Indoor robot high-precision positioning method and system based on multi-source data fusion

Abstract

The invention belongs to the technical field of mobile robots, and provides a multi-source data fusion indoor robot high-precision positioning method and system, and the method comprises the steps: extracting high-dimensional features in visual image information and point cloud information of a laser radar, carrying out the precise matching at a roughly matched position, and obtaining a fusion estimation pose; obtaining a first estimated pose by using the measurement data of the inertial measurement unit, obtaining a second estimated pose by using the measurement data of the wheel type odometer, and obtaining a third estimated pose by using the measurement data of the ultra wide band; carrying out pose difference measurement on the fusion estimation pose, the first estimation pose, the second estimation pose and the third estimation pose, screening to obtain an optimal estimation pose, and carrying out pose optimization according to the optimal estimation pose to obtain a final pose; the positioning and repositioning precision and efficiency of the indoor robot in a complex environment are improved, and the robustness of the whole positioning system is improved.

Description

Technical Field

[0001] The present invention relates to the field of mobile robot technology, and in particular to a high-precision positioning method for an indoor robot using multi-source data fusion, a high-precision positioning system for an indoor robot using multi-source data fusion, a high-precision repositioning method for an indoor robot using multi-sensor fusion, a computer device, a computer-readable storage medium, and a computer program product. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] In the field of robot positioning, multi-sensor fusion methods have become mainstream. Common sensors include lidar, visual sensors, inertial measurement units (IMUs), and ultra-wideband (UWB). These systems usually combine data from multiple sensors to improve the robustness and accuracy of positioning. For example, lidar can provide high-precision distance measurement, while visual sensors can obtain rich environmental information, and IMUs are used to compensate for posture changes in dynamic scenes.

[0004] Although existing robot positioning and relocalization schemes can improve robustness by fusing multiple sensor data, they still face many challenges in some dynamic environments: (1) The fusion algorithm may not be able to effectively handle data conflicts between sensors, resulting in unreliable final positioning results. When dealing with the estimation errors of different sensors, the existing fusion algorithm fails to effectively consider the characteristics and limitations of each sensor, which may lead to a decrease in overall positioning accuracy, especially when the environmental features are sparse or the lighting changes greatly. (2) When relocalizing, existing methods often need to traverse a large range of environmental data, which is time-consuming. Especially in similar scenarios, the relocalization accuracy is low, making it difficult for the robot to quickly find the correct position. (3) Existing multi-sensor fusion methods may have delays in data processing and calculation, affecting the real-time response capability of the system. This is particularly critical in dynamic environments and can easily lead to the robot's lack of sensitivity when performing tasks. Summary of the invention

[0005] In order to address the deficiencies in the prior art, the present invention provides a high-precision positioning method and system for an indoor robot with multi-source data fusion, which solves the problems of poor positioning accuracy, low repositioning success rate and efficiency, and a significant impact of single sensor distortion on the positioning accuracy of the positioning system of the existing robots in environments with large scene changes and similar scenes such as long indoor corridors, thereby improving the positioning and repositioning accuracy and efficiency of indoor robots in complex environments and improving the robustness of the entire positioning system.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a high-precision positioning method for an indoor robot by fusing multi-source data.

[0008] A high-precision positioning method for indoor robots based on multi-source data fusion includes the following processes:

[0009] Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position;

[0010] Extract high-dimensional features from visual image information and LiDAR point cloud information, perform precise matching at the roughly matched positions, and obtain fused estimated pose;

[0011] Using the measurement data of the inertial measurement unit to obtain a first estimated posture, using the measurement data of the wheel odometer to obtain a second estimated posture, and using the measurement data of the ultra-wideband to obtain a third estimated posture;

[0012] The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose are measured for pose difference, the optimal estimated pose is screened out, and the final pose is obtained after pose optimization is performed according to the optimal estimated pose.

[0013] As a further limitation of the first aspect of the present invention, the positional relationship features include: positional relationship between lines and planes, positional relationship between lines and positional relationship between planes.

[0014] As a further limitation of the first aspect of the present invention, the high-dimensional features include: color features and shape features.

[0015] As a further limitation of the first aspect of the present invention, the inertial measurement unit measures acceleration and angular velocity data based on the internal accelerometer and gyroscope sensor, and outputs a first estimated posture through integration operation; the wheel odometer detects the wheel rotation angle and speed information of the indoor robot and performs integration calculation to output a second estimated posture.

