High-precision positioning method and system based on multi-source fusion technology, and vehicle

Abstract

The present invention relates to a high-precision positioning method and system based on multi-source fusion technology, and a vehicle. By obtaining a point cloud around a vehicle and obtaining a point cloud map by means of point cloud mapping, position information of the vehicle in the point cloud map is obtained; actual point cloud position information is compared with pre-correction point cloud position information corresponding to a pre-correction position of the vehicle in the point cloud map to obtain a positioning offset; and the position of the vehicle is finally corrected by means of the positioning offset to obtain the final positioning of the vehicle. By means of fusing point clouds and combined navigation, a position deviation and a heading angle deviation of a vehicle under conditions where combined navigation fails are corrected, thereby solving the problem in the prior art of a multi-source fusion positioning method having poor applicability.

Description

A high-precision positioning method, system and vehicle based on multi-source fusion technology Technical Field

[0001] The present invention relates to a high-precision positioning method, system and vehicle based on multi-source fusion technology, belonging to the technical field of multi-source fusion positioning. Background Art

[0002] With the rise of the new energy vehicle industry in recent years, the prospects for autonomous driving technology have become increasingly promising among vehicle users and manufacturers. Accurate positioning is a key component of autonomous vehicle systems and a prerequisite for safe operation.

[0003] Existing precise positioning technologies for autonomous driving rely on single methods such as GPS positioning and environmental feature matching. However, these positioning technologies have significant drawbacks and limitations. Signal positioning cannot provide accurate positioning in obstructed environments, and environmental feature matching cannot achieve high-precision positioning when there are few positioning features or when the sensor is obscured. Therefore, accurate positioning cannot be achieved in obstructed environments, when the sensor is blocked, or when there are few matching positioning features, resulting in potential dangers and reduced safety.

[0004] In order to solve the problem of low positioning accuracy, the Chinese invention application publication document with application publication number CN116660961A provides a vehicle high-precision positioning method based on the fusion of GPS and on-board sensors. This solution connects the on-board sensors, external network detection module, navigation system and traffic light linkage module to the GPS locator respectively, solving the problem of poor positioning accuracy of traditional GPS. However, due to the setting of the traffic light linkage module, it is necessary to combine traffic light information when implementing this solution, and the coupling with the infrastructure is too high. The traffic light linkage module is meaningless when driving in areas without infrastructure. It has poor applicability on sections of highways without traffic lights and cannot support the needs of autonomous driving.

[0005] The Chinese invention patent application publication document with application publication number CN114199259A provides a multi-source fusion navigation and positioning method based on motion state and environmental perception based on the image feature point position, IMU, Kalman filtering, GNSS and visual loose combined fusion positioning algorithm. This solution gives full play to the role of vision in combined navigation and solves the problem of poor accuracy of traditional GPS positioning. However, since it is necessary to use a camera to capture image frames and eliminate dynamic points in the image frames, the positioning accuracy of this solution is poor in low visibility environments such as blizzards, heavy fog and other weather conditions. Even in slightly dark environments at night, rainy times, etc., the positioning accuracy will be affected. It has high requirements on the environment and poor applicability.

[0006] Summary of the Invention

[0007] The purpose of the present invention is to provide a high-precision positioning method, system and vehicle based on multi-source fusion technology to solve the problem of poor applicability of multi-source fusion positioning methods in the prior art.

[0008] To achieve the above object, the solution of the present invention includes:

[0009] The present invention provides a high-precision positioning method based on multi-source fusion technology, which obtains the current point cloud around the vehicle, matches the current point cloud with a pre-acquired point cloud map, obtains the actual point cloud position information of the current vehicle in the point cloud map, compares the actual point cloud position information with the pre-correction point cloud position information corresponding to the vehicle's pre-correction position in the point cloud map, obtains a positioning offset, corrects the vehicle's pre-correction position according to the positioning offset, obtains the final vehicle positioning, and outputs a fusion positioning result.

