various layers and regions, and the thickness of some layers and regions has been exaggerated for convenience of explanation.

[0058] In the description of the invention, when a part is described as "comprising" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0059] The terms “part,” “module,” or “part” as used herein perform at least one function or operation and may be implemented as hardware or software consisting of mechanical or electrical / electronic configurations, or as a combination of hardware and software; and a plurality of “parts,” “modules,” or a plurality of “parts” may be integrated into at least one module and implemented by at least one processor, except for the “parts,” “modules,” or “parts” that need to be implemented in specific hardware.

[0060] In addition, as a term used herein, "Node" refers to a location where a change in speed occurs while a vehicle is traveling on a road. Its types may include intersections, road intersections, bridge start / end points, traffic control points, road structure transition points, road operation transition points, traffic entry / exit points, overpass start / end points, road start / end points, underpass start / end points, tunnel start / end points, administrative boundaries, IC / JCs, etc., but the ITS project entity may define them differently as necessary.

[0061] In addition, as used in this institution, "Link" refers to a line connecting nodes that are points where speed changes occur, and is a line created by spacing road centerlines connecting adjacent nodes at regular intervals in each direction, containing information about actual road sections. These types of links can be classified into roads, bridges, overpasses, underpasses, tunnels, etc.

[0062] In an embodiment of the present invention, a "roadside device," e.g., an Edge RSU, can be understood as a system capable of recognizing individual traffic objects located on the road, such as vehicles, people, and motorcycles, in real time, and supporting status-sharing between infrastructure and vehicles. Additionally, the Edge RSU can provide information such as LDM Layer 4 (real-time detection / classification and attributes of objects, location information, and movement information) and Layer 3 (infrastructure status, surrounding weather, road surface conditions, various road condition information such as accidents and stopped vehicles). To collect such information, the entire system may be composed of an AI camera and a LiDAR sensor for real-time dynamic object analysis corresponding to Layer 4, various sensors for collecting environmental information corresponding to Layer 3, an MEC that aggregates information from multiple sensors and performs road condition recognition, and an RSU for sharing real-time information with vehicles.

[0063] In an embodiment of the present invention, an "intelligent transportation system," such as an ITS, is a system that utilizes V2X communication to enable vehicles to exchange information and connects them to infrastructure, thereby detecting hazards and unexpected situations in advance. It is configured to include V2X communication technology in intelligent transportation systems, such as real-time traffic information used in navigation, real-time signal control of roads, high-pass and lane control on highways, and bus information systems at bus stops, and can identify roadside conditions using sensors such as radar and cameras.

[0064] In this case, V2X is a technology in which a vehicle exchanges or shares information with surrounding vehicles, mobile devices, traffic infrastructure, etc., through wired or wireless communication networks, and includes vehicle-to-vehicle (V2V) communication, vehicle-to-infrastructure (V2I) communication, vehicle-to-pedestrian (V2P) communication, and vehicle-to-network (V2N) communication.

[0065] In an embodiment of the present invention, the "smart intersection system" can analyze video information from unidirectional and multidirectional cameras within the intersection in real time through a deep learning-based video analysis algorithm, analyze smart traffic information within the intersection (traffic volume, queues, vehicle type classification), etc., and generate traffic big data.

[0066] In the embodiments of the present invention, "autonomous driving vehicle" is defined as a Level 4 vehicle, in which the system can perform all core control of driving, monitoring of the driving environment, and response in emergencies according to the classification of the National Highway Traffic Safety Administration (NHTSA) under the U.S. Department of Transportation, and a Level 5 vehicle, in which the system always takes charge of driving in all road conditions and environments, and "non-autonomous driving vehicle" may be defined as a Level 0 to Level 3 vehicle as defined by NHTSA. The autonomous driving vehicle may be referred to as an autonomous vehicle or an autonomous car in the embodiments and drawings of the present invention, and the non-autonomous driving vehicle may be referred to as a general vehicle or a general car.

[0067]

[0068] Hereinafter, an embodiment of a system for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention will be described in detail with reference to the drawings.

[0069] FIGS. 1a and 1b are drawings illustrating a system for collecting vehicle information at links and nodes where autonomous vehicles and non-autonomous vehicles are mixed, respectively, according to an embodiment of the present invention, and FIG. 2 is a block diagram illustrating a system for collecting vehicle information on a road where autonomous vehicles and non-autonomous vehicles are mixed, according to an embodiment of the present invention. FIG. 3a is a drawing illustrating the configuration of a data processing unit according to an embodiment of the present invention, and FIG. 3b is a drawing illustrating the configuration of a data storage unit according to an embodiment of the present invention. FIG. 4 is a drawing illustrating an example of a traffic management index based on autonomous vehicles according to an embodiment of the present invention, and FIG. 5 is a flowchart of data classification according to traffic conditions according to an embodiment of the present invention. FIG. 6 is a flowchart showing an example of a traffic management indexing procedure according to an embodiment of the present invention, FIG. 7 is a table showing an example of data collected according to an embodiment of the present invention being classified according to a traffic management indexing process and organized into a database, FIG. 8 is a diagram showing an algorithm of an AI model for real-time large-scale data traffic management indexing according to an embodiment of the present invention.

