Autonomous mobile robot and indoor localization method thereof

Abstract

An autonomous mobile robot is provided, comprising a robot body, a distance sensor, and processing circuitry. The distance sensor is mounted on the robot body and configured to acquire current spatial data of an enclosed indoor space. The processing circuitry is coupled to the distance sensor and configured to identify multiple current anchor points from the current spatial data by performing feature extraction, and to determine the current position of the robot body within the enclosed indoor space based on the current anchor points by performing multi-point localization.

Description

Technical Field

[0001] This invention relates to autonomous navigation, and more particularly to autonomous mobile robots and their indoor positioning methods. Prior Technology

[0002] Autonomous navigation is a fundamental requirement for mobile robots in various applications, including goods delivery, home or office cleaning, security patrols, inventory management, and hospital services. Especially when delivering goods to different rooms within a multi-story building, autonomous mobile robots need to take elevators to reach other floors. During elevator rides, the autonomous mobile robot performs navigation operations regardless of whether there are people or non-human objects in the elevator. For safe navigation in enclosed indoor spaces (e.g., elevators with limited space and only one entrance), the requirements for autonomous mobile robots include: (1) high positioning accuracy better than 5cm, (2) readily available positioning data, (3) data remaining valid even if there are people or non-human objects in the elevator, and (4) low system cost.

[0003] Achieving high positioning accuracy in enclosed indoor spaces (such as elevators) is challenging due to the limited space and features available. Existing map-based positioning technologies (such as high-definition (HD) maps with semantic information) can provide centimeter-level accuracy, but are resource-intensive in terms of cost and computation. Grid-based mapping technologies (such as Gmapping, Hector SLAM, and Cartographer) require high-quality sensor data and a feature-rich environment, which elevators typically lack. Therefore, when using low-cost sensors in feature-scarce environments (such as elevators), these technologies may experience a decrease in positioning accuracy, leading to navigation failures for autonomous mobile robots.

[0004] Alternative methods (such as multi-QR code positioning using QR codes installed on the ceiling) have been proven to be more accurate than 5cm. However, implementing these methods in enclosed indoor spaces (such as elevators) requires installing multiple QR codes on the ceiling of each room on each floor, posing a logistical challenge. Furthermore, conditions with limited light sources (such as in older elevators) may negatively impact QR code detection, further reducing positioning accuracy.

[0005] Reinforcement learning-based localization methods can simultaneously learn localization and depth prediction tasks, providing improved feature representations. However, these methods have high computational complexity, limiting their practical applications.

[0006] Overall, existing positioning technologies for autonomous mobile robots face numerous limitations when applied to the specific constraints of indoor environments with limited space and characteristics. These limitations include decreased positioning accuracy, high computational complexity, increased system costs, and limited generalization. Therefore, autonomous mobile robots and their indoor positioning methods are needed to address these limitations. Summary of the Invention

[0007] An embodiment of the present invention provides an autonomous mobile robot, comprising a robot body; at least one distance sensor; and processing circuitry. The distance sensor is mounted on the robot body and configured to acquire current spatial data of an enclosed indoor space. The processing circuitry is coupled to the distance sensor and configured to identify multiple current anchor points from the current spatial data by performing feature extraction on the current spatial data; and to determine the current position of the robot body within the enclosed indoor space based on the current anchor points by performing multi-point localization processing.

[0008] In one embodiment, the current anchor point corresponds to multiple corners of an enclosed interior space.

[0009] In one embodiment, during the pre-positioning phase, the distance sensor is further configured to acquire pre-spatial data of the enclosed indoor space, and the processing circuit is further configured to identify multiple pre-anchor points corresponding to corners of the enclosed indoor space from the pre-spatial data by performing feature extraction on the pre-spatial data. During the positioning phase, the processing circuit is further configured to detect abnormal current anchor points by comparing the current anchor points and pre-anchor points in an aligned coordinate system using a transformation matrix between the pre-spatial data and the current spatial data; and to replace the abnormal current anchor points with corresponding pre-anchor points obtained by transforming the abnormal current anchor points using the transformation matrix, thereby determining the current position of the robot body within the enclosed indoor space.

