A map matching method, apparatus and device

Abstract

The application provides a map matching method, device and equipment, and relates to the technical field of Internet of Vehicles, and the method comprises the following steps: filtering vehicle position data in acquired vehicle state data by using a preset Kalman filtering model to obtain historical trajectory point position data of a vehicle; generating a vehicle trajectory fitting curve according to the historical trajectory point position data; generating road fitting curves of multiple roads and lane fitting curves of each lane in each road according to map information broadcast by a road side unit (RSU); and obtaining a map matching result of the vehicle according to the vehicle trajectory fitting curve, the road fitting curves of the multiple roads and the lane fitting curves of each lane. According to the application, the vehicle position data is filtered by using a Kalman filtering algorithm, the stability and reliability of the vehicle position data are increased, the vehicle position error is reduced, and the accuracy of map matching is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Vehicles, and in particular to a map matching method, device and equipment. BACKGROUND

[0002] The vehicle-road cooperative system based on Internet of Vehicles is an important application for developing intelligent traffic in the future, which can not only improve the safety driving and communication efficiency of traffic, reduce traffic congestion and traffic accidents in special sections or special periods, but also empower automatic driving and smart city, making traffic more intelligent and safe. In the Internet of Vehicles environment, not only basic information that can be collected by traditional detectors can be obtained, but also more updated, more various and more accurate vehicle-road related information can be obtained through vehicle-road communication. The positioning accuracy of traditional vehicle navigation is generally low. Under normal and non-interference conditions, the error of the civilian vehicle-mounted Global Positioning System (GPS) locator is usually within 10-20 meters. When the vehicle is in an environment with interference such as urban valley, underground garage and tunnel, the vehicle-mounted GPS locator will have greater error. Therefore, the traditional GPS navigation cannot provide accurate map matching results, so as to provide more accurate navigation suggestions. Therefore, how to give full play to the technical advantages of Vehicle to Everything (V2X) to provide more accurate map matching results and provide more accurate navigation suggestions for drivers has become a hot research problem.

[0003] Map matching refers to how to match the recorded geographic coordinates to the logical model of the real world, associate the ordered GPS positions of the vehicle to the electronic road network, and convert the position sampling sequence under the GPS coordinates to the road network coordinate sequence.

[0004] V2X is a communication method for information exchange between vehicles and the outside world, including direct communication between vehicles (Vehicle to Vehicle, V2V), communication between vehicles and pedestrians (Vehicle to Pedestrian, V2P), communication between vehicles and road infrastructure (Vehicle to Infrastructure, V2I), and communication between vehicles and cloud through mobile networks (Vehicle to Network, V2N). V2X is a key technology for future intelligent transportation systems. Through V2X technology, a series of traffic information such as real-time road condition information, road information and pedestrian information can be effectively obtained, so as to improve driving safety, reduce congestion and improve traffic efficiency.

[0005] The existing map matching method based on V2X technology often considers using comprehensive information, usually matching the entire vehicle trajectory with the road network, considering the predecessors and successors of the vehicle position points, and when the vehicle position update frequency is low and there is certain noise interference, the map matching method also uses a probability algorithm. The probability algorithm makes an explicit range division of the noise of the GPS position points by considering the probability of the GPS position points, and considers multiple possible paths through the road network to find the best matching path. However, in the existing map matching method based on V2X technology, there are still problems of large vehicle position error and low map matching accuracy. SUMMARY

[0006] Embodiments of the present application provide a map matching method, device and equipment to solve the problem of large vehicle position error and low map matching accuracy in the existing map matching method.

[0007] To solve the above technical problems, embodiments of the present application provide the following technical solutions:

[0008] The embodiment of the present application provides a map matching method, comprising:

[0009] The vehicle position data in the acquired vehicle state data is filtered by using a preset Kalman filter model to obtain historical trajectory point position data of the vehicle;

[0010] A vehicle trajectory fitting curve is generated according to the historical trajectory point position data;

[0011] According to the map information broadcast by the road side unit RSU, a road fitting curve of a plurality of roads and a lane fitting curve of each lane in each road are generated;

[0012] The map matching result of the vehicle is obtained according to the vehicle trajectory fitting curve, the road fitting curve of the plurality of roads and the lane fitting curve of each lane.

[0013] The embodiment of the present application also provides a map matching device, comprising:

[0014] The first processing module is configured to filter the vehicle position data in the acquired vehicle state data by using a preset Kalman filter model to obtain historical trajectory point position data of the vehicle;

[0015] The first curve generation module is configured to generate a vehicle trajectory fitting curve according to the historical trajectory point position data;

[0016] The second curve generation module is configured to generate a road fitting curve of a plurality of roads and a lane fitting curve of each lane in each road according to the map information broadcast by the road side unit RSU;

[0017] The second processing module is used to obtain the map matching result of the vehicle based on the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curve of each lane.

[0018] This invention also provides a map matching device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the map matching method as described above.

[0019] This invention also provides a readable storage medium storing a program that, when executed by a processor, implements the steps of the map matching method as described above.

[0020] The beneficial effects of this invention are:

[0021] The present invention utilizes a preset Kalman filter model to filter vehicle position data in acquired vehicle state data to obtain historical trajectory point position data of the vehicle. Specifically, the Kalman filter algorithm is used to filter the vehicle position data, increasing its stability and reliability and reducing position errors. Based on the historical trajectory point position data, a vehicle trajectory fitting curve is generated. Based on map information broadcast by the Roadside Unit (RSU), road fitting curves for multiple roads and lane fitting curves for each lane on each road are generated. Based on the vehicle trajectory fitting curve, the road fitting curves for the multiple roads, and the lane fitting curves for each lane, the map matching result for the vehicle is obtained, improving the accuracy of map matching. Attached Figure Description

[0022] Figure 1 A flowchart illustrating the map matching method provided in an embodiment of the present invention;

[0023] Figure 2 This is a schematic diagram illustrating the change in vehicle position before and after according to an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of the structure of the map matching device provided in an embodiment of the present invention;

[0025] Figure 4 This is a schematic diagram illustrating the structure of the map matching device provided in an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] This invention addresses the problems of large vehicle position errors and low map matching accuracy in existing map matching methods by providing a map matching method, apparatus, and device.

[0028] like Figure 1 As shown, an embodiment of the present invention provides a map matching method, including:

[0029] Step 101: Use a preset Kalman filter model to filter the vehicle position data in the acquired vehicle state data to obtain the historical trajectory point position data of the vehicle.

[0030] This invention provides a map matching system, including an onboard unit (OBU) located inside a vehicle and a roadside unit (RSU) located on the side of a road. The roadside unit (RSU) is primarily responsible for broadcasting map (MAP) road information, while the onboard unit (OBU) is primarily responsible for acquiring vehicle status information (i.e., vehicle status data), data processing, and map matching.

[0031] The aforementioned MAP road information is broadcast by the roadside unit (RSU) to transmit local area map information to vehicles. The local area map information includes intersection information (node), road segment information (link), lane information (lane), and the connection relationships between roads.

[0032] Furthermore, the on-board unit (OBU) includes at least a positioning module, an information processing module, and a map matching module.

[0033] In this step, the positioning module of the on-board unit (OBU) acquires vehicle status information in real time and provides the vehicle status information to the information processing module of the OBU. The information processing module uses a preset Kalman filter model to perform Kalman filtering on the vehicle position data in the vehicle status information to obtain the vehicle's historical trajectory point position data.

[0034] Since the accuracy of in-vehicle map matching navigation is low, it is difficult to provide accurate business applications. This embodiment of the invention uses the Kalman filtering algorithm to process the vehicle location data, thereby increasing the accuracy and reliability of the vehicle location data.

