Method for dynamic environment adaptive navigation, robot and central processing unit

Abstract

The application discloses a kind of dynamic environment self-adapting navigation method, robot and central processing unit, the method comprises: laser radar scans environment and generates current point cloud data;Electronic price tag unit sends measurement signal to base station and obtains angle of arrival positioning position;Nine-axis sensor measures robot nine-axis data;Central processing unit compares current point cloud data with pre-generated supermarket point cloud map data, and when determining that surrounding environment changes according to comparison result, mark and evaluate abnormal area;When determining that abnormal area is the state that needs to update immediately, first current position is obtained according to angle of arrival positioning position and nine-axis data fusion positioning;Determine the first path that returns to historical safe location from current position;Navigation unit controls robot to return to safe location according to the path;Laser radar reconstructs point cloud map according to abnormal area;Navigation control unit controls robot to complete task according to the reconstructed map.The application can realize dynamic environment self-adapting navigation.

Description

Technical Field

[0001] This invention relates to the field of supermarket technology, and in particular to a method for dynamic environment adaptive navigation, a robot, and a central processing unit. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] With the development of intelligent technology, intelligent devices are being used more and more widely in large supermarkets and warehouses, undertaking tasks such as picking, replenishing, and inventory management, greatly improving operational efficiency. However, due to environmental factors such as similar merchandise shelves, strong glare on the ground, or dim lighting, coupled with potential odometry errors in the positioning system itself, inspection robots often encounter positioning failures. This not only affects their normal navigation and movement but also increases the cost and frequency of manual maintenance.

[0004] Current autonomous mobile robot technologies commonly rely on LiDAR point cloud data for environmental mapping and navigation. However, this method faces significant limitations in dynamic environments, such as warehouses where shelf locations change. Specifically, when the physical layout of the environment changes, the original point cloud map may no longer be accurate, making the robot prone to getting lost. Furthermore, relying solely on LiDAR point cloud data makes it difficult to quickly adapt and update in complex or changing environments, thus affecting the robot's navigation efficiency and accuracy. Summary of the Invention

[0005] This invention provides a method for adaptive navigation in dynamic environments, enabling efficient and accurate adaptive navigation in dynamic environments. This method is applied to robots and includes: The lidar scans the surrounding environment and generates current point cloud data of the surrounding environment, which is then sent to the central processing unit. The electronic price tag unit sends a measurement signal to the base station; the base station is used to determine the angle of arrival based on the measurement signal, and sends the angle of arrival to the angle of arrival server; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. The nine-axis sensor measures the robot's nine-axis data and sends it to the central processing unit. The central processing unit compares the current point cloud data with the pre-generated supermarket point cloud map data. When it is determined that the surrounding environment has changed based on the comparison results, it marks the abnormal area on the map, evaluates the abnormal area, and determines the state of the abnormal area. When it is determined that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately based on the evaluation results, it performs fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position. It then calls the last confirmed safe position in the historical positioning data to determine the first optimal path from the first current positioning position back to that safe position. The navigation control unit controls the robot to return to the safe position according to the first optimal path; LiDAR captures environmental changes that have occurred since the last scan in anomaly areas and reconstructs point cloud maps based on these changes. The navigation control unit regenerates the navigation path based on the reconstructed point cloud map; based on the regenerated navigation path, the robot is controlled to continue performing the task from that safe position.

[0006] This invention provides a method for adaptive navigation in dynamic environments, enabling efficient and accurate adaptive navigation in dynamic environments. The method is applied to a central processing unit and includes: The current point cloud data is compared with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the lidar, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment. When changes in the surrounding environment are determined based on the comparison results, abnormal areas on the map are marked. Assess the abnormal region and determine its status; When the evaluation results determine that an abnormal area is in a state where navigation cannot be achieved normally and an immediate update is required, the robot's first current location is obtained by fusing the arrival angle positioning position and nine-axis data. The arrival angle positioning position is sent by the arrival angle server, which is used to determine the robot's arrival angle positioning position based on the arrival angle. The arrival angle is sent by the base station, which is used to determine the arrival angle based on the measurement signal sent by the electronic price tag unit. The nine-axis data is sent by the nine-axis sensor, which is used to measure the robot's nine-axis data. The last confirmed safe location in the historical positioning data is retrieved, and a first optimal path is determined to return from the first current positioning location to that safe location and sent to the navigation control unit. The navigation control unit controls the robot to return to the safe location according to the first optimal path. The lidar is also used to capture environmental changes that have occurred since the last scan from abnormal areas and reconstruct a point cloud map based on the environmental changes. The navigation path is regenerated based on the reconstructed point cloud map. Controlling the robot to continue performing the task from the safe location based on the regenerated navigation path is further configured to: regenerate the navigation path based on the reconstructed point cloud map; and control the robot to continue performing the task from the safe location based on the regenerated navigation path.