[0016] As a further limitation of the first aspect of the present invention, the fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose within the same time period are obtained, and the pose difference is measured using Euclidean distance, rotation difference or weighted difference to obtain the optimal estimated pose.

[0017] In a second aspect, the present invention provides a high-precision repositioning method for an indoor robot using multi-sensor fusion.

[0018] A multi-sensor fusion high-precision relocalization method for indoor robots includes the following processes:

[0019] Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position;

[0020] The high-dimensional features in the visual image information and the point cloud information of the laser radar are extracted, and precise matching is performed at the roughly matched position to obtain a fused estimated pose, which is used as the robot relocalization result.

[0021] In a third aspect, the present invention provides a high-precision positioning system for indoor robots with multi-source data fusion, comprising:

[0022] The coarse matching unit is configured to: extract position relationship features in the laser radar frame, match the position relationship features with the point cloud information in the global map, select the scene most similar to the current environment according to the matching rate, and obtain a coarse matching position;

[0023] The fusion pose estimation unit is configured to: extract high-dimensional features from the visual image information and the point cloud information of the laser radar, perform precise matching at the roughly matched position, and obtain a fusion estimated pose;

[0024] A posture estimation unit is configured to: obtain a first estimated posture using measurement data of an inertial measurement unit, obtain a second estimated posture using measurement data of a wheel odometer, and obtain a third estimated posture using measurement data of wireless carrier communication;

[0025] The posture optimization unit is configured to: perform posture difference measurement on the fused estimated posture, the first estimated posture, the second estimated posture and the third estimated posture, screen out the optimal estimated posture, and obtain the final posture after posture optimization according to the optimal estimated posture.

[0026] In a fourth aspect, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium;

[0027] a processor adapted to execute a computer program;

[0028] A computer-readable storage medium having a computer program stored therein, wherein when the computer program is executed by the processor, the method for high-precision positioning of an indoor robot by multi-source data fusion as described in the first aspect of the present invention is implemented; or, the method for high-precision repositioning of an indoor robot by multi-sensor fusion as described in the second aspect of the present invention is implemented.

[0029] In a fifth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor and executing the high-precision positioning method for an indoor robot using multi-source data fusion as described in the first aspect of the present invention; or executing the high-precision repositioning method for an indoor robot using multi-sensor fusion as described in the second aspect of the present invention.

[0030] In a sixth aspect, the present invention provides a computer program product, characterized in that the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the high-precision positioning method for an indoor robot using multi-source data fusion as described in the first aspect of the present invention; or, it implements the high-precision repositioning method for an indoor robot using multi-sensor fusion as described in the second aspect of the present invention.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention innovatively proposes a high-precision positioning and repositioning method for indoor robots with multi-source data fusion. Multiple sensors independently estimate the posture, and data fusion is performed after the output posture is correctly screened, which can significantly improve the positioning accuracy of the robot in complex environments and solve the problem of positioning failure caused by distortion of the wheel odometer caused by wheel slippage and radar point cloud feature matching errors in similar scenes. Geometric relationships are used for coarse positioning, and deep learning is used to extract high-dimensional features of point clouds and images to complete precise positioning, avoiding positioning errors and repositioning failures in similar scenes or changing scenes. Coarse matching can screen out most invalid scenes, which is conducive to rapid positioning of precise matching, improves computing efficiency and robustness in dynamic scenes, and reduces the impact of single sensor failure.

[0033] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0035] Figure 1 A schematic diagram of the process of a high-precision positioning method for an indoor robot using multi-source data fusion provided in Example 1 of the present invention;

[0036] Figure 2 A schematic diagram of a high-precision positioning system for an indoor robot using multi-source data fusion provided in Example 2 of the present invention;

[0037] Figure 3A schematic diagram of a computer device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0039] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0040] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0041] Embodiment 1:

[0042] Although multi-sensor fusion has achieved certain success, positioning and re-positioning in complex environments still face some challenges, such as easy loss of position in similar scenes and low re-positioning success rate; when a sensor is distorted, it will introduce a large error to the entire positioning system and have a great impact on the final positioning result; when the environment changes, the positioning result is often inaccurate; the positioning schemes applicable indoors and outdoors are not unified, etc. In view of this, this implementation proposes a high-precision positioning method for indoor robots with multi-source data fusion, including the following processes:

[0043] Sensor data collection: Use binocular cameras, lidar, IMU (inertial measurement unit), wheel odometers, and UWB (when the signal is stable) to collect environmental information;

[0044] Pose estimation: Each sensor outputs a stable pose estimate;

[0045] The laser radar extracts line features, surface features, positional relationship features between lines and surfaces, positional relationship features between lines, and positional relationship features between surfaces in each valid frame, and roughly matches the above-mentioned positional relationship features with the point cloud information in the global map. The scene most similar to the current environment is selected according to the matching rate to obtain the coarse matching position.

[0046] More specifically, the laser radar extracts the positional relationship features of lines, surfaces, and lines and surfaces, lines and lines, and surfaces and surfaces in each valid frame, including:

[0047] Each frame uses methods such as Hough transform and RANSAC to detect straight line extraction lines by identifying the continuity of line segments, extract the main directions of the local point cloud based on principal component analysis (PCA), extract features in the same plane, and merge these features with the point and surface features in the global point cloud to form complete line and surface features.

[0048] The spatial geometric relationship between the line and the surrounding surface features is used as a feature descriptor. The relationship between the line and the surface: use the point normal vector equation to determine whether the line and the surface intersect and determine the coordinates of the intersection, or whether the line is parallel to the surface;

[0049] Relationship between lines: Use the parametric equation of the line to determine the relationship between two lines, such as intersection, parallelism, etc., by calculating the minimum distance, etc.;

[0050] Relationship between faces: By calculating the face normal vector, we can deduce the relationship between faces, such as parallel, intersecting or having a certain angle, and calculate their angle or intersection line

[0051] Use deep learning methods to extract high-dimensional features such as color and shape from the images returned by the surrounding visual sensor and the point cloud information returned by the radar, perform precise matching at the rough matching position (matching can be completed based on tiny features such as the edge and texture of the environment described in the image, avoiding positioning failures in scenes with similar environments and improving matching efficiency), and output the fused estimated pose of the radar and visual sensors;

[0052] Since the robot's lidar and visual sensors can perceive features within a certain range, the environment map is segmented into large voxels at an adjustable distance, such as a few meters as a scene, and the feature descriptors in the current frame are quickly and roughly matched in several sub-areas of the map segmentation;

[0053] Calculate the matching rate: The extracted descriptors are matched using methods such as Euclidean distance and cosine similarity. Descriptor pairs with smaller distances or higher similarities are considered to be successfully matched. The matching ratio is determined by the ratio of the number of successfully matched descriptor pairs to the total number of descriptors extracted in the current frame. Finally, several scenes with similar matching rates are obtained, and the approximate positions in these scenes are determined to narrow the scope of precise matching.

[0054] In this implementation, the precise matching process is to perform precise matching in several rough matching positions obtained by rough matching, and finally obtain a pose estimation with high accuracy. The deep learning method is used to extract the image information returned by the visual sensor and the point cloud information returned by the radar, and high-dimensional features are extracted from them to complete high-dimensional feature matching. The process is as follows:

[0055] The feature information of the visual image (such as color, texture, edge, etc.) and the geometric structure of the LiDAR point cloud are encoded into a high-dimensional feature vector in a joint space. The visual features and geometric features are input in parallel to form a "multi-dimensional multi-modal joint feature vector";

[0056] Based on the above feature vectors, a “multi-level feature association network” is used for learning to extract more robust features, which includes three modules: (1) Local feature module: Local convolution is used to extract low-level association patterns of visual features and geometric features; (2) Global feature module: Based on the graph neural network (GNN), a global spatial relationship graph of visual features and geometric features is constructed to capture possible long-range spatial relationships in vision and point clouds; (3) Cross-modal interaction module: Using the cross-modal attention mechanism, an association mapping is established between visual and geometric features to extract common environmental structure information (for example, the interaction between edges in vision and surfaces in point clouds), and to enhance the feature expression ability in the matching process.

