Multi-sensor data fusion method and device, computer equipment and storage medium

Abstract

The invention discloses a multi-sensor data fusion method and device, computer equipment and a storage medium, and the method comprises the following steps: obtaining original data of each perception sensor, converting all original data into a 2D point cloud, and carrying out the coordinate conversion to the 2D point cloud under a vehicle body coordinate system, the perception sensor comprises a 3D laser radar, a 2D laser radar and an RGBD camera; coordinates of points in the 2D point cloud under the vehicle body coordinate system are converted from Cartesian coordinates to polar coordinates, and data compression is carried out through an angle discretization method; and fusing the data of all the perception sensors which are converted into polar coordinates under the polar coordinate system of the vehicle body. By compressing a large amount of original data into a small amount of polar coordinate points, the data transmission amount is greatly reduced, the problem of bandwidth occupation of a high-resolution sensor is solved, resources occupied by calculation and time consumed by calculation are greatly reduced, the data format is uniform after fusion, multiple sensor caches do not need to be maintained, and the data transmission efficiency is improved. The coding difficulty is reduced while memory resources are saved, and the development efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of robotics technology, and in particular to a method, device, computer equipment and storage medium for multi-sensor data fusion. Background Art

[0002] As the core component of a robot, the robot navigation system generally includes modules such as positioning, planning, control, and perception. The perception module is mainly responsible for receiving various sensor data and analyzing this data to determine whether there are obstacles near the robot and whether the robot will collide with obstacles.

[0003] See also Figure 1 The conventional processing method is to transmit various sensor data directly to the perception module separately. This method leads to the following problems: high-resolution sensors (such as RGBD cameras that output hundreds of thousands of 3D points and 3D lidars that output tens of thousands of 3D points) output a large amount of raw data that is directly transmitted to the perception module, occupying a large amount of bandwidth resources; the perception module needs to perform independent coordinate transformations on multi-source sensor data, which consumes a large amount of CPU resources; the perception module needs to maintain different data caches for different sensor types, which consumes a large amount of memory resources, increases coding difficulty for developers, and reduces development efficiency. These problems are particularly prominent in resource-constrained mobile robots (such as AGVs and service robots). Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a method, apparatus, computer equipment and storage medium for multi-sensor data fusion that can reduce resource occupation and time consumption.

[0005] To solve the above technical problems, the present invention adopts a technical solution: providing a method for multi-sensor data fusion, comprising the following steps:

[0006] Acquire the raw data from each perception sensor, including a 3D LiDAR, a 2D LiDAR, and an RGBD camera, convert all the raw data into a 2D point cloud, and then transform the coordinates into a 2D point cloud in the vehicle coordinate system.

[0007] The coordinates of the midpoints in the 2D point cloud in the vehicle coordinate system are converted from Cartesian coordinates to polar coordinates, and data compression is performed by using the angle discretization method;

[0008] All sensor data converted into polar coordinates are fused in the vehicle's polar coordinate system.

[0009] Furthermore, the steps of converting all raw data into 2D point clouds and then performing coordinate conversion into 2D point clouds in the vehicle coordinate system include the following sub-steps:

[0010] Taking the center of gravity or geometric center of the robot body as the origin, the default forward direction of the robot as the X-axis direction, the direction perpendicular to the X-axis in the horizontal plane as the Y-axis direction, and the direction perpendicular to the X-Y plane as the Z-axis direction, a vehicle body coordinate system of the robot is established;

[0011] Each sensing sensor has a corresponding reference coordinate system, and the original data obtained by each sensing sensor is converted into 2D point clouds in its corresponding reference coordinate system;

[0012] The 2D point clouds in the reference coordinate systems of all sensing sensors are converted into the vehicle body coordinate system.

[0013] Further, the method of converting the original data obtained by each sensor into 2D point clouds in its corresponding reference coordinate system specifically includes: projecting the 3D data of the 3D lidar onto the x-y plane of the reference coordinate system of the 3D lidar to generate the first 2D point cloud, projecting the 3D data of the RGBD camera onto the x-y plane of the reference coordinate system of the RGBD camera to generate the second 2D point cloud, taking the scan data of the 2D lidar as the third 2D point cloud, and the third 2D point cloud is located in the x-y plane of the reference coordinate system of the 2D lidar.