[0035] Step 102: Generate a vehicle trajectory fitting curve based on the historical trajectory point location data.

[0036] In this step, the information processing module of the on-board unit (OBU) uses the least squares method to fit the historical trajectory point location data generated in step 101 into a cubic curve, namely the vehicle trajectory fitting curve, based on the historical trajectory point location data.

[0037] Step 103: Based on the map information broadcast by the Roadside Unit (RSU), generate road fitting curves for multiple roads and lane fitting curves for each lane in each road.

[0038] In this step, the information processing module of the on-board unit (OBU) acquires the MAP road information broadcast by the roadside unit (RSU) in real time. The OBU processes the received MAP road information and fits the road segment information (link) in the MAP road information into a road fitting curve using the least squares method. It then fits all the lane information (lane) contained under each link into a lane fitting curve using the least squares method, forming a curve set.

[0039] It should be noted that "link" refers to road segment information with a direction, while "lane" refers to lane information with a direction.

[0040] Step 104: Obtain the map matching result of the vehicle based on the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curve of each lane.

[0041] In this step, the map matching module of the on-board unit (OBU) compares the similarity between the vehicle's fitted curve and the road fitted curve of each road and the lane fitted curve of each lane in the curve set, and takes the road fitted curve and vehicle fitted curve with the highest similarity as the map matching result for that vehicle.

[0042] In this embodiment of the invention, the Kalman filter algorithm is used to process vehicle position data, increasing the stability and reliability of vehicle position. Combined with V2X technology, high-precision map information of the road is obtained in real time, and the vehicle position is accurately located in the traffic flow, enhancing navigation applications and improving traffic efficiency and driving safety.

[0043] Optionally, the step of filtering the vehicle position data in the acquired vehicle state data using a preset Kalman filter model to obtain the vehicle's historical trajectory point position data includes:

[0044] Based on the preset Kalman filter model and the vehicle state data, the optimal estimated vehicle position data corresponding to the vehicle position data is obtained;

[0045] Based on the optimal estimated vehicle position data and the vehicle position data, the jump point position data and non-jump point position data in the vehicle position data are obtained;

[0046] The non-jump point location data and the vehicle's optimal estimated location data corresponding to the jump point location data are used as the vehicle's historical trajectory point location data.

[0047] When the information processing module performs Kalman filtering on the acquired vehicle position data using a preset Kalman filter model, it first predicts the optimal estimated vehicle position data using the preset Kalman filter model. Then, based on the optimal estimated vehicle position data, it determines whether the real-time observed value (i.e., the vehicle position data) has a jump. That is, based on the optimal estimated vehicle position data and the vehicle position data, it obtains the jump point position data and the non-jump point position data in the vehicle position data. If it is a jump point position data, the real-time observed vehicle position data is considered unreliable and can be discarded (filtered out), and the optimal estimated vehicle position data needs to be used instead. The estimated position data is used as the vehicle's historical trajectory point position data. If it is a non-jump point position data, and it is assumed that the real-time observed vehicle position data has not changed, then the non-jump point position data is used as the vehicle's historical trajectory point position data and saved to the historical trajectory. The real-time observed vehicle position data is then assigned to the vehicle's optimal estimated position data. In other words, the optimal estimated position data of the vehicle corresponding to the non-jump point position data and the jump point position data is used as the vehicle's historical trajectory point position data. That is, the vehicle's historical trajectory point position data includes the optimal estimated position data of the vehicle corresponding to the non-jump point position data and the jump point position data.

[0048] The following details the process of predicting the optimal estimated position data of a vehicle using a pre-defined Kalman filter model:

[0049] Optionally, obtaining the optimal estimated vehicle position data corresponding to the vehicle position data based on the preset Kalman filter model and the vehicle state data includes:

[0050] Based on the vehicle state data at the first time step and the vehicle state data at the second time step, the state transition matrix, the mapping matrix, and the state observation matrix at the first time step are determined in the preset Kalman filter model; the mapping matrix is ​​the matrix in the preset Kalman filter model that maps the vehicle state data at the second time step to the state vector from the control input matrix.

[0051] Based on the vehicle's optimal state data at the second time step, the state transition matrix at the second time step, and the mapping matrix at the second time step, the vehicle's predicted state data at the first time step is obtained.

[0052] Based on the covariance matrix at the second time step, the state transition matrix at the second time step, and the preset prediction process noise vector in the preset Kalman filter model, the prediction estimation covariance matrix at the first time step is obtained.

[0053] Based on the vehicle state data at the first moment, the vehicle predicted state data at the first moment, the state observation matrix at the first moment, and the prediction estimation covariance matrix at the first moment, the optimal vehicle state data at the first moment is determined. The optimal vehicle state data at the first moment includes the optimal estimated vehicle position data at the first moment corresponding to the vehicle position data at the first moment.

[0054] Wherein, the vehicle status data at the first moment is the vehicle status data at any moment other than the vehicle status data at the first moment.

[0055] The optimal vehicle state data at the second time moment is the vehicle state data at the second time moment, or the optimal vehicle state data at the second time moment is determined based on the vehicle state data at the second time moment, the predicted vehicle state data at the second time moment, the state observation matrix at the second time moment, and the prediction estimation covariance matrix at the second time moment.

[0056] The second moment is the moment preceding the first moment;

[0057] The vehicle position data at the first moment refers to the vehicle position data at any moment other than the vehicle position data at the first moment.

[0058] The covariance matrix at the second time step is a pre-defined covariance matrix, or the covariance matrix at the second time step is obtained based on the covariance matrix at the third time step, the state transition matrix at the third time step, and the pre-defined prediction process noise vector in the pre-defined Kalman filter model.

[0059] The third time point is a time point preceding the second time point.

[0060] After the information processing module acquires the vehicle state data, it enters the preset Kalman filter model for processing, i.e., the Kalman filter is used for processing. The Kalman filter is divided into two stages to complete the current position prediction and verification, namely the prediction stage and the update stage. In the prediction stage, based on the vehicle state data at the first moment and the vehicle state data at the previous moment (the second moment), the state transition matrix A, the mapping matrix, and the observation matrix H at the second moment are obtained. The control input matrix B in the preset Kalman filter model can convert the motion measurement value u t The effect is mapped onto the state vector, i.e., the mapping matrix, where the motion state data at the second time moment includes the motion measurement value u. t The vehicle state data at the second time step is further predicted forward using the state transition matrix A and mapping matrix B at the second time step to obtain the predicted vehicle state data at the first time step.

[0061] During the update phase, the covariance matrix P at the second time step is used. t-1 The state transition matrix A at the second time step and the prediction noise vector Q are used to obtain the prediction estimation covariance matrix P′ at the first time step. t Wherein, when the second time step is the first time step, the covariance matrix P′ at the second time step. t The covariance matrix is ​​a pre-defined value. Alternatively, if the second time step is not the first time step, the covariance matrix at the second time step is predicted based on the state transition matrix at the third time step and the prediction process noise vector. Then, based on the vehicle state data at the first time step, the predicted vehicle state data at the first time step, the state observation matrix at the first time step, and the prediction estimation covariance matrix at the first time step, the optimal estimated vehicle position data at the first time step is obtained.

[0062] Furthermore, the vehicle status data includes vehicle speed, vehicle heading angle, vehicle yaw rate, and data on the vehicle's position point;

[0063] The step of determining the state transition matrix, mapping matrix, and state observation matrix at the second time step in the preset Kalman filter model based on the vehicle state data at the first time step and the vehicle state data at the second time step includes:

[0064] If the vehicle yaw rate at the second time moment is not equal to 0, based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, the vehicle speed at the second time moment, and the vehicle yaw rate at the second time moment, the state transition matrix, the mapping matrix, and the state observation matrix at the first time moment in the preset Kalman filter model are obtained.