[0057] In the feature matching stage, a "semantic consistency metric" is introduced to measure the similarity between visual features and point cloud features, considering the consistency between the two in the global spatial relationship. For example, the consistency of "key visual edges and adjacent point cloud normal vectors" is calculated to make the matching more accurate and reduce the impact of noise and data deviation.

[0058] For noisy and dynamic environments, an adaptive weight optimization module is designed, which automatically adjusts the contribution ratio of visual and point cloud features during the matching process. Specifically, the attention mechanism in the graph neural network is used to adjust the weights of visual and geometric features in real time according to the stability of the visual and geometric features to reduce the impact of environmental complexity.

[0059] In this implementation, the IMU measures acceleration and angular velocity data based on the internal accelerometer and gyroscope sensor, and outputs the first estimated posture through integration operation; the wheel odometer performs integration calculation by detecting the wheel rotation angle and speed information, and outputs the second estimated posture; the UWB uses the third estimated posture when the signal is stable;

[0060] Posture difference measurement: The output frequency of laser and visual sensors is low, while the frequency of IMU and wheel odometer is high. According to the difference between the robot postures output by the sensors in the same time period, the Euclidean distance, rotation difference, weighted difference and other methods are used to screen the correctness of the data, ignoring postures with too large a difference. This method can effectively filter the impact of sensor data failure on robot posture estimation.

[0061] For example, if the wheel odometer slips, it will output an erroneous robot displacement distance, while the fused posture obtained based on the lidar and binocular vision is relatively correct. For another example, in environments with high scene similarity such as long corridors and computer rooms, the probability of laser sensor positioning errors will increase significantly. At this time, the output data of the wheel odometer and IMU should be more trusted.

[0062] More specifically, the methods for measuring posture differences include:

[0063] A weight distribution is given according to the current measurement status of each sensor. The higher the weight, the higher the reliability of the estimated pose. For example, a higher weight is given to the UWB signal when it is stable, while its positioning information is ignored when the signal stability is low. The IMU pose is given priority in a high vibration environment. A consensus screening mechanism is adopted to measure the deviation between each estimated pose and other poses using the Euclidean distance or Mahalanobis distance of the pose residual. The difference between the poses of the IMU, wheel odometer, UWB, radar and vision fusion is calculated. If the difference between two or more poses is less than the set threshold, they can be regarded as estimated poses close to each other, indicating a higher reliability. Otherwise, it is considered that at least one sensor is in a distorted state. By comparing the pose similarity, the distorted pose is determined and ignored. If there are multiple close estimated poses, the final pose optimization is performed by graph optimization (such as g2o) under the conditions of combining environmental constraints and error metrics.

[0064] In this implementation, the factor graph optimization process includes: passing the correct pose data after screening to the graph optimization module, and using the g2o algorithm to perform pose optimization, so as to obtain a more accurate pose estimation.

[0065] This implementation also provides a high-precision repositioning method for indoor robots with multi-sensor fusion, specifically, including: extracting line features, surface features, positional relationship features between lines and surfaces, positional relationship features between lines, and positional relationship features between surfaces in each effective frame by the laser radar, roughly matching the above-mentioned positional relationship features with the point cloud information in the global map, screening the scene most similar to the current environment according to the matching rate, and obtaining the roughly matched position; using a deep learning method to extract high-dimensional features in the images returned by the surrounding environment visual sensor and the point cloud information returned by the radar, such as color and shape, and accurately matching at the roughly matched position, outputting the fused estimated pose of the radar and the visual sensor, and taking the fused estimated pose as the final repositioning result.

[0066] Embodiment 2:

[0067] like Figure 2 As shown, this implementation provides a multi-source data fusion indoor robot high-precision positioning system, including:

[0068] The coarse matching unit is configured to: extract position relationship features in the laser radar frame, match the position relationship features with the point cloud information in the global map, select the scene most similar to the current environment according to the matching rate, and obtain a coarse matching position;

[0069] The fusion pose estimation unit is configured to: extract high-dimensional features from the visual image information and the point cloud information of the laser radar, perform precise matching at the roughly matched position, and obtain a fusion estimated pose;

[0070] A posture estimation unit is configured to: obtain a first estimated posture using measurement data of an inertial measurement unit, obtain a second estimated posture using measurement data of a wheel odometer, and obtain a third estimated posture using measurement data of wireless carrier communication;

[0071] The posture optimization unit is configured to: perform posture difference measurement on the fused estimated posture, the first estimated posture, the second estimated posture and the third estimated posture, screen out the optimal estimated posture, and obtain the final posture after posture optimization according to the optimal estimated posture.