[0014] Further, the calculation formula for converting the 2D point clouds in the reference coordinate systems of all sensing sensors into the vehicle body coordinate system is: [[ID= (]]

[0015] P' = T·P (I)

[0016] In formula (I), T represents a rigid body transformation matrix, which is composed of a 2x2 rotation matrix and a 2x1 translation matrix, representing the positional relationship of the reference coordinate system of the sensing sensor relative to the vehicle body coordinate system. Each sensing sensor has a corresponding T, which is determined by the position of each sensing sensor. P is a 2x1 vector representing the position of a point in the reference coordinate system of the sensing sensor, and P' represents the position of the point in the vehicle body coordinate system.

[0017] Further, in the step of converting the coordinates of the points in the 2D point cloud in the vehicle body coordinate system from Cartesian coordinates to polar coordinates and performing data compression by the method of angle discretization, the following sub-steps are included:

[0018] Calculating the corresponding polar coordinates according to the Cartesian coordinates of each point in the 2D point cloud in the vehicle body coordinate system;

[0019] Discretizing the continuous polar coordinate angles into an angle sequence with a fixed interval according to a preset angle increment;

[0020] For the original angle of each polar coordinate, taking the closest discrete angle as the polar angle in the polar coordinate by the rounding principle;

[0021] All discretized polar coordinate data are grouped by polar angle. If there are several polar diameters at the same angle, the polar coordinate data with the smallest polar diameter is retained and the rest are discarded.

[0022] Furthermore, the formula for calculating polar coordinates based on Cartesian coordinates is as follows:

[0023]

[0024] In formula (2), r represents the polar diameter in polar coordinates, θ represents the original angle in polar coordinates, x represents the abscissa value in Cartesian coordinates, and y represents the ordinate value in Cartesian coordinates.

[0025] Furthermore, the method of fusing the data of all perception sensors converted into polar coordinates in the vehicle polar coordinate system specifically includes: treating the polar coordinate data of each perception sensor after discrete processing as a group, comparing the data of each group, retaining the polar coordinate data with the smallest polar diameter at the same polar angle as the fused data, and discarding the remaining data.

[0026] To solve the above technical problems, another technical solution adopted by the present invention is to provide a multi-sensor data fusion device, comprising:

[0027] A conversion module is used to obtain the raw data of each perception sensor and convert all the raw data into a 2D point cloud, and then perform coordinate conversion to the 2D point cloud in the vehicle coordinate system. The perception sensors include 3D laser radar, 2D laser radar and RGBD camera;

[0028] The compression module is used to convert the coordinates of the midpoints of the 2D point cloud in the vehicle coordinate system from Cartesian coordinates to polar coordinates, and compress the data by using the angle discretization method;

[0029] The fusion module is used to fuse the data of all perception sensors converted into polar coordinates in the vehicle polar coordinate system.

[0030] To solve the above technical problems, another technical solution adopted by the present invention is: providing a computer device, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the multi-sensor data fusion method when executing the computer program.

[0031] To solve the above technical problem, another technical solution adopted by the present invention is: providing a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the multi-sensor data fusion method are implemented.

[0032] The method, device, computer equipment and storage medium for multi-sensor data fusion of the present invention have at least the following beneficial effects: the present invention compresses a large amount of raw data into a small number of polar coordinate points, greatly reducing the data transmission volume, solving the problem of bandwidth occupancy of high-resolution sensors and saving hardware costs; computationally intensive operations such as coordinate transformation are completed at the data fusion front end, and the perception module only needs to process a single lightweight data stream, freeing up computing power for real-time obstacle avoidance decision-making; the data format after fusion is unified, and there is no need to maintain multiple sensor caches, which saves memory resources while also reducing coding difficulty and improving development efficiency; combining polar coordinate discretization with minimum distance screening rules, greatly reducing the complexity of data processing, improving the robot's obstacle avoidance response speed, and meeting the millisecond-level obstacle avoidance requirements of high-speed mobile robots; through the fusion strategy of selecting the minimum distance at the same angle, the advantages of each sensor are integrated, non-minimum distance values ​​are discarded, distant interference objects are filtered, and the focus is on near-end obstacles with the highest collision risk; supporting any type / number of perception sensor extensions (such as ultrasound), only the output conversion to 2D point cloud and polar coordinate conversion processes need to be met. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0034] Figure 1 This is a diagram of the conventional sensor data transmission process.