[0065] When the vehicle yaw rate at the second time moment is equal to 0, the state transition matrix, mapping matrix, and state observation matrix at the first time moment are obtained in the preset Kalman filter model based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, and the vehicle speed at the second time moment.

[0066] In this embodiment of the invention, the acquired vehicle status data includes (longitude, latitude, speed, Ψ (heading angle), w (yaw rate)), where longitude and latitude are the data of the vehicle's position points.

[0067] The relevant parameters in Kalman filtering are as follows:

[0068] The latitude and longitude of vehicle location data can be transformed into (x, y) coordinates in the Mercator coordinate system through Mercator projection coordinate transformation, i.e., x (x-axis coordinate in the Mercator coordinate system) and y (y-axis coordinate in the Mercator coordinate system). Based on this, the model is established as follows:

[0069] Please see Figure 2 , Figure 2 This diagram illustrates the change in vehicle position. Based on the speed (speed), heading angle Ψ, x-position, y-position, and yaw rate w, the relationship between the vehicle position at the previous moment (second moment) t-1 and the vehicle position at the next moment (first moment) t can be derived:

[0070] When the vehicle's yaw rate w at the previous moment (the second moment) t-1 When not equal to 0:

[0071] x t =x t-1 +(sin(Ψ t-1 +w t-1 *dt)-sin(Ψ t-1 speed t-1 / w t-1

[0072] y t =y t-1 +(cos(Ψ t-1 )-cos(Ψ t-1 +w t-1 *dt))speed t-1 / w t-1

[0073] Where, x t This represents the x-axis coordinate of the vehicle's longitude position in the Mercator coordinate system at the first moment. t-1 Ψ represents the x-axis coordinate of the vehicle's longitude position in the Mercator coordinate system at the second moment. t-1 w represents the vehicle's heading angle at the second moment. t-1 Speed ​​represents the yaw rate of the vehicle at the second moment. t-1 y represents the vehicle's speed at the second moment. t This represents the y-coordinate of the vehicle's latitude position in the Mercator coordinate system at the first moment. t-1 This represents the y-axis coordinate of the vehicle's latitude position in the Mercator coordinate system at the second moment.

[0074] Based on this, a state transition model can be established:

[0075]

[0076] Among them, Ψ t This indicates the vehicle's heading angle at the first moment.

[0077] Therefore, the state transition matrix A at the second time step is:

[0078]

[0079] Mapping matrix B*u t for:

[0080]

[0081] The state observation matrix H at the first moment is:

[0082]

[0083] The vehicle's yaw rate w at the previous moment (the second moment) t-1 When equal to 0:

[0084] x t =x t-1 +speed t-1 *dt*cos(Ψ t-1 )

[0085] y t =y t-1 +speed t-1 *dt*sin(Ψ t-1 )

[0086] Where, x t This represents the x-axis coordinate of the vehicle's longitude position in the Mercator coordinate system at the first moment. t-1 Ψ represents the x-axis coordinate of the vehicle's longitude position in the Mercator coordinate system at the second moment. t-1 The speed represents the vehicle's heading angle at the second moment. t-1 y represents the vehicle's speed at the second moment. t This represents the y-coordinate of the vehicle's latitude position in the Mercator coordinate system at the first moment. t-1 This represents the y-axis coordinate of the vehicle's latitude position in the Mercator coordinate system at the second moment.

[0087] Based on this, a state transition model can be established:

[0088]

[0089] Among them, Ψ t This indicates the vehicle's heading angle at the first moment.

[0090] Therefore, the state transition matrix A at the second time step is:

[0091]

[0092] The mapping matrix B*u at the second time step t for:

[0093]

[0094] The state observation matrix H at the first moment is:

[0095]

[0096] It should be noted that the prediction noise vector Q and the observation noise vector R need to be calibrated, which can be determined when testing the optimization effect.

[0097] Further, determining the optimal vehicle state data at the first time based on the vehicle state data at the first time, the predicted vehicle state data at the first time, the state observation matrix at the first time, and the prediction estimation covariance matrix at the first time includes:

[0098] Based on the vehicle state data at the first moment, the predicted vehicle state data at the first moment, and the state observation matrix at the first moment, the measurement margin at the first moment is obtained.

[0099] Based on the predicted covariance moment at the first time, the state observation matrix at the first time, and the preset observation noise vector in the preset Kalman filter model, the optimal Kalman increment at the first time is obtained.

[0100] Based on the vehicle predicted state data at the first moment, the measurement margin at the first moment, and the optimal Kalman increment at the first moment, the optimal vehicle state data at the first moment is obtained; the optimal vehicle state data at the first moment includes the optimal estimated vehicle position data at the first moment.

[0101] After determining the state transition matrix, mapping matrix, and state observation matrix at the second time step, the Kalman filter is applied as follows:

[0102] First, the explanation is as follows:

[0103] This represents the measured value (vehicle state data) at time t (first time), i.e., the vehicle position and heading angle obtained from the positioning module, x t The x-axis coordinate of the vehicle's longitude position at time t in the Mercator coordinate system is y t The speed is the y-axis coordinate of the vehicle's latitude position in the Mercator coordinate system at time t. t It is the velocity at time t, Ψ t It is the heading angle at time t;

[0104] This represents the state estimate (predicted vehicle state data) at time t (first time). It is a prior estimate, that is, a state estimate made based on time (t-1) (second time) and the previous vehicle historical trajectory point location data.

[0105] This represents the optimal state estimate (optimal estimated vehicle position data) at time t (first time). It is a posterior estimate, that is, the optimal state estimate made based on the current time t and the vehicle's historical trajectory point position data.

[0106] P t Let represent the covariance matrix at time t (the first time).

[0107] In the prediction phase, the information processing module calculates the state estimate (vehicle optimal estimated position data) for the next moment (second moment) based on the optimal estimate (vehicle optimal estimated position data) corresponding to the vehicle position point data of the previous moment (second moment).

[0108] First, calculate the following two variables:

[0109] Calculate the measurement margin (residual) at the first moment. The formula is as follows:

[0110]

[0111] in, H represents the vehicle state data at the first moment, and H represents the state observation matrix at the first moment. This represents the predicted vehicle status data at the first moment.

[0112] The first time step prediction estimates the covariance matrix P′ t The calculation formula is as follows:

[0113] P′ t =AP t-1 A T +Q

[0114] Among them, P t-1 Let A represent the covariance matrix at the second time step, let A represent the state transition matrix at the second time step, and let Q represent the noise vector of the prediction process.

[0115] Calculate the optimal Kalman increment K at the first time step. t The formula is as follows:

[0116] K t =P′ t H T (HP′ tH T +R) -1

[0117] Among them, P′ t Let H represent the predicted covariance moment at the first moment, H represent the state observation matrix at the first moment, and R represent the preset observation noise vector in the preset Kalman filter model.

[0118] The formula for further forward prediction of the previous state during the prediction phase is as follows:

[0119]

[0120] in, A represents the predicted vehicle state data at the first time step, and A represents the state transition matrix at the second time step. B*u represents the optimal vehicle state data at the second time step. t-1 This represents the mapping matrix. In the update phase, the information processing module updates the optimal Kalman increment based on the measurements and calculates the posterior estimate.

[0121] The optimal estimate is given by updating the measurement residual (the measurement margin at the first time step) and the optimal Kalman increment at the first time step.

[0122]

[0123] in, This represents the optimal vehicle state data at the first moment. K represents the predicted vehicle state data at the first moment. t This represents the optimal Kalman increment at the first moment. This indicates the measurement margin at the first moment.