[0070] Referring to FIGS. 1a to 2, a multi-collection data fusion analysis-based autonomous vehicle mixed traffic situation determination system (100) according to an embodiment of the present invention includes components for supporting the driving safety of an autonomous vehicle (110) on a road where an autonomous vehicle (110) and a non-autonomous vehicle (120) are mixed.

[0071] The above system (100) collects data obtained from a roadside device (210), an intelligent traffic system (220), an autonomous vehicle (110), and a smart intersection system (230) through a wireless communication network (310) or a server (320).

[0072] The above roadside device (210) can collect real-time dynamic information of the road by having a communication function and being configured to include an image sensor and a lidar, and can provide an optimal driving method to an autonomous vehicle (110) to improve environmental performance, such as reducing exhaust emissions, and to ensure the safety of traffic flow or smooth movement conditions by adopting a deep learning-based algorithm. The above roadside device (210) may be, for example, an Edge RSU.

[0073] The intelligent transportation system (220) may be a Vehicle Detection System (VDS) of an Intelligent Transportation System that detects information such as traffic volume, occupancy rate, vehicle speed, length of queues in each lane, and vehicle length in real time, and detects traffic flow and unexpected situations on the road. The vehicle detection system may be configured by adopting at least one device selected from a loop detector, an image detector, and a radar detector.

[0074] The above-mentioned autonomous vehicle (110) may be a vehicle having Level 4 or higher autonomous driving performance equipped with a semiconductor processor (AP) for autonomous driving, a high-precision lidar, a radar, front and rear cameras, an external microphone, a humidity sensor, and a high-precision map. The above-mentioned autonomous vehicle (110) may be equipped with a communication module for performing communication between a vehicle and an object (V2X), between a vehicle and a peripheral device (V2D), and between a vehicle and a pedestrian (V2P) by utilizing driving data.

[0075] Meanwhile, the above wireless communication network (310) may use various communication networks such as Bluetooth, Wi-Fi, IEEE 802.15.4-based technologies (Thread, Zigbee), Z-Wave, low-power wide-area communication networks (NB-IoT, LTE-M), and low-power wide-area networks (LoRaWAN, Sigfox), but may also use other communication networks not defined here, and may use a wired communication network to exchange data as needed. The above server (320) may be composed of a web server, DB server, file server, application server, FTP server, virtual server, or cloud server, but is not limited thereto.

[0076] The above system (100) comprises a data collection unit (400) for collecting traffic condition data obtained from a roadside device (210), an intelligent traffic system (220), an autonomous vehicle (110), and a smart intersection system (230), a data calculation unit (500) for calculating the data collected from the data collection unit (400), and a data storage unit (600) for storing the result calculated from the data calculation unit (500).

[0077] The above data collection unit (400) includes a map matching unit (410) that displays location information of the autonomous vehicle (110) on a map and visually outputs operation information of the autonomous vehicle (110), a basic information generation unit (420), and a basic information storage unit (430).

[0078] The above map matching unit (410) is configured to visually indicate which point the autonomous vehicle (110) is heading toward on which path on the map. The dedicated map of the autonomous vehicle (110), which serves as the standard for map matching, includes precise information such as lanes, crosswalks, and intersections that are characteristic features of the road. Various feature information can be utilized as standard data, and it is expressed as network data composed of nodes indicating the location of intersections or important points, and links indicating roads connecting nodes adjacent to intersections, for example. Standard node links can be constructed by utilizing the nationwide standard node links provided by the ITS National Traffic Information Center. The location information of the above autonomous vehicle (110) can be displayed in latitude and longitude.

[0079] Meanwhile, since it is difficult to analyze traffic conditions using only raw data or metadata obtained from various sensors of the above-mentioned roadside device (210), intelligent traffic system (220), autonomous vehicle (110), and smart intersection system (230), an efficient classification system for real-time large-capacity traffic data must be established to provide faster and more accurate traffic operation optimization information to the autonomous vehicle (110) in traffic situations where autonomous vehicles (110) and non-autonomous vehicles (120) are mixed.

[0080] The above basic information generation unit (420) performs traffic management indexing using an AI model algorithm for each of the data provided from the roadside device (210), intelligent traffic system (220), autonomous vehicle (110), and smart intersection system (230) via the server (320) or wired / wireless communication network (310). The indexing can be performed according to the driving state of the autonomous vehicle (110), traffic conditions, operating time of the autonomous vehicle (110), weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the autonomous vehicle (110) based on the data provided from the roadside device (210), intelligent traffic system (220), autonomous vehicle (110), and smart intersection system (230).

[0081] The above basic information generation unit (420) can perform traffic management indexing using an MLOps (Machine Learning Operations) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM and CatBoost is applied.

[0082] Below, we will explain in more detail the traffic management indexing and the validity of AI models to establish an AI-based big data classification system.

[0083]

[0084] 1. Collection of big data to support autonomous driving

[0085] In order to support the driving of an autonomous vehicle (110) operating in a traffic situation where autonomous vehicles (110) and non-autonomous vehicles (120) are mixed, it is important to collect accurate information based on updates to traffic facility information and real-time information based on changes in traffic conditions.

