Cooperative vehicle data correction method for V2X

Abstract

The present invention discloses a cooperative vehicle data correction method for V2X, including: the vehicle regularly collects its own driving state information and sends it to the roadside unit; the vehicle predicts, judges the rationality and corrects its own driving state data at the current moment based on its own driving state information at the previous moment; the roadside unit is trained based on the historical vehicle driving state information within its management scope, judges the rationality of the current vehicle driving state information and corrects it; the vehicle compares and integrates the rationality judgment result and corrected information obtained by itself with the rationality judgment result and corrected information of the roadside unit, and obtains the current moment vehicle driving state information after cooperative correction. The vehicle data correction method can simultaneously perform real-time detection and correction of the driving state information generated by the vehicle itself and the roadside unit, thereby realizing the detection and correction of vehicle information in dense traffic scenes in cities, and ensuring road safety.

Description

Technical Field

[0001] The present invention relates to the field of vehicle networking information communication technology, and in particular provides a V2X-oriented cooperative vehicle data correction method. Background Art

[0002] As an important component of the transportation field, safe and high-quality networked communication of vehicles is a solid foundation for realizing intelligent transportation. V2X can realize information exchange between vehicles and all entities that may affect vehicles. The purpose is to use communication between entities to reduce accidents, alleviate traffic congestion and ensure road safety. The transmission of information in V2X can be achieved using hardware devices such as on-board units (OBU) and roadside units (RSU). Since the communication between vehicles in V2X uses wireless links, and the information reported by vehicles on the road may be erroneous due to unpredictable reasons, the basic safety information (BSM) exchanged between vehicles to ensure vehicles is unreliable.

[0003] In the known technologies in this field, researchers mostly aim to identify the sent error information. According to the timing and method of information verification by researchers, the algorithms they proposed can be roughly divided into the following categories: 1. Check the information at the physical layer: use the propagation law of physical signals in space to check whether the vehicle reports the real position; 2. Check the rationality of information at the application layer: use the rationality and consistency of vehicle driving information to check it; 3. Use machine learning algorithms at the application layer to check vehicle information and use the self-learning ability of machine learning algorithms to find the sent error information.

[0004] Existing technologies usually focus on individual vehicles, and only perform simple verification or testing after obtaining data to verify whether the data exists, or diagnose the data to find fault factors in the vehicle. This has limited practical effect on vehicles that are currently driving, and cannot correct the data of vehicles traveling on the road in real time to ensure that the vehicle can generate accurate driving data in real time and improve road safety. Summary of the invention

[0005] In view of this, an object of the present invention is to provide a V2X-oriented cooperative vehicle data correction method to solve the problem that the accuracy of existing vehicle information cannot be guaranteed.

[0006] The technical solution provided by the present invention is: a cooperative vehicle data correction method for V2X, comprising the following steps:

[0007] S1: The vehicle regularly collects its own driving status information and packages the driving status information together with time information and vehicle ID information and sends it to the roadside unit that manages the area where the vehicle is located, wherein the driving status information includes the vehicle's location information, speed information, heading angle information and acceleration information;

[0008] S2: The vehicle predicts its current driving state data based on its previous driving state information, and then makes a rationality judgment on the collected current driving state based on the prediction result and corrects unreasonable data;

[0009] S3: The roadside unit is trained based on the historical vehicle driving state information within its management range, determines the rationality of the currently received vehicle driving state information, and corrects the unreasonable data in the current driving state of the vehicle based on the vehicle's previous driving state information, and then returns the rationality judgment result and the corrected vehicle driving state information to the original vehicle;

[0010] S4: The vehicle compares and integrates its own rationality judgment result and corrected driving status information with the rationality judgment result and corrected driving status information of the roadside unit, obtains the current driving status information of the vehicle after cooperative correction, and sends it to the roadside unit and surrounding neighbors.

[0011] Preferably, S2: specifically comprises the following steps:

[0012] S21: The vehicle predicts its current driving state data based on its previous driving state information;

[0013] S22: Compare the predicted driving state with the collected driving state at the current moment. If the data difference of the corresponding attribute is greater than a preset threshold, determine that the data of the corresponding attribute in the collected driving state information at the current moment is unreasonable data, and use the data of the corresponding attribute in the predicted driving state information to correct the data of the corresponding attribute in the vehicle driving state information at the current moment.

[0014] Further preferably, a reinforcement learning algorithm is deployed on the roadside unit to learn past experience learned in driving status information judgment and apply it to current real-time information judgment.

