A vehicle platoon control method, device and medium based on artificial intelligence

Abstract

The embodiments of this specification disclose a vehicle platoon control method, device and medium based on artificial intelligence, which relates to the field of vehicle control technology and is used to solve the problem of unreliable control of existing vehicle platoons. The method includes: obtaining vehicle data and spatial environment data of each vehicle in the current fleet; determining the road traffic parameters of the safety area corresponding to each vehicle based on the spatial environment data in the continuous acquisition cycle and the safe passage data information corresponding to the current fleet; inputting the road traffic parameters into the preset long-short term memory network to obtain traffic development prediction data; determining the corresponding vehicle operation events based on the vehicle traffic data, so as to construct a vehicle association topology map based on the matching relationship between the basic vehicle data and each vehicle operation event; determining the adjustable driving data of the current fleet based on the traffic development prediction data and the vehicle association topology map, adjusting the driving planning data, and realizing intelligent control of the fleet.

Description

Technical Field

[0001] The present specification relates to the field of traffic control technology, and in particular to a vehicle platoon control method, device and medium based on artificial intelligence. Background Art

[0002] Vehicle platooning is a traffic management and driving strategy that groups multiple vehicles into an orderly convoy for coordinated driving. In this convoy, vehicles are connected to each other through communication technology and advanced control systems to maintain appropriate spacing, speed and driving paths. This control method aims to improve traffic efficiency, enhance traffic safety, reduce energy consumption, and realize intelligent vehicle management.

[0003] In the control framework, model predictive control is a commonly used method and has been applied by many scholars to fleet control. The collaborative control strategy based on model predictive control is used for the platooning of connected and automated vehicles (CAVs), which can be well applied to the control of the fleet. However, the fleet strategy chosen by them in existing studies generally adopts a fixed head strategy. Although this strategy can ensure the safety of the fleet, it is difficult to consider the traffic conditions of each vehicle in the fleet under complex traffic conditions, and it is impossible to fully perceive the spatial environment and traffic data. It is difficult to achieve intelligent traffic control of each vehicle in the vehicle team, which easily leads to waste of road resources. In addition, there is a lack of periodic planning and in-depth analysis for the collection of vehicle data, which makes it difficult to cope with complex and changing traffic conditions. In terms of collaborative driving, the fixed head strategy for the head vehicle only lacks global planning and dynamic adjustment for the entire fleet. The strategy adjustment method relies on the driver's personal judgment of road conditions and other vehicle behavior events, which will cause the vehicle team to react slowly when facing traffic events with sudden road conditions. Summary of the invention

[0004] In order to solve the above technical problems, one or more embodiments of this specification provide a vehicle platoon control method, device and medium based on artificial intelligence.

[0005] One or more embodiments of this specification adopt the following technical solutions:

[0006] One or more embodiments of this specification provide a vehicle platoon control method based on artificial intelligence, the method comprising:

[0007] Based on the on-board communication equipment and pre-installed on-board laser radar of each vehicle in the current fleet, the vehicle data and spatial environment data of each vehicle are obtained; wherein the vehicle data includes: basic vehicle data and vehicle traffic data;

[0008] Determine the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet according to the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet;

[0009] Inputting the road traffic parameters corresponding to each vehicle in the current fleet into a preset long short-term memory network to obtain traffic development prediction data corresponding to each vehicle;

[0010] Determine the vehicle operation event corresponding to each vehicle according to the vehicle traffic data, so as to construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event;

[0011] According to the traffic development prediction data and the vehicle association topology map, the adjustable driving data of the current fleet is determined, so as to adjust the driving planning data of the current fleet based on the adjustable driving data, and the adjusted driving planning data is sent to each vehicle through the on-board communication equipment of each vehicle, so as to realize intelligent control and coordinated driving of the current fleet.

[0012] Optionally, in one or more embodiments of the present specification, based on the vehicle-mounted communication equipment and the preset vehicle-mounted laser radar of each vehicle in the current fleet, the vehicle data and the spatial environment data of each vehicle are obtained, specifically including:

[0013] Determine the basic vehicle data of each vehicle in the current fleet based on the vehicle-mounted communication equipment of each vehicle, and obtain the initial vehicle traffic data and vehicle monitoring point cloud of each vehicle according to the preset vehicle-mounted laser radar of each vehicle; wherein the initial vehicle traffic data at least includes: vehicle speed, vehicle acceleration, vehicle steering angle, and vehicle braking;

[0014] preprocessing the initial vehicle traffic data of each vehicle based on a preset processing strategy to obtain processed vehicle data;

[0015] Extracting the vehicle monitoring point cloud based on a preset collection interval to obtain a plurality of vehicle point cloud images of the vehicle;

[0016] Comparing the vehicle point cloud images in sequence to use the mutation point cloud images in the vehicle point cloud images as key point cloud images;

[0017] Acquire multiple adjacent point cloud images of the key point cloud image, and based on the time sequence of the key point cloud image and the adjacent point cloud images, sequentially obtain differential point cloud images and use the connected areas of the differential point cloud images as mutation targets, and annotate the mutation targets in the key point cloud images to obtain spatial environment data; wherein the mutation targets include: static mutation targets and dynamic mutation targets.

[0018] Optionally, in one or more embodiments of the present specification, preprocessing the initial vehicle traffic data of each vehicle based on a preset processing strategy to obtain processed vehicle data specifically includes:

[0019] Acquire attribute information corresponding to each of the initial vehicle traffic data; wherein the attribute information includes: status transmission time, status occurrence location and status data value;

[0020] Arranging the condition data values ​​based on the condition occurrence time and the condition occurrence location corresponding to each of the initial vehicle traffic data to obtain a first condition data value and a second condition data value, and using the first condition data value and the second condition data value corresponding to the same vehicle traffic data as a vehicle operation event of the corresponding vehicle to determine a plurality of vehicle operation events of each of the vehicles;

[0021] Based on the first status data value and the second status data value corresponding to each vehicle operation event and the matching relationship with other vehicle operation events, determine the companion vehicles corresponding to each vehicle operation event of the vehicle, so as to determine the companion relationship topology corresponding to each of the vehicles;

[0022] Based on the preset behavior table, the developable conditions corresponding to each condition data value are determined, and based on the condition data value, the developable conditions and the companion relationship topology map, the probability of each vehicle operation event occurring to the vehicle is determined, and the initial condition data corresponding to the vehicle operation event is filtered based on the preset probability to obtain processed vehicle traffic data.

