Collaborative positioning method for intelligent connected vehicles based on V2X communication and traffic sign recognition

Abstract

The present invention relates to a collaborative positioning method for intelligent networked vehicles based on V2X communication and traffic sign recognition, and belongs to the field of intelligent vehicle positioning. Obtain road images and point cloud data, obtain traffic sign detection results in the image, and extract features of targets in the point cloud according to traffic sign features; classify traffic signs in the point cloud according to image detection results and point cloud features, and obtain point cloud traffic sign detection results; use the position of the intelligent networked vehicle and the distance between the vehicle and the traffic sign to locate the traffic sign, and fuse the positioning results of the two intelligent networked vehicles for the traffic sign; in the current frame, use the located traffic sign position and the position of the intelligent networked vehicle to estimate the position of the intelligent networked vehicle to be located; use the positioning method based on local constraints of the previous and next frames to obtain the position of the intelligent networked vehicle to be located; finally, use V2X communication technology and adaptive weight fusion method to fuse the multi-party positioning results.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent vehicle positioning and relates to a collaborative positioning method for intelligent networked vehicles based on V2X communication and traffic sign recognition. Background Art

[0002] With the development of science and technology, the automotive industry has ushered in a new round of industrial transformation and upgrading led by intelligent manufacturing. At present, the automotive industry is in an era of great change and development. Intelligent connected vehicles are one of the "smart" high points that countries are vying for. The rapid development of intelligent connected vehicles is of great significance to the country's industrial transformation and upgrading and economic development. Intelligent connected vehicles rely on technologies such as computer vision, deep learning, and wireless network communications, and are equipped with a variety of sensors, so that intelligent connected vehicles have basic functions such as environmental perception, path planning, and decision control, thereby realizing autonomous driving. Among them, accurate positioning of itself is the prerequisite for intelligent connected vehicles to achieve autonomous driving, which can provide accurate posture information for autonomous navigation map construction, path planning, etc.

[0003] In order to obtain its own position, intelligent connected vehicles can obtain accurate information through GPS positioning systems and inertial navigation systems. However, in actual applications, the positioning results of a single sensor are often relatively simple and not robust. In addition, when an intelligent connected vehicle is driving in a real traffic environment such as an underground parking lot, the positioning may fail, which leads to the unknown movement of the vehicle coordinate system before and after the movement of the intelligent connected vehicle. At present, a large number of studies have been carried out on the positioning technology of intelligent connected vehicles at home and abroad, but most positioning algorithms only consider the situation of a single vehicle with a single sensor or a single vehicle with multiple sensors, and there is little research on the collaborative positioning of multiple vehicles. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a collaborative positioning method for intelligent connected vehicles based on V2X communication and traffic sign recognition, so as to realize the precise positioning of intelligent connected vehicles in the application of positioning sensor failure. The method uses V2X communication technology and long-term fixed and evenly distributed traffic signs in the city to provide a solution for the auxiliary positioning of intelligent connected vehicles, so that it can realize accurate map construction and path planning, and then complete autonomous navigation.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for collaborative positioning of intelligent connected vehicles based on V2X communication and traffic sign recognition is firstly assumed that there are intelligent connected vehicle 1 (Connected and Intelligent Vehicle, CIV1) and intelligent connected vehicle 2 (CIV2). At a certain moment, the position of CIV1 needs to be determined, which specifically includes the following steps:

[0007] S1: Use cameras and lidar to obtain road images and point cloud data. Use deep learning traffic sign detection methods to obtain traffic sign detection results in the image. Then preprocess the point cloud data and extract features of the targets in the point cloud based on the traffic sign features;

[0008] S2: Based on the image detection results and the target features extracted from the point cloud, the traffic signs in the point cloud are classified using the traffic sign detection method based on image and point cloud fusion to obtain accurate point cloud traffic sign detection results;

[0009] S3: Use the position of the intelligent connected vehicle and the distance between the vehicle and the traffic sign to locate the traffic sign, and use V2X communication technology and an adaptive weight fusion method based on angular trust to fuse the positioning results of the two intelligent connected vehicles for the traffic sign;

[0010] S4: In the current frame, the location of the intelligent connected vehicle to be located is estimated using the located traffic sign location and the location of the intelligent connected vehicle; the location of the intelligent connected vehicle to be located is obtained using the positioning method based on the local constraints of the previous and next frames. Finally, the multi-party positioning results are fused through V2X communication technology and adaptive weight fusion method.