[0007] This invention also provides a robot for adaptive navigation in dynamic environments, which can efficiently and accurately adaptively navigate in dynamic environments. The robot includes: LiDAR is used to scan the surrounding environment, generate current point cloud data of the surrounding environment and send it to the central processing unit; it captures environmental changes that have occurred since the last scan from abnormal areas and reconstructs the point cloud map based on the environmental changes. An electronic price tag unit is used to send measurement signals to a base station; the base station is used to determine the angle of arrival based on the measurement signals and send the angle of arrival to an angle of arrival server; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. A nine-axis sensor is used to measure data from the robot's nine axes and send it to the central processing unit. The central processing unit compares the current point cloud data with pre-generated supermarket point cloud map data. When the comparison results indicate that the surrounding environment has changed, it marks abnormal areas on the map, evaluates the abnormal areas, and determines their status. When the evaluation results indicate that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, it performs fusion positioning based on the angle of arrival and nine-axis data to obtain the robot's first current positioning position. It then calls the last confirmed safe position from the historical positioning data and determines the first optimal path from the first current positioning position back to that safe position. The navigation control unit is used to control the robot to return to the safe position according to the first optimal path; to regenerate the navigation path according to the reconstructed point cloud map; and to control the robot to continue to perform the task from the safe position according to the regenerated navigation path.

[0008] This invention provides a central processing unit for adaptive navigation in dynamic environments, enabling efficient and accurate adaptive navigation in dynamic environments. The central processing unit includes: The comparison module is used to compare the current point cloud data with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the lidar, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment. The marking module is used to mark abnormal areas on the map when changes in the surrounding environment are determined based on comparison results; The evaluation module is used to evaluate the abnormal region and determine the state of the abnormal region; The fusion positioning module is used to perform fusion positioning based on the angle of arrival (AHA) and nine-axis data when the evaluation results determine that an abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, to obtain the robot's first current positioning position. The AHA is sent by the angle of arrival server, which is used to determine the robot's AHA positioning position based on the angle of arrival. The angle of arrival is sent by the base station, which is used to determine the angle of arrival based on the measurement signal sent by the electronic price tag unit. The nine-axis data is sent by the nine-axis sensor, which is used to measure the robot's nine-axis data. The optimal path determination module is used to call the last confirmed safe location in the historical positioning data, determine a first optimal path from the first current positioning location back to the safe location, and send it to the navigation control unit; the navigation control unit is used to control the robot to return to the safe location according to the first optimal path; the lidar is also used to capture environmental changes that have occurred since the last scan from abnormal areas, and reconstruct a point cloud map according to the environmental changes; the regeneration of the navigation path according to the reconstructed point cloud map; and controlling the robot to continue to complete the task from the safe location according to the regenerated navigation path are further configured to: regenerate the navigation path according to the reconstructed point cloud map; and control the robot to continue to complete the task from the safe location according to the regenerated navigation path.

[0009] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for dynamic environment adaptive navigation.

[0010] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for dynamic environment adaptive navigation.

[0011] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for dynamic environment adaptive navigation.

[0012] In this embodiment of the invention, the dynamic environment adaptive navigation scheme involves: a lidar scanning the surrounding environment to generate current point cloud data, which is then sent to a central processing unit; an electronic price tag unit sending measurement signals to a base station; the base station determining the angle of arrival based on the measurement signals and sending the angle of arrival to an angle of arrival server; the angle of arrival server determining the robot's positioning position based on the angle of arrival and returning it to the central processing unit; a nine-axis sensor measuring nine-axis data of the robot and sending it to the central processing unit; the central processing unit comparing the current point cloud data with pre-generated supermarket point cloud map data; when a change in the surrounding environment is determined based on the comparison results, an abnormal area on the map is marked, the abnormal area is evaluated, and its state is determined; if the evaluation results determine that the abnormal area is not present... When navigation requires immediate updates, the robot performs fusion positioning based on the angle of arrival and nine-axis data to obtain its first current location. It then retrieves the last confirmed safe location from historical positioning data and determines the first optimal path from the first current location back to that safe location. The navigation control unit controls the robot to return to the safe location based on this first optimal path. The lidar captures environmental changes that have occurred since the last scan from abnormal areas and reconstructs a point cloud map based on these changes. The navigation control unit regenerates the navigation path based on the reconstructed point cloud map. Following the regenerated navigation path, the robot is controlled to continue its task from the safe location, achieving adaptive navigation in dynamic environments and improving the robot's navigation efficiency and accuracy. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating the dynamic environment adaptive navigation method applied to a robot in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the principle of dynamic environment adaptive navigation in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the dynamic environment adaptive navigation method applied to the central processing unit in an embodiment of the present invention. Figure 4 This is a schematic diagram of the structure of the dynamic environment adaptive navigation robot in an embodiment of the present invention; Figure 5 This is a schematic diagram of the central processing unit for dynamic environment adaptive navigation in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0015] The acquisition, storage, use, and processing of data in this application comply with relevant laws and regulations.

[0016] Before introducing the embodiments of the present invention, the terms involved in the present invention will be introduced first.

[0017] ESL: Electronic Shelf Label or Electronic Price Tag. Electronic shelf labels (ESLs) are electronic display devices used in retail stores, typically installed on shelves, to display product prices, promotional information, inventory status, etc.

[0018] AOA server: Used to manage and control wireless devices, such as ESL, to ensure that tags can receive and send information.

[0019] AOA: Angle of Arrival, the angle at which a wireless signal arrives at the receiving device.

[0020] AOA positioning is a technique that uses the angle of arrival of a received signal to determine the location of the signal source.

[0021] Nine-axis sensor MEMS: Inertial sensing devices, including accelerometers, gyroscopes, and geomagnetic sensors.