[0023] Optionally, in one or more embodiments of the present specification, determining the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet according to the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet, specifically includes:

[0024] Determine the vehicle driving section corresponding to the current vehicle fleet according to the location data of the current vehicle fleet and the preset road network data, determine the driving scene corresponding to the current vehicle fleet according to the road type and vehicle flow of the vehicle driving section, and determine the corresponding collection period based on the driving scene;

[0025] In a continuous collection cycle, the sudden change target in the collected spatial environment data is identified to determine the dynamic change data of the sudden change target according to the position data of the sudden change target in the adjacent spatial environment data; wherein the dynamic change data includes: moving speed, speed direction and moving trajectory;

[0026] Determine the safety zone corresponding to each vehicle according to the vehicle traffic data and multi-dimensional traffic margin data of each vehicle in the current fleet;

[0027] Determine a first mutation target located in the safety area according to the position data of the mutation target, and determine a second mutation target close to the safety area and matching a specified critical condition according to the position data of the mutation target and the dynamic change data;

[0028] The position data corresponding to the first sudden change target and the second sudden change target, the dynamic change data and the road data of the driving scene are used as road traffic parameters of the safe area.

[0029] Optionally, in one or more embodiments of the present specification, the road traffic parameters corresponding to each vehicle in the current fleet are input into a preset long short-term memory network to obtain traffic development prediction data corresponding to each vehicle, specifically including:

[0030] According to the position information in the road traffic parameters of the vehicle safety area and the traffic flow direction of the current fleet, the safety area is divided to obtain the safety path grid corresponding to each of the vehicles;

[0031] Determine an initial risk path grid corresponding to the safety area according to the position of the sudden change target in the road traffic parameter of the vehicle safety area and the connection line of the safety path grid;

[0032] Determining the grid type of the initial risk path grid according to the coordinate points of each of the vehicles in the safety area and the traffic flow direction of the current fleet; wherein the grid type includes an upstream grid located in front of the vehicle and a downstream grid located behind the vehicle;

[0033] Determining a first weight value of the initial risk path grid based on a grid type of the initial risk path grid, determining a second weight value of the initial risk path grid according to a distance of the initial risk path grid from the vehicle, and performing weighted processing on the first weight value and the second weight value to obtain a weight coefficient;

[0034] According to the corresponding relationship between the initial risk path grid and the road traffic parameters, the road traffic parameters corresponding to the weight coefficients are determined, so that the road traffic parameters and the weight coefficients corresponding to the road traffic parameters are input into the preset long short-term memory network to obtain the traffic development prediction data corresponding to each vehicle.

[0035] Optionally, in one or more embodiments of the present specification, determining the vehicle operation event corresponding to each vehicle according to the vehicle traffic data, so as to construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event, specifically includes:

[0036] Determining the probability of each vehicle operation event occurring to the vehicle according to the vehicle traffic data, so as to screen and determine the vehicle operation event corresponding to each vehicle according to the probability;

[0037] Determine each vehicle node according to the basic vehicle data, and determine the edges between each vehicle node according to the association relationship between each vehicle operation event;

[0038] A vehicle association topology graph is constructed based on the edges between the vehicle node and each vehicle node.

[0039] Optionally, in one or more embodiments of the present specification, determining the adjustable driving data of the current fleet according to the traffic development prediction data and the vehicle association topology map specifically includes:

[0040] Predicting the vehicle position and vehicle travel data of each vehicle in the next time period according to the traffic development prediction data;

[0041] Predicting traffic conflict data and conflicting vehicles in the next time period of the current fleet based on the vehicle association topology map and the vehicle positions and vehicle driving data of each vehicle in the next time period;

[0042] Quantitatively processing the traffic conflict data to determine a risk value corresponding to the traffic conflict data, and determining first to-be-adjusted driving data corresponding to the conflicting vehicle according to the risk value and the conflict type corresponding to the traffic conflict data;

[0043] Determine the association relationship between the first to-be-adjusted driving data corresponding to the conflicting vehicle and the adjacent vehicle of the conflicting vehicle, so as to determine the second to-be-adjusted driving data of the adjacent vehicle of the conflicting vehicle according to the association relationship;

[0044] The first driving data to be adjusted and the second driving data to be adjusted are integrated to determine adjustable driving data of the current vehicle fleet.

[0045] Optionally, in one or more embodiments of the present specification, sending the adjusted driving plan data to each vehicle through the vehicle-mounted communication device of each vehicle specifically includes:

[0046] According to the transmission format requirements of the vehicle-mounted communication device, the adjusted driving plan data is encoded to obtain data to be transmitted, and the data to be transmitted is encapsulated based on the communication protocol of the vehicle-mounted communication device to obtain encapsulated data to be transmitted;

[0047] Calling the vehicle-mounted communication equipment of each vehicle through the central control unit corresponding to the current fleet, establishing a communication connection with each vehicle, and sending the encapsulated data to be transmitted based on the communication connection;

[0048] According to the feedback data of each vehicle in the current fleet, it is determined whether the packaged data to be transmitted has been sent.

[0049] One or more embodiments of this specification provide a vehicle platoon control device based on artificial intelligence, the device comprising:

[0050] at least one processor; and,

[0051] a memory communicatively connected to the at least one processor; wherein,

[0052] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform any of the above methods.

[0053] One or more embodiments of the present specification provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute any of the above-described methods.