[0072] It is understandable that the above-mentioned various units can be separately or all combined into one or several other units to constitute, or some (some) units therein can also be split into multiple smaller units in function to constitute, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present application. The above-mentioned units are divided based on logical functions. In practical applications, the function of a unit can also be realized by multiple units, or the function of multiple units can be realized by one unit. In other embodiments of the present application, the system can also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented by the collaboration of multiple units.

[0073] According to another embodiment of the present application, the system described in this embodiment can be constructed, and the method of Example 1 of the present application can be implemented by running a computer program (including program code) capable of executing the steps involved in the corresponding method described in Example 1 on a general computing device such as a computer including processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, and loaded into the above-mentioned computing device through the computer-readable recording medium and run therein.

[0074] Embodiment 3:

[0075] like Figure 3 As shown, this implementation provides an electronic device, which includes a processor 1001, a communication interface 1002, and a computer-readable storage medium 1003. The processor 1001, the communication interface 1002, and the computer-readable storage medium 1003 may be connected via a bus or other means.

[0076] Among them, the communication interface 1002 is used to receive and send data, the computer-readable storage medium 1003 can be stored in the memory of the electronic device, the computer-readable storage medium 1003 is used to store a computer program, the computer program includes program instructions, and the processor 1001 is used to execute the program instructions stored in the computer-readable storage medium 1003.

[0077] Processor 1001 (or CPU (Central Processing Unit)) is the computing core and control core of the electronic device, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to implement corresponding method processes or corresponding functions.

[0078] The processor 1001 is configured to execute the following process:

[0079] Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position;

[0080] Extract high-dimensional features from visual image information and LiDAR point cloud information, perform precise matching at the roughly matched positions, and obtain fused estimated pose;

[0081] Using the measurement data of the inertial measurement unit to obtain a first estimated posture, using the measurement data of the wheel odometer to obtain a second estimated posture, and using the measurement data of the ultra-wideband to obtain a third estimated posture;

[0082] The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose are measured for pose difference, the optimal estimated pose is screened out, and the final pose is obtained after pose optimization is performed according to the optimal estimated pose.

[0083] Embodiment 4:

[0084] This implementation provides a computer-readable storage medium (Memory), which is a memory device in an electronic device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the electronic device and, of course, extended storage media supported by the electronic device. The computer-readable storage medium provides a storage space that stores the processing system of the electronic device.

[0085] In addition, the storage space also stores one or more instructions suitable for being loaded and executed by the processor, and these instructions may be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here may be a high-speed RAM memory, or a non-volatile memory, such as at least one disk memory; optionally, it may also be at least one computer-readable storage medium located away from the aforementioned processor.

[0086] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to implement the following process:

[0087] Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position;

[0088] Extract high-dimensional features from visual image information and LiDAR point cloud information, perform precise matching at the roughly matched positions, and obtain fused estimated pose;

[0089] Using the measurement data of the inertial measurement unit to obtain a first estimated posture, using the measurement data of the wheel odometer to obtain a second estimated posture, and using the measurement data of the ultra-wideband to obtain a third estimated posture;

[0090] The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose are measured for pose difference, the optimal estimated pose is screened out, and the final pose is obtained after pose optimization is performed according to the optimal estimated pose.

[0091] Embodiment 5:

[0092] The present implementation provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the following process:

[0093] Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position;

[0094] Extract high-dimensional features from visual image information and LiDAR point cloud information, perform precise matching at the roughly matched positions, and obtain fused estimated pose;

[0095] Using the measurement data of the inertial measurement unit to obtain a first estimated posture, using the measurement data of the wheel odometer to obtain a second estimated posture, and using the measurement data of the ultra-wideband to obtain a third estimated posture;

[0096] The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose are measured for pose difference, the optimal estimated pose is screened out, and the final pose is obtained after pose optimization is performed according to the optimal estimated pose.

