Self-Calibrating Intelligent Traffic Lights and Self-Calibrating Intelligent Traffic Light Control Method

Abstract

The present invention relates to a self-calibrating intelligent traffic light and a self-calibrating intelligent traffic light control method, which are applied to the field of intelligent transportation technology. The self-calibrating intelligent traffic light includes: a lamp post main body and a mobile device cabin, the lamp post main body is installed on the mobile device cabin, traffic lights, a data acquisition mechanism and a communication device are installed on the lamp post main body, a calculation and control system is arranged in the mobile device cabin, the data acquisition mechanism is electrically connected to the communication device and the calculation and control system, and the calculation and control system is electrically connected to the communication device and the traffic lights. This application has the effect of enabling traffic signal lights to be flexibly and quickly put into use.

Description

Technical Field

[0001] This application relates to the technical field of intelligent transportation, and in particular, to a self-calibrating intelligent traffic light and a self-calibrating intelligent traffic light control method. Background Art

[0002] Traffic signal lights are an essential part of the traffic system. In the current traffic system, traffic signal lights usually adopt a fixed setting method, that is, fixed traffic signal lights are set at intersections, relevant road information is calibrated in the traffic signal lights, and detection devices are installed on the traffic signal lights.

[0003] However, since more and more people choose to travel by car now, the traffic conditions have become more and more complex, and the fixed traffic signal lights in use gradually cannot meet the complex and changing traffic conditions. There is an urgent need for a traffic signal light that can be put into use flexibly and quickly. Summary of the Invention

[0004] To meet the need for traffic signal lights to be put into use flexibly and quickly, this application provides a self-calibrating intelligent traffic light and a self-calibrating intelligent traffic light control method.

[0005] In a first aspect, this application provides a self-calibrating intelligent traffic light, adopting the following technical solution:

[0006] A self-calibrating intelligent traffic light includes:

[0007] A lamp post main body and a mobile device cabin. The lamp post main body is installed on the mobile device cabin. Traffic lights, a data acquisition mechanism, and a communication device are installed on the lamp post main body. A calculation and control system is arranged in the mobile device cabin. The data acquisition mechanism is electrically connected to the communication device and the calculation and control system, and the calculation and control system is electrically connected to the communication device and the traffic lights.

[0008] By adopting the above technical solution, the mobile device cabin and the lamp post main body are combined to obtain a movable traffic light, and a data acquisition mechanism and a communication device are provided. When in use, it can be moved to any position as needed. At the same time, the data acquisition mechanism and the calculation and control system collect and control data, and automatically perform calculation and processing according to the actual application environment, so as to meet the need for traffic signal lights to be put into use flexibly and quickly.

[0009] Optionally, the data acquisition mechanism includes a lidar, a position and pose detection device, an image detection device, and a millimeter wave radar. The lidar, the position and pose detection device, the image detection device, and the millimeter wave radar are all installed on the top of the lamp post main body.

[0010] Optionally, the calculation and control system includes a calculation unit and a controller, and the calculation unit is electrically connected to the controller.

[0011] In a second aspect, the present application provides a self-calibrating intelligent traffic light control method, adopting the following technical solutions:

[0012] A self-calibrating intelligent traffic light control method is applied to a self-calibrating intelligent traffic light. The self-calibrating intelligent traffic light control method includes:

[0013] Obtain the initial pose information and surrounding environment information of the target mapping device;

[0014] Construct an environmental point cloud map based on the initial pose information and the surrounding environment information;

[0015] Introduce the target traffic light;

[0016] The target traffic light performs self-localization based on the environmental point cloud map to generate traffic light self-localization information.

[0017] By adopting the above technical solutions, an environmental point cloud map is constructed according to the initial pose information and the surrounding environment information of the mapping device. After placing the target traffic light at the corresponding position during use, the target traffic light performs self-localization processing, which can quickly determine its position in the environmental point cloud map without manual adjustment, and can flexibly locate and use according to the actual use environment, thus meeting the requirement that the traffic signal light can be put into use flexibly and quickly.