[0010] In one embodiment, during the preliminary stage, the corners of the enclosed interior space are free of obstructions and reflective materials.

[0011] In one embodiment, the distance sensor is a LiDAR, and the current spatial data acquired through the LiDAR is a two-dimensional point cloud. Each point in the two-dimensional point cloud is represented by polar coordinates, which include radial coordinates and angular coordinates.

[0012] In one embodiment, the processing circuit is further configured to derive the relative coordinates of each of the current anchor points from the corresponding polar coordinates, and determine the current position of the robot body within the enclosed indoor space based on the relative coordinates and radial coordinates of the current anchor points.

[0013] In one embodiment, the processing circuit is further configured to determine the current position of the robot body within the enclosed indoor space by solving equations based on the relative and radial coordinates of the current anchor point using a nonlinear optimizer.

[0014] In one embodiment, the processing circuitry is further configured to identify the current anchor point from the current spatial data by minimizing the convex function representing the current spatial data.

[0015] In one embodiment, the processing circuitry is further configured to navigate the robot body to the target location within the enclosed indoor space based on the determined current position and target position of the robot body.

[0016] In one embodiment, multi-point localization processing is used to derive the current position of the robot body using four or more current anchor points.

[0017] An embodiment of the present invention provides an indoor positioning method for an autonomous mobile robot. The method includes acquiring current spatial data of an enclosed indoor space using at least one distance sensor mounted on the robot body; identifying multiple current anchor points from the current spatial data by performing feature extraction on the current spatial data; and determining the current position of the robot body within the enclosed indoor space based on the current anchor points by performing multi-point positioning processing.

[0018] This paper presents an autonomous mobile robot and indoor positioning method aimed at improving positioning accuracy, computational efficiency, and robustness, making the system more reliable and efficient in a wide range of indoor environments. By utilizing advanced feature extraction and localization technologies, the provided solution ensures accurate navigation even under feature constraints or dynamic conditions (such as occluded corners). Furthermore, its high generalization capability allows the autonomous mobile robot to seamlessly adapt to various indoor spaces, such as elevators, offices, warehouses, and public buildings, without requiring large-scale customization or additional facilities. Simple Explanation of the Diagram

[0019] The invention will be best understood from the following detailed description and examples when read in conjunction with the accompanying drawings, wherein: Figure 1 is a block diagram of an autonomous mobile robot according to one embodiment of the present invention. Figure 2 is a flowchart of the indoor positioning method applied to the autonomous mobile robot in Figure 1. Figure 3 illustrates a typical example of spatial data according to one embodiment of the present invention. Figures 4A to 4E illustrate an example of the derivation process of the relative coordinates of the anchor point according to one embodiment of the present invention. Figure 4F illustrates the information required to determine the current position of the robot body RB according to one embodiment of the present invention. Figure 5A illustrates an example of spatial data with a shield in a corner of an enclosed indoor space according to one embodiment of the present invention. Figure 5B illustrates an example of spatial data with reflective elements in a corner of an enclosed indoor space according to one embodiment of the present invention. Figure 6A is a flowchart of the preliminary stage of an indoor positioning method according to one embodiment of the present invention. Figure 6B is a flowchart of the positioning stage of an indoor positioning method according to one embodiment of the present invention. Implementation

[0020] The following description is for illustrative purposes and should not be construed as limiting. The scope of the invention can be best determined by referring to the claims.

[0021] In each of the following embodiments, the same reference numerals represent the same or similar elements or components.

[0022] It should be understood that the terms "comprising" and "including" used in the specification are used to indicate the presence of specific technical features, values, method steps, process operations, components, and / or parts, but do not exclude additional technical features, values, method steps, process operations, components, parts, or any combination thereof.