[0054] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects:

[0055] By obtaining basic vehicle data, vehicle traffic data and spatial environment data, the detailed information of the fleet can be fully grasped, providing a rich data basis for subsequent decision-making. The limitations of a single data source are avoided, and the reliability of subsequent processing is improved. By using the long short-term memory network to process road traffic parameters and obtain traffic development forecast data, potential traffic change trends can be discovered in advance, providing forward-looking guidance data for fleet planning; according to the traffic development forecast data and the vehicle association topology map, the adjustable driving data can be determined, and the driving planning data of the fleet can be adjusted, which reflects a high degree of flexibility and adaptability, so that the fleet can flexibly change the driving plan according to the actual situation, and adjust the driving planning data of the entire fleet, realizing the overall planning and dynamic adjustment of the entire fleet, and also helping to improve the fleet's response ability when facing sudden road conditions. The adjusted driving planning data is sent to each vehicle through the on-board communication equipment to realize the intelligent control and coordinated driving of the fleet, ensuring that the fleet is coordinated as a whole, improving transportation efficiency while reducing traffic congestion and accident risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0057] Figure 1 A schematic diagram of a method flow of a vehicle platoon control method based on artificial intelligence provided in an embodiment of this specification;

[0058] Figure 2 A schematic diagram of a companion relationship topology diagram in a scenario provided in an embodiment of this specification;

[0059] Figure 3 A schematic diagram of the structure of a vehicle platoon control device based on artificial intelligence provided in an embodiment of this specification;

[0060] Figure 4 A schematic diagram of the structure of a non-volatile storage medium provided in an embodiment of this specification. DETAILED DESCRIPTION

[0061] The embodiments of this specification provide a vehicle platoon control method, device and medium based on artificial intelligence.

[0062] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0063] like Figure 1 As shown, the embodiment of this specification provides a method flow chart of a vehicle platoon control method based on artificial intelligence. Figure 1 It can be seen that in one or more embodiments of this specification, a vehicle platoon control method based on artificial intelligence includes the following steps:

[0064] S101: Based on the on-board communication equipment and pre-installed on-board laser radar of each vehicle in the current fleet, the vehicle data and spatial environment data of each vehicle in the current fleet are obtained; wherein the vehicle data includes: basic vehicle data and vehicle traffic data.

[0065] In the current vehicle fleet control process, it is difficult to consider the traffic conditions of each vehicle in the fleet under complex traffic conditions, it is impossible to fully perceive the spatial environment and traffic data, and it is difficult to achieve intelligent traffic control of each vehicle in the vehicle fleet, which easily leads to a waste of road resources. Moreover, there is a lack of periodic planning and in-depth analysis for the collection of vehicle data, which makes it difficult to cope with complex and changeable traffic conditions. Therefore, in order to solve this problem, in one or more embodiments of this specification, the vehicle data and spatial environment data of each vehicle in the current fleet will be collected and obtained respectively according to the information collection equipment of each vehicle in the current fleet, that is, the vehicle-mounted communication equipment of each vehicle and the pre-installed vehicle-mounted laser radar. It should be noted that vehicle data includes basic vehicle data and vehicle traffic data. By collecting vehicle data and spatial environment data in real time, the management system of the current fleet can quickly obtain the traffic conditions and environmental information of the current fleet, providing comprehensive data support for subsequent decision-making and control, and improving the comprehensive coverage of data.

[0066] Specifically, in one or more embodiments of the present specification, based on the vehicle-mounted communication equipment and the preset vehicle-mounted laser radar of each vehicle in the current fleet, the vehicle data and the spatial environment data of each vehicle are obtained, which specifically includes the following process:

[0067] First, the vehicle-mounted communication equipment of each vehicle in the fleet is used to determine the basic vehicle data of each vehicle. The basic vehicle data may include information such as the vehicle model, identification number, and the fleet to which it belongs. This information is generally obtained through the interaction between the vehicle-mounted communication equipment and the external system to identify and distinguish different vehicles. At the same time, the initial vehicle traffic data and vehicle monitoring point cloud of each vehicle are obtained according to the preset vehicle-mounted laser radar of each vehicle; wherein, the initial vehicle traffic data at least includes: vehicle speed, vehicle acceleration, vehicle steering angle, vehicle braking and other key information, which reflect the real-time operation status of the vehicle. In order to process the original data to avoid the impact of erroneous data on the subsequent, the initial vehicle traffic data of each vehicle will be preprocessed according to the preset processing strategy to improve the data quality, so as to obtain the processed vehicle data. Then, the vehicle monitoring point cloud is extracted according to the preset acquisition interval to obtain multiple vehicle point cloud images of the vehicle. The collected vehicle point cloud images are compared in turn. Through the image comparison algorithm, the mutation part in the image is found, and the image containing the mutation is determined as the key point cloud image. It can be understood that the mutation in this scene may be manifested as an obstacle suddenly appearing in front of the vehicle, the color of the traffic light changes, and the vehicle driving scene changes significantly. After determining the key frame image, multiple adjacent images of the key frame image are obtained. Then, based on the acquisition time sequence of the key point cloud image and the adjacent point cloud images, the differences between the adjacent point cloud images are calculated in turn to obtain the differential point cloud image. Since the differential point cloud image highlights the changes between the adjacent point cloud images, it can more clearly show the dynamic changes in the scene, so the differential point cloud image is processed and the connected areas therein are identified as mutation targets. Then, these mutation targets are annotated in the key point cloud image to finally form the spatial environment data; it can be understood that the mutation may be manifested as an obstacle suddenly appearing in front of the vehicle, the color of the traffic light changes, and the vehicle driving scene changes significantly, so the mutation targets include: static mutation targets and dynamic mutation targets.

[0068] In this process, the vehicle monitoring video is segmented according to the preset acquisition interval, and the vehicle point cloud image is obtained according to the preset acquisition interval. While ensuring the acquisition of sufficient information, the amount of data can be effectively controlled, and the acquisition of too many redundant images can be avoided, thereby improving data processing efficiency. By determining the key point cloud image, it is possible to focus on the important information in the video, and the vehicle point cloud image containing mutations is used as the key point cloud image, avoiding the processing of a large number of irrelevant point cloud images and saving computing resources. By calculating the differential point cloud image between the key point cloud image and its adjacent point cloud images, the changes between the adjacent point cloud images are highlighted, so that the dynamic changes in the scene are more clearly displayed. The connected areas in the differential point cloud image are identified as mutation targets, and are annotated in the key point cloud image to form spatial environment data, which intuitively displays the key change information in the vehicle driving environment, which is conducive to the subsequent more comprehensive control of the vehicle team.