[0086] The accurate information resulting from the update of the above information includes precise location and attribute information (lane-level, section-level, point-level) for each traffic safety facility, and also includes accurate information on changes in road regulations and controls (such as control sections and times for lane closures due to road construction). The real-time information resulting from changes in the above traffic conditions includes real-time information for use during the driving of autonomous vehicles, real-time information for driving control of autonomous vehicles in the event of unexpected situations such as traffic accidents, and real-time traffic signal information during the driving of autonomous vehicles.

[0087] In the present invention, data is defined as shown in [Table 1] by classifying the traffic information collected based on the driving state of the autonomous vehicle (110) into autonomous vehicle driving information, Edge RSU perception information, road section traffic condition information, and intersection traffic condition information.

[0088]

[0089]

[0090] To support the operation of autonomous vehicles in traffic conditions where autonomous and conventional vehicles coexist in a Living Lab city, we aim to collect big data in real time by linking autonomous vehicle operation data, Edge RSU and vehicle detection system data, as well as traffic information provided by relevant organizations. After automatically classifying and analyzing the collected data through AI modeling, we intend to provide traffic strategy information tailored to the traffic situation to autonomous vehicle service centers via a message broker.

[0091]

[0092] 2. Development of Traffic Management Index

[0093] The ODD (Operable Destination Diameter) of an autonomous vehicle is a concept that defines the basic range for the operation of an autonomous driving system, and it is required to be proposed by the vehicle manufacturer. ISO 34503, which is currently undergoing international standardization, classifies the ODD into three main categories: scene elements, environment elements, and dynamic elements, while NHTSA classifies it into physical infrastructure, operational constraints, objects, connectivity, environmental conditions, and zones. In Korea, according to Article 111-2 (Designation of ODD for Autonomous Driving Systems) of the Rules on Performance and Standards of Automobiles and Automotive Parts, the ODD is specified to include driving environments such as roads and weather, the operating limits of the autonomous driving system, and other conditions related to the safe operation of the vehicle.

[0094] In this invention, a data classification system was constructed as a traffic management index to include traffic management zones not included in the ODD classification system of autonomous vehicles, in order to secure driving data of autonomous vehicles on actual roads and to collect fused traffic information data reflecting traffic situations where autonomous vehicles and general vehicles are mixed.

[0095] To develop an efficient traffic management index in a mixed traffic environment of autonomous and conventional vehicles within a Living Lab city, the driving status of autonomous vehicles is assessed, and road traffic conditions are considered as a priority. Since road traffic conditions affect the operational state of autonomous vehicles during normal driving, 'traffic conditions' were established as the second criterion for the traffic management index. Traffic conditions are generally classified into 'smooth,' 'delayed,' and 'congested' for application in traffic operation management, and congested congestion is further divided into recurring congestion and non-recurring (sudden) congestion.

[0096] - Recurring Congestion: A phenomenon in which traffic demand exceeds road capacity, resulting in periodic congestion; specifically, congestion occurring periodically in specific sections at specific times (peak hours).

[0097] - Non-recurring congestion: Congestion caused by traffic accidents, vehicle breakdowns, road construction, natural disasters, etc., resulting from emergencies or sudden situations.

[0098] The traffic management index indexed traffic flow conditions by fusing autonomous vehicle operating status information with traffic flow data from ITS centers and Edge RSU infrastructure information, and was also indexed based on road and traffic conditions. Furthermore, based on the spatiotemporal information of autonomous vehicles, the index was generated by reflecting the mixing ratio of autonomous vehicles in situations where autonomous and conventional vehicles coexist.

[0099] Information necessary for traffic operation optimization is extracted from the diverse data collected in the Living Lab, classified using the Traffic Management Index method—a big data classification technique—and stored in a database. Accordingly, the Traffic Management Index collects data required for optimizing traffic operations in situations where autonomous and conventional vehicles coexist, and performs index classification methods through an AI system for the automatic indexing of large volumes of data collected in real-time.

[0100] This traffic management indexing proceeds in a total of 9 stages based on the driving status of the autonomous vehicle, as shown in Fig. 4.

[0101] [Step 1] Classification of Autonomous Vehicles Based on Driving Conditions - Classifying situations that occur while an autonomous vehicle is driving on the road into normal driving, slow driving, stopping, accident, and vehicle breakdown.

[0102] [Step 2] Data Classification Based on Traffic Conditions - Classifying data based on road traffic conditions by integrating the driving speed and location of autonomous vehicles with traffic information from Edge RSUs and ITS centers.

[0103] [Step 3] Data Classification by Operating Time - Divide the time periods during which autonomous vehicles operate in the Living Lab into peak and off-peak hours, and - Reflecting the operating characteristics of each time period, classify peak hours into AM and PM, and off-peak hours into dawn, AM, PM, and night.

[0104] [Step 4] Data Classification Based on Weather Conditions - Since the risk of traffic accidents varies depending on weather conditions during the operation of autonomous vehicles, data is classified by reflecting weather sensor information from the Edge RSU and weather information provided by the Korea Meteorological Administration.

[0105] [Step 5] Data Classification by Road Function - Based on the Road Act and considering road operation information, data is classified into expressways, arterial roads, secondary arterial roads, collector roads, and local roads.