[0023] The ordinal numbers used in the scope of the patent application, such as "first," "second," and "third," are for ease of interpretation only and do not indicate any arbitrary priority relationship between them.

[0024] The embodiments of the apparatus or system provided below can also be used in the embodiments of the method, and vice versa.

[0025] Figure 1 is a block diagram of an autonomous mobile robot (AMR) according to one embodiment of the present invention. As shown in Figure 1, the autonomous mobile robot (AMR) includes a robot body RB, at least one distance sensor DS, and a processing circuit PCC.

[0026] Autonomous mobile robots (AMRs) are self-navigating robot systems capable of operating in various indoor environments and autonomously performing a variety of tasks. Applications of AMRs include, but are not limited to, goods delivery, home or office cleaning, security patrols, inventory management, and hospital services. In some implementations, during navigation through indoor spaces, AMRs can be designed to adapt to dynamic environments and interact with surrounding people and objects to efficiently and safely complete assigned tasks.

[0027] The robot body (RB) serves as the physical structure and foundation of the autonomous mobile robot (AMR). Furthermore, the robot body (RB) supports various components, including the distance sensor (DS), and houses necessary elements such as the processing circuitry (PCC), power supply, motor system, and communication module. In embodiments of this invention, the robot body (RB) acts as the carrier of the distance sensor and also serves as the primary entity for indoor positioning. By accurately determining the current position of the robot body (RB) within an enclosed indoor space, the autonomous mobile robot (AMR) can ensure precise navigation and task execution in dynamic environments.

[0028] A distance sensor DS is a sensing device used to measure the distance to surrounding walls or objects in an environment. In various embodiments, the distance sensor DS operates using technologies such as Light Detection and Ranging (LiDAR), ultrasound, or infrared to acquire spatial data related to the surrounding environment. In embodiments of the present invention, the distance sensor DS is mounted on the robot body RB and configured to acquire current spatial data of the enclosed indoor space where the robot body RB is located. It should be understood that although only one distance sensor DS is shown in Figure 1, in some embodiments, multiple distance sensors may be used to overcome the field of view (FoV) limitations of a single sensor and provide more comprehensive and accurate spatial data acquisition. Further details regarding the function of the distance sensor DS will be described below.

[0029] The processing circuit PCC is a specially designed circuit that includes one or more electronic circuits, such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or system-on-chips (SoCs), but the present invention is not limited thereto. In embodiments of the present invention, the processing circuit PCC is coupled to a distance sensor DS and configured to identify multiple current anchor nodes from the current spatial data by performing feature extraction, and to determine the current position of the robot body RB within the enclosed indoor space by performing multilateration based on the current anchor nodes. Further details will be described below.

[0030] Figure 2 is a flowchart of a method 20 for indoor positioning of the autonomous mobile robot (AMR) shown in Figure 1 according to an embodiment of the present invention. As shown in Figure 2, method 20 includes steps S201 to S203.

[0031] In step S201, the distance sensor DS installed on the robot body RB of the autonomous mobile robot (AMR) is used to acquire the current spatial data of the enclosed indoor space. Then, the method proceeds to step S202.

[0032] In step S202, multiple current anchor points are identified from the current spatial data by performing feature extraction on the spatial data acquired by the distance sensor DS. Then, the method proceeds to step S203.

[0033] In step S203, based on the current anchor point, the current position of the robot body RB in the enclosed indoor space is determined by performing multi-point positioning processing.

[0034] Spatial data refers to real-time information acquired by the distance sensor DS, representing the geometric and positional features of the surrounding indoor environment, including details about the position and distance of objects, walls, and other structural features relative to the robot body RB. Spatial data can be such as a 2D point cloud, a panoramic depth map, or a series of distance measurements, but this invention is not limited to these.