[0018] Optionally, the constructing the environmental point cloud map based on the initial pose information and the surrounding environment information includes:

[0019] Generate initial point cloud data based on the surrounding environment information;

[0020] Perform data preprocessing on the initial pose information to generate corrected pose information;

[0021] Obtain the motion state of the target mapping device;

[0022] Perform point cloud prediction processing based on the motion state, the corrected pose information, and the initial point cloud data to generate a predicted point cloud map;

[0023] Perform optimization processing on the predicted point cloud map to generate an environmental point cloud map.

[0024] Optionally, the performing point cloud prediction processing based on the corrected pose information and the initial point cloud data to generate a predicted point cloud map includes:

[0025] Estimate the displacement information of the target mapping device based on the initial point cloud data and a preset registration algorithm;

[0026] Generate an odometry estimation result based on the corrected pose information and the displacement information;

[0027] Generate predicted pose information based on a preset motion model and the corrected position information;

[0028] Perform state correction on the predicted pose information based on the initial point cloud data and a preset filter to generate an accurate predicted pose;

[0029] Register the initial point cloud data, the accurate predicted position, and the odometry estimation result for each frame to generate an estimated point cloud map.

[0030] Optionally, the self - localizing the target traffic light based on the environmental point cloud map to generate traffic light self - localization information includes:

[0031] Obtain the localization information and traffic light point cloud data of the target traffic light;

[0032] Register the traffic light point cloud data with the environmental point cloud map to generate a registration result;

[0033] Generate traffic light self - localization information based on the registration result and the localization information.

[0034] Optionally, the registering the traffic light point cloud data with the environmental point cloud map to generate a registration result includes:

[0035] Construct a pose transformation matrix based on the traffic light point cloud data and the environmental point cloud map;

[0036] Register the traffic light point cloud data and the environmental point cloud map based on the pose transformation matrix and a maximum likelihood estimation model to generate a registration result.

[0037] Optionally, after constructing the environmental point cloud map based on the initial pose information and the surrounding environment information, it further includes:

[0038] In response to the user's box - selection calibration operation, obtain the box - selected area and the area identifier corresponding to the box - selected area;

[0039] Determine the area coordinates of the box - selected area based on the environmental point cloud map;

[0040] Obtain the area type of the box - selected area;

[0041] Store the box - selected area, the area identifier, the area coordinates, and the area type, and mark them on the environmental point cloud map.

[0042] Optionally, after self - positioning the target traffic light based on the environmental point cloud map to generate traffic light self - positioning information, the following steps are further included:

[0043] Obtain the structural pose relationship between the data acquisition mechanism and the main body of the lamp post;

[0044] Construct a local coordinate system, where the local coordinate system includes the lamp post main body coordinate system and the data acquisition mechanism coordinate system;

[0045] Construct a transformation matrix between the lamp post main body coordinate system and the data acquisition mechanism;

[0046] Determine the device self - positioning information of the data acquisition mechanism based on the transformation matrix.

[0047] In summary, the present application includes at least one of the following beneficial technical effects:

[0048] 1. Combine the mobile device cabin and the main body of the lamp post to obtain a movable traffic light, and be equipped with a data acquisition mechanism and communication equipment. When in use, it can be moved to any position as needed. At the same time, the data acquisition mechanism and the calculation control system collect and control data, and automatically perform calculation processing according to the actual application environment, so as to meet the requirement that traffic signal lights can be put into use flexibly and quickly;

[0049] 2. Construct an environmental point cloud map based on the initial pose information of the mapping device and the surrounding environment information. After placing the target traffic light at the corresponding position during use, the target traffic light performs self - positioning processing, can quickly determine its own position in the environmental point cloud map, without manual adjustment, and can be quickly positioned and used according to the actual use environment, so as to meet the requirement that traffic signal lights can be put into use flexibly and quickly. Description of the Drawings

[0050] Figure 1 is a schematic structural diagram of a self - calibrating intelligent traffic light provided by an embodiment of the present application.