[0124] The optimal state data of the vehicle at the first moment includes the coordinates of the vehicle's latitude and longitude position in the Mercator coordinate system at the first moment.

[0125] Furthermore, after obtaining the optimal vehicle state data at the first time based on the vehicle predicted state data at the first time, the measurement margin at the first time, and the optimal Kalman increment at the first time, the method further includes:

[0126] Based on the preset unit matrix in the preset Kalman filter model, the optimal Kalman increment at the first time, the state observation matrix at the first time, and the prediction estimation covariance matrix at the first time, the covariance matrix at the first time is obtained.

[0127] After obtaining the optimal vehicle state data at the first moment, the covariance matrix at the first moment is updated according to the following formula:

[0128] P t =(IK t H)P′ t

[0129] Among them, P t Let I represent the covariance matrix at the first time step, and K represent the unit matrix. t Let P' represent the optimal Kalman increment at time 1, H represent the state observation matrix at time 1, and P' represent the state observation matrix at time 1. t This represents the predicted covariance matrix at the first moment.

[0130] The following details the process of obtaining the transition point position data and non-transition point position data from the vehicle position data based on the optimal estimated vehicle position data and the vehicle position data:

[0131] Optionally, obtaining the transition point position data and non-transition point position data from the vehicle position data based on the optimal estimated vehicle position data and the vehicle position data includes:

[0132] Based on the vehicle's optimal estimated position data and the vehicle's position data, the distance between the vehicle's optimal estimated position and the vehicle's position data is obtained;

[0133] If the distance between the optimal estimated position of the vehicle and the vehicle position data is greater than a preset distance, the vehicle position data is determined to be the jump point position data;

[0134] If the distance between the optimal estimated position of the vehicle and the vehicle position data is less than or equal to a preset distance, the vehicle position data is determined to be the jump point position data.

[0135] To determine if there has been a jump in vehicle position data, calculate the distance difference DeltaDis (the distance between the optimal estimated vehicle position and the vehicle position data) between the position observation (vehicle position data) and the optimal state estimate (optimal estimated vehicle position data). The calculation formula is as follows:

[0136]

[0137] Among them, (x t y t () represents the coordinates of the vehicle's location data. The coordinates represent the optimal estimated position data of the vehicle.

[0138] Preferably, a preset distance threshold between the optimal estimated vehicle position and the vehicle position data is set to 5 meters. This preset distance can be calibrated according to the positioning system. If DeltaDis > threshold, the observed value (vehicle position data) is considered unreliable and discarded, and the optimal estimated vehicle position data is used instead. If DeltaDis ≤ threshold, the observed value (vehicle position data) is considered reliable and no jump has occurred. The position observed value (vehicle position data) is then assigned to the optimal state estimate (optimal estimated vehicle position data).

[0139]

[0140] That is, after the prediction and update phases are completed, the optimal vehicle state data for the next moment will be obtained. Because this optimal vehicle state data takes into account factors such as measurement uncertainty and white noise, it will be used in the next iterative prediction step.

[0141] In summary, the information processing module performs Kalman filtering on the vehicle position data in the vehicle status data. When the vehicle position data changes abruptly, the position data will be filtered out, and the position data predicted by the Kalman filter model will be used instead as the vehicle's current position data, and saved as the vehicle's historical trajectory position data.

[0142] Optionally, generating a vehicle trajectory fitting curve based on the historical trajectory point location data includes:

[0143] Based on the historical trajectory point location data, an initial fitting curve for the vehicle trajectory and the sum of squared errors of the historical trajectory point location data are generated.

[0144] The vehicle trajectory fitting curve is obtained based on the sum of squared errors between the initial fitted curve of the vehicle trajectory and the historical trajectory point location data.

[0145] After obtaining the vehicle's historical trajectory point location data, the information processing module uses the least squares method to fit the historical trajectory point location data into a cubic fitting curve F. ve h icle That is, the vehicle trajectory fitting curve.

[0146] Specifically, the historical trajectory point location data set VechilePos can be obtained through the Kalman filter model and vehicle location point data. t (x t y t (t=1,2,3,4…n), assuming the optimal cubic fitting curve (initial fitting curve of vehicle trajectory) that conforms to the historical trajectory is:

[0147]

[0148] Where a, b, c, and d are coefficients.

[0149] The sum of squared errors for the data points within the historical trajectory location dataset of the vehicle is:

[0150]

[0151] Where S represents the sum of squared errors of data points within the historical trajectory point location dataset, f(x) t ) represents a data point within the historical trajectory point location dataset, y t This represents the mean of the data points within the historical trajectory point location dataset.

[0152] The least squares method posits that the coefficients of the optimal function should minimize the sum of squared errors, S. For the optimal function, the partial derivatives of the sum of squared errors, S, with respect to coefficients a, b, c, and d yield the following formula:

[0153]

[0154] Substituting the data points from the historical trajectory location data into the above formula and solving it yields the coefficients a, b, c, and d, thus deriving the optimal historical trajectory cubic fitting curve (vehicle trajectory fitting curve) F. ve hicle.

[0155] Optionally, the map information includes the center point coordinates of multiple roads and the center point coordinates of each lane in each road;

[0156] The process of generating road fitting curves for multiple roads and lane fitting curves for each lane in each road based on map information broadcast by the Roadside Unit (RSU) includes:

[0157] Based on the center point coordinates of each road in the multiple roads, generate the initial fitting curve for each road and the sum of squared errors of the center point coordinates of each road;

[0158] Based on the sum of squared errors of the initial road fitting curve of each road and the corresponding center point coordinate data of each road, the road fitting curve of each road in the multiple road fitting curves is obtained;

[0159] Based on the center point coordinates of each lane in each road, generate the initial lane fitting curve for each lane and the sum of squared errors of the center point coordinates of each lane;

[0160] The lane fitting curve for each lane is obtained by summing the squared errors of the initial lane fitting curve for each lane and the corresponding center point coordinate data for each lane.

[0161] The on-board unit (OBU) acquires map (MAP) information broadcast by the roadside unit (RSU) in real time. The MAP information includes the center point coordinates of multiple roads and the center point coordinates of each lane in each road. The information processing module in the OBU processes the received MAP information to obtain the road fitting curve for each road and the lane fitting curve for each lane.

[0162] In this embodiment of the invention, the coordinate data of the center point of each road (link) are fitted using the least squares method to generate a set of cubic curves: F link (linkID i (i = 1, 2, 3, 4…n), this set of cubic curves includes the road fitting curve for each road, where i represents the number of each road, and linkID i This represents the identity document (ID) of links in different directions within a road. For the center point coordinates of all lanes contained under each link, a set of cubic curves F is generated by fitting the data using the least squares method. lane (linkID i laneID j (i = 1, 2, 3, 4…n, j = 1, 2, 3, 4…n), this set of cubic curves includes the lane fitting curve for each lane in each road, where i represents the road number, j represents the lane number, and linkID... i LaneID represents the ID of a link in different directions on a road. j Indicates that it belongs to the road linkID i The ID of a lane in the system.

[0163] Please refer to the above process for generating vehicle trajectory fitting curves. Based on the center point coordinate data of each road in multiple roads, generate the initial road fitting curve for each road and the sum of squared errors of the center point coordinate data for each road. Substitute the link center point coordinate data from the MAP information into the corresponding partial derivative formula to obtain the set F of road fitting curves for each link. link (linkID i(i = 1, 2, 3, 4…n), based on the center point coordinates of each lane in each road, generate the initial lane fitting curve for each lane and the sum of squared errors of the center point coordinates of each lane. Substitute the road center point coordinates from the MAP information into the corresponding partial derivative formulas to obtain the set F of lane fitting curves for each lane in each link. lane (linkID i laneID j (i = 1, 2, 3, 4…n, j = 1, 2, 3, 4…n), where linkID i LaneID represents the ID of a link in different directions on a road. j Indicates that it belongs to the road linkID i The ID of a lane in the system.