[0106] [Step 6] Data Classification by Road Scale - Based on the location information of autonomous vehicles, data is classified into multi-lane roads (6 lanes or more), 4-lane roads, 2-lane roads, and mixed-use roads, considering the number of lanes and road width.

[0107] [Step 7] Data Classification by Road Type - Map-match the location of the autonomous vehicle to distinguish the points where the vehicle is situated into links and nodes; links are classified into main roads, connecting roads, junctions, junctions, overpasses, underpasses, tunnels, and bridges, while nodes are classified into signalized intersections, unsignalized intersections, and roundabouts.

[0108] [Step 8] Data Classification Based on Traffic Conditions - Based on the location information of autonomous vehicles, data is classified by linking information from other agencies to reflect directional traffic volume, lane-specific traffic volume, traffic signal timing, and surrounding road environment information regarding road types; the surrounding road environment information is then classified into illegal parking, bus lanes, and child protection zones.

[0109] [Step 9] Data Classification Based on Autonomous Vehicle Operation Distribution - Classify data based on the distribution rate of autonomous vehicles within the total traffic volume of links and nodes to reflect traffic situations where autonomous and conventional vehicles are mixed.

[0110] Data that has undergone the 9-step traffic management indexing process as described above is classified into service levels A to F of links or intersections on the road where autonomous vehicles are traveling, and stored in a database.

[0111]

[0112] In an embodiment of the present invention, FIG. 5 exemplarily shows the flow of data classification according to traffic conditions, and FIG. 6 shows a process for deriving the service level of a link or node according to [step 1] to [step 9] described above.

[0113] Referring to FIG. 5, the system (100) obtains driving data of an autonomous vehicle (vehicle ID, time and date, link ID, distance from the link starting point, speed) and obtains information on the traffic conditions of the road by combining traffic information from the Edge RSU and the ITS center. Based on this information, it distinguishes whether the road on which the autonomous vehicle is driving is a highway according to the Road Act or a road other than a highway (Off-highway road).

[0114] In the case of a highway, if the speed of the autonomous vehicle is 70 kph or higher, it is determined to be normal driving with smooth traffic flow; if it is between 70 kph and 40 kph, it is classified as a delayed situation (slow driving); and if it is less than 40 kph, it is classified as a congested situation. The congested situation is distinguished as either recurring congestion or non-recurring congestion caused by sudden events. In the case of roads other than highways, if the speed is 40 kph or higher, it is determined to be normal driving with smooth traffic flow; if it is between 40 kph and 20 kph, it is classified as a delayed situation (slow driving); and if it is less than 20 kph, it is classified as a congested situation. The congested situation is distinguished as either recurring congestion or non-recurring congestion caused by sudden events.

[0115] Referring to FIG. 6, the process of deriving the road service level according to [step 1] to [step 9] is exemplarily shown when the autonomous vehicle is moving slowly, in a congested situation, during the morning peak time, when the weather is clear, when it is a main road, when it is a multi-lane road with six or more lanes, when it is a main road (signaled intersection), when it is a traffic condition (traffic volume by direction and lane, bus lane), and when the operation distribution rate of the autonomous vehicle is 20%. FIG. 7 is a table showing an example of a database configured by classifying collected data according to a traffic management indexing process according to an embodiment of the present invention.

[0116]

[0117] 3. Traffic Management Index AI Modeling

[0118] In this invention, an AI model was used to automatically classify large amounts of data collected in real-time in traffic situations where autonomous vehicles and conventional vehicles are mixed, according to a traffic management indexing process. The AI ​​is a self-learning system based on data, and it was modeled using supervised learning, a method of training the model based on labeled data.

[0119] Data classified as a traffic management index is built into a training database for an AI model that selects traffic strategies to support full driving of autonomous vehicles. The algorithm of the AI ​​model for traffic management indexing of real-time large-capacity data collected and stored in the main storage of the living lab operating platform is illustrated in Figure 8.

[0120]

[0121] 4. Traffic Management Indexing Test

[0122] Tests were conducted using traffic data secured in Hwaseong-si, Gyeonggi-do, which was selected as an autonomous driving living lab city. Since autonomous vehicles are scheduled to begin operation in 2026, autonomous driving data was generated using VISSIM, and traffic condition data, signal information data, and sudden event information data provided by the Hwaseong-si ITS Traffic Center, as well as weather environment data provided by the Korea Meteorological Administration, were used.

[0123] - VDS Data: Travel speed (km / h), Traffic volume (vehicles / 5 minutes)

[0124] - Smart Intersection Data

[0125] : Traffic speed by approach (km / h), turning traffic volume by approach and vehicle type (vehicles / 5 minutes)

[0126] - Signalized Intersection TOD Data

[0127] - Incident information data: Location and time of occurrence, type of incident

[0128] - ITS Standard Link Data Collection

[0129] - Weather environment data (Korea Meteorological Administration): Temperature, wind speed, wind direction, rainfall, etc.

[0130] - Generation of autonomous vehicle driving status information: autonomous vehicle location information, traffic speed, etc.

[0131] - Generation of Map Matching Data: Road Function, Road Scale, Road Type, Traffic Conditions, etc.

[0132]

[0133] The data definition of the traffic management index for the collected link and node data is as shown in [Table 2].