[0011] Further, in step S1, a camera and a laser radar are used to obtain road images and point cloud data. A deep learning traffic sign detection method is used to obtain traffic sign detection results in the image. The point cloud data is then preprocessed, and features are extracted for targets in the point cloud that resemble traffic signs. The specific steps include:

[0012] S11: Use cameras and laser radar to obtain road images and point cloud data. First, install the equipment bracket on the top of the intelligent connected car. The camera is a fixed video acquisition and is installed in front of the top of the intelligent connected car. The installation position is 1.79 meters from the ground and the image in front of the vehicle is collected at a rate of 30 frames per second. The laser radar is installed in the center of the roof of the intelligent connected car, 1.99 meters from the ground, and the road point cloud data is collected at a rate of 10 frames per second.

[0013] S12: Use the traffic sign dataset Tsinghua-Tencent 100K to train the YOLOv3 neural network, and use the trained model to obtain image-based traffic sign detection results.

[0014] S13: According to the image detection results, when the image contains traffic signs, the point cloud data and image in the area are captured with a radius of r = 30m around the GPS positioning position as the origin, and then the points in the point cloud of the area are projected onto the image through the following transformation formula according to the correspondence between the image and the point cloud.

[0015]

[0016] in, and They represent the homogeneous coordinates of the midpoints in the world coordinate system and the image coordinate system respectively, s is the scaling factor, G is the intrinsic parameter of the camera, and [R|t] is the extrinsic parameter of the camera, including the rotation parameter R and the translation parameter t.

[0017] These points that can be projected onto the image are considered as potential points where traffic signs exist, and these points are clustered into a rough area containing traffic signs.

[0018] S14: Based on the prior knowledge of traffic signs, the point cloud data in the rough area is first clustered, and then the objects other than the traffic signs are filtered out in the rough area by means of point set counting filtering, height filtering, and shape filtering, so as to obtain a more accurate point cloud area containing the traffic signs.

[0019] S15: In the point cloud region obtained in step S14, based on the shape features of the traffic sign that are different from other targets and the point cloud reflection intensity, feature extraction is performed on the target in the point cloud that is similar to the traffic sign, and a rough point cloud traffic sign detection result is obtained.

[0020] Furthermore, in step S2, according to the image detection results and the target features extracted from the point cloud, the traffic signs in the point cloud are classified using a traffic sign detection method based on image and point cloud fusion to obtain accurate point cloud traffic sign detection results. The specific steps include:

[0021] S21: Based on the detection results of traffic signs in the image obtained in step S12, a classifier is used to first classify traffic signs based on the image. Then, a multi-source adaptive matching method based on Euclidean distance is used to assign the confidence of the traffic sign type in each image according to the Euclidean distance between the intelligent connected vehicle and the traffic sign detected in step S15 in the laser radar coordinate system, and then the label type with the highest probability is selected for the point cloud, as shown in equations (2), (3), and (4). Then, the traffic signs in the point cloud are classified and the false detection results are removed to obtain accurate traffic sign detection results based on the point cloud.

[0022]

[0023]

[0024]

[0025] Among them, L i Indicates the possible types of tags, is the weighting function The maximum value of , indicates the most likely label type in the point cloud. Is the tag type L i Number of repetitions. Is a repeating tag type L i The average confidence of . Δ1 and Δ2 are the weights of these two factors. (d CIV1 ,d CIV2 ) respectively represent the distances between the intelligent connected vehicles CIV1 and CIV2 and the rough traffic sign targets detected by S15.