[0164] Optionally, obtaining the map matching result of the vehicle based on the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curve of each lane includes:

[0165] Using a preset Framincher distance algorithm, the first similarity between the vehicle fitting curve and the road fitting curve of each road is obtained;

[0166] The road corresponding to the fitted curve of the road with the highest similarity is selected as the target road;

[0167] Using the preset Framinger distance algorithm, a second similarity is obtained between the vehicle fitting curve and the lane fitting curve of each lane in the target road;

[0168] The lane corresponding to the lane fitting curve with the highest second similarity is selected as the target lane;

[0169] The target road and the target lane are used as the map matching results for the vehicle.

[0170] In this embodiment of the invention, the map matching module of the OBU calculates the vehicle trajectory fitting curve F using the Frechet distance algorithm. vehicle and the set of road curves F link The similarity of each curve in the (including the road fitting curve for each road and the lane fitting curve for each lane in each road) is calculated. The road fitting curve with the highest similarity is the road fitting curve of the link where the vehicle is located, and the linkID of that link is recorded. i This link is the target road, and then the vehicle trajectory is fitted to curve F. vehicle and corresponding linkID iThe set of lane fitting curves F for each lane below lane Similarity calculation is performed, and the similarity score is the laneID of the fitted curve for that lane. j The target lane is the target road and the target lane, which are used as the result of map matching for the vehicle.

[0171] Specifically, the vehicle trajectory is fitted to curve F. vehicle The cubic fitting curves F with respect to the center line of the link respectively link The (road fitting curve) is input into the Frechet distance algorithm library for calculation to obtain the similarity return value. The road with the largest similarity return value is the linkID of the most similar road. i This road is the target road;

[0172] From the linkID of the nearest road i Extracting the cubic fitting curve F of the lane centerline lane (Lane fitting curve), and the vehicle trajectory fitting curve F vehicle The similarity values ​​are calculated using the Frechet distance algorithm. The lane ID with the highest similarity value is the closest lane ID. j This lane is the target lane;

[0173] Furthermore, the map matching results are as follows:

[0174] After similarity calculation, the final map matching result is output as [linkID]. i ,lane j This is used for upper-layer business applications, such as navigation, green wave speed guidance, and red light violation warning, for further logic development.

[0175] The map matching method provided in this invention uses a Kalman filter model to predict and filter vehicle location data, reducing location errors caused by environmental interference, thereby providing reliable and stable vehicle historical trajectories. It acquires road and lane information around the vehicle using V2X technology, and uses a curve fitting algorithm to fit discrete road and lane information into continuous traffic flow information, increasing the recognizability of road, lane, and intersection information. Simultaneously, it fits the discrete historical trajectory of the vehicle to generate a continuous trajectory. Finally, considering the overall trajectory trend of the vehicle and road, and combining it with a trajectory similarity algorithm, it improves the accuracy of map matching in complex road network environments. In other words, it combines V2X technology to acquire high-precision road map information in real time, accurately locating the vehicle's position within the traffic flow, enhancing navigation applications, and improving traffic efficiency and driving safety.

[0176] In this embodiment of the invention, a cubic curve is generated by fitting the discrete historical trajectory of the vehicle and the discrete lane information of the road through a fitting algorithm. This can reduce vehicle position error, enhance the overall trend of vehicle trajectory and traffic lane flow trajectory, and more accurately represent the positional relationship between the vehicle and the road. Furthermore, trajectory similarity matching is performed on the vehicle trajectory and the road traffic flow trajectory. By comparing the matching degree scores, the optimal vehicle driving road trajectory is obtained. Compared with geometric road matching of points and lines, this further improves the accuracy of map matching.

[0177] The map matching method provided in this embodiment of the invention does not require cloud-based backend to participate in calculation and information processing. The method is simple and has a low cost.

[0178] like Figure 3 As shown, embodiments of the present invention also provide a map matching device, comprising:

[0179] The first processing module 301 is used to filter the vehicle position data in the acquired vehicle state data using a preset Kalman filter model to obtain the historical trajectory point position data of the vehicle.

[0180] The first curve generation module 302 is used to generate a vehicle trajectory fitting curve based on the historical trajectory point location data.

[0181] The second curve generation module 303 is used to generate road fitting curves for multiple roads and lane fitting curves for each lane in each road based on the map information broadcast by the roadside unit (RSU).

[0182] The second processing module 304 is used to obtain the map matching result of the vehicle based on the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curve of each lane.

[0183] Optionally, the first processing module 301 includes:

[0184] The first processing submodule is used to obtain the optimal estimated vehicle position data corresponding to the vehicle position data based on the preset Kalman filter model and the vehicle state data.

[0185] The second processing submodule is used to obtain the jump point position data and non-jump point position data in the vehicle position data based on the vehicle's optimal estimated position data and the vehicle position data.

[0186] The third processing submodule is used to take the non-jump point position data and the vehicle's optimal estimated position data corresponding to the jump point position data as the vehicle's historical trajectory point position data.

[0187] Optionally, the first processing submodule includes:

[0188] The first determining unit is used to determine, based on the vehicle state data at the first time and the vehicle state data at the second time, the state transition matrix at the second time, the mapping matrix at the first time, and the state observation matrix at the first time in the preset Kalman filter model; the mapping matrix is ​​the matrix in the preset Kalman filter model that maps the vehicle state data at the second time to the state vector from the control input matrix at the second time.

[0189] The first processing unit is used to obtain the predicted vehicle state data at the first time based on the optimal vehicle state data at the second time, the state transition matrix at the second time, and the mapping matrix at the second time.

[0190] The second processing unit is used to obtain the prediction estimation covariance matrix at the first time based on the covariance matrix at the second time, the state transition matrix at the second time, and the preset prediction process noise vector in the preset Kalman filter model.

[0191] The second determining unit is used to determine the optimal vehicle state data at the first time based on the vehicle state data at the first time, the vehicle predicted state data at the first time, the state observation matrix at the first time, and the prediction estimation covariance matrix at the first time. The optimal vehicle state data at the first time includes the optimal estimated vehicle position data at the first time corresponding to the vehicle position data at the first time.

[0192] Wherein, the vehicle status data at the first moment is the vehicle status data at any moment other than the vehicle status data at the first moment.

[0193] The optimal vehicle state data at the second time moment is the vehicle state data at the second time moment, or the optimal vehicle state data at the second time moment is determined based on the vehicle state data at the second time moment, the predicted vehicle state data at the second time moment, the state observation matrix at the second time moment, and the prediction estimation covariance matrix at the second time moment.

[0194] The second moment is the moment preceding the first moment;

[0195] The vehicle position data at the first moment refers to the vehicle position data at any moment other than the vehicle position data at the first moment.

[0196] The covariance matrix at the second time step is a pre-defined covariance matrix, or the covariance matrix at the second time step is obtained based on the covariance matrix at the third time step, the state transition matrix at the third time step, and the pre-defined prediction process noise vector in the pre-defined Kalman filter model.

[0197] The third time point is a time point preceding the second time point.

[0198] Optionally, the vehicle status data includes vehicle speed, vehicle heading angle, vehicle yaw rate, and data on the vehicle position point;

[0199] The first determining unit is specifically used for:

[0200] If the vehicle yaw rate at the second time moment is not equal to 0, based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, the vehicle speed at the second time moment, and the vehicle yaw rate at the second time moment, the state transition matrix, the mapping matrix, and the state observation matrix at the first time moment in the preset Kalman filter model are obtained.