[0069] Furthermore, in the actual vehicle traffic data collection process, due to factors such as sensor errors and communication interference, the initial data may contain errors or data that do not conform to the actual traffic scene. Therefore, in one or more embodiments of this specification, the initial vehicle traffic data of each vehicle is pre-processed based on a preset processing strategy to obtain processed vehicle data, which specifically includes the following process:

[0070] First, the attribute information corresponding to each initial vehicle traffic data is obtained; wherein the attribute information includes: the time when the condition is sent, the location where the condition occurs, and the condition data value. Based on the obtained time when the condition occurs and the location where the condition occurs, the condition data values ​​are arranged to obtain a first condition data value and a second condition data value. These two data values ​​come from the same vehicle traffic data, and they together constitute a vehicle operation event of the corresponding vehicle. For example, taking the vehicle speed as an example, the speed values ​​recorded at two different time points at a certain location can be regarded as a vehicle operation event, and this event may reflect the speed change of the vehicle at this location. In this way, multiple vehicle operation events of each vehicle can be determined. The first condition data value and the second condition data value corresponding to each vehicle operation event are analyzed to find out the matching relationship between them and other vehicle operation events. For example, if two vehicles have deceleration operation events at similar times and locations, then there is a certain matching relationship between the two events. Through the analysis of this matching relationship, the companion vehicle corresponding to each vehicle operation event, that is, other vehicles associated with the vehicle operation event, is determined, and then the following can be determined. Figure 2 The companion relationship topology corresponding to the vehicle shown. Then, based on the preset behavior table, the developable conditions corresponding to each condition data value are determined, and based on the condition data value, the developable conditions and the companion relationship topology, the probability of each vehicle operation event occurring in the vehicle is determined, and the initial condition data corresponding to the vehicle operation event is filtered based on the preset probability to obtain the processed vehicle traffic data. Among them, it can be understood that the preset behavior table is formulated based on a large amount of historical data, traffic rules, and vehicle operation characteristics, etc., and it contains the possible evolution results of different condition data values ​​in different situations. For example, for a certain speed value, combined with the current position of the vehicle such as approaching an intersection and traffic rules, the preset behavior table may give developable conditions such as deceleration and parking.

[0071] In this process, by arranging the status data values ​​according to time and place and constructing vehicle operation events, the scattered vehicle traffic data can be integrated into a meaningful event sequence, making the data organization more orderly and structured. For example, a combination of status data values ​​of frequent sudden acceleration and sudden braking of vehicles at a specific location within a certain period of time can be clearly identified as a vehicle operation event with potential risks, avoiding the confusion and irrelevance of the original data. Then, based on the matching relationship with other vehicle operation events, a companion relationship topology map is constructed to mine the behavioral similarities and correlations between vehicles. For vehicles that often have similar operation events at the same time and place, they can be determined to have a high correlation, so that in subsequent analysis, the data collected by the vehicle that may be erroneous can be corrected based on the behavior of the companion vehicle, further improving the relevance and analyzability of the data. By using the preset behavior table to determine the developable status and combining the companion relationship topology map to determine the probability of event occurrence, the data can be effectively screened. The initial status data corresponding to the vehicle operation events with low probability, which may be abnormal or erroneous, are filtered out, and only the data with high credibility and representativeness are retained, which greatly improves the accuracy and reliability of the data and provides a high-quality data foundation for subsequent precise analysis and decision-making. In addition, identifying multiple vehicle operation events of a vehicle and its companion vehicles can also help discover potential dangerous driving behavior patterns. For example, if multiple companion vehicles frequently have oversteering or unstable speed operation events on a certain road section, it may indicate that there are special road conditions or driving risks on that road section, so that measures can be taken in advance, such as issuing warning information or adjusting the fleet's driving route.

[0072] S102: Determine the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet based on the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet.

[0073] Since traffic conditions are constantly changing, a suitable collection cycle can enable the collected data to better reflect this dynamic change. Therefore, when determining the collection cycle corresponding to the vehicle data, the road traffic parameters of the safety boundary area corresponding to each vehicle in the current fleet will be determined based on the spatial environment data within the continuous collection cycle and the safety traffic data information corresponding to the current fleet. The road traffic parameters of the safety boundary area are crucial to determining the safe driving range of the vehicle on the road. These parameters include the safe distance between the vehicle and surrounding obstacles and other vehicles, and the safe driving range of the vehicle under different road conditions such as curves and slopes. Therefore, by determining the road traffic parameters corresponding to each vehicle, the safety risks during vehicle driving can be more accurately assessed.

[0074] Specifically, in one or more embodiments of the present specification, determining the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet according to the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet, specifically includes the following process:

[0075] The vehicle driving section corresponding to the current fleet is determined based on the location data of the current fleet and the preset road network data. The driving scene corresponding to the current vehicle is determined based on the road type and vehicle flow of the vehicle driving section, and the corresponding collection cycle is determined based on the driving scene. For example, on urban roads or highways, the vehicle data collection cycle is determined by comprehensively considering factors such as vehicle driving speed, road condition complexity, and sensor performance, and this cycle serves as the time window for subsequent data analysis and processing.

[0076] In a continuous collection cycle, the collected spatial environment data is analyzed to identify the sudden change targets. These sudden change targets may include suddenly appearing pedestrians, vehicles, obstacles, or changes in traffic lights, etc. They are key factors affecting vehicle driving safety. After identifying the sudden change target, the position data of the sudden change target in the adjacent spatial environment data will be compared, that is, the position data of the sudden change target in the adjacent spatial environment data will be used to calculate the dynamic change data such as the moving speed, speed direction and moving trajectory of the sudden change target. Then, according to the vehicle traffic data and multi-dimensional traffic margin data of each vehicle in the current fleet, the safety area corresponding to each of the vehicles is determined. Among them, the multi-dimensional traffic margin data may include: road width margin, time margin, speed margin, etc. By comprehensively considering the vehicle traffic data and multi-dimensional traffic margin data, a relatively reasonable safety area can be determined for each vehicle in the fleet to ensure the safety and smoothness of vehicle driving. Then, according to the position data of the sudden change target, the first sudden change target located in the safety area is found. These targets are factors that have a direct impact on vehicle driving safety. At the same time, the second mutation targets close to the safety area and matching the specified critical conditions are determined. Although they are not in the safety area temporarily, they may also pose a potential threat to vehicle driving because they are close to the safety area. Then the position data, dynamic change data and road data of the driving scene corresponding to the first mutation target and the second mutation target obtained above are used as the road traffic parameters of the safety area. Among them, the specified critical condition is a preset judgment standard for determining which mutation targets close to the safety area belong to the second mutation target. For example, the distance from the boundary of the safety area is less than 10 meters, and the speed exceeds a certain threshold, such as more than 40km / h on urban roads, or the angle with the vehicle's driving direction is within a specific range, such as near an intersection, and the angle with the vehicle's driving direction is between 30°-60°.