[0026] Further, in step S3, the traffic sign is located using the position of the intelligent connected vehicle and the distance between the intelligent connected vehicle and the traffic sign, and the positioning results of the traffic sign of the two intelligent connected vehicles are fused through V2X communication technology and an adaptive weight fusion method based on angle trust, so as to obtain an accurate position of the traffic sign. The specific steps include:

[0027] S31: CIV1 and CIV2 can both obtain accurate traffic sign detection results by using the method in step S2, and can use laser radar to obtain the distance between the traffic sign, thereby locating the traffic sign according to its own position, and transmitting the positioning result to CIV1 through V2X communication technology. Then, based on the adaptive weight fusion method of angular trust, the confidence is calculated according to the angular relationship between the two smart cars and the traffic sign, so as to determine the weights of CIV1 and CIV2 for the traffic sign positioning results, and then the traffic sign positioning results are fused according to the weights, as shown below:

[0028]

[0029]

[0030] in, is the fusion result of the traffic sign position, and are the estimation results of traffic sign positions by CIV1 and CIV2 respectively. σ1 and σ2 are the adaptive weights of the positioning results of CIV1 and CIV2 respectively. They are the angular confidence of CIV1 and CIV2 for traffic signs respectively.

[0031] Further, in step S4, CIV1 is positioned, and in the current frame, the position of CIV1 is estimated using the located traffic sign position and the position of the intelligent connected vehicle CIV2. Then, the position of CIV1 is obtained using a positioning method based on local constraints of the previous and next frames. Finally, the multi-party positioning results are fused using V2X communication technology and an adaptive weight fusion method. The specific steps are as follows:

[0032] S41: In the current frame, the position of CIV1 is estimated. 1) Using the laser radar, the distance between CIV1 and the traffic sign is obtained. Since the traffic sign is fixed, CIV1 can estimate its own position based on the position of the traffic sign obtained in step S3; 2) Then, CIV2 estimates the position of CIV1 based on its own position and the distance between it and CIV1, and transmits the positioning result to CIV1 through V2X technology.

[0033] S42: Since the movement of the intelligent connected car changes little between the two frames, the relative distance and angle between the two cars remain unchanged, and remain parallel to the local structure of the previous frame. CIV2 uses the local constraints between the two cars in the previous and next frames to obtain the position of CIV1 in the current frame based on the local constraints of the previous and next frames, and transmits the positioning result to CIV1 through V2X technology. The transformation relationship of the local structure of the previous and next frames is as follows:

[0034]

[0035] in, represents the local structural features between CIV1 and CIV2 in the kth frame (current frame), represents the local distance feature of the kth frame, Represents the local angle feature of the kth frame. represents the local structural features between the k-1th frame (the previous frame CIV1 and CIV2. α represents the transformation relationship matrix.

[0036] S43: After receiving multiple positioning results, CIV1 fuses these results through a multi-source adaptive weight fusion method to obtain an accurate positioning result. The details are as follows:

[0037]

[0038] in, is the fusion result of the three positioning results in steps S41 and S42, are the two positioning results in step S41, is the positioning result obtained in step S42, (σ I ,σ II ,σ III ) are the adaptive weights of the three positioning results S41 and S42 respectively.

[0039] The beneficial effects of the present invention are:

[0040] 1) The present invention uses existing and fixed traffic signs in the city as reference objects, which can reduce the implementation costs of installing specific markers, and at the same time pre-processes the laser radar point cloud, avoiding the processing of huge data when using the laser radar;

[0041] 2) Compared with other existing collaborative positioning methods, the present invention uses traffic signs on the road as reference objects. Traffic signs are stable, in fixed positions for a long time, evenly distributed, and without serious obstruction. They are natural reference objects. When it is necessary to locate the intelligent connected vehicle, it can quickly assist it to find its own position.

[0042] 3) The present invention utilizes V2X communication technology and traffic sign recognition to estimate the position of the intelligent connected vehicle to be located according to the relative distance between the intelligent connected vehicle and the traffic sign in the current frame and the unchanged local constraints between the two intelligent connected vehicles in the previous and next frames, and integrates multiple positioning results, which can effectively improve the positioning accuracy and robustness.

[0043] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0045] Figure 1 This is a main flow chart of the collaborative positioning method according to an embodiment of the present invention;

[0046] Figure 2 A diagram of the installation location of the equipment described in the embodiment of the invention;

[0047] Figure 3 It is a schematic diagram of the positioning method described in an embodiment of the invention. DETAILED DESCRIPTION

[0048] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0049] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0050] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "back" and the like indicate directions or positional relationships, they are based on the directions or positional relationships shown in the drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0051] Specific implementation method Figure 1 As shown, first assume that there are connected and intelligent vehicle 1 (CIV1) and connected and intelligent vehicle 2 (CIV2). At a certain moment, the location of CIV1 needs to be determined, including the following steps:

[0052] Step 1: Use cameras and LiDAR to obtain road images and point cloud data. Use deep learning traffic sign detection methods to obtain traffic sign detection results in the image. Then preprocess the point cloud data and extract features of the targets in the point cloud based on the traffic sign features. The specific steps include:

[0053] 101: Use cameras and laser radar to obtain road images and point cloud data. The camera and laser radar are installed in the following positions: Figure 2 As shown. First, the equipment fixing bracket is installed on the top of the intelligent networked car. The camera is for fixed video acquisition and is installed in front of the top of the intelligent networked car. The installation position is 1.79 meters from the ground and the image in front of the vehicle is collected at a rate of 30 frames per second. The laser radar is installed in the center of the roof of the intelligent networked car. The installation position is 1.99 meters from the ground and the road point cloud data is collected at a rate of 10 frames per second.

[0054] 102: Use the traffic sign dataset Tsinghua-Tencent 100K to train the YOLOv3 neural network, and use the trained model to obtain image-based traffic sign detection results.

[0055] 103: According to the image detection results, when the image contains traffic signs, with r = 30m as the radius, around the GPS positioning position as the origin, capture the point cloud data and image in the area, and then according to the correspondence between the image and the point cloud, project the points in the point cloud of the area onto the image through the following transformation formula.

[0056]

[0057] in, and They represent the homogeneous coordinates of the midpoints in the world coordinate system and the image coordinate system respectively, s is the scaling factor, G is the intrinsic parameter of the camera, and [R|t] is the extrinsic parameter of the camera, including the rotation parameter R and the translation parameter t.

[0058] These points that can be projected onto the image are considered as potential points where traffic signs exist, and these points are clustered into a rough area containing traffic signs.

[0059] 104: Based on the prior knowledge of traffic signs, the point cloud data in the rough area is first clustered, and then the objects other than the traffic signs are filtered out in the rough area by point set counting filtering, height filtering and shape filtering, so as to obtain a more accurate point cloud area containing the traffic signs.

[0060] 105: In the above point cloud area, according to the shape characteristics of the traffic sign that are different from other targets and the point cloud reflection intensity, the target similar to the traffic sign in the point cloud is extracted, and a rough point cloud traffic sign detection result is obtained.

[0061] Step 2: Based on the image detection results and the target features extracted from the point cloud, the traffic signs in the point cloud are classified using the traffic sign detection method based on image and point cloud fusion to obtain accurate point cloud traffic sign detection results;

[0062] Based on the detection results of traffic signs in the image obtained in step 102, a classifier is used to first classify traffic signs based on the image. Then, a multi-source adaptive matching method based on Euclidean distance is used to assign the confidence of the traffic sign type in each image according to the Euclidean distance between the intelligent connected vehicle and the traffic sign detected in step 105 in the laser radar coordinate system, and then the label type with the highest probability is selected for the point cloud, as shown in equations (2), (3), and (4). Then, the traffic signs in the point cloud are classified, and the false detection results are removed to obtain accurate traffic sign detection results based on the point cloud.

[0063]

[0064]

[0065]

[0066] Among them, L i Indicates the possible types of tags, is the weighting function The maximum value of , indicates the most likely label type in the point cloud. Is the tag type L i Number of repetitions. Is a repeating tag type L i The average confidence of . Δ1 and Δ2 are the weights of these two factors. (d CIV1 ,d CIV2 ) respectively represent the distances between the intelligent connected vehicles CIV1 and CIV2 and the rough traffic sign targets detected by 105.

[0067] Step 3: Use the position of the intelligent connected vehicle and the distance between the vehicle and the traffic sign to locate the traffic sign, and use V2X communication technology and an adaptive weight fusion method based on angular trust to fuse the positioning results of the two intelligent connected vehicles for the traffic sign, so as to obtain the accurate position of the traffic sign, specifically including: CIV1 and CIV2 can obtain accurate traffic sign detection results using the method of step 2, and can use laser radar to obtain the distance between the traffic sign, so as to locate the traffic sign according to its own position, and transmit the positioning result to CIV1 through V2X communication technology. Then, based on the adaptive weight fusion method of angular trust, the confidence is calculated according to the angular relationship between the two intelligent vehicles and the traffic sign, so as to determine the weights of CIV1 and CIV2 for the positioning results of the traffic sign, and then the positioning results of the traffic sign are fused according to the weights, as shown below:

[0068]

[0069]

[0070] in, is the fusion result of the traffic sign position, and are the estimation results of traffic sign positions by CIV1 and CIV2 respectively. σ1 and σ2 are the adaptive weights of the positioning results of CIV1 and CIV2 respectively. They are the angular confidence of CIV1 and CIV2 for traffic signs respectively.