[0051] Figure 2 is a structural block diagram of a self - calibrating intelligent traffic light provided by an embodiment of the present application.

[0052] Figure 3 is a schematic flowchart of a self - calibrating intelligent traffic light method provided by an embodiment of the present application.

[0053] Explanation of the reference numerals: 1. Main body of the lamp post; 2. Mobile device cabin; 3. Traffic lights; 4. Data acquisition mechanism; 41. Lidar; 42. Position and pose detection device; 43. Image detection device; 44. Millimeter - wave radar; 5. Communication equipment; 6. Calculation control system; 61. Calculation unit; 62. Controller. Detailed implementation manners

[0054] The following further elaborates on this application with reference to the accompanying drawings.

[0055] An embodiment of this application discloses a self-calibrating intelligent traffic light. Referring to Figure 1 and Figure 2 , the self-calibrating intelligent traffic light mainly includes two main parts: a lamp post main body 1 and a mobile device cabin 2. Among them, the mobile device cabin 2 is a box body equipped with mobile wheels. The box body is a rectangular box such as a cube or a cuboid. The lamp post main body 1 is installed above the mobile device cabin 2. By pushing the mobile device cabin 2, the lamp post main body 1 is driven to move together, so as to realize the switching of the usage position of the self-calibrating intelligent traffic light. A traffic signal light 3 for traffic indication, a data acquisition mechanism 4 and a communication device 5 are installed on the lamp post main body 1. Both the data acquisition mechanism 4 and the communication device 5 are installed at one end of the lamp post main body 1 away from the ground. A calculation and control system 6 is arranged in the mobile device cabin 2. The data acquisition mechanism 4 is electrically connected to the communication device 5 and the calculation and control system 6. The calculation and control system 6 is electrically connected to the communication device 5 and the traffic signal light 3. Among them, the communication device 5 is specifically a roadside unit RSU, which is used to realize basic vehicle identity recognition and electronic point deduction and other functions.

[0056] Referring to Figure 1 , the data acquisition mechanism 4 includes a lidar 41 for constructing a point cloud map, a position and pose detection device 42 for determining the position and pose, an image detection device 43 for collecting environmental images, and a millimeter wave radar 44 for detecting objects and obstacles in the environment. Among them, the position and pose detection device 42 is specifically a nine-axis gyroscope, and the image detection device 43 is specifically a camera.

[0057] Referring to Figure 1 and Figure 2 , the calculation and control system 6 is installed inside the mobile device cabin 2, and mainly includes a calculation unit 61 and a controller 62. The calculation unit 61 is mainly used for constructing a point cloud map and realizing self-calibration. The controller 62 is electrically connected to the calculation unit 61. At the same time, the controller 62 is electrically connected to the lidar 41, the position and pose detection device 42, the image detection device 43, the millimeter wave radar 44 and the communication device 5. The controller 62 transmits the data collected by the above devices to the calculation unit 61, and the calculation unit 61 performs calculation and processing.

[0058] An embodiment of the present application provides a self-calibrating intelligent traffic light control method. This self-calibrating intelligent traffic light control method can be executed by an electronic device, which can be a server or a terminal device. The server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smart phone, a tablet computer, a desktop computer, etc., but is not limited thereto.

[0059] Figure 1 It is a schematic flowchart of a self-calibrating intelligent traffic light control method provided by an embodiment of the present application.

[0060] As Figure 1 shown, the main process of this method is described as follows (Steps S101 - S104):

[0061] Step S101, obtain the initial pose information and surrounding environment information of the target mapping device.

[0062] In this embodiment, the target mapping device is also an intelligent traffic light with self-calibrating function. When creating an environmental map, one traffic light can be used for creation, or another traffic light can be created and put into use through one traffic light.