[0035] Anchor points refer to specific feature points within an enclosed indoor space identified from spatial data. These feature points can represent important structural elements, such as corners, edges, distinguishable landmarks, or other special structural features that can be used as reference points for locating the robot's body (RB).

[0036] The feature extraction performed in step S202 may include processing the current spatial data to identify specific geometric features within the enclosed indoor space. Specifically, this processing includes detecting specific elements, such as corners, edges, distinguishable landmarks, or other specific structural features that can be used as references for locating the robot body RB. Various implementations of feature extraction include Random Sample Consensus (RANSAC), Iterative Closest Point (ICP), Harris Corner Detector, or Principal Component Analysis (PCA), but the invention is not limited thereto.

[0037] The multi-point localization process in step S203 may include calculating the current position of the robot body RB within an enclosed indoor space by utilizing the geometric relationship between anchor points and the robot body RB. This operation is based on the principle that the position of an object can be identified if the distance to multiple known reference points is available. While there is no strict limitation on the number of anchor points required for the multi-point localization process, in a preferred embodiment, the multi-point localization process includes four or more anchor points to derive the current position of the robot body RB. Compared to triangulation relying on only three anchor points, multi-point localization using four or more anchor points provides additional reference information, resulting in improved error correction and more robust uncertainty handling, and helps increase the accuracy of the calculated position, especially in certain cases, such as when the spatial data is a 2D point cloud with limited accuracy.

[0038] Figure 3 illustrates a typical example of spatial data 30 according to one embodiment of the present invention. In this embodiment, anchor points (labeled as anchor points N1 to N4 in Figure 3) correspond to the four corners of the enclosed indoor space (e.g., inside an elevator) where the robot body RB is located. The multi-point positioning process in step S203 includes using anchor points N1 to N4 to derive the relative position of the robot body RB.

[0039] It should be understood that the enclosed indoor space shown in Figure 3 is generally a rectangle with four corners, but this is only an example and not a limitation. In other cases, the enclosed indoor space where the robot body RB is located can be of other shapes, and the number of anchor points is not limited. For example, a triangular indoor space can have three anchor points, and a hexagonal indoor space can have six anchor points.

[0040] In one embodiment, the distance sensor DS is a LiDAR, and the spatial data obtained by the LiDAR is a 2D point cloud. Since the LiDAR scans the surrounding environment by rotating, each point in the 2D point cloud is represented in polar coordinates (r,θ), where the polar coordinates include the radial coordinate r and the angular coordinate θ, which respectively represent the distance from the robot body RB to the point and the angle of the point relative to the reference direction of the robot body RB.

[0041] In another embodiment, the operation of identifying anchor points from the spatial data can be achieved by minimizing a function representing the spatial data. In other words, the problem of extracting features (corners) in the enclosed indoor space is a minimization problem of a convex function. The 2D point cloud of the enclosed indoor space is represented in the form of a convex function, and then a minimization operation is performed to extract the corner vector. The mathematical expression of this process is as follows: Let F be a function representing the 2D point cloud of the enclosed indoor space, and G∩F (i.e., G is a sub - function of F). Then, if for all s G, s ≠ r, F(r) < F(s), then r F is a local minimum, where r, s ∈ R. Then, the extracted corner vector can be expressed as , where, 、 are radial coordinates, and is the corresponding angular coordinate of the extracted corner.

[0042] In one embodiment, step S203 further includes deriving the relative coordinates of each of the current anchor points from the corresponding polar coordinates, and determining the current position of the robot body RB within the enclosed indoor space based on the relative coordinates and the radial coordinates of the current anchor point. Further details will be described below with reference to Figures 4A through 4E.