[0010] Furthermore, the point cloud map is pre-grid-divided to obtain a blocked point cloud map; before matching the current point cloud with the point cloud map, the corresponding block in the blocked laser point cloud map where the vehicle is currently located is found through preliminary positioning, and the matching of the current point cloud with the block of the corresponding point cloud map is completed.

[0011] Furthermore, the vehicle RTK information is used to perform the preliminary positioning.

[0012] Furthermore, the point cloud map is obtained by fusing the basic positioning information of the vehicle integrated navigation system and the corresponding point cloud data.

[0013] Furthermore, after obtaining the final vehicle positioning, the final vehicle positioning and the combined navigation positioning are converted into the same coordinate system for comparison, and the actual deviation between the final vehicle positioning and the combined navigation positioning is calculated. If the actual deviation is greater than the deviation threshold, the combined navigation positioning is used as the fusion positioning result; if the actual deviation is less than the deviation threshold, the final vehicle positioning and the combined navigation positioning are fused to obtain a fusion result, and the fusion result is used as the fusion positioning result.

[0014] Furthermore, after obtaining the fusion positioning result of the current frame, the fusion positioning results of the previous frame are compared with the current frame, and the difference between the fusion positioning result of the previous frame and the fusion positioning result of the current frame is calculated. If the difference is greater than the difference threshold, the fusion positioning result of the previous frame is used to obtain the final fusion positioning result; if the difference is less than the difference threshold, the fusion positioning result of the current frame is used as the final fusion positioning result to complete the vehicle positioning.

[0015] Furthermore, the processing method using the fused positioning result of the previous frame is to use IMU information for extrapolated positioning.

[0016] Furthermore, the final vehicle positioning and the combined navigation positioning are fused using a weighted average approach.

[0017] Furthermore, before obtaining the fused positioning result, when the actual deviation is greater than the deviation threshold, the combined navigation positioning is processed using Kalman filtering to obtain the fused positioning result; and when the actual deviation is less than the deviation threshold, the fusion result is processed using Kalman filtering to obtain the fused positioning result.

[0018] Furthermore, after the Kalman filtering, the result of the Kalman filtering is fused with the IMU information to obtain the fused positioning result.

[0019] A high-precision positioning system based on multi-source fusion technology of the present invention includes a memory and a processor, wherein the processor is used to execute instructions of the high-precision positioning method based on multi-source fusion technology stored in the memory.

[0020] A vehicle of the present invention includes a high-precision positioning system based on multi-source fusion technology, and the high-precision positioning system based on multi-source fusion technology is a high-precision positioning system based on multi-source fusion technology as described above.

[0021] The beneficial effects of the present invention are:

[0022] By obtaining a point cloud around the vehicle and mapping it through the point cloud, a point cloud map is generated, along with the vehicle's position within the point cloud map. The actual point cloud position is then compared with the pre-correction point cloud position corresponding to the vehicle's pre-correction position in the point cloud map to determine a positioning offset. Finally, the vehicle's position is corrected using the positioning offset to achieve its final position. By fusing the point cloud with integrated navigation, the system corrects for positional and heading deviations in the event of an integrated navigation failure, addressing the limited applicability of existing multi-source fusion positioning methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] FIG1 is a flow chart of a fusion positioning technology solution based on a laser SLAM algorithm of the present invention;

[0024] FIG2 is a flow chart of the fusion positioning based on the laser SLAM algorithm and combined navigation of the present invention. DETAILED DESCRIPTION

[0025] The idea of ​​the present invention is to obtain a point cloud map by obtaining a point cloud around the vehicle and mapping it through the point cloud, thereby obtaining the position information of the vehicle in the point cloud map, comparing the actual point cloud position information with the pre-correction point cloud position information corresponding to the pre-correction position of the vehicle in the point cloud map, obtaining a positioning offset, and finally correcting the vehicle position by the positioning offset to obtain the final positioning of the vehicle.

[0026] This embodiment uses an inertial navigation system as an integrated navigation system and uses vehicle RTK as a preliminary positioning method to illustrate the present invention in more detail.