[0201] When the vehicle yaw rate at the second time moment is equal to 0, the state transition matrix, mapping matrix, and state observation matrix at the first time moment are obtained in the preset Kalman filter model based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, and the vehicle speed at the second time moment.

[0202] Optionally, the second determining unit is specifically used for:

[0203] Based on the vehicle state data at the first moment, the predicted vehicle state data at the first moment, and the state observation matrix at the first moment, the measurement margin at the first moment is obtained.

[0204] Based on the predicted covariance moment at the first time, the state observation matrix at the first time, and the preset observation noise vector in the preset Kalman filter model, the optimal Kalman increment at the first time is obtained.

[0205] Based on the vehicle predicted state data at the first time step, the measurement margin at the first time step, and the optimal Kalman increment at the first time step, the optimal vehicle state data at the first time step is obtained.

[0206] Optionally, the second determining unit is further specifically used for:

[0207] Based on the preset unit matrix in the preset Kalman filter model, the optimal Kalman increment at the first time, the state observation matrix at the first time, and the prediction estimation covariance matrix at the first time, the covariance matrix at the first time is obtained.

[0208] Optionally, the second processing submodule includes:

[0209] The third processing unit is used to obtain the distance between the optimal estimated position of the vehicle and the vehicle position data based on the optimal estimated position data of the vehicle and the vehicle position data.

[0210] The third determining unit is used to determine the vehicle position data as the jump point position data when the distance between the optimal estimated position of the vehicle and the vehicle position data is greater than a preset distance.

[0211] The fourth determining unit is used to determine the vehicle position data as the jump point position data when the distance between the optimal estimated position of the vehicle and the vehicle position data is less than or equal to a preset distance.

[0212] Optionally, the first curve generation module 302 includes:

[0213] The first generation unit is used to generate an initial fitting curve for the vehicle trajectory and a sum of squared errors of the historical trajectory point location data based on the historical trajectory point location data.

[0214] The fourth processing unit is used to obtain the vehicle trajectory fitting curve based on the sum of squared errors between the initial fitting curve of the vehicle trajectory and the historical trajectory point position data.

[0215] Optionally, the map information includes the center point coordinates of multiple roads and the center point coordinates of each lane in each road;

[0216] The second curve generation module 303 includes:

[0217] The second generation unit is used to generate the initial fitting curve of each road and the sum of squared errors of the center point coordinate data of each road based on the center point coordinate data of each road in the multiple roads.

[0218] The fifth processing unit is used to obtain the road fitting curve of each road from the road fitting curves of multiple roads based on the sum of squared errors of the initial road fitting curve of each road and the corresponding center point coordinate data of each road.

[0219] The third generation unit is used to generate the initial lane fitting curve for each lane and the sum of squared errors of the center point coordinate data of each lane based on the center point coordinate data of each lane in each road.

[0220] The sixth processing unit is used to obtain the lane fitting curve for each lane based on the sum of squared errors of the initial lane fitting curve for each lane and the corresponding center point coordinate data of each lane.

[0221] Optionally, the second processing module 304 includes:

[0222] The seventh processing unit is used to obtain the first similarity between the vehicle fitting curve and the road fitting curve of each road using a preset Framincher distance algorithm;

[0223] The first selection unit is used to select the road corresponding to the road fitting curve with the largest similarity as the target road;

[0224] The eighth processing unit is used to obtain a second similarity between the vehicle fitting curve and the lane fitting curve of each lane in the target road using the preset Framinger distance algorithm.

[0225] The second selection unit is used to select the lane corresponding to the lane fitting curve with the highest similarity as the target lane;

[0226] The ninth processing unit is used to use the target road and the target lane as the map matching result of the vehicle.

[0227] It should be noted that the map matching device provided in the embodiments of the present invention is a device capable of executing the above-described map matching method. Therefore, all embodiments of the above-described map matching method are applicable to this device and can achieve the same or similar technical effects.

[0228] like Figure 4 As shown, this embodiment of the invention also provides a map matching device, including: a processor 400; and a memory 410 connected to the processor 400 via a bus interface, the memory 410 being used to store programs and data used by the processor 400 when performing operations, and the processor 400 calling and executing the programs and data stored in the memory 410.

[0229] The map matching device further includes a transceiver 420, which is connected to a bus interface and is used to receive and send data under the control of the processor 400.

[0230] Specifically, the processor 400 performs the following processes:

[0231] The vehicle position data in the acquired vehicle state data is filtered using a preset Kalman filter model to obtain the historical trajectory point position data of the vehicle.

[0232] Based on the historical trajectory point location data, a vehicle trajectory fitting curve is generated;

[0233] Based on the map information broadcast by the Roadside Unit (RSU), road fitting curves for multiple roads and lane fitting curves for each lane in each road are generated.

[0234] The map matching result of the vehicle is obtained based on the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curve of each lane.

[0235] Optionally, the processor 400 is configured to:

[0236] Based on the preset Kalman filter model and the vehicle state data, the optimal estimated vehicle position data corresponding to the vehicle position data is obtained;

[0237] Based on the optimal estimated vehicle position data and the vehicle position data, the jump point position data and non-jump point position data in the vehicle position data are obtained;

[0238] The non-jump point location data and the vehicle's optimal estimated location data corresponding to the jump point location data are used as the vehicle's historical trajectory point location data.

[0239] Optionally, the processor 400 is specifically used for:

[0240] Based on the vehicle state data at the first time step and the vehicle state data at the second time step, the state transition matrix, the mapping matrix, and the state observation matrix at the first time step are determined in the preset Kalman filter model; the mapping matrix is ​​the matrix in the preset Kalman filter model that maps the vehicle state data at the second time step to the state vector from the control input matrix.

[0241] Based on the vehicle's optimal state data at the second time step, the state transition matrix at the second time step, and the mapping matrix at the second time step, the vehicle's predicted state data at the first time step is obtained.

[0242] Based on the covariance matrix at the second time step, the state transition matrix at the second time step, and the preset prediction process noise vector in the preset Kalman filter model, the prediction estimation covariance matrix at the first time step is obtained.

[0243] Based on the vehicle state data at the first moment, the vehicle predicted state data at the first moment, the state observation matrix at the first moment, and the prediction estimation covariance matrix at the first moment, the optimal vehicle state data at the first moment is determined. The optimal vehicle state data at the first moment includes the optimal estimated vehicle position data at the first moment corresponding to the vehicle position data at the first moment.

[0244] Wherein, the vehicle status data at the first moment is the vehicle status data at any moment other than the vehicle status data at the first moment.

[0245] The optimal vehicle state data at the second time moment is the vehicle state data at the second time moment, or the optimal vehicle state data at the second time moment is determined based on the vehicle state data at the second time moment, the predicted vehicle state data at the second time moment, the state observation matrix at the second time moment, and the prediction estimation covariance matrix at the second time moment.

[0246] The second moment is the moment preceding the first moment;

[0247] The vehicle position data at the first moment refers to the vehicle position data at any moment other than the vehicle position data at the first moment.

[0248] The covariance matrix at the second time step is a pre-defined covariance matrix, or the covariance matrix at the second time step is obtained based on the covariance matrix at the third time step, the state transition matrix at the third time step, and the pre-defined prediction process noise vector in the pre-defined Kalman filter model.

[0249] The third time point is a time point preceding the second time point.

[0250] Optionally, the vehicle status data includes vehicle speed, vehicle heading angle, vehicle yaw rate, and data on the vehicle position point;

[0251] The processor 400 is specifically used for:

[0252] If the vehicle yaw rate at the second time moment is not equal to 0, based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, the vehicle speed at the second time moment, and the vehicle yaw rate at the second time moment, the state transition matrix, the mapping matrix, and the state observation matrix at the first time moment in the preset Kalman filter model are obtained.