[0071] Step 4: Position CIV1. In the current frame, use the located traffic sign position and the position of the intelligent connected car CIV2 to estimate the position of CIV1. Then, use the positioning method based on the local constraints of the previous and next frames to get the position of CIV1. Finally, use V2X communication technology and adaptive weight fusion method to fuse the multi-party positioning results. The overall scheme diagram is as follows: Figure 3The specific steps are as follows:

[0072] 401: In the current frame, the position of CIV1 is estimated. 1) Using the laser radar, the distance between CIV1 and the traffic sign is obtained. Since the traffic sign is fixed, CIV1 can estimate its own position based on the position of the traffic sign obtained in step 3; 2) Then, CIV2 estimates the position of CIV1 based on its own position and the distance between it and CIV1, and transmits the positioning result to CIV1 through V2X technology.

[0073] 402: Since the movement of the intelligent connected car changes little between the two frames, the relative distance and angle between the two cars remain unchanged, and remain parallel to the local structure of the previous frame. CIV2 uses the local constraints between the two cars in the previous and next frames to obtain the position of CIV1 in the current frame based on the local constraints of the previous and next frames, and transmits the positioning result to CIV1 through V2X technology. The transformation relationship of the local structure of the previous and next frames is as follows:

[0074]

[0075] in, represents the local structural features between CIV1 and CIV2 in the kth frame (current frame), represents the local distance feature of the kth frame, Represents the local angle feature of the kth frame. represents the local structural features between the k-1th frame (the previous frame CIV1 and CIV2. α represents the transformation relationship matrix.

[0076] 403: After receiving multiple positioning results, CIV1 fuses these results through a multi-source adaptive weight fusion method to obtain an accurate positioning result. The details are as follows:

[0077]

[0078] in, is the fusion result of the three positioning results in steps 401 and 402. are the two positioning results in step 401, is the positioning result obtained in step 402, (σ I ,σ II ,σ III ) are the adaptive weights of the three positioning results in steps 401 and 402 respectively.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A collaborative positioning method for intelligent connected vehicles based on V2X communication and traffic sign recognition, characterized in that: Based on two intelligent networked vehicles CIV1 and CIV2, at a certain moment, positioning one of the intelligent networked vehicles includes the following steps: S1: Use cameras and lidar to obtain road images and point cloud data, use deep learning traffic sign detection methods to obtain traffic sign detection results in the image, then preprocess the point cloud data and extract features of the targets in the point cloud based on traffic sign features; S2: Based on the image detection results and the target features extracted from the point cloud, the traffic signs in the point cloud are classified using the traffic sign detection method based on image and point cloud fusion to obtain accurate point cloud traffic sign detection results; S3: using the position of the intelligent connected vehicle and the distance between the vehicle and the traffic sign, positioning the traffic sign, and integrating the positioning results of the two intelligent connected vehicles for the traffic sign through V2X communication technology and an adaptive weight fusion method based on angle trust; the step S3 specifically includes: CIV1 and CIV2 use the method in step S2 to obtain accurate traffic sign detection results, and use laser radar to obtain the distance between the traffic sign, so as to locate the traffic sign according to their own position, and transmit the positioning result to CIV1 through V2X communication technology; then, based on the adaptive weight fusion method of angular trust, the confidence is calculated according to the angular relationship between CIV1, CIV2 and the traffic sign, so as to determine the weights of CIV1 and CIV2 for the traffic sign positioning results, and then the positioning results of the traffic sign are fused according to the weights, as shown below: in, is the fusion result of the traffic sign position, and are the estimation results of traffic sign positions by CIV1 and CIV2 respectively; σ1 and σ2 are the adaptive weights of the positioning results of CIV1 and CIV2 respectively; are the angular confidences of CIV1 and CIV2 for traffic signs, respectively; S4: In the current frame, the position of the intelligent connected vehicle to be located is estimated using the located traffic sign position and the intelligent connected vehicle position; the position of the intelligent connected vehicle to be located is obtained using a positioning method based on local constraints of previous and next frames; and finally, multiple positioning results are fused using V2X communication technology and an adaptive weight fusion method. The step S4 specifically includes the following steps: S41: In the current frame, estimate the position of CIV1: 1) Using laser radar, the distance between CIV1 and the traffic sign is obtained, and CIV1 estimates its own position according to the position of the traffic sign obtained in step S3; 2) Then, CIV2 estimates the location of CIV1 based on its own location and the distance between it and CIV1, and transmits the positioning result to CIV1 through V2X technology; S42: CIV2 uses the unchanged local constraints between the two vehicles in the previous and next frames, and obtains the position of CIV1 in the current frame based on the local constraints of the previous and next frames. The positioning result is transmitted to CIV1 through V2X technology. The transformation relationship of the local structure of the previous and next frames is as follows: in, represents the local structural features between the kth frames CIV1 and CIV2, represents the local distance feature of the kth frame, represents the local angle feature of the kth frame, represents the local structural features between the k-1th frame CIV1 and CIV2, and α represents the transformation relationship matrix; S43: After receiving multiple positioning results, CIV1 fuses these results through a multi-source adaptive weight fusion method to obtain an accurate positioning result, as shown below: in, is the fusion result of the three positioning results in steps S41 and S42, are the two positioning results in step S41, is the positioning result obtained in step S42, (σ I ,σ II ,σ III ) are the adaptive weights of the three positioning results S41 and S42 respectively.