[0077] In another scenario, the safety area corresponding to each vehicle can also be determined based on the vehicle traffic data of each vehicle in the current fleet and the preset following model to determine the longitudinal safety boundary area corresponding to each vehicle. At the same time, the vehicle width corresponding to each vehicle is determined based on the basic vehicle data of each vehicle in the current fleet, so as to determine the lateral safety boundary area corresponding to each vehicle based on the vehicle width, the lane width of the vehicle driving section and the preset parallel driving safety margin. The lane width determines the available range of the vehicle in the lateral space, and the parallel driving safety margin is to ensure that the vehicle has sufficient safety interval when running parallel with the vehicle in the adjacent lane to avoid accidents such as scratches. For example: Assuming that vehicle A is driving in the right lane, half of the lane width is 3.5 / 2=1.75 meters based on the center line of the lane. The width of vehicle A is 1.8 meters, and its half width is 1.8 / 2=0.9 meters. Adding the parallel driving safety margin of 0.5 meters, the safety boundary area on the left side of vehicle A is 1.75-0.9-0.5=0.35 meters away from the center line of the lane. This means that the left side of vehicle A needs to maintain a safety distance of at least 0.35 meters to avoid a collision with the vehicle in the left lane when running parallel. The safety area on the right side of vehicle A is 1.8 / 2+0.5 meters from the right edge of the lane. That is, the right side of vehicle A needs to maintain a safety distance of at least 1.4 meters to deal with possible situations on the right side, such as other vehicles merging from the right, pedestrians or non-motor vehicles approaching, etc. The longitudinal safety boundary area and the lateral safety boundary area are combined to comprehensively determine the safety area corresponding to each vehicle.

[0078] In this process, the vehicle driving section is determined by the fleet location data and the preset road network data, and the driving scene is determined by combining the road type and vehicle flow to determine the collection cycle. In this way, spatial environment data can be collected at an appropriate frequency according to different road conditions such as highways, urban roads, etc. and traffic flow conditions such as congestion and smoothness. In addition, identifying the sudden change targets in the spatial environment data within the continuous collection cycle and determining their dynamic change data will help to timely discover objects that may affect the safety of the fleet, such as pedestrians who suddenly break into the lane, other vehicles that accelerate or change lanes, etc., and provide key information for subsequent safety boundary analysis and risk assessment. According to vehicle traffic data and multi-dimensional traffic margin data, the multi-dimensional safety factors of the vehicle are considered to determine a relatively reasonable safety area for each vehicle in the fleet to ensure the safety and smoothness of vehicle driving. The location data and dynamic change data corresponding to the first sudden change target and the second sudden change target are used as road traffic parameters of the safety area, which will help to achieve the control of vehicle formation in the future and reduce the incidence of traffic accidents.

[0079] S103: Inputting the road traffic parameters corresponding to each vehicle in the current fleet into a preset long short-term memory network to obtain traffic development prediction data corresponding to each vehicle.

[0080] In order to avoid the problem that the current process of adjusting vehicle fleets can only be based on current vehicle information and relies on the driver's personal driving experience, which makes it difficult to predict the next moment and leads to slow response when facing sudden road conditions, the road traffic parameters of the safety boundary area corresponding to each vehicle in the current fleet will be input into a pre-set long short-term memory network in the embodiment of this specification, so as to obtain the traffic development prediction data corresponding to each vehicle.

[0081] Specifically, in one or more embodiments of the present specification, the road traffic parameters corresponding to each vehicle in the current fleet are input into a preset long short-term memory network to obtain the traffic development prediction data corresponding to each vehicle, which specifically includes the following process:

[0082] According to the location information in the road traffic parameters of the vehicle safety area and the traffic flow direction of the current fleet, the safety area is divided to obtain the safety path grid corresponding to each of the vehicles. That is, assuming that the traffic flow direction is from south to north, the safety boundary area of ​​the vehicle may contain the front, rear, left and right boundary coordinates of the vehicle, and the area is divided into a plurality of equidistant small grids, which together constitute the safety path grid of the vehicle. Through such division, the impact of traffic elements at different positions on the vehicle can be better located and analyzed. For example, with vehicle 1 as the center, the safety areas in front, behind, and on the left and right sides are divided into a plurality of small grids similar to a chessboard. In the same way, according to the location information and traffic flow direction of the safety area of ​​vehicle 2, the safety path grid corresponding to vehicle 2 is divided, and the safety path grid corresponding to each vehicle in the current fleet is obtained by analogy.

[0083] Then, the position information of the sudden target in the road traffic parameters of the vehicle safety boundary area is used to connect the positions of these sudden targets with the safe path grid. The safe path grids that intersect or cover these lines are determined as the initial risk path grids. For example, in the road traffic parameters of the safety area of ​​vehicle 1, a sudden target is detected to be located to the left in front of the vehicle. A line is drawn from the position of this sudden target to the safe path grid of vehicle 1. Since the traffic conditions in these grid areas may pose potential risks to the vehicle, the safe path grid that intersects with the line is the initial risk path grid corresponding to the boundary grid position. This line passes through several grids to the left in front of vehicle 1, and these grids are determined as the initial risk path grid. Then, based on the position of the sudden target in the road traffic parameters of the vehicle safety boundary area and the line connecting the safe path grid, the corresponding initial risk path grid is determined.

[0084] According to the coordinate points of each vehicle in the safety area and the traffic flow direction of the current fleet, the grid type of the initial risk path grid is determined; wherein the grid type includes an upstream grid located in front of the vehicle and a downstream grid located behind the vehicle, because the situation in front of the vehicle, such as an obstacle that suddenly appears in front, usually has a greater impact on the subsequent driving of the vehicle, so this distinction helps to evaluate the grid according to different degrees of influence in the future. Then, based on the grid type of the initial risk path grid, the first weight value of the initial risk path grid is determined, and according to the distance of the initial risk path grid from the vehicle, the second weight value of the initial risk path grid is determined. By weighting the first weight value and the second weight value, the weight coefficient can be obtained. By normalizing the road traffic parameters of the safety boundary area corresponding to each vehicle in the current fleet, data of different ranges and dimensions can be unified into a standard range, thereby obtaining the processed road traffic parameters. Then, in order to convert the data into a format that can be received and processed by the long short-term memory network, the processed road traffic parameters are encoded according to the preset long short-term memory network input format, thereby obtaining the model input data. According to the correspondence between the initial risk path grid and the road traffic parameters, the road traffic parameters corresponding to the weight coefficients are determined, so that the model input data and the weight coefficients corresponding to the road traffic parameters are input into the preset long short-term memory network to obtain the traffic development forecast data corresponding to each vehicle.