[0253] When the vehicle yaw rate at the second time moment is equal to 0, the state transition matrix, mapping matrix, and state observation matrix at the first time moment are obtained in the preset Kalman filter model based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, and the vehicle speed at the second time moment.

[0254] Optionally, the processor 400 is specifically used for:

[0255] Based on the vehicle state data at the first moment, the predicted vehicle state data at the first moment, and the state observation matrix at the first moment, the measurement margin at the first moment is obtained.

[0256] Based on the predicted covariance moment at the first time, the state observation matrix at the first time, and the preset observation noise vector in the preset Kalman filter model, the optimal Kalman increment at the first time is obtained.

[0257] Based on the vehicle predicted state data at the first time step, the measurement margin at the first time step, and the optimal Kalman increment at the first time step, the optimal vehicle state data at the first time step is obtained.

[0258] Optionally, the processor 400 is further specifically used for:

[0259] Based on the preset unit matrix in the preset Kalman filter model, the optimal Kalman increment at the first time, the state observation matrix at the first time, and the prediction estimation covariance matrix at the first time, the covariance matrix at the first time is obtained.

[0260] Optionally, the processor 400 is specifically used for:

[0261] Based on the vehicle's optimal estimated position data and the vehicle's position data, the distance between the vehicle's optimal estimated position and the vehicle's position data is obtained;

[0262] If the distance between the optimal estimated position of the vehicle and the vehicle position data is greater than a preset distance, the vehicle position data is determined to be the jump point position data;

[0263] If the distance between the optimal estimated position of the vehicle and the vehicle position data is less than or equal to a preset distance, the vehicle position data is determined to be the jump point position data.

[0264] Optionally, the processor 400 is specifically used for:

[0265] Based on the historical trajectory point location data, an initial fitting curve for the vehicle trajectory and the sum of squared errors of the historical trajectory point location data are generated.

[0266] The vehicle trajectory fitting curve is obtained based on the sum of squared errors between the initial fitted curve of the vehicle trajectory and the historical trajectory point location data.

[0267] Optionally, the map information includes the center point coordinates of multiple roads and the center point coordinates of each lane in each road;

[0268] The processor 400 is used for:

[0269] Based on the center point coordinates of each road in the multiple roads, generate the initial fitting curve for each road and the sum of squared errors of the center point coordinates of each road;

[0270] Based on the sum of squared errors of the initial road fitting curve of each road and the corresponding center point coordinate data of each road, the road fitting curve of each road in the multiple road fitting curves is obtained;

[0271] Based on the center point coordinates of each lane in each road, generate the initial lane fitting curve for each lane and the sum of squared errors of the center point coordinates of each lane;

[0272] The lane fitting curve for each lane is obtained by summing the squared errors of the initial lane fitting curve for each lane and the corresponding center point coordinate data for each lane.

[0273] Optionally, the processor 400 is configured to:

[0274] Using a preset Framincher distance algorithm, the first similarity between the vehicle fitting curve and the road fitting curve of each road is obtained;

[0275] The road corresponding to the fitted curve of the road with the highest similarity is selected as the target road;

[0276] Using the preset Framinger distance algorithm, a second similarity is obtained between the vehicle fitting curve and the lane fitting curve of each lane in the target road;

[0277] The lane corresponding to the lane fitting curve with the highest second similarity is selected as the target lane;

[0278] The target road and the target lane are used as the map matching results for the vehicle.

[0279] Among them, Figure 4 In this context, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 400) and memory (memory 410). The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides a user interface 430. A transceiver 420 may be multiple elements, including transmitters and receivers, providing units for communicating with various other devices over a transmission medium. Processor 400 is responsible for managing the bus architecture and general processing, and memory 410 may store data used by processor 400 during operation.

[0280] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a computer program instructing the relevant hardware to implement them. The computer program includes instructions to perform some or all of the steps of the above methods; and the computer program can be stored in a readable storage medium, which can be any form of storage medium.

[0281] In addition, embodiments of the present invention also provide a computer-readable storage medium storing a program. When executed by a processor, this program implements the various processes of the map matching method embodiments described above and achieves the same technical effect. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0282] Furthermore, it should be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above-described series of processes can naturally be executed in the order described or in chronological order, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0283] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a known general-purpose device. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code implementing the method or apparatus. That is, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future.

[0284] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0285] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principles described in this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A map matching method characterized by, The method comprises the following steps: filtering vehicle position data in acquired vehicle state data by using a preset Kalman filter model to obtain historical trajectory point position data of the vehicle; generating a vehicle trajectory fitting curve according to the historical trajectory point position data; generating road fitting curves of multiple roads and lane fitting curves of each lane in each road according to map information broadcast by a road side unit (RSU); obtaining a map matching result of the vehicle according to the vehicle trajectory fitting curve, the road fitting curves of the multiple roads and the lane fitting curves of each lane; wherein the filtering of the vehicle position data in the acquired vehicle state data by using the preset Kalman filter model to obtain the historical trajectory point position data of the vehicle comprises: obtaining vehicle optimal estimation position data corresponding to the vehicle position data according to the preset Kalman filter model and the vehicle state data; obtaining jump point position data and non-jump point position data in the vehicle position data according to the vehicle optimal estimation position data and the vehicle position data; taking the vehicle optimal estimation position data corresponding to the non-jump point position data and the jump point position data as the historical trajectory point position data of the vehicle; wherein the obtaining of the vehicle optimal estimation position data corresponding to the vehicle position data according to the preset Kalman filter model and the vehicle state data comprises: determining a state transition matrix, a mapping matrix and a state observation matrix of a first time in the preset Kalman filter model according to vehicle state data of the first time and vehicle state data of a second time; the mapping matrix is a matrix in which a control input matrix in the preset Kalman filter model maps the vehicle state data of the second time to a state vector; obtaining vehicle prediction state data of the first time according to vehicle optimal state data of the second time, the state transition matrix of the second time and the mapping matrix of the second time; obtaining a prediction estimation covariance matrix of the first time according to a covariance matrix of the second time, the state transition matrix of the second time and a preset prediction process noise vector in the preset Kalman filter model; determining vehicle optimal state data of the first time according to vehicle state data of the first time, the vehicle prediction state data of the first time, the state observation matrix of the first time and the prediction estimation covariance matrix of the first time, wherein the vehicle optimal state data of the first time comprises the vehicle optimal estimation position data of the first time corresponding to the vehicle position data of the first time; wherein the vehicle state data of the first time is vehicle state data of any time except the vehicle state data of the first time. The optimal vehicle state data at the second time moment is the vehicle state data at the second time moment, or the optimal vehicle state data at the second time moment is determined based on the vehicle state data at the second time moment, the predicted vehicle state data at the second time moment, the state observation matrix at the second time moment, and the prediction estimation covariance matrix at the second time moment. The second moment is the moment preceding the first moment; The vehicle position data at the first moment refers to the vehicle position data at any moment other than the vehicle position data at the first moment. The covariance matrix at the second time point is a pre-defined covariance matrix, or the covariance matrix at the second time point is obtained based on the covariance matrix at the third time point, the state transition matrix at the third time point, and the pre-defined prediction process noise vector in the pre-defined Kalman filter model. The third time point is a time point preceding the second time point.