[0063] Step S102, construct an environmental point cloud map based on the initial pose information and surrounding environment information.

[0064] Regarding Step S102, generate initial point cloud data based on the surrounding environment information; perform data preprocessing on the initial pose information to generate corrected pose information; obtain the motion state of the target mapping device; perform point cloud prediction processing based on the motion state, corrected pose information, and initial point cloud data to generate a predicted point cloud map; perform optimization processing on the predicted point cloud map to generate an environmental point cloud map.

[0065] Furthermore, estimate the displacement information of the target mapping device based on the initial point cloud data and a preset registration algorithm; generate a mileage prediction result based on the corrected pose information and displacement information; generate predicted pose information based on a preset motion model and corrected position information; perform state correction on the predicted pose information based on the initial point cloud data and a preset filter to generate an accurate predicted pose; register the initial point cloud data, accurate predicted position, and mileage prediction result of each frame to generate a predicted point cloud map.

[0066] Obtain the initial pose information of the target mapping device through a nine-axis gyroscope, i.e., a position and pose detection device. The initial pose information includes acceleration, angular velocity, magnetic field data, etc., and provides the motion state of the target mapping device. Scan the surrounding environment during the movement process using a lidar to generate high-precision three-dimensional point cloud data, and use this unit point cloud data as the initial point cloud data. It should be noted that the initial point cloud data here is not the point cloud data at a single position, but the point cloud data during the entire movement process, that is, the point cloud data containing an entire movement environment.

[0067] Filter and calibrate the initial pose information collected by the position and pose detection device to remove noise and errors, and obtain accurate pose information, that is, corrected pose information. Synchronize and register the initial point cloud data collected by the lidar to ensure that the corrected pose information is aligned with the initial point cloud data in time.

[0068] Use the initial point cloud data of the lidar to perform point cloud registration through the ICP (Iterative Closest Point) or NDT (Normal Distributions Transform) algorithm to estimate the relative displacement and rotation of the target mapping device in each frame. Combine the corrected pose information provided by the position and pose detection device with the displacement information of the lidar to obtain a more accurate odometry estimation result.

[0069] After that, use a preset motion model and corrected position information to make predictions and generate predicted pose information. The predicted position information is the position and pose of the target mapping device at the next moment. Fuse the predicted pose information with the odometry estimation result, and use a Kalman Filter (KF) or Extended Kalman Filter (EKF) to perform state correction to obtain an accurate predicted pose. Register the point cloud data of each frame according to the result of the odometry estimation, and accumulate the accurate predicted pose of each frame frame by frame to construct a predicted point cloud map. Use graph optimization technology to optimize the predicted point cloud map and correct the cumulative error to obtain an accurate environmental map. Since the target mapping device is moving in real time, the LIO-SAM algorithm runs in real time. Every time a new frame of data is obtained, the above-mentioned odometry estimation, state prediction and correction, and map update are performed. Through continuous iteration, the pose information of the target mapping device and the environmental map are updated in real time until the final location is reached. The accurate environmental map obtained after stopping moving is used as the environmental point cloud map.

[0070] Step S103, introduce the target traffic light.

[0071] In this embodiment, the target traffic light is the traffic light for official use, which can be the same as the target mapping device or a new traffic light that is separate and does not participate in the environmental point cloud map. The introduced target traffic light has the same functions and hardware facilities as the mapping device, that is, it has a lamp post main body, a mobile device cabin, traffic lights, a data acquisition mechanism, and a communication device.

[0072] Step S104: The target traffic light performs self-localization based on the environmental point cloud map to generate traffic light self-localization information.

[0073] Regarding step S104, obtain the positioning information of the target traffic light and the traffic light point cloud data; register the traffic light point cloud data with the environmental point cloud map to generate a registration result; generate traffic light self-localization information based on the registration result and the positioning information.

[0074] Furthermore, construct a pose transformation matrix based on the traffic light point cloud data and the environmental point cloud map; register the traffic light point cloud data and the environmental point cloud map based on the pose transformation matrix and the maximum likelihood estimation model to generate a registration result.