[0085] Taking the location of the sudden change target into consideration and determining the initial risk path grid can help to timely discover potential risk factors that may affect vehicle driving and improve the ability to respond to emergencies. The first weight value is determined according to the upstream grid and the downstream grid of the grid type, and the second weight value is determined according to the distance from the vehicle, and the weight coefficient is obtained by weighted processing. In this way, the difference in risk levels at different locations and distances can be more reasonably reflected, making the obtained model input data more targeted and accurate. The road traffic parameters are normalized and encoded to conform to the input format of the long short-term memory network, giving full play to the advantages of the network in processing time series data, and can better capture the dynamic trend of traffic conditions, thereby more accurately predicting the development of traffic conditions for each vehicle, providing a more reliable decision-making basis for traffic management and vehicle traffic control.

[0086] S104: Determine the vehicle operation event corresponding to each vehicle according to the vehicle traffic data, so as to construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event.

[0087] In order to grasp the real-time operation data of each vehicle and quickly locate vehicles with key functions, such as vehicles at key nodes of the transportation route, vehicles closely related to multiple other vehicles, etc., the vehicle operation events corresponding to each vehicle will be determined according to the vehicle traffic data in the embodiments of this specification, so as to construct a vehicle association topology map based on the basic data of the vehicle and the matching relationship between the operation events of each vehicle. By constructing the vehicle association topology map, the association relationship between each vehicle can be clearly displayed, such as following, overtaking, parallel, etc., which provides support for the subsequent vehicle team control.

[0088] Specifically, in one or more embodiments of the present specification, the vehicle operation events corresponding to each vehicle are determined according to the vehicle traffic data, so as to construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event, which specifically includes the following processes:

[0089] The probability of each vehicle operation event occurring in the vehicle is determined according to the vehicle traffic data, so as to screen and determine the vehicle operation event corresponding to each vehicle according to the probability. For example, based on the naive Bayes classifier, the vehicle traffic data is classified and the probability of each category is output. For example, for a certain vehicle, the algorithm outputs the probability of normal driving as 0.8, the probability of sudden braking as 0.1, the probability of sudden acceleration as 0.05, and the probability of turning as 0.05. At this time, since the probability of normal driving is the highest at 0.8 and exceeds the set threshold of 0.75, it can be considered that the vehicle is currently in a normal driving state. Then, each vehicle node is determined according to the basic data of the vehicle, and the edges between each vehicle node are determined according to the association relationship between each vehicle operation event. Then, according to the edges between the vehicle node and each vehicle node, a vehicle association topology map is constructed. In this process, by calculating the probability of each vehicle operation event occurring in the vehicle, and screening and determining the vehicle operation event based on these probabilities, the actual operation status of the vehicle can be more accurately identified, which is more flexible and accurate than simple threshold judgment, and can capture more details and changes. By determining the vehicle nodes based on the basic vehicle data and determining the edges between the vehicle nodes based on the correlation between vehicle operation events, a vehicle correlation topology graph can be dynamically constructed. The vehicle correlation topology graph not only displays the vehicle nodes, but also represents the correlation between vehicles through edges. This comprehensive information display helps users better understand the operating laws of traffic flow and the interactions between vehicles.

[0090] S105: Determine the adjustable driving data of the current fleet based on the traffic development prediction data and the vehicle association topology map, adjust the driving planning data of the current fleet based on the adjustable driving data, and send the adjusted driving planning data to each vehicle through the on-board communication device of each vehicle, so as to realize intelligent control and coordinated driving of the current fleet.

[0091] After obtaining the traffic development prediction data and the vehicle association topology map based on the above steps, the adjustable driving data of the current fleet will be determined, so as to adjust the driving planning data of the current fleet according to the adjustable driving data, and send the adjusted driving planning data to each vehicle through the on-board communication equipment of each vehicle, so as to realize the intelligent control and coordinated driving of the current fleet. Traditional fleet driving planning is often difficult to cope with complex verification conditions in real time, but with the help of traffic development prediction data and intelligent adjustment of driving planning, possible dangerous traffic incidents can be effectively avoided. Intelligent control plans driving in advance according to the predicted data, avoiding dangerous behaviors such as sudden lane changes and emergency braking of vehicles under complex traffic conditions, and solving the safety hazard problem during the driving of the fleet.

[0092] Specifically, in one or more embodiments of the present specification, the adjustable driving data of the current fleet is determined according to the traffic development prediction data and the vehicle association topology map, specifically including:

[0093] According to the traffic development forecast data, the vehicle position and vehicle driving data of each vehicle in the next time period are predicted, and then the traffic conflict data and conflicting vehicles in the next time period of the current fleet are predicted according to the vehicle association topology map and the vehicle position and vehicle driving data of each vehicle in the next time period. For example, by analyzing the current traffic flow and vehicle speed, the trend of vehicle speed changes within a certain period of time can be predicted, and then the distance that the vehicle may travel in the next time period can be calculated. Combined with the current position information of the vehicle, the vehicle position of each vehicle in the next time period can be predicted. In addition, the predicted vehicle driving data may also include vehicle speed, acceleration, driving direction, etc. The predicted position and driving data of each vehicle in the next time period are mapped to the vehicle association topology map. By analyzing the dynamic changes of vehicles in the topology map, the situation where traffic conflicts may occur can be identified. For example, if the predicted positions of two or more vehicles in the next time period will be very close, and their driving directions may cause mutual interference such as driving in opposite directions or crossing driving, then it can be judged that these vehicles have the risk of traffic conflicts. At this time, the traffic conflict data may include the time and place of the conflict, the number of vehicles involved, the type of conflict, and the estimated severity of the conflict. Therefore, by quantifying the traffic conflict data, the risk value corresponding to the traffic conflict data can be determined, and then the first driving data to be adjusted corresponding to the conflict vehicle can be determined according to the risk value and the conflict type corresponding to the traffic conflict data, and the correlation between the first driving data to be adjusted corresponding to the conflict vehicle and the adjacent vehicles of the conflict vehicle can be determined at the same time, so as to determine the second driving data to be adjusted of the adjacent vehicles of the conflict vehicle according to the correlation. By determining the second driving data to be adjusted, it is ensured that the adjustment is not limited to the conflict vehicle itself, but from the perspective of overall coordination, the adjacent vehicles also make corresponding reasonable adjustments to ensure the smoothness and coordination of the operation of the entire fleet, and avoid local adjustments causing chain negative reactions. Then the first driving data to be adjusted and the second driving data to be adjusted are integrated to determine the adjustable driving data of the current fleet, that is, the adjustment needs of the conflict vehicle and its adjacent vehicles are comprehensively considered, and an overall adjustment strategy covering all relevant vehicles in the fleet is generated, so that the adjusted driving plan can not only solve the current traffic conflict, but also maintain the overall operation order of the fleet, and realize the systematic and holistic intelligent control of the current fleet.

[0094] Specifically, in one or more embodiments of the present specification, sending the adjusted driving plan data to each vehicle through the vehicle-mounted communication device of each vehicle specifically includes:

[0095] Different on-board communication devices may have specific transmission format requirements. In order to ensure that the adjusted driving planning data can be correctly received and processed by the on-board communication device, it is necessary to encode it according to its requirements. Therefore, in the embodiment of this specification, according to the transmission format requirements of the on-board communication device, the adjusted driving planning data is encoded to obtain the data to be transmitted, and the data to be transmitted is encapsulated based on the communication protocol of the on-board communication device to obtain the encapsulated data to be transmitted. Then the central control unit uses its own communication function to interact with the on-board communication device of each vehicle, completes a series of operations such as handshake according to the predetermined communication protocol process, thereby establishing a reliable communication connection, and then sends the encapsulated data to be transmitted based on the communication connection. The data is transmitted to the corresponding vehicle bit by bit or packet by packet through a wireless or wired communication link in the manner specified by the communication protocol. After receiving the encapsulated data to be transmitted, the on-board communication device of each vehicle will send the return data to the central control unit according to the preset program. The return data contains information about the data reception situation, such as whether it is successfully received, whether the data is complete, whether there is an error, etc. The central control unit determines whether the encapsulated data to be transmitted is successfully sent based on the return data received from each vehicle in the current fleet. If the feedback data shows that all vehicles have received the data correctly, then the data can be considered to have been sent successfully; if the feedback data of a certain vehicle shows reception abnormalities such as incorrect verification code, missing data, etc., the central control unit may initiate a retransmission mechanism and send data to the vehicle again to ensure that each vehicle can accurately obtain the adjusted driving plan data, thereby achieving effective intelligent control of the entire fleet.

[0096] like Figure 3 As shown, in one or more embodiments of this specification, a structural schematic diagram of a vehicle platoon control device based on artificial intelligence is provided. Figure 3 It can be seen that in one or more embodiments of this specification, a vehicle platoon control device based on artificial intelligence includes:

[0097] at least one processor; and,

[0098] a memory communicatively connected to the at least one processor; wherein,

[0099] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform any of the above methods.

[0100] like Figure 4 As shown, the present specification provides a schematic diagram of the structure of a non-volatile storage medium. Figure 4As shown, in one or more embodiments of the present specification, a non-volatile storage medium stores computer executable instructions 401, and the computer executable instructions 401 can execute any of the methods described above.

[0101] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0102] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A vehicle platoon control method based on artificial intelligence, characterized in that: The method comprises: Based on the on-board communication equipment and pre-installed on-board laser radar of each vehicle in the current fleet, the vehicle data and spatial environment data of each vehicle are obtained; wherein the vehicle data includes: basic vehicle data and vehicle traffic data; Determine the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet according to the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet; Inputting the road traffic parameters corresponding to each vehicle in the current fleet into a preset long short-term memory network to obtain traffic development prediction data corresponding to each vehicle; Determine the vehicle operation event corresponding to each vehicle according to the vehicle traffic data, so as to construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event; According to the traffic development prediction data and the vehicle association topology map, the adjustable driving data of the current fleet is determined, so as to adjust the driving planning data of the current fleet based on the adjustable driving data, and the adjusted driving planning data is sent to each vehicle through the on-board communication equipment of each vehicle, so as to realize intelligent control and coordinated driving of the current fleet.

2. The vehicle platoon control method based on artificial intelligence according to claim 1, characterized in that: Based on the on-board communication equipment and pre-installed on-board laser radar of each vehicle in the current fleet, the vehicle data and spatial environment data of each vehicle are obtained, including: Determine the basic vehicle data of each vehicle in the current fleet based on the vehicle-mounted communication equipment of each vehicle, and obtain the initial vehicle traffic data and vehicle monitoring point cloud of each vehicle according to the preset vehicle-mounted laser radar of each vehicle; wherein the initial vehicle traffic data at least includes: vehicle speed, vehicle acceleration, vehicle steering angle, and vehicle braking; preprocessing the initial vehicle traffic data of each vehicle based on a preset processing strategy to obtain processed vehicle data; Extracting the vehicle monitoring point cloud based on a preset collection interval to obtain a plurality of vehicle point cloud images of the vehicle; Comparing the vehicle point cloud images in sequence to use the mutation point cloud images in the vehicle point cloud images as key point cloud images; Acquire multiple adjacent point cloud images of the key point cloud image, and based on the time sequence of the key point cloud image and the adjacent point cloud images, sequentially obtain differential point cloud images and use the connected areas of the differential point cloud images as mutation targets, and annotate the mutation targets in the key point cloud images to obtain spatial environment data; wherein the mutation targets include: static mutation targets and dynamic mutation targets.