[0097] A person skilled in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0098] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instruction can be stored in a computer-readable storage medium or transmitted by a computer-readable storage medium. The computer instruction can be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (for example, coaxial cable, optical fiber, digital line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server, a data center, etc. that contains one or more available media integration. Available media can be magnetic media (for example, floppy disk, hard disk, tape), optical media (for example, DVD), or semiconductor media (for example, solid-state drive (Solid State Disk, SSD)) and the like.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A high-precision positioning method for indoor robots based on multi-source data fusion, characterized in that: The process includes: Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position; Extract high-dimensional features from visual image information and LiDAR point cloud information, perform precise matching at the roughly matched positions, and obtain fused estimated pose; Using the measurement data of the inertial measurement unit to obtain a first estimated posture, using the measurement data of the wheel odometer to obtain a second estimated posture, and using the measurement data of the ultra-wideband to obtain a third estimated posture; The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose are measured for pose difference, the optimal estimated pose is screened out, and the final pose is obtained after pose optimization is performed according to the optimal estimated pose.

2. The high-precision positioning method for indoor robots based on multi-source data fusion as claimed in claim 1, characterized in that: The positional relationship features include: the positional relationship between lines and surfaces, the positional relationship between lines, and the positional relationship between surfaces.

3. The high-precision positioning method for indoor robots based on multi-source data fusion as claimed in claim 1, characterized in that: The high-dimensional features include: color features and shape features.

4. The high-precision positioning method for indoor robots based on multi-source data fusion as claimed in claim 1, characterized in that: The inertial measurement unit measures acceleration and angular velocity data based on the internal accelerometer and gyroscope sensor, and outputs a first estimated posture through integration operation; the wheel odometer detects the rotation angle and speed information of the indoor robot's wheels and performs integration calculation to output a second estimated posture.

5. The high-precision positioning method for indoor robots based on multi-source data fusion as claimed in claim 1, characterized in that: The fused estimated pose, the first estimated pose, the second estimated pose and the third estimated pose within the same time period are obtained, and the pose difference is measured by using Euclidean distance, rotation difference or weighted difference to obtain the optimal estimated pose.

6. A multi-sensor fusion indoor robot high-precision repositioning method, characterized in that: The process includes: Extracting positional relationship features from the laser radar frame, matching the positional relationship features with the point cloud information in the global map, selecting the scene most similar to the current environment according to the matching rate, and obtaining a rough matching position; The high-dimensional features in the visual image information and the point cloud information of the laser radar are extracted, and precise matching is performed at the roughly matched position to obtain a fused estimated pose, which is used as the robot relocalization result.

7. A high-precision positioning system for indoor robots based on multi-source data fusion, characterized in that: include: The coarse matching unit is configured to: extract position relationship features in the laser radar frame, match the position relationship features with the point cloud information in the global map, select the scene most similar to the current environment according to the matching rate, and obtain a coarse matching position; The fusion pose estimation unit is configured to: extract high-dimensional features from the visual image information and the point cloud information of the laser radar, perform precise matching at the roughly matched position, and obtain a fusion estimated pose; A posture estimation unit is configured to: obtain a first estimated posture using measurement data of an inertial measurement unit, obtain a second estimated posture using measurement data of a wheel odometer, and obtain a third estimated posture using measurement data of wireless carrier communication; The posture optimization unit is configured to: perform posture difference measurement on the fused estimated posture, the first estimated posture, the second estimated posture and the third estimated posture, screen out the optimal estimated posture, and obtain the final posture after posture optimization according to the optimal estimated posture.

8. A computer device, characterized in that: include: a processor and a computer readable storage medium; a processor adapted to execute a computer program; A computer-readable storage medium having a computer program stored therein, wherein when the computer program is executed by the processor, the method for high-precision positioning of an indoor robot using multi-source data fusion as described in any one of claims 1 to 5 is implemented; or, the method for high-precision repositioning of an indoor robot using multi-sensor fusion as described in claim 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which is suitable for being loaded by a processor and executing the high-precision positioning method for an indoor robot using multi-source data fusion as described in any one of claims 1 to 5; or executing the high-precision repositioning method for an indoor robot using multi-sensor fusion as described in claim 6.

10. A computer program product, characterized in that The computer program product includes a computer program, which, when executed by a processor, implements the high-precision positioning method for an indoor robot using multi-source data fusion as described in any one of claims 1 to 5; or, implements the high-precision repositioning method for an indoor robot using multi-sensor fusion as described in claim 6.

Images