[0027] An embodiment of a high-precision positioning method based on multi-source fusion technology:

[0028] The present invention provides a high-precision positioning method based on multi-source fusion technology, including laser SLAM (Simultaneous Localization And Mapping) mapping and positioning.

[0029] As shown in Figure 1, the data acquisition module collects combined navigation information, vehicle information, lidar information and vehicle RTK (Real-Time Kinematic) information, and then performs laser SLAM mapping to obtain a laser point cloud map, i.e., a high-precision map.

[0030] After obtaining the high-precision map in advance, the high-precision map is divided into blocks using the high-precision map segmentation tool to obtain a segmented laser point cloud map (i.e., a segmented high-precision map). The vehicle is initially positioned using the map range search module combined with the RTK information collected by the vehicle and the segmented high-precision map to obtain the block in the segmented high-precision map where the vehicle is currently located as the high-precision map module to be used. Since RTK obtains the position through carrier phase difference technology and RTK is real-time and dynamic, it continuously determines whether the current position needs to be updated. When the position is updated, the current high-precision map is updated again through laser SLAM mapping and the high-precision map segmentation tool, and the high-precision map module to be used is continuously output.

[0031] Since the amount of data in the laser point cloud map constructed by collecting the combined navigation information, vehicle information, lidar information and vehicle RIK information is very large, the amount of calculation during positioning will also increase accordingly. Therefore, after constructing the laser point cloud map on a large scale, it is necessary to grid the map to form a block map module with high precision and small data volume to reduce the amount of calculation during positioning.

[0032] At the start of the vehicle's operation and after it has traveled a certain distance, the LiDAR emits laser light and receives reflected laser light to obtain a point cloud around the vehicle. At the start of the vehicle's operation and after it has traveled a certain distance, the LiDAR emits and receives laser light to obtain the vehicle's distance from the same landmark at the start of the operation and after it has traveled a certain distance. This allows the vehicle's location and heading angle to be determined. The vehicle's heading angle and corrected point cloud are compared with the point cloud corresponding to the vehicle's heading angle and position before correction in the laser point cloud map to obtain the translation offset and yaw angle.

[0033] Among them, the optimal translation offset is obtained by comparing many translation offsets, and the optimal translation offset is pushed to the position fitting module. The vehicle heading angle is corrected by the yaw angle. Before correction, the vehicle position is obtained according to the translation offset and the corrected heading angle to obtain the vehicle SLAM fusion positioning information.

[0034] The vehicle SLAM fusion positioning information is the coordinate information (x, y, theta), where x is the horizontal axis coordinate, y is the vertical axis coordinate, and theta is the heading angle.

[0035] In order to make the SLAM fusion positioning technical solution introduced in this embodiment clearer, a specific application scenario is provided below to explain the SLAM fusion positioning in this embodiment in more detail:

[0036] When a vehicle enters a tunnel, due to the severe obscuration of the tunnel, traditional GPS positioning cannot accurately obtain the precise location of the vehicle. When the vehicle enters the tunnel, the GPS signal is lost. The combined navigation uses the speed of the vehicle when it enters the tunnel as the estimated average speed, and multiplies the average speed by the time the vehicle has been traveling in the tunnel to roughly estimate the vehicle's location. At a certain moment when the vehicle is running in the tunnel, the combined navigation estimates the approximate position and heading angle of the vehicle at this moment as the position of the vehicle before correction. In the SLAM fusion positioning method provided in this embodiment, a laser radar is used to emit laser to establish a point cloud around the vehicle body, and the point cloud is matched with the corresponding point cloud map blocks constructed to obtain the vehicle's current actual point cloud position and heading angle in the point cloud map.

[0037] The actual point cloud position and heading angle in the point cloud map are compared with the point cloud position and heading angle before correction of the vehicle in the point cloud map to obtain the translation offset and the vehicle's yaw angle. The combined navigation positioning of the translation offset and yaw angle correction is used to obtain the actual position and heading angle of the vehicle in the tunnel.

[0038] Among them, vehicle information includes the vehicle's own basic positioning system and vehicle coding; integrated navigation uses the inertial navigation system and the optimal estimation method of Kalman filtering.