[0035] Figure 2 This is a flow chart of an embodiment of a method for multi-sensor data fusion according to the present invention.

[0036] Figure 3 for Figure 2 Flowchart of an implementation method of step S100.

[0037] Figure 4 for Figure 2 Flowchart of an implementation method of step S200.

[0038] Figure 5 This is a structural block diagram of an embodiment of a multi-sensor data fusion device of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings.

[0040] See also Figure 2 , which is a flow chart of an embodiment of a method for multi-sensor data fusion of the present invention.

[0041] In this embodiment, the following steps are included:

[0042] S100, data conversion.

[0043] Specifically, the raw data of each perception sensor is obtained and all the raw data is converted into a 2D point cloud, and then the coordinates are converted into the 2D point cloud in the vehicle coordinate system. The perception sensors include 3D laser radar, 2D laser radar and RGBD camera. It should be understood that the perception sensors are not limited to the above sensors, and may also include ultrasonic sensors. This embodiment is used for obstacle avoidance in robot navigation. The 2D laser radar only scans the horizontal plane and may miss low obstacles and high obstacles, while the 3D laser radar can fill this loophole through vertical scanning. The 3D laser radar has a long detection distance, but the single cost is high. The RGBD camera can detect low and high obstacles and is low-cost, but the detection distance is limited. Therefore, this embodiment selects 3D laser radar, 2D laser radar and RGBD camera to work together.

[0044] See also Figure 3 , this step S100 includes the following sub-steps:

[0045] S110: Establishing a vehicle body coordinate system.

[0046] Specifically, the robot's body coordinate system is established with the center of gravity or geometric center of the robot body as the origin, the robot's default forward direction as the X-axis direction, the direction perpendicular to the X-axis in the horizontal plane as the Y-axis direction, and the direction perpendicular to the XY plane as the Z-axis direction.

[0047] S120: Generate a 2D point cloud.

[0048] Specifically, each perception sensor has a corresponding reference coordinate system, and the raw data acquired by each perception sensor is converted into a 2D point cloud in its corresponding reference coordinate system. The operation method is as follows: 3D data from the 3D lidar is projected onto the xy plane of the 3D lidar's reference coordinate system to generate a first 2D point cloud; 3D data from the RGBD camera is projected onto the xy plane of the RGBD camera's reference coordinate system to generate a second 2D point cloud; and the scan data from the 2D lidar is used as a third 2D point cloud, which is located within the xy plane of the 2D lidar's reference coordinate system.

[0049] In this embodiment, the reference coordinate system of a 3D lidar is composed of the x-axis, which is the scanning direction directly in front of the sensor and represents the horizontal longitudinal distance of obstacles; the y-axis, which is perpendicular to the left side of the sensor and represents the horizontal lateral distance of obstacles; and the z-axis, which is perpendicular to the ground and upward, represents the obstacle's height. The 3D point cloud is projected onto the xy plane, and the z-axis height data is discarded, retaining the horizontal obstacle position. In the reference coordinate system of a 2D lidar, the x-axis is directly in front of the sensor's scanning plane and represents the horizontal longitudinal distance of obstacles; the y-axis is perpendicular to the left side of the sensor's scanning plane and represents the horizontal lateral distance of obstacles. In the reference coordinate system of an RGBD camera, the x-axis is parallel to the camera's optical axis and represents the scene depth, indicating the distance of objects; the y-axis is horizontal to the left of the image plane and represents the horizontal position of the scene; and the z-axis is perpendicular to the ground and upward, representing the obstacle's height. The z-axis of each sensor represents obstacle height information. Since robot navigation only requires the distribution of horizontal obstacles, actively discarding z-axis data during the data preprocessing stage can significantly reduce computational complexity.