[0043] Figures 4A to 4E illustrate an example of the process for deriving the relative coordinates of anchor points according to one embodiment of the present invention, and Figure 4F illustrates the information required to determine the current position of the robot body RB according to one embodiment of the present invention. In this example, the four corners are extracted, and in Figures 4A to 4F, the corresponding anchor points are labeled as anchor points N1 to N4. Furthermore, anchor point N1 is set as the origin. Therefore, the positioning problem can be described as follows: Let A≡(x,y) be the unknown coordinates of the robot body RB within the enclosed indoor space, and These are the coordinates of anchor points N2, N3, and N4 relative to anchor point N1, respectively.

[0044] Referring to Figure 4A, according to the Law of Cosines, the coordinates of the current anchor point N2 are... The calculation is as follows: in, Obtain from Z.

[0045] Then, referring to Figure 4B, and based on the law of cosines, the distance ρ between anchor point N1 and anchor point N3 is calculated as follows: in, Obtain from Z.

[0046] Then, referring to Figure 4C, and based on the law of cosines, the distance ρ between anchor point N2 and anchor point N3 is calculated as follows: in, Obtain from Z.

[0047] Then, referring to Figure 4D, and based on the Law of Cosines, the coordinates of anchor point N3 are... The calculation is as follows:

[0048] Then, referring to Figure 4E, and based on the Law of Cosines, the coordinates of anchor point N4 are... The calculation is as follows:

[0049] Then, refer to Figure 4F. As shown in Figure 4F, the information required to determine the coordinates (x, y) of the robot body RB includes the coordinates of anchor points N2, N3, and N4 relative to anchor point N1. , ,and , and are the radial coordinates from anchor point N1 to anchor point N4. , ,and Once all the information is available, multi-point positioning can be used to determine the current position of the robot body RB, represented by coordinates (x, y).

[0050] In another embodiment, the aforementioned multi-point localization includes using a nonlinear optimizer to solve equations based on the relative and radial coordinates of the current anchor points. In the example shown in Figure 4F, the Euclidean distances between the robot body RB and anchor points N1 to N4 are {A, }、{A, }、{A, }、and {A, } can be expressed as the following equation: It should be noted that the above equation is nonlinear, and the radial coordinates are measured by the distance sensor DS. , ,and This may contain errors. In such cases, a nonlinear optimizer (such as the Levenberg-Marquardt algorithm or the Gauss-Newton algorithm) can be used to find the best approximate solution that satisfies all the equations.

[0051] Figure 5A illustrates an example of spatial data 50A when corners C1 and C2 of an enclosed indoor space have occlusions according to one embodiment of the present invention. Occlusions at corners C1 and C2 can be caused by various reasons, such as someone standing in the corner or an object temporarily placed in the corner area. Such occlusions may prevent the distance sensor DS from accurately acquiring spatial data of these corners, resulting in an incomplete or distorted representation of the indoor environment. This can lead to errors in anchor point identification, thereby affecting the accuracy of the robot body RB's positioning.

[0052] Figure 5B illustrates an example of spatial data 50B when corners C1 and C2 of an enclosed indoor space, according to one embodiment of the present invention, have reflective objects. Reflective objects (e.g., mirrors or other reflective surfaces) at corners C1 and C2 can cause multiple reflections of the LiDAR signal, leading to inaccurate or misleading spatial data. As shown in Figure 5B, the spatial data in the area near corners C1 and C2 shows scattered and inconsistent points, representing the presence of reflection artifacts. Because reflections may be misinterpreted as actual environmental features, these artifacts can lead to incorrect identification of the current anchor point. Therefore, the positioning accuracy of the robot body RB will be affected.

[0053] In one embodiment, prior information about the indoor environment can be used to address the problems shown in Figures 5A and 5B. Specifically, in step S203 of the embodiment in Figure 2, the current position of the robot body can be determined by referring to pre-stored reference spatial data containing the expected positions of anchor points. During the positioning phase, the current spatial data and the prior spatial data are compared to identify differences caused by obstructions or reflective objects. If an anomaly is detected in the current anchor point, such as a missing or misaligned point, the abnormal anchor point is replaced with the corresponding anchor point derived from the prior data. Further details will be described below with reference to Figures 6A and 6B.