[0015] Further preferably, the reinforcement learning algorithm is a one-step Q-learning algorithm.

[0016] Further preferably, in S4, the vehicle compares and integrates the rationality judgment result and the corrected driving state information obtained by itself with the rationality judgment result and the corrected driving state information of the roadside unit to obtain the current moment vehicle driving state information after cooperative correction as follows:

[0017] Determine whether the rationality judgment result obtained by the vehicle itself is the same as the rationality judgment result of the roadside unit. If they are the same, take the average value of the corresponding attribute data in the corrected driving state information of the two as the final corrected data of the attribute. Otherwise, select the corresponding attribute data of the two with the smaller difference with the corresponding attribute in the driving state information of the vehicle at the previous moment as the final corrected data of the attribute.

[0018] The V2X-oriented cooperative vehicle data correction method provided by the present invention adopts a cooperative data detection and correction scheme. The vehicle node and the roadside unit are used to process the vehicle's own data and all vehicle data within the control range respectively. The vehicle can obtain the position, speed, heading angle and acceleration information of its own node, and can judge the correctness of the current information and correct it through its own information at the last moment. In combination with the currently received vehicle information and the historical vehicle information within the control range, the roadside unit can judge the correctness of the vehicle information and correct it. By using the collaborative work of the vehicle node itself and the roadside unit, a vehicle driving parameter model can be constructed, thereby realizing vehicle data correction in a high-density vehicle environment in the city, and ensuring vehicle driving safety and road safety in a high-density vehicle environment. DETAILED DESCRIPTION

[0019] The present invention will be further explained below in conjunction with specific implementation schemes, but the present invention is not limited thereto.

[0020] The application environment of the present invention is as follows: vehicle nodes traveling on a highway network are connected through wireless short-range communication technology, and data packets are transmitted between vehicle nodes and between vehicle nodes and roadside units (RSUs) within their range. The main content of the data packets is the driving status information of the vehicle. The communication between vehicles and between vehicles and RSUs is wireless.

[0021] The present invention provides a V2X-oriented cooperative vehicle data correction method, comprising the following steps:

[0022] S1: The vehicle periodically collects its own driving state information and encapsulates the driving state information together with time information and vehicle ID information and sends it to a roadside unit (RSU) that manages the area where the vehicle is located, wherein the driving state information includes the vehicle's position information, speed information, heading angle information and acceleration information. In this application, the position information, speed information, heading angle information and acceleration information are referred to as attributes of the driving state information;

[0023] S2: The vehicle predicts its current driving state data based on its previous driving state information, and then makes a rationality judgment on the collected current driving state based on the prediction result and corrects unreasonable data;

[0024] S3: The roadside unit (RSU) is trained based on the historical vehicle driving state information within its management range, determines the rationality of the currently received vehicle driving state information, and corrects the unreasonable data in the current driving state of the vehicle based on the vehicle's previous driving state information, and then returns the rationality judgment result and the corrected vehicle driving state information to the original vehicle;

[0025] S4: The vehicle compares and integrates its own rationality judgment results and corrected driving status information with those of the roadside unit (RSU), obtains the current vehicle driving status information after cooperative correction, and sends it to the roadside unit (RSU) and surrounding neighbors.

[0026] This V2X-oriented cooperative vehicle data correction method adopts a cooperative data detection and correction scheme. The vehicle node and roadside unit are used to process the vehicle's own data and all vehicle data within the control range respectively. The vehicle can obtain the position, speed, heading angle and acceleration information of its own node. The vehicle can judge the correctness of the current information and correct it through its own information at the last moment. Combined with the currently received vehicle information and the historical vehicle information within the control range, the roadside unit can judge the correctness of the vehicle information and correct it. By using the collaborative work of the vehicle node itself and the roadside unit, a vehicle driving parameter model can be constructed, thereby realizing vehicle data correction in a high-density vehicle environment in the city, ensuring vehicle driving safety and road safety in a high-density vehicle environment.

[0027] Specifically, the vehicle node in S1 uses sensors to continuously capture the vehicle's position, speed, heading angle and acceleration information during driving, and encapsulates the captured information together with the information acquisition time and vehicle ID into the vehicle's basic safety information BSM. Therefore, a BSM message contains timestamp, ID, lat, long, v, acc and angle, where timestamp is the timestamp sent by the BSM, that is, the time when each sensor in the vehicle captures the sensor data, ID is the vehicle ID, lat is the current latitude of the vehicle, long is the current longitude of the vehicle, v is the speed of the vehicle, acc is the acceleration of the vehicle, and angle is the heading angle of the vehicle. The vehicle will save its own BSM at the previous moment and the current moment to judge the rationality and correct the driving status information. At the same time, the vehicle will also send the BSM information to the RSU, and the RSU will judge the correctness of the information and make corresponding corrections. Among them, each attribute of the BSM can be distinguished by adding a time information subscript to the BSM at different times, such as using v t represents the speed of the vehicle at time t, v t-1 represents the speed of the vehicle at time t-1.