[0050] S130: Coordinate system conversion.

[0051] Specifically, the 2D point clouds in the reference coordinate systems of all perception sensors are converted to the vehicle coordinate system. In this step, the calculation formula used is as follows:

[0052] P'=T·P

[0053] Among them, T represents the rigid body transformation matrix, which consists of a 2x2 rotation matrix and a 2x1 translation matrix. It represents the position relationship of the reference coordinate system of the perception sensor relative to the vehicle coordinate system. Each perception sensor has a corresponding T, which is determined by the position of each perception sensor. P is a 2x1 vector, which represents the position of the point in the reference coordinate system of the perception sensor, and P' represents the position of the point in the vehicle coordinate system.

[0054] S200: Data compression.

[0055] Specifically, the coordinates of the midpoints in the 2D point cloud in the vehicle coordinate system are converted from Cartesian coordinates to polar coordinates, and data compression is performed using the angle discretization method. Figure 4 , this step S200 includes the following sub-steps:

[0056] S210: Calculate polar coordinates.

[0057] Specifically, the polar coordinates of each point in the 2D point cloud are calculated based on the Cartesian coordinates of the vehicle body coordinate system. The formula for calculating polar coordinates based on Cartesian coordinates is:

[0058]

[0059] Where r represents the polar diameter in polar coordinates, θ represents the original angle in polar coordinates, x represents the abscissa value in Cartesian coordinates, and y represents the ordinate value in Cartesian coordinates.

[0060] S220: Discrete processing.

[0061] To reduce computational complexity without sacrificing data accuracy, in this embodiment, the polar coordinate angles need to be discretized. Angle discretization involves dividing a continuous angle (a total of 360°) into a certain number of portions according to a preset angle increment. Specifically, the continuous polar coordinate angles are discretized into a fixed-interval angle sequence according to a preset angle increment. In this embodiment, the preset angle increment is 0.2°, and the angle is divided into 1800 portions in total. The resulting discrete angle sequence is -180°, -179.8°, ..., 179.8°, and 180°.

[0062] S230: Obtain angle value.

[0063] Specifically, for each original angle of the polar coordinate, the closest discrete angle is taken as the polar angle in the polar coordinate. For example, when the calculated original angle is 179.96°, the polar angle is taken as 180°.

[0064] S240, data discarded

[0065] Specifically, all discretized polar coordinate data is grouped by polar angle. If there are multiple polar diameters at the same angle, the polar coordinate data with the smallest polar diameter is retained, and the remaining data is discarded. Robots avoid obstacles by prioritizing the nearest obstacle. The smallest polar diameter represents the distance to the nearest obstacle, so retaining the polar coordinate data with the smallest polar diameter is more meaningful.

[0066] S300, data fusion.

[0067] Specifically, the data from all perception sensors converted to polar coordinates is fused within the vehicle's polar coordinate system. The specific operation involves grouping the discretized polar coordinate data from each perception sensor. Each group of data is compared, and the polar coordinate data with the smallest polar diameter at the same polar angle is retained as the fused data, while the remaining data is discarded. Multiple perception sensors may have obstacle information at different distances at the same angle. Selecting the one with the smallest polar diameter allows for the integration of the nearest obstacle information from each sensor. This fused data is then transmitted to the robot's perception module, ensuring the module obtains the safest obstacle avoidance data.

[0068] Experimental results show that the fused data volume is 1,800 polar coordinate points, far less than tens of thousands or even hundreds of thousands of 3D points, significantly reducing computational resources and time. Furthermore, the fused data is equivalent to a single sensor data point transmitted to the perception module, significantly simplifying the perception module's programming.