[0134]

[0135]

[0136] In the embodiment of the present invention, since the traffic management indexing model must be constructed using highly imbalanced data in which the service level analysis result, which is the dependent variable of the data used, is 99% or more suitable and less than 1% unsuitable, the model must be constructed after applying a sampling technique to the data to sample it into balanced data.

[0137] Therefore, a database was constructed by classifying large amounts of collected data through automatic indexing using AI models. Five types of AI training models—Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost—were built and their performance was compared.

[0138] - Classified into service levels A to F using traffic volume, vehicle travel speed, number of lanes, travel time, road type, etc.

[0139] - Since the complexity of the problem is not high, a representative machine learning classification algorithm is used.

[0140] In addition, automatic traffic management indexing was performed by selecting the model with the best performance through a performance comparison between the basic traffic management indexing model and the basic model without tuning, using a tuned model that applied hyperparameter tuning with Grid Search and Random Search tuning algorithms to enhance performance.

[0141] As shown in Figure 12, the performance of the AI ​​model was evaluated using test data from the Living Lab City, and it was confirmed that the data indexing processing accuracy for four algorithms, excluding Logistic Regression, was over 98%.

[0142] Since the data handled in this invention is highly imbalanced, the performance of the model cannot be judged solely by accuracy, and precision is important because the goal is to classify non-fits with very low probability. All four algorithms showed good performance, similar to accuracy. Therefore, the AI ​​modeling of a traffic management index for real-time big data to support autonomous driving of Level 4 or higher autonomous vehicles in mixed traffic situations was evaluated as appropriate.

[0143]

[0144] Referring again to FIGS. 2 and FIGS. 3, the basic information storage unit (430) is equipped with a temporary storage unit (430a) and a permanent storage unit (430b), so that data can be stored by appropriately utilizing the temporary storage unit (430a) and the permanent storage unit (430b) according to the importance and utility of the data. The basic information storage unit (430) stores data input from a roadside device (210), an intelligent traffic system (220), an autonomous vehicle (110), and a smart intersection system (230), and can store a traffic management index generated by the basic information generation unit (420). In addition, the basic information storage unit (430) stores the results classified according to the traffic management indexing process and the AI ​​model algorithm of the collected data.

[0145] The data calculation unit (500) comprises a comparison judgment unit (510), a LOS calculation unit (520), and a congestion index calculation unit (530). The comparison judgment unit (510) compares the lane-by-lane speed measured by the Edge RSU with the link-by-link average travel speed and traffic volume of the vehicle detection system (VDS) to determine if there is a speed difference. For example, the data calculation unit (500) can determine whether there is a speed difference or if data measured by one of the devices is missing by comparing the data measured by the Edge RSU and the data measured by the VDS for the same lane at the same time. If the difference between the link-by-link average speeds measured by the Edge RSU and the VDS exceeds a threshold, the data measured by one of the devices, for example, the VDS measurement data, can be used as reference data to determine the service level of the road.

[0146] In addition, the comparison judgment unit (510) can determine whether there is a difference between the speed of the autonomous vehicle (110) and the average travel speed per link, and if there is a difference, it can provide information related to this to the autonomous vehicle (110).

[0147] The above LOS calculation unit (520) can calculate the level of service (LOS) of the road based on the vehicle density, lane occupancy rate, and traffic speed in the link, based on the status information of the autonomous vehicle, road function, and link attribute information from the data collected by the data collection unit (400). To explain in more detail, the above LOS calculation unit (520) calculates the level of service by road type and can calculate the level of service by road type from the speed of the autonomous vehicle (110), the speed of the Edge RSU by lane, and the average traffic speed and traffic volume data by VDS link.

[0148] The above congestion index calculation unit (530) calculates the congestion index (C) according to [Mathematical Formula 1]. nThe current traffic situation can be determined using ), and if the congestion index value is 0.85 or higher, it can be determined as congested, and if the congestion index value is less than 0.85, it can be determined as smooth or slow.

[0149]

[0150] Here, CS n The congestion reference speed for the current speed can be defined, and CF n is defined as the current traffic volume relative to the congestion-based traffic volume, and CO n It can be defined as the current share relative to the share based on stagnation.

[0151] ( )

[0152]

[0153] The above data storage unit (600) is equipped with a temporary storage unit (600a), a permanent storage unit (600b), and a message broker (610), and stores judgment results generated by the data processing unit (500) and traffic status information by link. Additionally, the above data storage unit (600) can transmit or display the current traffic situation judgment results to various objects through the message broker (610). The objects may be a service center for an autonomous vehicle, a router information provider, a road management agency, or a traffic information center.

[0154] For example, the message broker may provide the route and detour of the autonomous vehicle (110) to the service center of the autonomous vehicle, or provide the results of the current traffic situation judgment to navigation, applications, variable message signs (VMS), homepages, etc.

[0155] Therefore, the present invention can support the safe driving of autonomous vehicles in road environments where autonomous and non-autonomous vehicles are mixed by generating a real-time road congestion index and providing it to autonomous vehicles.

[0156]

[0157] Next, an embodiment of the method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention will be described in detail with reference to the drawings.