[0075] In the positioning modeling, the core idea is to obtain the pose information of the lamp post main body in the environmental point cloud map in real time, that is, x, y, z, roll (roll angle), yaw (yaw angle), pitch (pitch angle). Since the pose information can be regarded as the target mapping device and the target traffic making translational and rotational movements, given a motion model, parameters need to be calculated to construct a 3x4 pose transformation matrix T. Therefore, the method adopted in this application is to solve the local optimal solution through MLE (Maximum Likelihood Estimation).

[0076] Since it is a local optimal solution, an initial value needs to be given to make the final result approach the global optimal solution. In this application, relatively rough positioning information provided by roadside units (RSUs) is usually used, or other additional devices with positioning functions are used to obtain positioning information. This positioning information is used as the starting point for searching, so as to calculate the pose of the lamp post body. The maximum likelihood estimation MLE(xi|θ) model, that is, the maximum probability of the occurrence of a given sample x under different model parameters θ. In the maximum likelihood estimation MLE model, xi has been set to the pose transformation matrix T. After the definition of xi is given, the definition of θ also needs to be given. Since point cloud data is compared, each frame of point cloud data has at least hundreds of thousands of n-dimensional data (n>=3). In order to ensure real-time performance and not lose the characteristics of the data, a rasterization method is adopted. The space formed by the entire point cloud is divided into a large number of grids, and the attribute of each grid is the centroid of the point cloud within the grid, thus completing data distillation. In order to improve the practicality and applicability of the technical implementation, when calculating the probability density function, a Gaussian distribution method is used for calculation, which can effectively reduce the calculation amount of registration. Since each grid has a probability density function, all the probability density functions are multiplied to obtain the product of the probability density functions, thus obtaining the final maximum likelihood estimation MLE mathematical model. Solving the maximum likelihood estimation MLE model can obtain the required pose transformation matrix T after maximum likelihood estimation. Therefore, by taking the logarithm on both sides, the power can be quickly reduced to obtain the optimization objective function, which is convenient for obtaining the first and second derivatives and solving. In the process of numerical calculation, the Gauss-Newton method is used in this application to complete the least squares regression of the parameters. Through the above calculation and processing, the traffic light point cloud data and the environmental point cloud map are registered, so that the coordinates of each point cloud data in the traffic light point cloud data correspond to the coordinates of each point cloud data in the environmental point cloud map. For example, the coordinates of point cloud data A in the traffic light point cloud data are the same as the coordinates of point cloud data B in the environmental point cloud map. According to the registration result and the positioning information, the pose of the traffic light in the environmental point cloud map is determined, and this pose is used as the self-positioning information of the traffic light.

[0077] In this embodiment, in response to the user's box selection and calibration operation, the box selection area and the area identifier corresponding to the box selection area are obtained; the area coordinates of the box selection area are determined based on the environmental point cloud map; the area type of the box selection area is obtained; the box selection area, the area identifier, the area coordinates, and the area type are stored and marked on the environmental point cloud map.

[0078] To provide more complete functions such as route planning for autonomous vehicles, a high-precision map is constructed based on the created environmental point cloud map. The high-precision map includes road information such as lane lines, crosswalks, and road signs, and the monitoring area information is entered, that is, the monitoring area information is marked on the high-precision map. The monitoring area can be a lane, a crosswalk, etc. When marking on the high-precision map, it is not necessary to measure and input intersection data in the actual environment, and the marking process can be achieved simply by manual selection on the map.