[0069] See also Figure 5 , is a structural block diagram of an embodiment of a multi-sensor data fusion device of the present invention. The multi-sensor data fusion device of this embodiment is used to implement the multi-sensor data fusion method described in the above embodiment. Specifically, the multi-sensor data fusion device of this embodiment includes a conversion module 100, a compression module 200, and a fusion module 300. Among them:

[0070] The conversion module 100 is used to obtain the raw data of each perception sensor and convert all the raw data into a 2D point cloud, and then perform coordinate conversion to the 2D point cloud in the vehicle coordinate system. The perception sensors include 3D laser radar, 2D laser radar and RGBD camera. The conversion module 100 is also used to establish the vehicle coordinate system of the robot with the center of gravity or geometric center of the robot body as the origin, the default forward direction of the robot as the X-axis direction, the direction perpendicular to the X-axis in the horizontal plane as the Y-axis direction, and the direction perpendicular to the XY plane as the Z-axis direction; it is also used to convert the raw data obtained by each perception sensor into a 2D point cloud in its corresponding reference coordinate system; finally, it is used to convert the 2D point clouds in the reference coordinate systems of all perception sensors into the vehicle coordinate system.

[0071] The compression module 200 is used to convert the coordinates of the points in the 2D point cloud in the vehicle body coordinate system from Cartesian coordinates to polar coordinates, and simultaneously perform data compression using an angle discretization method. Specifically, the compression module 200 is used to calculate the corresponding polar coordinates of each point in the 2D point cloud in the vehicle body coordinate system based on the Cartesian coordinates of the point; further used to discretize the continuous polar coordinate angles into a fixed-interval angle sequence according to a preset angle increment; and for each original angle of the polar coordinate, the closest discrete angle is taken as the polar angle in the polar coordinate; finally, the module is used to group all discretized polar coordinate data by polar angle. If there are multiple polar diameters at the same angle, the polar coordinate data with the smallest polar diameter is retained and the remaining data is discarded.

[0072] The fusion module 300 is responsible for fusing the data from all sensory sensors, converted to polar coordinates, within the vehicle's polar coordinate system. Specifically, the module groups the discretized polar coordinate data from each sensor, comparing each data set. The polar coordinate data with the smallest polar diameter at the same polar angle is retained as the fused data, while the remaining data is discarded. The resulting fused data is then transmitted to the robot's perception module.

[0073] The present invention also discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the multi-sensor data fusion method when executing the computer program.

[0074] The present invention also discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the multi-sensor data fusion method are implemented.

[0075] The present invention compresses a large amount of raw data into a small number of polar coordinate points, greatly reducing the amount of data transmission, solving the problem of bandwidth occupancy of high-resolution sensors and saving hardware costs. Computational intensive operations such as coordinate transformation are completed at the data fusion front end, and the perception module only needs to process a single lightweight data stream, freeing up computing power for real-time obstacle avoidance decision-making. The data format is unified after fusion, and there is no need to maintain multiple sensor caches, which saves memory resources while also reducing coding difficulty and improving development efficiency. Combining polar coordinate discretization with minimum distance screening rules greatly reduces the complexity of data processing, improves the robot's obstacle avoidance response speed, and meets the millisecond-level obstacle avoidance requirements of high-speed mobile robots. The advantages of each sensor are integrated through a fusion strategy that selects the minimum distance at the same angle, discards non-minimum distance values, filters distant interference, and focuses on near-end obstacles with the highest collision risk. It supports any type / number of perception sensor extensions (such as ultrasound), and only needs to meet the output conversion to 2D point cloud and polar coordinate conversion processes.

[0076] The above description merely expresses the preferred embodiments of the present invention, and its description is relatively specific and detailed, but it should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art may make a number of variations and improvements without departing from the concept of the present invention, and these variations and improvements fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent of the present invention shall be based on the appended claims.

Claims

1. A method for multi-sensor data fusion, characterized in that: It includes the following steps: Obtain the original data of each perception sensor, convert all the original data into 2D point clouds, and then perform coordinate transformation to obtain 2D point clouds in the vehicle body coordinate system. The perception sensors include 3D lidar, 2D lidar, and RGBD cameras; Convert the coordinates of the points in the 2D point clouds in the vehicle body coordinate system from Cartesian coordinates to polar coordinates, and at the same time perform data compression by means of angle discretization; Fuse the data of all perception sensors that have been converted to polar coordinates in the vehicle body polar coordinate system.