[0022] Figure 1 This is a flowchart illustrating the dynamic environment adaptive navigation method applied to a robot in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes the following steps: Step 101: The lidar scans the surrounding environment and generates current point cloud data of the surrounding environment, which is then sent to the central processing unit. Step 102: The electronic price tag unit sends a measurement signal to the base station; the base station is used to determine the angle of arrival based on the measurement signal, and sends the angle of arrival to the angle of arrival server; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. Step 103: The nine-axis sensor measures the robot's nine-axis data and sends it to the central processing unit; Step 104: The central processing unit compares the current point cloud data with the pre-generated supermarket point cloud map data. When it is determined that the surrounding environment has changed based on the comparison results, it marks the abnormal area on the map, evaluates the abnormal area, and determines the state of the abnormal area. When it is determined that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately based on the evaluation results, it performs fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position. It calls the last confirmed safe position in the historical positioning data and determines the first optimal path from the first current positioning position back to the safe position. Step 105: The navigation control unit controls the robot to return to the safe position according to the first optimal path; Step 106: The lidar captures environmental changes that have occurred since the last scan from the abnormal area, and reconstructs the point cloud map based on the environmental changes; Step 107: The navigation control unit regenerates the navigation path based on the reconstructed point cloud map; based on the regenerated navigation path, the robot is controlled to continue performing the task from the safe position.

[0023] The dynamic environment adaptive navigation method provided in this embodiment of the invention operates as follows: A lidar scans the surrounding environment, generating current point cloud data of the surrounding environment and sending it to a central processing unit; an electronic price tag unit sends measurement signals to a base station; the base station determines the angle of arrival based on the measurement signals and sends the angle of arrival to an angle of arrival server; the angle of arrival server determines the robot's positioning position based on the angle of arrival and returns it to the central processing unit; a nine-axis sensor measures the robot's nine-axis data and sends it to the central processing unit; the central processing unit compares the current point cloud data with pre-generated supermarket point cloud map data; when a change in the surrounding environment is determined based on the comparison result, an abnormal area on the map is marked, the abnormal area is evaluated, and the state of the abnormal area is determined; if the abnormal area is determined to be a... When navigation cannot be achieved normally and an immediate update is required, the robot's first current location is obtained by fusing the arrival angle positioning position and nine-axis data. The last confirmed safe location from the historical positioning data is retrieved to determine the first optimal path from the first current location to that safe location. The navigation control unit controls the robot to return to the safe location according to the first optimal path. The LiDAR captures environmental changes that have occurred since the last scan from abnormal areas and reconstructs a point cloud map based on these changes. The navigation control unit regenerates the navigation path based on the reconstructed point cloud map. Based on the regenerated navigation path, the robot continues to perform the task from the safe location, achieving adaptive navigation in dynamic environments and improving the robot's navigation efficiency and accuracy. A detailed description follows.

[0024] To address the problem of existing robots easily getting lost in dynamic environments when relying on LiDAR point cloud mapping, this invention proposes a robot-assisted navigation and LiDAR remapping scheme that combines Angular Arrival (AOA) localization technology and a nine-axis sensor. The specific components and process of this system are as follows: To facilitate understanding of how this invention is implemented, the following first combines... Figure 4 Introduce the components of the robot system.

[0025] 1. Robot

[0026] a. LiDAR: Used to continuously scan the environment, generate point cloud data of surrounding objects, and achieve preliminary positioning and navigation.

[0027] b. Nine-axis MEMS sensor: measures the robot's acceleration, angular velocity, and magnetic field strength, providing data on the robot's speed and direction of movement, and assisting in more accurate real-time positioning.

[0028] c. Central Processing Unit: Receives data from lidar, AOA positioning module and accelerometer, processes and fuses it to achieve precise positioning and path planning.

[0029] d. such as Figure 4 The “Storage Unit 09” shown contains point cloud maps and historical navigation data of the storage environment for reference and updating.

[0030] e. Navigation Control Unit: Guides robot movement based on processed data, controls robot behavior and path selection.

[0031] f. such as Figure 4 The “camera 08” shown is the image acquisition device in this embodiment of the invention: the camera plays a crucial role as a visual sensor in the robot system, used to capture visual information of the environment and assist the robot in performing more complex navigation and interaction tasks.

[0032] g. ESL: ESL is used in robotic systems to interact with smart tags in retail environments, providing real-time product information and location services. Based on this, in one embodiment, the aforementioned dynamic environment adaptive navigation method may further include: The electronic price tag unit interacts with the smart electronic tags in the supermarket to obtain the location information of the product currently bound to the smart electronic tag. Based on the location information of the product currently bound to the smart electronic tag, the robot performs assisted positioning.

[0033] In specific implementation, embodiments of the present invention also use the ESL unit to interact with smart electronic tags in supermarkets to perform robot-assisted positioning, thereby improving the accuracy and flexibility of robot positioning.

[0034] 2. AOA positioning module: a. AOA server: Used to manage and control wireless devices, such as ESL, to ensure that tags can receive and send information.

[0035] b. Base station: It uses wireless signal transmission to determine the robot's position relative to the signal source by the angle of arrival of the received signal.

[0036] Figure 2 This is a schematic diagram illustrating the principle of dynamic environment adaptive navigation in an embodiment of the present invention. The following is in conjunction with... Figure 2 This paper introduces the process and methodology of adaptive navigation in dynamic environments.