[0158] FIG. 9 is a data processing flowchart for determining a traffic situation in which autonomous vehicles are mixed according to an embodiment of the present invention, FIG. 10 is a diagram for explaining a detailed algorithm for determining LOS by road type according to an embodiment of the present invention, and FIG. 11 is a diagram for explaining a detailed algorithm for calculating a mixing index of a data operation unit according to an embodiment of the present invention.

[0159] Referring to FIG. 9, the method for determining a mixed traffic situation involving autonomous vehicles according to the present invention includes the step of collecting traffic situation data from a roadside device (210), an intelligent transportation system (220), and an autonomous vehicle (110) (S100). The roadside device (210) transmits data that can be collected by Edge RSU equipment to the system (100), the intelligent transportation system (220) transmits data that can be collected by ITS equipment to the system (100), and the autonomous vehicle (110) transmits data regarding vehicle ID, speed, time and date, link ID, distance from the link starting point, destination, etc. to the system (100).

[0160] After the above step (S100), a map matching step is performed to visually output driving information of the autonomous vehicle (110) by displaying the location information of the autonomous vehicle (110) on a map based on traffic condition data transmitted from the roadside device (210), intelligent traffic system (220), and autonomous vehicle (110).

[0161] After the above step (S110), basic analysis information is generated (S120). Since it is difficult to analyze traffic conditions using only raw data or metadata obtained from various sensors of the roadside device (210), intelligent traffic system (220), autonomous vehicle (110), and smart intersection system (230), an efficient classification system for real-time large-capacity traffic data must be established to provide faster and more accurate traffic operation optimization information to the autonomous vehicle (110) in traffic situations where autonomous vehicles (110) and non-autonomous vehicles (120) are mixed.

[0162] Accordingly, the above-mentioned basic analysis information is generated by performing traffic management indexing using an AI model algorithm based on data provided from a roadside device (210), an intelligent traffic system (220), and an autonomous vehicle (110), according to the driving state of the autonomous vehicle (110), traffic flow conditions, the time of operation of the autonomous vehicle, weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the autonomous vehicle. In this case, the AI ​​model includes an MLOps (Machine Learning Operations) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost is applied.

[0163] In addition, the driving status of the autonomous vehicle (110) is classified into normal driving, slow driving, stopping, accident, and vehicle breakdown, and the traffic situation is classified using the driving speed and location of the autonomous vehicle (110), and data collected by the roadside device (210) and the intelligent traffic system (220).

[0164] The operating time of the above-mentioned autonomous vehicle (110) is classified into peak hours (morning, afternoon) and off-peak hours (dawn, morning, afternoon, night), and the above-mentioned weather environment is classified by including weather sensor data from the roadside device (210).

[0165] The above road scale is classified into a multi-lane road with 6 or more lanes, a 4-lane road, a 2-lane road, and a mixed-use road for pedestrians and vehicles based on the number of lanes and road width of the road based on the location information of the autonomous vehicle (110), and the above road type is classified into links and nodes based on the location of the autonomous vehicle displayed in the map matching unit (410), and the above traffic conditions are classified including traffic volume by direction, traffic volume by lane, traffic signal display, and surrounding environment information of the road type.

[0166] The operation distribution of the above-mentioned autonomous vehicle (110) includes data classification based on the operation distribution rate of the autonomous vehicle (110) among the total traffic volume of links and nodes to reflect traffic conditions in which the autonomous vehicle (110) and non-autonomous vehicle (120) are mixed.

[0167] After the above step (S120), the generated basic information is stored in the temporary storage unit (430a) or permanent storage unit (430b) of the basic information storage unit (430) (S130).

[0168] Additionally, after the above step (S120), the data processing unit (500) uses the generated basic information as raw data to determine the traffic condition by fusion analyzing the state of the autonomous vehicle (110), Edge RSU information, VDS equipment of the ITS center, and smart intersection information. The determination of the traffic condition can be performed by including link-specific traffic information generated through location information and map matching.

[0169] To explain in more detail, depending on the device, the generated basic information may omit some details or produce different results even for the same data. For example, the lane speeds generated by the Edge RSU device may differ from those generated by the VDS device, or lane speeds in a specific area at a specific point in time may not be detected by the Edge RSU device but are detected by the VDS device; therefore, the speed differences must be corrected by comparing them.

[0170] Accordingly, after the above step (S120), the speeds of the Edge RSU and the VDS device are compared (S140), and if the difference exceeds a reference value, either the value measured by the Edge RSU or the VDS device is selected to determine the Level of Service (LOS) for each road type (S150) and used to calculate the congestion index (S160). The calculation of the Level of Service and congestion index for each road type will be explained in more detail below.

[0171] After the above step (S160), a judgment result is generated (S170) and stored in a data storage unit (600) (S180). In this case, the data storage unit (600) includes a message broker (610) so that the current traffic situation judgment result can be provided to a service center of an autonomous vehicle, a router information provider, a road management agency, or a traffic information center.

[0172] Meanwhile, after the above step (S120), it is determined whether the autonomous vehicle is driving normally (S190). If it is determined that the autonomous vehicle is not driving normally, the information of the autonomous vehicle is used to determine the service level by road type. If it is determined that the autonomous vehicle is driving normally, the speed of the autonomous vehicle is compared with the difference in average traffic speed by link (S200). If the difference value exceeds a threshold value, it is determined whether information provision is necessary and a result is generated (S210). The information may be provided to the autonomous vehicle or to the autonomous vehicle's service center, router information provider, road management agency, or traffic information center.