[0039] In order to more accurately obtain the specific location of the vehicle at a certain moment of travel, and to provide more accurate map data for autonomous driving to ensure the safety of autonomous driving, as shown in Figure 2, the coordinates (x, y, theta) obtained above are fused with the coordinates of the combined navigation positioning, making the fused positioning method more accurate. The output data of the previous frame and the next frame are then compared to ensure the accuracy of positioning.

[0040] First, the fused positioning result (x, y, theta) is obtained. Then, the vehicle coordinates in the integrated navigation system (the earth coordinate system) are obtained through integrated navigation. The ICP (Iterative Closest Point) algorithm is used to convert the vehicle coordinates in the earth coordinate system into the coordinate system of the fused positioning result (x, y, theta).

[0041] After the coordinate system is converted through ICP, the coordinates (x, y, theta) in the fused positioning coordinate system are compared with the coordinates of the vehicle in the fused coordinate system in the integrated navigation. The horizontal deviation between (x, y, theta) and the coordinates of the vehicle in the fused coordinate system in the integrated navigation is obtained, and a deviation threshold of 1m is set based on experience. The deviation threshold is set before the vehicle leaves the factory.

[0042] If the horizontal deviation between (x, y, theta) and the coordinates of the vehicle in the fused coordinate system in the combined navigation is greater than 1m, the positioning result of the combined navigation is directly output; if the horizontal deviation between (x, y, theta) and the coordinates of the vehicle in the fused coordinate system in the combined navigation is less than 1m, the weighted average of (x, y, theta) and the coordinates of the vehicle in the fused coordinate system in the combined navigation is performed, and the weighted average value is output.

[0043] After determining horizontal deviations, the Kalman filter method is then used. Using the linear system state equation and observation data from the system input and output, the system state is optimally estimated, acting as a filter. The vehicle's angular rate and acceleration are then measured using the IMU (Inertial Measurement Unit). The IMU data collected from the vehicle is fused with the Kalman filtered positioning data, and the filtered fused positioning result is output. This ensures minimal fluctuations between data frames, improving system stability.

[0044] Finally, compare the filtered fusion positioning results of the previous frame and the current frame of the current frame, and compare the filtered fusion positioning result of the previous frame with the filtered fusion positioning result of the current frame to obtain the difference between the filtered fusion positioning results of the previous frame and the current frame, and set the differential threshold of the filtered fusion positioning results of the previous frame and the current frame, where the differential threshold is set before the vehicle leaves the factory.

[0045] If the difference between the filtered fusion positioning results of the previous frame and the current frame is greater than the differential threshold, the filtered fusion positioning result of the previous frame is combined with the IMU information of the current frame for extrapolated positioning, and the positioning result after extrapolated positioning is used as the final fusion positioning result; if the difference between the filtered fusion positioning results of the previous frame and the current frame is less than the differential threshold, the filtered fusion positioning result of the current frame is directly used as the final fusion positioning result.

[0046] By comparing the data of the previous frame with the current frame, even if the positioning data drifts while the vehicle is driving, it will not affect the final fusion positioning result.

[0047] An embodiment of a high-precision positioning system based on multi-source fusion technology:

[0048] This embodiment provides a high-precision positioning system using multi-source fusion technology, including a processor and a memory, wherein the memory stores a computer program that can be run on the processor, and the processor executes the computer program to complete the high-precision positioning method based on multi-source fusion technology mentioned in the embodiment of the above-mentioned high-precision positioning method based on multi-source fusion technology.

[0049] That is to say, a high-precision positioning method based on multi-source fusion technology described in detail in an embodiment of a high-precision positioning method based on multi-source fusion technology is written through a computer program and burned into the memory, and the processor receives the information required by the high-precision positioning method based on multi-source fusion technology and processes it.

[0050] Among them, a high-precision positioning method based on multi-source fusion technology has been described in detail in an embodiment of a high-precision positioning method based on multi-source fusion technology, and will not be repeated here.