[0079] Through the box selection tool for area selection, using the graphical user interface (GUI) tool or program function, the user can accurately select the position and boundary of the monitoring area on the environmental point cloud map with a mouse or touch screen. When the user uses the box selection tool to select an area on the point cloud map, the area coordinates of all selected points are recorded. Among them, the selected coordinate points should be based on the map coordinate system to ensure the accuracy and consistency of the position. An area identifier is assigned to each selected area. The area identifier is the area ID of the selected area, and each area identifier is unique and cannot be reused. When assigning the area identifier, it can be assigned by the user, randomly generated and assigned, or generated according to the area type of the selected area by the user, and it needs to be set according to actual needs, which is not specifically limited here. Among them, the area type includes lanes, sidewalks, intersections, etc. The user selects according to the preset area type list, and the area identifier can be a single number, a single letter, or a combination of numbers and letters.

[0080] After the area selection is made to determine the selected area, the corresponding area identifier and area type, the above data is stored. The information of each selected area is stored using a data structure or a database table. After storage, the stored information can also be processed for addition, deletion, modification, and query. Thus, it can effectively support the perception and analysis tasks of specific areas and enhance the application value and reliability of real-time monitoring and data analysis.

[0081] In this embodiment, the structural pose relationship between the data acquisition mechanism and the lamp post body is obtained; a local coordinate system is constructed, where the local coordinate system includes the lamp post body coordinate system and the data acquisition mechanism coordinate system; a transformation matrix between the lamp post body coordinate system and the data acquisition mechanism is constructed; and the device self-positioning information of the data acquisition mechanism is determined based on the transformation matrix.

[0082] After the target traffic light completes self-positioning in the environmental point cloud map, the position and pose of the target traffic light are obtained. Based on the structural pose relationship between the lidar, the position and pose detection device, the image detection device, the millimeter-wave radar, and the lamp post body, the pose relationship of each device is calculated using the transformation matrix, so as to achieve the automatic calibration of the overall traffic light without the need to calibrate each device separately.

[0083] First, construct an initial transformation matrix and define a coordinate system. A local coordinate system is defined for each device. The local coordinate system includes a pole coordinate system and a data acquisition mechanism coordinate system, specifically the pole coordinate system (Pole, P), the lidar coordinate system (LIDAR, L), the millimeter-wave radar coordinate system (Millimeter Wave Radar, MW), the position and pose detection device coordinate system (IMU, I), and the image detection device system (Camera, C). After determining the local coordinate system, calculate the transformation matrix. Based on the known geometric relationship between the pole body and the data acquisition mechanism, calculate the initial transformation matrix. The transformation matrix T_PL from the pole to the lidar is: (T_{PL} = \begin{bmatrix}R_{PL}&t_{PL}\\0&1\end{bmatrix}), where R_PL is the rotation matrix, calculated from the angles θx, θy, θz, and t_PL is the translation vector (x, y, z). Similarly, calculate the transformation matrices between other devices. The transformation matrix T_PMW from the pole body to the millimeter-wave radar is T_{PMW} = \begin{bmatrix}R_{PMW}&t_{PMW}\\0&1\end{bmatrix}, the transformation matrix T_PI from the pole body to the position and pose detection device is T_{PI} = \begin{bmatrix}R_{PI}&t_{PI}\\0&1\end{bmatrix}, and the transformation matrix T_PC from the pole body to the image detection device is T_{PC} = \begin{bmatrix}R_{PC}&t_{PC}\\0&1\end{bmatrix}.

[0084] After that, use the initial transformation matrix to perform mutual transformation for each device. For example, transform the pose of the lidar to the pose of the pole body, that is, transform the pose P_L of the lidar to the pose P_P of the pole through the transformation matrix T_PL, P_{P} = T_{PL} * P_{L}. For example, transform the lidar pose to the millimeter-wave radar pose: P_{MW} = T_{PMW} * P_{P}, etc.

[0085] The term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0086] The above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the application involved in the present application is not limited to the technical solution formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned application concept. For example, the technical solution formed by mutually replacing the above features with the technical features (but not limited to) having similar functions applied in the present application.