[0173] Next, with reference to FIGS. 10 and FIGS. 11, the detailed processes for determining the level of service and calculating the mixture index by road type will be explained, respectively.

[0174] Referring to FIG. 10, state information of an autonomous vehicle including driving speed, location, and time is provided from basic information which is raw data (S300), and it is determined whether the autonomous vehicle is driving on an expressway and a motor-only road or on a road with four or more lanes (S310, S320, S330). If it is an expressway and a motor-only road, it is determined to be a motor-only road (S340), and if it is driving on a road with four or more lanes, it is determined to be a multi-lane road (S350).

[0175] After the above steps (S340, S350), attribute information of the link including traffic volume, travel speed, and number of lanes is provided (S360), and the level of service of the road is determined as LOS A to LOS F for vehicle density (pcpkmpl) or traffic volume to capacity ratio (V / C), respectively (S370, S380, S371 to S375, S381 to S385).

[0176] Meanwhile, in the above step (S330), if it is determined that the autonomous vehicle is driving on a road with fewer than 4 lanes, for example, a 2-lane road (S400), the road service level is determined to be LOS A ~ LOS F according to the traffic speed (kph) (S410, S411 ~ S415).

[0177] Referring to FIG. 11, the congestion index calculation procedure includes a step of collecting real-time traffic information for each road section, including travel speed, traffic volume, and occupancy rate (S500). After the above step (S500), according to [Equation 1] as described above, a congested area (C n Calculate the number (S510), and if the congestion index value is 0.85 or higher, determine that it is delayed (S515), and if the congestion index value is less than 0.85, determine that it is smooth or slow (S520).

[0178] After the above step (S515), CS n , CF n , CO nIt is determined whether the value of is negative or positive (S530), and if the value is all positive, it is determined that the congestion situation has worsened (S540), and if the value is all negative, it is determined that the congestion situation has eased (S550), CS n , CF n , CO n If the values ​​of are all negative and all positive, it is determined that the stagnation is maintained (S560), and a determination result is generated.

[0179] Although the invention made by the inventors has been specifically described according to the above embodiments, the present invention is not limited to the above embodiments and can be modified in various ways without departing from the gist thereof.

[0180] By using the system and method for determining mixed traffic conditions of autonomous vehicles based on multi-collected data fusion analysis according to the present invention, safe driving of autonomous vehicles can be supported in road environments where autonomous vehicles and non-autonomous vehicles are mixed.

Claims

1. In a traffic situation judgment system based on multi-collected data fusion analysis for roads where autonomous and non-autonomous vehicles are mixed, A data collection unit that receives traffic condition data from roadside devices, intelligent traffic systems, and autonomous vehicles; A data calculation unit that determines the traffic status by link by calculating the service level and congestion index for each road from the data collected by the above data collection unit; and It includes a data storage unit that stores traffic status information for each link and provides traffic status information determined by the data processing unit to the autonomous vehicle, The above data collection unit A map matching unit that displays location information of the autonomous vehicle on a map and visually outputs operation information of the autonomous vehicle; and A traffic situation judgment system characterized by including a basic information generation unit that performs traffic management indexing using an AI model algorithm for each of the data provided from the above roadside device, the above intelligent traffic system, and the above autonomous vehicle.

2. In Paragraph 1, A traffic situation judgment system characterized by the above basic information generation unit performing indexing of data received from the above roadside device, the above intelligent traffic system, and the above autonomous vehicle according to the driving state of the autonomous vehicle, traffic flow conditions, the time of operation of the above autonomous vehicle, weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the above autonomous vehicle.

3. In Paragraph 2, The driving conditions of the above autonomous vehicle are classified into normal driving, slow driving, stopping, accident, and vehicle breakdown, and The above traffic conditions are classified using the driving speed and location of the autonomous vehicle, the roadside device, and data collected by the intelligent traffic system, and The operating hours of the above autonomous vehicles are classified into peak hours (morning, afternoon) and off-peak hours (dawn, morning, afternoon, night), and The above weather environment classifies data by reflecting weather sensor data from the roadside device and weather information provided by the Korea Meteorological Administration, and The above road scale is classified into multi-lane roads with 6 or more lanes, 4-lane roads, 2-lane roads, and mixed pedestrian and vehicle roads based on the number of lanes and road width according to the location information of the autonomous vehicle, and The above road type classifies the location of the autonomous vehicle displayed in the above map matching unit into links and nodes, wherein the links classify data into main roads, connecting roads, junctions, branching points, overpasses, underpasses, tunnels, and bridges, and the nodes classify data into signalized intersections, non-signaled intersections, and roundabouts. The above traffic conditions are classified by including traffic volume by direction, traffic volume by lane, traffic signal timing, and surrounding road environment information regarding road types, and A traffic situation judgment system characterized by including data classification based on the operation distribution rate of the autonomous vehicle among the total traffic volume of the links and nodes to reflect a traffic situation in which the autonomous vehicle and the non-autonomous vehicle are mixed.

4. In Paragraph 3, A traffic situation judgment system characterized by the above-mentioned basic information generation unit including an MLOps (Machine Learning Operations) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost is applied.