2. The map matching method according to claim 1, characterized in that, The vehicle status data includes vehicle speed, vehicle heading angle, vehicle yaw rate, and vehicle position data; The step of determining the state transition matrix, mapping matrix, and state observation matrix at the second time step in the preset Kalman filter model based on the vehicle state data at the first time step and the vehicle state data at the second time step includes: If the vehicle yaw rate at the second time moment is not equal to 0, based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, the vehicle speed at the second time moment, and the vehicle yaw rate at the second time moment, the state transition matrix, the mapping matrix, and the state observation matrix at the first time moment in the preset Kalman filter model are obtained. When the vehicle yaw rate at the second time moment is equal to 0, the state transition matrix, mapping matrix, and state observation matrix at the first time moment are obtained in the preset Kalman filter model based on the vehicle position data at the first time moment, the vehicle heading angle at the first time moment, the vehicle position data at the second time moment, the vehicle heading angle at the second time moment, and the vehicle speed at the second time moment.

3. The map matching method according to claim 1, characterized in that, The step of determining the optimal vehicle state data at the first moment based on the vehicle state data at the first moment, the predicted vehicle state data at the first moment, the state observation matrix at the first moment, and the prediction estimation covariance matrix at the first moment includes: Based on the vehicle state data at the first moment, the predicted vehicle state data at the first moment, and the state observation matrix at the first moment, the measurement margin at the first moment is obtained. Based on the predicted covariance moment at the first time, the state observation matrix at the first time, and the preset observation noise vector in the preset Kalman filter model, the optimal Kalman increment at the first time is obtained. Based on the vehicle predicted state data at the first time step, the measurement margin at the first time step, and the optimal Kalman increment at the first time step, the optimal vehicle state data at the first time step is obtained.

4. The map matching method according to claim 2, characterized in that, After obtaining the first-time vehicle optimal state data according to the vehicle predicted state data at the first time, the measurement residual at the first time and the first-time optimal Kalman increment, the method further comprises: obtaining the first-time covariance matrix according to the preset unit matrix in the preset Kalman filtering model, the first-time optimal Kalman increment, the first-time state observation matrix and the first-time predicted estimation covariance matrix.

5. The map matching method according to claim 1, characterized in that, The obtaining of the jump point position data and the non-jump point position data in the vehicle position data according to the vehicle optimal estimated position data and the vehicle position data comprises: obtaining the distance between the vehicle optimal estimated position and the vehicle position data according to the vehicle optimal estimated position data and the vehicle position data; determining the vehicle position data as the jump point position data when the distance between the vehicle optimal estimated position and the vehicle position data is greater than a preset distance; determining the vehicle position data as the jump point position data when the distance between the vehicle optimal estimated position and the vehicle position data is less than or equal to a preset distance.

6. The map matching method according to claim 1, characterized in that, The generating of the vehicle trajectory fitting curve according to the historical trajectory point position data comprises: generating a vehicle trajectory initial fitting curve and an error sum of squares of the historical trajectory point position data according to the historical trajectory point position data; obtaining the vehicle trajectory fitting curve according to the vehicle trajectory initial fitting curve and the error sum of squares of the historical trajectory point position data.

7. The map matching method according to claim 1, characterized in that, The map information comprises center point coordinate data of a plurality of roads and center point coordinate data of each lane in each road; The generating of the road fitting curve of each road in the plurality of roads and the lane fitting curve of each lane in each road according to the map information broadcast by the road side unit (RSU) comprises: generating a road initial fitting curve of each road and an error sum of squares of the center point coordinate data of each road according to the center point coordinate data of each road in the plurality of roads; obtaining the road fitting curve of each road in the road fitting curves of the plurality of roads according to the road initial fitting curve of each road and the error sum of squares of the corresponding center point coordinate data of each road; generating a lane initial fitting curve of each lane and an error sum of squares of the center point coordinate data of each lane according to the center point coordinate data of each lane in each road; obtaining the lane fitting curve of each lane according to the lane initial fitting curve of each lane and the error sum of squares of the corresponding center point coordinate data of each lane.

8. The map matching method of claim 1, wherein, The obtaining of the map matching result of the vehicle according to the vehicle trajectory fitting curve, the road fitting curve of each road in the plurality of roads and the lane fitting curve of each lane comprises: obtaining a first similarity between the vehicle trajectory fitting curve and the road fitting curve of each road by using a preset Fréchet distance algorithm; selecting a road corresponding to a road fitting curve with the maximum first similarity as a target road; The preset Franschise distance algorithm is used to obtain a second similarity between the vehicle fitting curve and a lane fitting curve of each lane in the target road; A lane corresponding to the lane fitting curve with the largest second similarity is selected as a target lane; The target road and the target lane are taken as a map matching result of the vehicle.

9. A map matching device, characterized by The device is applied to the map matching method in any one of claims 1 to 8, and the device comprises: A first processing module is configured to filter vehicle position data in acquired vehicle state data by using a preset Kalman filter model to obtain historical trajectory point position data of the vehicle; A first curve generation module is configured to generate a vehicle trajectory fitting curve according to the historical trajectory point position data; A second curve generation module is configured to generate road fitting curves of multiple roads and lane fitting curves of each lane in each road according to map information broadcast by a road side unit (RSU); A second processing module is configured to obtain a map matching result of the vehicle according to the vehicle trajectory fitting curve, the road fitting curves of the multiple roads, and the lane fitting curves of each lane; The first processing module comprises: A first processing submodule is configured to obtain vehicle optimal estimation position data corresponding to the vehicle position data according to the preset Kalman filter model and the vehicle state data; A second processing submodule is configured to obtain jump point position data and non-jump point position data in the vehicle position data according to the vehicle optimal estimation position data and the vehicle position data; A third processing submodule is configured to take vehicle optimal estimation position data corresponding to the non-jump point position data and the jump point position data as the historical trajectory point position data of the vehicle; The first processing submodule comprises: A first determination unit is configured to determine a state transition matrix at a second time, a mapping matrix, and a state observation matrix at a first time in the preset Kalman filter model according to vehicle state data at the first time and vehicle state data at the second time; the mapping matrix is a matrix in which a control input matrix in the preset Kalman filter model maps the vehicle state data at the second time to a state vector; A first processing unit is configured to obtain vehicle predicted state data at the first time according to vehicle optimal state data at the second time, the state transition matrix at the second time, and the mapping matrix at the second time; A second processing unit is configured to obtain a predicted estimation covariance matrix at the first time according to a covariance matrix at the second time, the state transition matrix at the second time, and a preset prediction process noise vector in the preset Kalman filter model. The second determining unit is configured to determine the optimal state data of the vehicle at the first time according to the state data of the vehicle at the first time, the predicted state data of the vehicle at the first time, the state observation matrix at the first time, and the predicted estimation covariance matrix at the first time, wherein the optimal state data of the vehicle at the first time comprises optimal estimated position data of the vehicle at the first time corresponding to the position data of the vehicle at the first time. The state data of the vehicle at the first time is the state data of the vehicle at any time except the first time. The optimal state data of the vehicle at the second time is the state data of the vehicle at the second time, or the optimal state data of the vehicle at the second time is determined according to the state data of the vehicle at the second time, the predicted state data of the vehicle at the second time, the state observation matrix at the second time, and the predicted estimation covariance matrix at the second time. The second time is one time before the first time. The position data of the vehicle at the first time is the position data of the vehicle at any time except the first time. The covariance matrix at the second time is a preset covariance matrix, or the covariance matrix at the second time is obtained according to the covariance matrix at the third time, the state transition matrix at the third time, and a preset predicted process noise vector in the preset Kalman filtering model. The third time is one time before the second time.

10. A map matching device, characterized by The method comprises the following steps: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program is executed by the processor to implement the map matching method according to any one of claims 1 to 8.

11. A readable storage medium, characterized by, The program is stored in the readable storage medium and is executed by the processor to implement the steps in the map matching method according to any one of claims 1 to 8.

Images