[0028] S2: The vehicle predicts its current driving state data based on its previous driving state information, and then makes a rationality judgment on the collected current driving state based on the prediction result and corrects unreasonable data. The specific steps include the following:

[0029] S21: The vehicle predicts its current driving state data based on its previous driving state information;

[0030] S22: comparing the predicted driving state with the collected driving state at the current moment, if the data difference of the corresponding attribute is greater than a preset threshold, determining that the data of the corresponding attribute in the collected driving state information at the current moment is unreasonable data, and using the data of the corresponding attribute in the predicted driving state information to correct the data of the corresponding attribute in the vehicle driving state information at the current moment;

[0031] Among them, the specific process of S2 is as follows:

[0032] According to industry standards, the frequency of sending basic vehicle safety information (BSM) is 10Hz. Under the maximum speed limit of 60km / h on urban roads, the movement of the vehicle between each message transmission can be simplified to uniformly accelerated linear motion. The vehicle's position prediction information at time t is It is calculated as follows:

[0033]

[0034] Among them, v t-1 *|sin(angle t-1 )|、v t-1 *|cos(angle t-1 )| are the speeds of the vehicle along the latitude and longitude directions, a t-1 *|sin(angle t-1 )|、a t-1 *|cos(angle t-1 )| are the acceleration of the vehicle in the latitude and longitude directions respectively, and Δt is the time interval between two messages sent by the vehicle, which is 0.1 seconds.

[0035] The difference (i.e., distance) between the predicted position and the actual reported position is calculated using formula (2). If the difference between the two points is less than the set threshold, that is, less than the positioning error, the vehicle is considered to have reported the correct position information. Otherwise, the vehicle is considered to have reported the wrong position information and the predicted position information is used instead of the reported information to complete the position correction.

[0036]

[0037] The predicted speed and acceleration are calculated according to formula (3) and formula (4) respectively. Similar to the judgment of position information, if the difference between the predicted speed (acceleration) and the speed (acceleration) information captured by the sensor is greater than the set threshold, it is considered that the BSM contains wrong speed (acceleration) information, and the predicted speed (acceleration) information is used instead of the reported information to complete the correction of speed (acceleration).

[0038]

[0039]

[0040] The roadside unit RSU has greater computing power than the vehicle, so more complex calculations can be performed to determine the correctness of the BSM sent by the vehicle.

[0041] In S3, a reinforcement learning algorithm is deployed on the roadside unit RSU to learn from the past experience in BSM judgment and apply it to the current real-time information judgment. This not only makes fuller use of the computing resources of the roadside unit RSU, but also effectively combines the historical driving status of vehicles within the management range of the roadside unit RSU to more accurately determine whether the vehicle has sent a BSM containing erroneous information, that is, to determine the rationality of the vehicle's driving status information and correct the unreasonable data in the vehicle's current driving status.

[0042] Among them, the reinforcement learning algorithm can adopt a one-step Q-learning algorithm, which is expressed as follows:

[0043]

[0044] By setting a higher learning rate α and a lower discount factor γ, the one-step Q-learning algorithm can better focus on dynamic traffic changes and learn real-time vehicle status information.

[0045] Specifically: according to the distance between the vehicle and the center line of the road (less than 1 / 2 lane width, greater than 1 / 2 lane width), vehicle speed (low speed: 0-40km / h, normal speed: 40-60km / h, speeding: >60km / h), vehicle acceleration (<-4.6m / s 2 、-4.6~3.4m / s 2 、>3.4m / s 2 ) and the vehicle’s heading angle (0°~90°, 90°~180°, 180°~270°, 270°~360°) divide the state space in the Q-learning algorithm into the form of Table 1.