[0054] Figure 6A is a flowchart of the preliminary stage 60A of an indoor positioning method according to one embodiment of the present invention. As shown in Figure 6A, the preliminary stage 60A includes steps S601 and S602.

[0055] In step S601, a distance sensor is used to acquire pre-spatial data of the enclosed indoor space. Then, in step S602, feature extraction is performed on the pre-spatial data to identify multiple pre-anchor points corresponding to the corners of the enclosed indoor space.

[0056] During the localization phase, pre-identified anchor points identified during pre-phase 60A are referenced as expected anchor locations to correct any abnormal current anchor points. In a preferred embodiment, during pre-phase 60A, the corners of the enclosed indoor space are free of obstructions and reflective objects. This allows the autonomous mobile robot system to obtain accurate and unobstructed anchor point location references and provides a reliable benchmark for comparing and correcting the current anchor point during the localization phase.

[0057] Figure 6B is a flowchart of the positioning stage 60B of an indoor positioning method according to one embodiment of the present invention. As shown in Figure 6B, the positioning stage 60B includes steps S611 and S612.

[0058] In step S611, by utilizing the transformation matrix between the pre-existing spatial data and the current spatial data, the current anchor point and the pre-existing anchor point are compared in the aligned coordinate system to detect abnormal current anchor points. In other words, the transformation matrix is ​​used to align the current anchor point and the pre-existing anchor point to the same coordinate system to perform the comparison, thereby detecting any differences or anomalies. The transformation matrix can be obtained through various methods, such as using point cloud alignment techniques (e.g., ICP, or feature pairing algorithms). The above methods calculate the optimal transformation matrix by minimizing the error between corresponding points in the pre-existing and current spatial data. Alternatively, tools such as the ROS navigation component provide built-in functions for generating transformation matrices between different coordinate systems.

[0059] In step S612, the abnormal current anchor point is replaced with a corresponding pre-anchor point obtained by transforming the abnormal current anchor point using a transformation matrix, in order to determine the current position of the robot body within the enclosed indoor space. Specifically, a transformation matrix is ​​first applied to align the pre-anchor point with the current anchor point, ensuring that the coordinates of the pre-anchor point correspond to the same reference coordinates as the current spatial data. Then, the abnormal current anchor point is replaced with a reliable, pre-verified pre-anchor point. Therefore, by avoiding the use of erroneous current anchor points that may be affected by environmental factors (such as obstructions or reflective objects), the localization of the autonomous mobile robot becomes more reliable and robust.

[0060] In one embodiment, the processing circuit PCC is further configured to navigate the robot body to the target location within an enclosed indoor space based on the determined current position and target location of the robot body. Specifically, the processing circuit PCC uses the current position of the robot body RB as the starting position and generates the optimal path to the target location by considering obstacles and limitations in the indoor environment. The navigation process may include calculating a series of waypoints or using path planning algorithms (such as A* or Dijkstra's algorithm) to ensure efficient and safe movement. Furthermore, the processing circuit PCC continuously monitors the current position of the robot body RB relative to the planned path and dynamically adjusts the path if unexpected obstacles are detected or environmental conditions change.

[0061] In another embodiment, the target location can be determined based on the robot's specific task, and can be dynamically adjusted according to current spatial data. For example, for a delivery robot, the initial target location can be set near the elevator door, allowing the robot to quickly exit the elevator after reaching the designated floor. However, if the current spatial data indicates that the location is already occupied, the processing circuit PCC can adjust the target location to another suitable location within the elevator to avoid delays. For a cleaning robot, the initial target location can be set to a corner of the elevator to minimize interference with passengers or other objects. However, if the current spatial data indicates that the corner is already occupied, the processing circuit PCC can adjust the target location to another corner. Furthermore, the processing circuit PCC can incorporate a priority mechanism for determining the target location, prioritizing certain areas within an enclosed indoor space based on the robot's task. For example, in scenarios with high foot traffic, the cleaning robot can prioritize moving to less crowded areas while optimizing its path to ensure task efficiency. By combining specific mission objectives, dynamic adjustments based on current spatial data, and priority-based decision-making, the autonomous mobile robot (AMR) can be ensured to navigate efficiently and operate in complex indoor environments.