[0037] 1. Base station deployment and configuration

[0038] (1) Location selection and deployment: a. Deploy multiple base stations in key areas of dynamic environments such as shopping malls and supermarkets to ensure accurate AOA signals are available throughout the entire area. Select appropriate frequency points and adjust signal transmission intervals and power to optimize base station response time and signal quality.

[0039] b. Consider the layout of the base stations to ensure their coverage fully covers all areas requiring location services. The installation height and location of the base stations need to be determined based on the actual environment to maximize coverage efficiency.

[0040] (2) Base station configuration: All base stations must be installed horizontally and equipped with high-precision level detection devices to compensate for errors caused by possible installation tilt or uneven ground. Figure 2 The terms "AOA base station deployment" and "AOA positioning" are shown.

[0041] 2. Deploy robots

[0042] LiDAR Scanning and Navigation Initialization: When the robot starts, it first uses a LiDAR to scan the environment and generate a high-precision initial point cloud map. This will serve as the primary basis for basic navigation, i.e., the LiDAR scans the surrounding environment and generates the current point cloud data of the surrounding environment in step 101 above. Simultaneously, the robot's navigation control unit plans the initial navigation path according to a preset program in order to begin executing the task.

[0043] 3. Environmental scanning and preliminary mapping

[0044] LiDAR system: The robot uses a lidar system to scan its surroundings, generate an initial point cloud map, and perform basic navigation, such as... Figure 2 The image shows "Generate initial point cloud map of LiDAR scanning environment".

[0045] 4. AOA plus sensor fusion positioning and navigation system.

[0046] AOA plus sensor fusion positioning: Navigation using LiDAR (i.e., Figure 2 While using LiDAR for navigation and positioning, the robot also uses an AOA (Automatic Area of ​​Array) positioning module, such as... Figure 2 The "AOA positioning" shown, and the nine-axis (accelerometer, gyroscope, magnetometer) sensor based on known supermarket store maps (such as... Figure 2 The "Store Map Information" shown is used for independent positioning, such as... Figure 2 The "nine-axis sensor MEMS" and "nine-axis sensor positioning" shown demonstrate the implementation of a fusion positioning and navigation system, such as... Figure 2 The diagram illustrates "fusion localization." In practice, a Kalman filter algorithm can be used for fusion localization. The AOA (Automatic Anchored Array) provides observations to initially determine the position; the nine-axis sensor provides state estimates to predict the robot's next position. The filter combines the data from both to output the final position information.

[0047] Specifically, the fusion positioning process, which in one embodiment involves performing fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position, may include: Obtain the robot's current position and velocity, as well as the covariance matrix of that position and velocity; Using acceleration and angular velocity data from a nine-axis sensor, along with the robot's motion equations, the robot's position and velocity at the next moment are predicted to obtain the current predicted state. Based on the prediction error and noise model, the covariance matrix is ​​updated to obtain the uncertainty of the current predicted state. Based on the current prediction state, calculate the predicted observation value of the nine-axis sensor; take the angle of arrival positioning position as the actual observation value, and calculate the difference between the actual observation value and the predicted observation value; according to the uncertainty of the current prediction state and the uncertainty of the actual measurement state, dynamically adjust the weight of the angle of arrival positioning position and the nine-axis positioning position (which is obtained from the nine-axis data) in the final fusion result. Based on the differences and the weights, the robot's current position and speed are corrected to obtain the first current positioning position.

[0048] The above-mentioned fusion positioning method will be described in detail below.

[0049] In this embodiment of the invention, by combining the system's motion model and observation model in two stages of prediction and updating, data from multiple sensors (such as a nine-axis sensor, AOA positioning, and lidar) can be fused to achieve accurate positioning in dynamic environments.

[0050] This invention utilizes a two-stage process of prediction and update, dynamically adjusting the weights of sensor data to achieve state estimation and error minimization. State variables include position and velocity. The prediction stage updates the state based on a motion model, while the update stage refines the prediction results by incorporating measured values.

[0051] (1) Initialization

[0052] Objective: Initialize the robot's position, velocity, and its uncertainties (if it is the current moment, we can obtain the robot's current position and velocity (i.e., current state), as well as the covariance matrix of that position and velocity).

[0053] Define state variables, including position and velocity.

[0054] Set initial values ​​for state variables, such as the robot's initial position and velocity.

[0055] Define a covariance matrix (the data in this covariance matrix represents the accuracy of the state and the correlation between the states) to represent the uncertainty of the initial state. At the same time, set the covariance matrices of the nine-axis sensor and process noise to ensure the algorithm's adaptability to different data sources.

[0056] (2) Prediction steps

[0057] State prediction: In this stage, using the acceleration and angular velocity data provided by the nine-axis sensor, the robot's position and velocity at the next moment are predicted according to the motion model (such as attitude prediction model, velocity prediction model, position prediction model), thus obtaining the current predicted state.

[0058] Uncertainty update: Combining prediction error and noise model (e.g. Gaussian noise model), update the covariance matrix (which can be seen as the credibility of the state prediction) to evaluate the credibility of the state prediction and obtain the uncertainty of the current predicted state.

[0059] (3) Update steps

[0060] The AOA positioning results were used as observation data to correct the prediction results.

[0061] ① Measurement prediction: Based on the current predicted state (velocity and position), calculate the data that the sensor may observe (predicted observations) and update the uncertainty of the predicted state.

[0062] ② Measurement residuals: Calculate the difference between the actual observed values ​​(AOA results) and the predicted observed values.