2. The intelligent connected vehicle collaborative positioning method based on V2X communication and traffic sign recognition according to claim 1, characterized in that: The step S1 specifically includes the following steps: S11: Install the camera in front of the top of the intelligent connected car to collect images in front of the vehicle; install the laser radar in the center of the roof of the intelligent connected car to collect road point cloud data; S12: Use the traffic sign dataset Tsinghua-Tencent 100K to train the YOLOv3 neural network, and use the trained model to obtain image-based traffic sign detection results; S13: According to the image detection results, when the image contains traffic signs, take r as the radius and the GPS positioning position as the origin to capture the point cloud data and image in the area, and then project the points in the point cloud of the area onto the image through the following transformation formula according to the correspondence between the image and the point cloud: in, and Represent the homogeneous coordinates of the point in the world coordinate system and the image coordinate system respectively, s is the scaling factor, G is the intrinsic parameter of the camera, [R|t] is the extrinsic parameter of the camera, including the rotation parameter R and the translation parameter t; the points projected on the image are considered to be potential points where traffic signs exist, and these points are aggregated into a rough area containing traffic signs; S14: Based on the prior knowledge of traffic signs, the point cloud data in the rough area is first clustered, and then the objects other than the traffic signs are filtered out in the rough area to obtain the precise point cloud area containing the traffic signs; S15: In the point cloud region obtained in step S14, based on the shape characteristics of the traffic sign that are different from other targets and the point cloud reflection intensity, features of the target that resembles the traffic sign in the point cloud are extracted, and a rough point cloud traffic sign detection result is obtained.

3. The intelligent connected vehicle collaborative positioning method based on V2X communication and traffic sign recognition according to claim 1, characterized in that: The step S2 specifically includes: based on the detection result of the traffic sign in the image obtained in step S12, using a classifier to classify the traffic sign based on the image; then using a multi-source adaptive matching method based on Euclidean distance, according to the Euclidean distance between the intelligent connected vehicle and the traffic sign detected in step S15 in the laser radar coordinate system, assigning the confidence of the traffic sign type in each image, and then selecting the label type with the highest probability for the point cloud, such as formula (2), (3), (4); and then classifying the traffic signs in the point cloud and removing the false detection results to obtain an accurate point cloud-based traffic sign detection result. Among them, L i Indicates the possible types of tags, is the weighting function W Li The maximum value of indicates the most likely label type in the point cloud; Is the tag type L i Number of repetitions; Is a repeating tag type L i The average confidence of; Δ1 and Δ2 are the weights of these two factors; (d CIV1 ,d CIV2 ) respectively represent the distances between the two intelligent connected vehicles CIV1 and CIV2 and the rough traffic sign target detected in step S15.

Images