[0062] This paper presents an autonomous mobile robot and indoor positioning method aimed at improving positioning accuracy, computational efficiency, and robustness, making the system more reliable and efficient in a wide range of indoor environments. By utilizing advanced feature extraction and localization technologies, the provided solution ensures accurate navigation even under feature constraints or dynamic conditions (such as occluded corners). Furthermore, its high generalization capability allows the autonomous mobile robot to seamlessly adapt to various indoor spaces, such as elevators, offices, warehouses, and public buildings, without requiring large-scale customization or additional facilities.

[0063] The present invention has been described above with reference to various embodiments. It is obvious that the teachings of this specification can be performed in many ways. The specific structures or functions disclosed in the examples are merely representative. Based on the teachings of this specification, those skilled in the art to which this invention pertains should understand that any of the disclosed embodiments can be performed alone, or two or more embodiments can be performed together.

[0064] Although the invention has been described by way of example and preferred embodiments, it should be understood that the invention is not limited to the disclosed embodiments. Rather, the above embodiments are intended to cover various modifications and similar configurations (as will be apparent to those skilled in the art to which this invention pertains). Therefore, the following claims should be given the broadest interpretation to cover all such modifications and similar configurations.

[0065] 20: Method 30, 50A, 50B: Spatial Data 60A: Preliminary Stage 60B: Positioning Phase AMR: Autonomous Mobile Robot C1, C2: Corner DS: Distance Sensor N1, N2, N3, N4: Anchor points PCC: Processing Circuit RB: Robot Body r, r1, r2, r3, r4: radial coordinates S201, S202, S203, S601, S602, S611, S612: Steps θ: angular coordinates ρ: distance

Claims

1. An autonomous mobile robot, comprising: a robot body; at least one distance sensor mounted on the robot body and configured to acquire current spatial data of an enclosed indoor space, the current spatial data including geometric and positional features of the enclosed indoor space; and a processing circuit coupled to the distance sensor and configured to: identify a plurality of current anchor points from the current spatial data by performing feature extraction on the current spatial data; and determine a current position of the robot body within the enclosed indoor space based on the current anchor points by performing a multi-point localization process.

2. The autonomous mobile robot as described in claim 1, wherein the aforementioned current anchor point corresponds to multiple corners of the aforementioned enclosed indoor space.

3. The autonomous mobile robot as described in claim 2, wherein: During a preliminary stage, the distance sensor is further configured to acquire preliminary spatial data of the enclosed indoor space, and the processing circuit is further configured to identify a plurality of preliminary anchor points corresponding to the corners of the enclosed indoor space from the preliminary spatial data by performing feature extraction on the preliminary spatial data; and during a positioning stage, the processing circuit is further configured to: compare the current anchor points and the preliminary anchor points in an aligned coordinate system using a transformation matrix between the preliminary spatial data and the current spatial data to detect a plurality of abnormal current anchor points; and replace the abnormal current anchor points with the corresponding preliminary anchor points obtained by transforming the abnormal current anchor points using the transformation matrix to determine the current position of the robot body in the enclosed indoor space.

4. The autonomous mobile robot as described in claim 3, wherein during the aforementioned preliminary stage, the aforementioned corners of the enclosed indoor space are free of obstructions and reflective objects.

5. The autonomous mobile robot as described in claim 2, wherein the distance sensor is a LiDAR, and the current spatial data acquired by the LiDAR is a two-dimensional point cloud, wherein each point in the two-dimensional point cloud is represented by polar coordinates, the polar coordinates including a radial coordinate and an angular coordinate.