[0063] Kalman gain calculation: Based on the uncertainty of the predicted state and the measurement uncertainty updated in ① above (which can be controlled by preset parameters), the weights of the two (nine-axis positioning results and AOA positioning results) in the final fusion result are dynamically adjusted.

[0064] ③ State Update: Combine the observed data with the prediction results (the "difference" in ② above) to correct the robot's current position and velocity, and obtain the first current positioning position. To illustrate this step, consider the following example: Assume the current state is X, the AOA result is M, the AOA weights calculated from the Kalman gain are K, the residual is y, y=MX, and Xnew = X +Ky.

[0065] ④ Covariance Update: Update the state uncertainty matrix to reflect the new state confidence level.

[0066] (4) Output

[0067] Output: The fused current position and velocity of the robot are used as the final navigation result.

[0068] Further applications: This result will be used for path planning and navigation control, while also providing initial values ​​for state prediction at the next time step.

[0069] 5. Dual-system positioning

[0070] (1) Robot navigation and positioning: a. LiDAR data processing: The point cloud data from the LiDAR is processed independently by the central processing unit for updating the point cloud map and for navigation.

[0071] b. Real-time Navigation and Map Updates: During robot movement, the LiDAR continuously scans the environment. Newly acquired point cloud data is compared and fused with existing map data to update the map in real time and optimize the navigation path. Comparison and fusion involves matching the point cloud information obtained from the current LiDAR scan, such as distance and angle, with the point cloud information in the already constructed map to determine the robot's current location.

[0072] c. Position estimation and path planning: Based on the current point cloud map, the robot calculates the optimal path to avoid obstacles and dynamically adjusts its route according to changes in the environment.

[0073] (2) AOA and nine-axis sensor data processing: AOA positioning and nine-axis sensor data are processed synchronously and integrated with the store map to provide continuous location updates and corrections, independent of point cloud data. Specifically: a. Signal Reception and Processing: In step 102 above, the electronic shelf label unit sends a measurement signal to the base station; the base station receives the signal from the robot and measures the angle of arrival (AOA). This data is sent to the AOA server, which calculates the position using the AOA, i.e., the AOA positioning result (AOA positioning position). In step 102 above, the autonomous mobile device (robot) is equipped with an electronic shelf label (ESL) and an ESL directional antenna. The device sends a request signal for a specific angle to the base station of the electronic shelf label system through the antenna. The base station responds to these request signals (determining the AOA based on the measurement signal), and the autonomous mobile device receives these response signals (AOA positioning position).

[0074] b. Position Calculation: The robot's position is calculated using a nine-axis sensor, and the AOA (Optical Object Analysis) positioning result is obtained simultaneously. The fusion positioning system combines these two positions to obtain the robot's final position. Nine-axis sensor positioning has cumulative errors, which accumulate over time. AOA positioning does not have cumulative errors, but it can only be used when the robot is within the base station's coverage area. Therefore, outside the base station's coverage area, the nine-axis sensor is used for positioning, and once the AOA positioning result is available, it is used to correct the current positioning result.

[0075] As can be seen from the above, in one embodiment, the robot's first current location is obtained by fusing the location based on the angle of arrival and the nine-axis data. This may include: outside the base station coverage area, using the location obtained from the nine-axis data for positioning; and within the base station coverage area, using the angle of arrival obtained from the measurement signal emitted by the electronic price tag unit to correct the nine-axis location.

[0076] In practice, the above-mentioned method of fusing positioning based on AOA positioning position and positioning position obtained from nine-axis data improves the accuracy of robot positioning.

[0077] A nine-axis sensor can calculate the current attitude using a gyroscope and integrate using an accelerometer to estimate velocity and position. Since the nine-axis sensor's position calculation has cumulative errors, these can be corrected using map information and AOA (Area of ​​Effect) positioning. For example, if the nine-axis sensor is located in an inaccessible area (shelf area), its position needs to be corrected to the nearest accessible area based on map information. Furthermore, if an AOA positioning result is available and the signal strength is above a threshold, the AOA positioning accuracy is considered high, and the AOA positioning result can be directly used to correct the nine-axis sensor's positioning result.

[0078] As can be seen from the above, in one embodiment, the robot's current location is obtained by fusing the AOA positioning position and the nine-axis positioning position. This includes: synchronously processing the AOA positioning position and the nine-axis positioning position and fusing them with the store map to obtain the robot's current location.

[0079] In practice, AOA positioning and data from the nine-axis sensor are processed synchronously and integrated with the store map, such as... Figure 2 The "store map information" shown provides continuous location updates and corrections, independent of point cloud data, further improving the accuracy of robot positioning.

[0080] As can be seen from the above, in one embodiment, the above-mentioned dynamic environment adaptive navigation method may further include: when the nine-axis positioning position is located in an inaccessible shelf area, correcting the nine-axis positioning position according to map information, and correcting the nine-axis positioning position to the nearest accessible area.

[0081] In practice, the above-described implementation method, which corrects the nine-axis positioning position to the nearest reachable area, improves the accuracy of robot positioning.

[0082] (3) In the embodiments of the inventors, the AOA positioning and the nine-axis sensor can be replaced by a vision positioning system, or the positioning accuracy can be improved by combining the positioning results of the vision positioning system.

[0083] The following section provides a detailed introduction to the visual positioning scheme.