[0051] An embodiment of an autonomous driving vehicle:

[0052] This embodiment provides an autonomous driving vehicle, comprising a vehicle body and a high-precision positioning system based on multi-source fusion technology. A high-precision positioning system based on multi-source fusion technology has been described in detail in an embodiment of a high-precision positioning system based on multi-source fusion technology and will not be repeated here.

Claims

1. A high-precision positioning method based on multi-source fusion technology, characterized in that: Obtain the current point cloud around the vehicle, match the current point cloud with the pre-acquired point cloud map, obtain the actual point cloud position information of the current vehicle in the point cloud map, compare the actual point cloud position information with the pre-correction point cloud position information corresponding to the vehicle's pre-correction position in the point cloud map, obtain the positioning offset, correct the vehicle's pre-correction position based on the positioning offset, obtain the final vehicle positioning, and output the fused positioning result.

2. The high-precision positioning method based on multi-source fusion technology according to claim 1 is characterized in that: The point cloud map is pre-grid-divided to obtain a block-based point cloud map; before matching the current point cloud with the point cloud map, the corresponding block in the block-based laser point cloud map where the vehicle is currently located is found through preliminary positioning, and the matching of the current point cloud with the point cloud map is completed by matching the blocks between the current point cloud and the corresponding point cloud map.

3. The high-precision positioning method based on multi-source fusion technology according to claim 2 is characterized in that: The preliminary positioning is performed using vehicle RTK information.

4. The high-precision positioning method based on multi-source fusion technology according to claim 1, characterized in that: The point cloud map is obtained by fusing the basic positioning information of the vehicle integrated navigation system and the corresponding point cloud data.

5. The high-precision positioning method based on multi-source fusion technology according to claim 1, 2, 3 or 4, characterized in that: After obtaining the final vehicle positioning, the final vehicle positioning and the combined navigation positioning are converted into the same coordinate system for comparison, and the actual deviation between the final vehicle positioning and the combined navigation positioning is calculated. If the actual deviation is greater than the deviation threshold, the combined navigation positioning is used as the fused positioning result; if the actual deviation is less than the deviation threshold, the final vehicle positioning and the combined navigation positioning are fused to obtain a fusion result, and the fusion result is used as the fused positioning result.

6. The high-precision positioning method based on multi-source fusion technology according to claim 5, characterized in that: After obtaining the fusion positioning result of the current frame, the fusion positioning results of the previous frame are compared with the current frame, and the difference between the fusion positioning result of the previous frame and the fusion positioning result of the current frame is calculated. If the difference is greater than the difference threshold, the fusion positioning result of the previous frame is used to obtain the final fusion positioning result; if the difference is less than the difference threshold, the fusion positioning result of the current frame is used as the final fusion positioning result to complete the vehicle positioning.

7. The high-precision positioning method based on multi-source fusion technology according to claim 6, characterized in that: The processing method using the fusion positioning result of the previous frame is to use IMU information for extrapolation positioning.

8. The high-precision positioning method based on multi-source fusion technology according to claim 5, characterized in that: The final vehicle positioning and the combined navigation positioning are fused using a weighted average approach.

9. The high-precision positioning method based on multi-source fusion technology according to claim 5, characterized in that: Before obtaining the fused positioning result, when the actual deviation is greater than the deviation threshold, the combined navigation positioning is processed using Kalman filtering to obtain the fused positioning result; and when the actual deviation is less than the deviation threshold, the fusion result is processed using Kalman filtering to obtain the fused positioning result.

10. The high-precision positioning method based on multi-source fusion technology according to claim 9, characterized in that: After the Kalman filtering, the result of the Kalman filtering is fused with the IMU information to obtain the fused positioning result.

11. A high-precision positioning system based on multi-source fusion technology, comprising a memory and a processor, characterized in that: The processor is used to execute instructions stored in the memory to implement the high-precision positioning method based on multi-source fusion technology as claimed in any one of claims 1 to 10.

12. A vehicle, characterized in that: It includes a high-precision positioning system based on multi-source fusion technology, and the high-precision positioning system based on multi-source fusion technology is a high-precision positioning system based on multi-source fusion technology as described in claim 11.

Images