2. The method for multi-sensor data fusion according to claim 1, wherein: In the step of converting all the original data into 2D point clouds and then performing coordinate transformation to obtain 2D point clouds in the vehicle body coordinate system, the following sub-steps are included: Establish a vehicle body coordinate system of the robot with the center of gravity or geometric center of the robot body as the origin, the default forward direction of the robot as the X-axis direction, the direction perpendicular to the X-axis in the horizontal plane as the Y-axis direction, and the direction perpendicular to the X-Y plane as the Z-axis direction; Each perception sensor has a corresponding reference coordinate system. Convert the original data obtained by each perception sensor into 2D point clouds in its corresponding reference coordinate system; Convert the 2D point clouds in the reference coordinate systems of all perception sensors into the vehicle body coordinate system.

3. The method for multi-sensor data fusion according to claim 2, wherein: The method of converting the original data obtained by each sensor into 2D point clouds in its corresponding reference coordinate system specifically includes: projecting the 3D data of the 3D lidar onto the x-y plane of the reference coordinate system of the 3D lidar to generate the first 2D point cloud, projecting the 3D data of the RGBD camera onto the x-y plane of the reference coordinate system of the RGBD camera to generate the second 2D point cloud, and using the scan data of the 2D lidar as the third 2D point cloud, and the third 2D point cloud is located in the x-y plane of the reference coordinate system of the 2D lidar.

4. The method for multi-sensor data fusion according to claim 2, wherein: The calculation formula for converting the 2D point clouds in the reference coordinate systems of all perception sensors into the vehicle body coordinate system is: P' = T·P (I) In formula (I), T represents a rigid body transformation matrix, which is composed of a 2x2 rotation matrix and a 2x1 translation matrix, representing the positional relationship of the reference coordinate system of the perception sensor relative to the vehicle body coordinate system. Each perception sensor has a corresponding T, which is determined by the position of each perception sensor. P is a 2x1 vector representing the position of a point in the reference coordinate system of the perception sensor, and P' represents the position of the point in the vehicle body coordinate system.

5. The method for multi-sensor data fusion according to claim 1, wherein: In the step of converting the coordinates of the points in the 2D point clouds in the vehicle body coordinate system from Cartesian coordinates to polar coordinates and at the same time performing data compression by means of angle discretization, the following sub-steps are included: Calculate the corresponding polar coordinates according to the Cartesian coordinates of each point in the 2D point clouds in the vehicle body coordinate system; Discretize the continuous polar coordinate angles into an angle sequence with a fixed interval according to a preset angle increment; For each original angle of the polar coordinates, round it to the closest discrete angle as the polar angle in the polar coordinates; Group the discretized polar coordinate data by polar angle. If there are several radial distances at the same angle, retain the polar coordinate data with the smallest radial distance and discard the rest of the data.

6. The multi-sensor data fusion method according to claim 5, characterized in that: The formula for calculating polar coordinates according to Cartesian coordinates is as follows: In formula (2), r represents the polar diameter in polar coordinates, θ represents the original angle in polar coordinates, x represents the abscissa value in Cartesian coordinates, and y represents the ordinate value in Cartesian coordinates.

7. The multi-sensor data fusion method according to claim 1, wherein: The method for fusing the data of all perception sensors converted into polar coordinates in the vehicle polar coordinate system specifically includes: treating the polar coordinate data of each perception sensor after discrete processing as a group, comparing the data of each group, retaining the polar coordinate data with the smallest polar diameter at the same polar angle as the fused data, and discarding the remaining data.

8. A multi-sensor data fusion device, characterized in that: include: A conversion module is used to obtain the raw data of each perception sensor and convert all the raw data into a 2D point cloud, and then perform coordinate conversion to the 2D point cloud in the vehicle coordinate system. The perception sensors include 3D laser radar, 2D laser radar and RGBD camera; The compression module is used to convert the coordinates of the midpoints of the 2D point cloud in the vehicle coordinate system from Cartesian coordinates to polar coordinates, and compress the data by using the angle discretization method; The fusion module is used to fuse the data of all perception sensors converted into polar coordinates in the vehicle polar coordinate system.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-sensor data fusion method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-sensor data fusion method according to any one of claims 1 to 7 are implemented.

Images