[0084] a. Location Selection and Deployment: Deploy multiple high-precision cameras in key areas of dynamic environments such as shopping malls and supermarkets to ensure accurate visual information is obtained throughout the entire area. Select appropriate camera angles and adjust camera resolution to optimize image clarity and recognition performance. Consider the camera layout to ensure their field of view fully covers all areas requiring positioning. The installation height and position of the cameras need to be determined based on the actual environment to maximize visual coverage efficiency. All cameras require precise focus adjustment and regular calibration to ensure image quality and positioning accuracy.

[0085] b. Visual positioning: Signal reception and processing: Deployed cameras continuously capture images of the environment and identify the robot and other distinctive objects using image recognition technology. This image data is sent to the central processing unit in real time.

[0086] Position Calculation: The central processing unit uses image recognition results to determine the robot's precise position through machine vision algorithms. It synthesizes the robot's position information from multiple perspectives using techniques such as feature matching and geometric transformation.

[0087] c. Location anomaly handling and relocation: Anomaly Detection and Response: During robot task execution, the system continuously monitors and analyzes data from the LiDAR and visual positioning modules. If the robot's actual position deviates significantly from the expected trajectory, the system will automatically trigger the anomaly handling mechanism.

[0088] Determine the current position: The system uses visual positioning to accurately estimate the robot's current position.

[0089] Guiding the robot back to a safe location: If the system determines that the robot is lost, it will retrieve the last confirmed safe location from its historical positioning data. Using this location data as a reference, the system calculates the optimal path back to that point.

[0090] Regarding a scheme to further combine visual positioning results to improve positioning accuracy, in one embodiment, the above-mentioned dynamic environment adaptive navigation method may further include: The image acquisition device collects current visual image data of the surrounding environment; When the central processing unit detects a change in the surrounding environment, it matches the current visual image data with a pre-stored supermarket image dataset to obtain target image information on the shelves; it then performs image recognition on the target image information and performs robot-assisted positioning based on the preset location information of the identified goods in the supermarket.

[0091] In specific implementation, this embodiment of the invention uses image-assisted localization: an image acquisition device collects environmental images; when a localization anomaly occurs, the device matches the collected images with a pre-stored image dataset. Through image matching, target image information on the shelf is obtained, thereby assisting in localization. This embodiment of the invention also combines current visual image data of the surrounding environment collected by the image acquisition device for robot-assisted localization, further improving localization accuracy.

[0092] 6. Navigation and Environmental Change Monitoring

[0093] (1) Dual-system independent monitoring: Each system independently monitors environmental changes and the robot's position status, improving system redundancy and reliability. Specifically: a. Continuous monitoring and response: During the robot's task execution, the system continuously monitors environmental changes and the robot's real-time position status.

[0094] b. If significant environmental changes are detected (e.g., shelf movement or the appearance of new obstacles), the system will automatically mark the abnormal areas on the map that need to be updated or reassessed, i.e. Figure 2 The navigation and monitoring system indicates that if environmental changes are detected, the current area needs to be updated.

[0095] In practice, the abnormal area is assessed, including the following three scenarios: Scenario 1: Both map and location are abnormal and require immediate update. When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold, or the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold, and the navigation path fails, the abnormal area is determined to be in a state where navigation cannot be achieved normally, and it needs to be updated immediately.

[0096] As can be seen from the above, in one embodiment, evaluating the abnormal region and determining the state of the abnormal region may include: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold (e.g., 70%), or the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold (e.g., 15 cm), and the navigation path fails, the abnormal area is identified as a state where navigation cannot be achieved normally and needs to be updated immediately.

[0097] In practice, the above-mentioned implementation method, which identifies abnormal areas as states where navigation cannot be achieved normally and requires immediate updates, further improves navigation accuracy.

[0098] In step 104 above, the central processing unit compares the current point cloud data with the pre-generated supermarket point cloud map data. When it is determined that the surrounding environment has changed based on the comparison results, it marks the abnormal area on the map, evaluates the abnormal area, and determines the state of the abnormal area. When it is determined based on the evaluation results that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, it performs fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position. It calls the last confirmed safe position in the historical positioning data and determines the first optimal path from the first current positioning position back to the safe position.

[0099] In practice, when the abnormal area is determined to be in a state where navigation cannot be achieved normally and needs to be updated immediately based on the evaluation results, the robot's first current positioning position is obtained by fusing positioning based on the angle of arrival and nine-axis data. The last confirmed safe position in the historical positioning data is then called to determine the first optimal path from the first current positioning position back to the safe position. This solves the problem of the robot getting lost and enables adaptive and accurate navigation in dynamic environments, improving navigation accuracy and efficiency.

[0100] Scenario 2: Map and location are normal, no update required.

[0101] When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is higher than the second overlap rate threshold (e.g., 90%), and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold (e.g., 5 cm), it is considered that the changes in some areas have not exceeded the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold. The abnormal area is then determined to be in a state where normal navigation can be achieved without updating.

[0102] As can be seen from the above, in one embodiment, evaluating the abnormal region and determining the state of the abnormal region includes: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than the second overlap rate threshold, and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold, and the changes in some areas do not exceed the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold, the abnormal area is determined to be in a state where normal navigation can be achieved without updating. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0103] In practice, the above-mentioned implementation method, which identifies abnormal areas as being in a state where navigation can be achieved normally without updates, further improves navigation efficiency.