3. The vehicle platoon control method based on artificial intelligence according to claim 2 is characterized in that: The preprocessing of the initial vehicle traffic data of each vehicle based on a preset processing strategy to obtain processed vehicle data specifically includes: Acquire attribute information corresponding to each of the initial vehicle traffic data; wherein the attribute information includes: status transmission time, status occurrence location and status data value; Arranging the condition data values ​​based on the condition occurrence time and the condition occurrence location corresponding to each of the initial vehicle traffic data to obtain a first condition data value and a second condition data value, and using the first condition data value and the second condition data value corresponding to the same vehicle traffic data as a vehicle operation event of the corresponding vehicle to determine a plurality of vehicle operation events of each of the vehicles; Based on the first status data value and the second status data value corresponding to each vehicle operation event and the matching relationship with other vehicle operation events, determine the companion vehicles corresponding to each vehicle operation event of the vehicle, so as to determine the companion relationship topology corresponding to each of the vehicles; Based on the preset behavior table, the developable conditions corresponding to each condition data value are determined, and based on the condition data value, the developable conditions and the companion relationship topology map, the probability of each vehicle operation event occurring to the vehicle is determined, and the initial condition data corresponding to the vehicle operation event is filtered based on the preset probability to obtain processed vehicle traffic data.

4. The vehicle platoon control method based on artificial intelligence according to claim 2, characterized in that: Determining the collection period corresponding to the vehicle data, so as to determine the road traffic parameters of the safety area corresponding to each vehicle in the current fleet according to the spatial environment data in the continuous collection period and the safe passage data information corresponding to the current fleet, specifically including: Determine the vehicle driving section corresponding to the current vehicle fleet according to the location data of the current vehicle fleet and the preset road network data, determine the driving scene corresponding to the current vehicle fleet according to the road type and vehicle flow of the vehicle driving section, and determine the corresponding collection period based on the driving scene; In a continuous collection cycle, the sudden change target in the collected spatial environment data is identified to determine the dynamic change data of the sudden change target according to the position data of the sudden change target in the adjacent spatial environment data; wherein the dynamic change data includes: moving speed, speed direction and moving trajectory; Determine the safety area corresponding to each vehicle according to the vehicle traffic data and multi-dimensional traffic margin data of each vehicle in the current fleet; Determine a first mutation target located in the safety area according to the position data of the mutation target, and determine a second mutation target close to the safety area and matching a specified critical condition according to the position data of the mutation target and the dynamic change data; The position data corresponding to the first sudden change target and the second sudden change target, the dynamic change data and the road data of the driving scene are used as road traffic parameters of the safe area.

5. The vehicle platoon control method based on artificial intelligence according to claim 1, characterized in that: The road traffic parameters corresponding to each vehicle in the current fleet are input into a preset long short-term memory network to obtain traffic development prediction data corresponding to each vehicle, specifically including: According to the position information in the road traffic parameters of the vehicle safety area and the traffic flow direction of the current fleet, the safety area is divided to obtain the safety path grid corresponding to each of the vehicles; Determine an initial risk path grid corresponding to the safety area according to the position of the sudden change target in the road traffic parameter of the vehicle safety area and the connection line of the safety path grid; Determining the grid type of the initial risk path grid according to the coordinate points of each of the vehicles in the safety area and the traffic flow direction of the current fleet; wherein the grid type includes an upstream grid located in front of the vehicle and a downstream grid located behind the vehicle; Determining a first weight value of the initial risk path grid based on a grid type of the initial risk path grid, determining a second weight value of the initial risk path grid according to a distance of the initial risk path grid from the vehicle, and performing weighted processing on the first weight value and the second weight value to obtain a weight coefficient; According to the corresponding relationship between the initial risk path grid and the road traffic parameters, the road traffic parameters corresponding to the weight coefficients are determined, so that the road traffic parameters and the weight coefficients corresponding to the road traffic parameters are input into the preset long short-term memory network to obtain the traffic development prediction data corresponding to each vehicle.

6. The vehicle platoon control method based on artificial intelligence according to claim 3 is characterized in that: Determine the vehicle operation event corresponding to each vehicle according to the vehicle traffic data, and construct a vehicle association topology map according to the matching relationship between the vehicle basic data and each vehicle operation event, specifically including: Determining the probability of each vehicle operation event occurring to the vehicle according to the vehicle traffic data, so as to screen and determine the vehicle operation event corresponding to each vehicle according to the probability; Determine each vehicle node according to the basic vehicle data, and determine the edges between each vehicle node according to the association relationship between each vehicle operation event; A vehicle association topology graph is constructed based on the edges between the vehicle node and each vehicle node.

7. The vehicle platoon control method based on artificial intelligence according to claim 1, characterized in that: Determining adjustable driving data of the current fleet according to the traffic development prediction data and the vehicle association topology map, specifically including: Predicting the vehicle position and vehicle travel data of each vehicle in the next time period according to the traffic development prediction data; Predicting traffic conflict data and conflicting vehicles in the next time period of the current fleet based on the vehicle association topology map and the vehicle positions and vehicle driving data of each vehicle in the next time period; Quantitatively processing the traffic conflict data to determine a risk value corresponding to the traffic conflict data, and determining first to-be-adjusted driving data corresponding to the conflicting vehicle according to the risk value and the conflict type corresponding to the traffic conflict data; Determine the association relationship between the first to-be-adjusted driving data corresponding to the conflicting vehicle and the adjacent vehicle of the conflicting vehicle, so as to determine the second to-be-adjusted driving data of the adjacent vehicle of the conflicting vehicle according to the association relationship; The first driving data to be adjusted and the second driving data to be adjusted are integrated to determine adjustable driving data of the current vehicle fleet.

8. The vehicle platoon control method based on artificial intelligence according to claim 1, characterized in that: The adjusted driving plan data is sent to each vehicle through the vehicle-mounted communication device of each vehicle, specifically including: According to the transmission format requirements of the vehicle-mounted communication device, the adjusted driving plan data is encoded to obtain data to be transmitted, and the data to be transmitted is encapsulated based on the communication protocol of the vehicle-mounted communication device to obtain encapsulated data to be transmitted; Calling the vehicle-mounted communication equipment of each vehicle through the central control unit corresponding to the current fleet, establishing a communication connection with each vehicle, and sending the encapsulated data to be transmitted based on the communication connection; According to the feedback data of each vehicle in the current fleet, it is determined whether the packaged data to be transmitted has been sent.

9. A vehicle platoon control device based on artificial intelligence, comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any one of the methods described in claims 1-8.

10. A non-volatile storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 8.

Images