6. The autonomous mobile robot as described in claim 5, wherein the processing circuit is further configured to derive multiple relative coordinates of each of the current anchor points from the corresponding polar coordinates, and to determine the current position of the robot body within the enclosed indoor space based on the relative coordinates and radial coordinates of the current anchor points.

7. The autonomous mobile robot as described in claim 6, wherein the processing circuit is further configured to determine the current position of the robot body within the enclosed indoor space by using a nonlinear optimizer to solve multiple equations based on the relative coordinates and radial coordinates of the current anchor point.

8. The autonomous mobile robot as described in claim 5, wherein the processing circuitry is further configured to identify the current anchor point from the current spatial data by minimizing a convex function representing the current spatial data.

9. The autonomous mobile robot as described in claim 1, wherein the processing circuitry is further configured to navigate the robot body to the target position within the enclosed indoor space based on the determined current position of the robot body and a target position.

10. The autonomous mobile robot as described in claim 1, wherein the multi-point localization process is used to derive the current position of the robot body using four or more of the current anchor points.

11. An indoor positioning method applied to an autonomous mobile robot, the method comprising: acquiring current spatial data of an enclosed indoor space using at least one distance sensor mounted on the robot body of one of the autonomous mobile robots, the current spatial data including geometric and positional features of the enclosed indoor space; identifying a plurality of current anchor points from the current spatial data by performing feature extraction on the current spatial data; and determining a current position of the robot body within the enclosed indoor space based on the current anchor points by performing a multi-point positioning process.

12. The indoor positioning method as described in claim 11, wherein the aforementioned current anchor point corresponds to a plurality of corners of the aforementioned enclosed indoor space.

13. The indoor positioning method as described in claim 12 further comprises: during a preliminary stage, acquiring preliminary spatial data of the enclosed indoor space using the distance sensor, and identifying a plurality of preliminary anchor points corresponding to the corners of the enclosed indoor space from the preliminary spatial data by performing feature extraction on the preliminary spatial data; and during a positioning stage, comparing the current anchor points and the preliminary anchor points in an aligned coordinate system using a transformation matrix between the preliminary spatial data and the current spatial data to detect a plurality of abnormal current anchor points, and replacing the abnormal current anchor points with the corresponding preliminary anchor points obtained by transforming the abnormal current anchor points using the transformation matrix, so as to determine the current position of the robot body in the enclosed indoor space.

14. The indoor positioning method as described in claim 13, wherein during the aforementioned preliminary stage, the aforementioned corners of the enclosed indoor space are free of obstructions and reflective objects.

15. The indoor positioning method as described in claim 12, wherein the distance sensor is an optical radar, and the current spatial data acquired by the optical radar is a two-dimensional point cloud, wherein each point in the two-dimensional point cloud is represented by polar coordinates, the polar coordinates including a radial coordinate and an angular coordinate.

16. The indoor positioning method as described in claim 15, wherein determining the current position of the robot body within the enclosed indoor space further comprises deriving multiple relative coordinates of each of the current anchor points from the corresponding polar coordinates, and determining the current position of the robot body within the enclosed indoor space based on the relative coordinates of the current anchor points and the radial coordinates.

17. The indoor positioning method as described in claim 16, wherein determining the current position of the robot body within the enclosed indoor space further comprises determining the current position of the robot body within the enclosed indoor space by using a nonlinear optimizer to solve multiple equations based on the relative coordinates and radial coordinates of the current anchor point.

18. The indoor positioning method as described in claim 15 further includes identifying the current anchor point from the current spatial data by minimizing a convex function representing the current spatial data.

19. The indoor positioning method as described in claim 11 further includes navigating the robot body to the target location within the enclosed indoor space based on the determined current location of the robot body and a target location.

20. The indoor positioning method as described in claim 11, wherein the multi-point positioning process is used to derive the current position of the robot body using four or more of the current anchor points.

Images