Claims

1. A self-calibrating intelligent traffic light control method, characterized in that, Applied to self-calibrating intelligent traffic lights, the self-calibrating intelligent traffic lights include: A lamp post main body (1) and a mobile device cabin (2), the lamp post main body (1) is installed on the mobile device cabin (2), a traffic light (3), a data acquisition mechanism (4) and a communication device (5) are installed on the lamp post main body (1), a calculation and control system (6) is arranged in the mobile device cabin (2), the data acquisition mechanism (4) is electrically connected to the communication device (5) and the calculation and control system (6), and the calculation and control system (6) is electrically connected to the communication device (5) and the traffic light (3); The data acquisition mechanism (4) includes a lidar (41), a position and pose detection device (42), an image detection device (43) and a millimeter wave radar (44), and the lidar (41), the position and pose detection device (42), the image detection device (43) and the millimeter wave radar (44) are all installed on the top of the lamp post main body (1); The calculation and control system (6) includes a calculation unit (61) and a controller (62), and the calculation unit (61) is electrically connected to the controller (62); The self-calibrating intelligent traffic light control method includes: Obtain the initial pose information and surrounding environment information of the target mapping device; Construct an environmental point cloud map based on the initial pose information and the surrounding environment information; Introduce the target traffic light; The target traffic light performs self-positioning based on the environmental point cloud map to generate traffic light self-positioning information; After constructing the environmental point cloud map based on the initial pose information and the surrounding environment information, it further includes: In response to the user's box selection and calibration operation, obtain the box selection area and the area identifier corresponding to the box selection area; Determine the area coordinates of the box selection area based on the environmental point cloud map; Obtain the area type of the box selection area; Store the box selection area, the area identifier, the area coordinates and the area type, and mark them on the environmental point cloud map; After performing self-positioning on the target traffic light based on the environmental point cloud map to generate traffic light self-positioning information, it further includes: Obtain the structural pose relationship between the data acquisition mechanism and the lamp post main body; Construct a local coordinate system, where the local coordinate system includes a lamp post main body coordinate system and a data acquisition mechanism coordinate system; Construct a transformation matrix between the lamp post main body coordinate system and the data acquisition mechanism; Determine the device self-positioning information of the data acquisition mechanism based on the transformation matrix.

2. The method according to claim 1, characterized in that, Constructing the environmental point cloud map based on the initial pose information and the surrounding environment information includes: Generate initial point cloud data based on the surrounding environment information; Perform data preprocessing on the initial pose information to generate corrected pose information; Obtain the motion state of the target mapping device; Perform point cloud prediction processing based on the motion state, the corrected pose information and the initial point cloud data to generate a predicted point cloud map; Perform optimization processing on the predicted point cloud map to generate an environmental point cloud map.

3. The method according to claim 2, wherein Performing point cloud prediction processing based on the corrected pose information and the initial point cloud data to generate a predicted point cloud map includes: Predicting the displacement information of the target mapping device based on the initial point cloud data and a preset registration algorithm; Generating a mileage prediction result based on the corrected pose information and the displacement information; Generating predicted pose information based on a preset motion model and the corrected pose information; Performing state correction on the predicted pose information based on the initial point cloud data and a preset filter to generate an accurate predicted pose; Registering the initial point cloud data, the accurate predicted pose, and the mileage prediction result of each frame to generate a predicted point cloud map.

4. The method according to claim 1, wherein Performing self-localization on the target traffic light based on the environmental point cloud map to generate traffic light self-localization information includes: Obtaining the localization information of the target traffic light and traffic light point cloud data; Registering the traffic light point cloud data with the environmental point cloud map to generate a registration result; Generating traffic light self-localization information based on the registration result and the localization information.

5. The method according to claim 4, characterized in that, The registering the traffic light point cloud data with the environmental point cloud map to generate a registration result includes: Constructing a pose transformation matrix based on the traffic light point cloud data and the environmental point cloud map; Registering the traffic light point cloud data and the environmental point cloud map based on the pose transformation matrix and a maximum likelihood estimation model to generate a registration result.

Images