5. In Paragraph 1, The above data processing unit A traffic situation judgment system characterized by calculating the level of service (LOS) of a road based on vehicle density, lane occupancy rate, and traffic speed in a link, based on the state information of the autonomous vehicle, road function, and attribute information of the link.

6. In Paragraph 1, The above data processing unit Congestion index (C) according to the [mathematical formula] below n A traffic situation judgment system characterized by determining the current traffic situation using ), determining congestion if the congestion index value is 0.85 or higher, and determining smooth or slow traffic if the congestion index value is less than 0.

85. [Mathematical Formula] ( ) 7. In Paragraph 1, A traffic situation judgment system characterized in that the above-mentioned roadside device is an Edge RSU and the above-mentioned intelligent traffic system is an ITS.

8. In a method for determining traffic conditions based on fusion analysis of multiple collected data on roads where autonomous and non-autonomous vehicles are mixed, i) A step of receiving traffic condition data from roadside devices, intelligent traffic systems, and autonomous vehicles from a data collection unit; ii) a step of displaying the location information of the autonomous vehicle on a map to visually output the operation information of the autonomous vehicle; iii) A step of performing traffic management indexing using an AI model algorithm for each of the data received from the roadside device, the intelligent traffic system, and the autonomous vehicle; iv) a step of determining the traffic status by link in the data processing unit by calculating the level of service and congestion index for each road based on the results of performing the above traffic management indexing; and v) A method for determining a traffic situation characterized by including the step of storing traffic status information for each link and providing traffic status information determined by the data processing unit to the autonomous vehicle from the data storage unit.

9. In Paragraph 8, The above step iii) is a method for determining traffic conditions characterized by performing indexing on data received from the roadside device, the intelligent traffic system, and the autonomous vehicle according to the driving state of the autonomous vehicle, traffic flow conditions, the time of operation of the autonomous vehicle, weather environment, road function, road scale, road type, traffic conditions, and the operation distribution of the autonomous vehicle.

10. In Paragraph 9, The driving conditions of the above autonomous vehicle are classified into normal driving, slow driving, stopping, accident, and vehicle breakdown, and The above traffic conditions are classified using the driving speed and location of the autonomous vehicle, the roadside device, and data collected by the intelligent traffic system, and The operating hours of the above autonomous vehicles are classified into peak hours (morning, afternoon) and off-peak hours (dawn, morning, afternoon, night), and The above weather environment classifies data by reflecting weather sensor data from the roadside device and weather information provided by the Korea Meteorological Administration, and The above road scale is classified into multi-lane roads with 6 or more lanes, 4-lane roads, 2-lane roads, and mixed pedestrian and vehicle roads based on the number of lanes and road width according to the location information of the autonomous vehicle, and The above road type classifies the location of the autonomous vehicle displayed in the above map matching unit into links and nodes, wherein the links classify data into main roads, connecting roads, junctions, branching points, overpasses, underpasses, tunnels, and bridges, and the nodes classify data into signalized intersections, non-signaled intersections, and roundabouts. The above traffic conditions are classified by including traffic volume by direction, traffic volume by lane, traffic signal timing, and surrounding road environment information regarding road types, and A method for determining traffic conditions characterized by including data classification based on the distribution rate of operation of the autonomous vehicle among the total traffic volume of the links and nodes to reflect traffic conditions in which the autonomous vehicle and the non-autonomous vehicle are mixed.

11. In Paragraph 10, A method for determining traffic conditions, characterized in that the above AI model includes a Machine Learning Operations (MLOps) system to which at least one model selected from Logistic Regression, KNN, Random Forest, LightGBM, and CatBoost is applied.

12. In Paragraph 8, A method for determining traffic conditions, characterized in that step iv) calculates the level of service (LOS) of the road based on vehicle density, lane occupancy rate, and traffic speed in the link, based on the state information of the autonomous vehicle, road function, and attribute information of the link.

13. In Paragraph 12, The service level of the above road is A step of determining the driving speed, location, time information, and type of road information of the autonomous vehicle; A step of analyzing attribute information of a link including traffic volume, travel speed, and number of lanes when the autonomous vehicle is traveling on an expressway or a dedicated automobile road, or on a multi-lane road with four or more lanes; and It includes a step of determining the level of service based on vehicle density (pcpkmpl) or the traffic-to-capacity ratio (V / C), and A method for determining traffic conditions characterized by determining the level of service based on the traffic speed when the autonomous vehicle is driving on a road with fewer than four lanes.

14. In Paragraph 8, The above step iv) is Congestion index (C) according to the [mathematical formula] below n A method for determining traffic conditions characterized by determining the current traffic condition using ), wherein if the congestion index value is 0.85 or higher, it is determined to be delayed, and if the congestion index value is less than 0.85, it is determined to be smooth or slow. [Mathematical Formula] ( ) 15. In Paragraph 14, The above CS n , the above CF n and the above CO n Determine whether the value of is negative or positive; if the values ​​are all positive, determine that the congestion situation has worsened, and if the values ​​are all negative, determine that the congestion situation has eased, and CS n , CF n , CO n A method for determining traffic conditions characterized by determining that congestion is maintained when the values ​​of are all negative and all positive.

Images