[0104] As can be seen from the above, in one embodiment, the dynamic environment adaptive navigation method may further include: When the central processing unit determines, based on the evaluation results, that the abnormal area is in a state where normal navigation can be achieved without updating, it sends a command to the navigation control unit to continue normal navigation. When the navigation control unit receives an instruction to continue normal navigation, it continues to control the robot to complete the task.

[0105] In practice, when the abnormal area is determined to be in a state where normal navigation can be achieved without updating based on the evaluation results, a command to continue normal navigation is sent to the navigation control unit. Upon receiving the command to continue normal navigation, the navigation control unit continues to control the robot to complete the task, further improving the efficiency of navigation.

[0106] Scenario 3: There is some discrepancy between the map and location, but no immediate update is required.

[0107] One of the following conditions must be met: Sub-case 1: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is between the first overlap rate threshold (e.g., 70%) and the second overlap rate threshold (e.g., 90%) (including equal to the first overlap rate threshold and the second overlap rate threshold), and the deviation between the robot's current positioning position and the expected position is between the navigation tolerance threshold (e.g., 15 cm) and the allowable boundary threshold (e.g., 5 cm) (including equal to the allowable boundary threshold and the navigation tolerance threshold).

[0108] Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is between the navigation tolerance threshold and the allowable boundary threshold (including equal to the allowable boundary threshold and the navigation tolerance threshold).

[0109] Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is between the first overlap rate threshold and the second overlap rate threshold (including equal to the first overlap rate threshold and the second overlap rate threshold), and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold.

[0110] When it is determined that changes in some areas exceed the change threshold but the overall navigation effect is not affected, the abnormal area is identified as a state that needs to be updated but does not need to be updated immediately.

[0111] Wherein: the first overlap rate threshold is less than the second overlap rate threshold; the allowable boundary threshold is less than the navigation tolerance threshold.

[0112] As can be seen from the above, in one embodiment, evaluating the abnormal region and determining the state of the abnormal region includes: If, under any of the following sub-conditions, a region changes beyond a change threshold but does not affect the overall navigation performance, the abnormal region is identified as needing to be updated but not immediately: Sub-case 1: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is less than the first overlap rate threshold, and the deviation between the robot's current location and the expected location is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0113] In practice, the above-mentioned implementation method of identifying abnormal areas as those that need updating but do not need to be updated immediately further improves navigation efficiency.

[0114] In one embodiment, the above-described dynamic environment adaptive navigation method may further include: When the central processing unit determines that an abnormal area needs to be updated but not immediately based on the evaluation results, it acquires the current point cloud data generated by the LiDAR scanning of the surrounding environment; determines the robot's second current location based on the current point cloud data; and generates a second optimal path based on the second current location. The navigation control unit controls the robot to continue executing the task from the second current location according to the second optimal path; after completing the task, the robot is controlled to return to the abnormal area; After completing its mission, the lidar captures environmental changes that have occurred since the last scan in the abnormal area and reconstructs a point cloud map based on these environmental changes.

[0115] In practice, when the abnormal area is determined to be in a state that needs to be updated but does not need to be updated immediately based on the evaluation results, the robot's second current positioning position is determined based on the current point cloud data; a second optimal path to avoid the abnormal area is generated based on the second current positioning position, which further improves the efficiency and accuracy of navigation.

[0116] In summary, if the map has two thresholds, the location also has two thresholds, as shown in Table 1 below: Table 1

[0117] There are a total of nine combinations: M1 N1 – Case 1; M1 N2 – Case 1; M1 N3 – Case 1; M2 N1 – Case 3; M2 N2 – Case 3; M2 N3 – Case 1; M3 N1 – Case 2; M3 N2 – Case 3; M3 N3 – Case 1.

[0118] 7. Location anomaly handling and relocation

[0119] Anomaly Handling: If the LiDAR system fails to navigate due to point cloud mismatch, the AOA plus sensor system can independently provide precise location or guide the robot back to a known location. Assisted Localization: Once the mobile device enters the calibration range of a base station based on a predicted path, or obtains accurate location information through image matching, it uses the base station's location information for assisted localization. The device then uses this information to correct its own positioning system and resume normal navigation. Specifically: Anomaly Detection and Response (1) Identify location anomalies: During the robot's task execution, the system continuously monitors and analyzes data from LiDAR, AOA positioning module, and nine-axis sensors. If the robot's actual position deviates significantly from the expected trajectory (e.g., the robot's position is more than 10 meters away from the expected route (configurable), or the robot's position is located in an inaccessible area (where the shelf is located)), the system will automatically trigger an exception handling mechanism.

[0120] Anomalies caused by changes in the surrounding environment can be due to various factors, such as changes in the environmental layout (e.g., moving shelves or the appearance of new obstacles), leading to significant deviations in LiDAR positioning. Anomaly identification methods: By comparing the LiDAR positioning results with map data, if the positioning result indicates that the location is in an inaccessible area (such as a shelf area), then the current positioning is considered abnormal. Judgment is made by matching degree: by comparing the matching degree of real-time point cloud data and point cloud map data, if the matching degree is less than the threshold, the current positioning is considered abnormal.

[0121] Anomaly Marking: Compare the current fused positioning result with the previous valid LiDAR positioning result and calculate the travel distance. Draw a circular area with the previous valid LiDAR positioning result as the center and the travel distance as the radius, and mark it as an area where anomalies have occurred.