[0046] Table 1 State space

[0047]

[0048] There are two set action spaces, no-action means that there is no error information in the received BSM; correct means that the received BSM contains error information and needs to be corrected. Different actions have different corresponding reward functions. When no-action is selected, it means that the roadside unit RSU believes that the received BSM is correct. The reward function at this time is related to the BSM at the previous moment and the BSM currently received. If the attributes of the two received BSMs meet the rationality prediction (consistent with the vehicle's own prediction method), flag1=1 is marked, and the reward value is 1; otherwise, the RSU made a wrong judgment, flag1=0 is marked, and the reward value is -1. When correct is selected, it means that the RSU believes that the received BSM contains wrong information. At this time, the BSM information of the previous moment is used to make a reasonable prediction of the information at the current moment (consistent with the vehicle's own correction method). If the difference between the attributes of the corrected BSM information (recorded as BSM') and the received BSM information is within the set threshold range, it is considered that the RSU has made an incorrect judgment, and flag2 is marked as 0, and the reward value is -1; otherwise, it is considered that the RSU has made a correct judgment, and flag2 is marked as 1, the reward value is 1, and the corrected BSM is BSM'. The reward functions corresponding to the no-action action and the correct action are shown in formulas (6) and (7), respectively.

[0049]

[0050]

[0051] S4: The vehicle compares and integrates the rationality judgment result and corrected driving state information obtained by itself with the rationality judgment result and corrected driving state information of the roadside unit (RSU) to obtain the current driving state information of the vehicle after cooperative correction. The process is as follows:

[0052] Determine whether the rationality judgment result obtained by the vehicle itself is the same as the rationality judgment result of the roadside unit RSU. If they are the same, take the average value of the corresponding attribute data in the corrected driving state information of the two as the final corrected data of the attribute. Otherwise, select the corresponding attribute data of the two with the smaller difference with the corresponding attribute in the driving state information of the vehicle at the previous moment as the final corrected data of the attribute.

[0053] The correctness judgment and correction results of the BSM information generated at the current moment by the vehicle and the roadside unit RSU will be returned to the original vehicle. The original vehicle will decide whether the currently generated BSM information contains unreasonable data and save the final result in the vehicle itself and broadcast it to surrounding neighbors and RSU.

[0054] The detailed description of the present invention is written in a progressive manner, emphasizing the differences between the various implementations, and similar parts thereof can be referred to each other.

[0055] The embodiments of the present invention are described in detail above, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A cooperative vehicle data correction method for V2X, characterized in that: The steps include: S1: The vehicle regularly collects its own driving status information and packages the driving status information together with time information and vehicle ID information and sends it to the roadside unit that manages the area where the vehicle is located, wherein the driving status information includes the vehicle's location information, speed information, heading angle information and acceleration information; S2: The vehicle predicts its current driving state data based on its previous driving state information, and then makes a rationality judgment on the collected current driving state based on the prediction result and corrects unreasonable data; S3: The roadside unit is trained based on the historical vehicle driving state information within its management range, determines the rationality of the currently received vehicle driving state information, and corrects the unreasonable data in the current driving state of the vehicle based on the vehicle's previous driving state information, and then returns the rationality judgment result and the corrected vehicle driving state information to the original vehicle; S4: The vehicle compares and integrates the rationality judgment result and corrected driving state information obtained by itself with the rationality judgment result and corrected driving state information of the roadside unit, obtains the current driving state information of the vehicle after cooperative correction, and sends it to the roadside unit and surrounding neighbors; A reinforcement learning algorithm is deployed on the roadside unit to learn the experience learned in the past in driving state information judgment and apply it to the current real-time information judgment; the reinforcement learning algorithm is a one-step Q-learning algorithm.

2. The V2X-oriented cooperative vehicle data correction method according to claim 1, characterized in that: S2: specifically includes the following steps: S21: The vehicle predicts its current driving state data based on its previous driving state information; S22: Compare the predicted driving state with the collected driving state at the current moment. If the data difference of the corresponding attribute is greater than a preset threshold, determine that the data of the corresponding attribute in the collected driving state information at the current moment is unreasonable data, and use the data of the corresponding attribute in the predicted driving state information to correct the data of the corresponding attribute in the vehicle driving state information at the current moment.

3. The V2X-oriented cooperative vehicle data correction method according to claim 1, characterized in that: In S4, the vehicle compares and integrates the rationality judgment result and corrected driving state information obtained by itself with the rationality judgment result and corrected driving state information of the roadside unit to obtain the current driving state information of the vehicle after cooperative correction as follows: Determine whether the rationality judgment result obtained by the vehicle itself is the same as the rationality judgment result of the roadside unit. If they are the same, take the average value of the corresponding attribute data in the corrected driving state information of the two as the final corrected data of the attribute. Otherwise, select the corresponding attribute data of the two with the smaller difference with the corresponding attribute in the driving state information of the vehicle at the previous moment as the final corrected data of the attribute.

Images