[0122] (2) Determine the current location: Since the lidar positioning has become abnormal at this time, the system needs to stop using the lidar positioning results and switch to using AOA and nine-axis data and combining them with store information for fusion positioning, so as to accurately estimate the robot's current location.

[0123] (3) Guide the return to a safe location: a. If the robot is determined to be lost, the system will retrieve the last confirmed safe location from its historical positioning data. Using this safe location as the endpoint, and the current fused positioning results as the starting point, the optimal path back to that safe location will be calculated.

[0124] b. The robot will move along the path guided by the system. Simultaneously, the AOA and nine-axis fusion positioning systems continuously update the positioning, while the LiDAR positioning system continuously performs positioning to ensure the robot avoids obstacles. Once the LiDAR locates a reasonable position (a reachable position with a distance from the previous safe position less than a certain threshold), it can be determined that the robot has returned to the known position, thus ensuring the robot can safely and effectively return to the known position. Figure 2 The example shown is "Anomaly handling and repositioning: using AOA and nine-axis fusion positioning to assist the robot in returning to a known position".

[0125] In one embodiment, the navigation control unit controls the robot to return to the safe location according to the first optimal path, including: During the process of the navigation control unit controlling the robot to return to the safe position according to the first optimal path, the distance between the robot and the safe position is continuously monitored; Based on the distance between the robot and the safe location, control the robot to approach the safe location along a first optimal path using a preset process.

[0126] In practice, the system continuously monitors the distance between the robot and the target's safe location, ensuring that the robot gradually approaches the target according to the planned path, thus guaranteeing the accuracy of navigation.

[0127] In one embodiment, the above-described dynamic environment adaptive navigation method may further include: When the central processing unit detects that the robot has reached a safe position, it uses the angle of arrival and nine-axis positioning to correct the safe position and obtain the corrected safe position; it verifies whether the current environmental data is consistent with the preset data of the safe position. If the current environmental data is inconsistent with the preset data of the safe position, it initiates a map reconstruction command to the LiDAR. When the lidar receives a map reconstruction command, it rescans the area within the preset range of the safe location and reconstructs the point cloud map according to the environmental changes.

[0128] In practice, after reaching a safe position, the current position is corrected using AOA and nine-axis sensor data. The system verifies that the current environment data matches the preset data for the safe position. If the current environment data does not match the preset data for the safe position, a map reconstruction command is sent to the LiDAR, initiating subsequent steps.

[0129] The above "4. AOA plus sensor fusion positioning and navigation system" to "7. Positioning anomaly handling and repositioning" are steps 102 to 105.

[0130] 8. Rebuild and update the point cloud map, i.e., step 106 above.

[0131] Navigation based on the AOA plus sensor system: When the LiDAR system needs re-matching or calibration, the position data from the AOA plus sensor system is used to help determine the robot's correct position and guide it in performing necessary map reconstruction. This involves the LiDAR rescanning the abnormal area, capturing environmental changes that have occurred since the last scan, and reconstructing a point cloud map based on these changes. Figure 2 The text indicates a "re-mapping" process. Specifically: (1) Restart lidar scanning: a. Once the robot returns to a known safe location or completes its navigation route, the status of certain areas will be marked as needing updating. The robot will then begin rescanning these areas.

[0132] b. Rescanning includes not only a thorough check of the directly returned area, but also a reassessment and verification of surrounding areas that may affect navigation.

[0133] c. If the robot successfully completes navigation, it needs to return to the areas marked as needing updating and rescan.

[0134] d. If the robot encounters a situation where drastic environmental changes prevent it from navigating normally, once it returns to a known safe location, it needs to rescan and update the point cloud map from that safe location. After the update is complete, it should attempt to restart navigation.

[0135] As can be seen from the above, in one embodiment, the dynamic environment adaptive navigation method may further include: The central processing unit reassesses and verifies the areas surrounding the abnormal region that affect navigation, and reconstructs the point cloud map based on the reassessment and verification results.

[0136] In specific implementation, in this embodiment of the invention, the central processing unit will also re-evaluate and verify the area around the abnormal area that affects navigation, and reconstruct the point cloud map based on the re-evaluation and verification results, thereby further improving the accuracy of robot positioning.

[0137] (2) Update the point cloud map: a. The updated point cloud data is transmitted back to the central processing unit, where the system compares and integrates this new data with the existing point cloud map.

[0138] b. During this process, the system utilizes advanced map matching technology to ensure seamless integration of old and new data, while identifying and correcting any inaccuracies in the known data.

[0139] (3) Ensure the accuracy and reliability of the map: a. The updated point cloud map is immediately applied to the robot's navigation system. This ensures that the robot continues to perform its tasks using the most accurate and up-to-date map information.

[0140] b. In this way, the system not only corrects navigation errors caused by environmental changes, but also enhances the robot's adaptability and accuracy in the face of future environmental changes.

[0141] 9. Path optimization and continuous navigation, i.e., step 107 above.

[0142] Dual-system path optimization: After the robot returns to its new position and completes map updates, the system continues to optimize its route. Algorithms continuously adjust the path to ensure navigation efficiency and safety while reducing sensitivity to environmental changes. The two systems operate independently but assist each other when necessary to ensure navigation continuity and accuracy. Specifically, the navigation control unit regenerates the navigation path based on the reconstructed point cloud map; based on the regenerated path, the robot is controlled to continue its task from that safe location.

[0143] Recording and Adjustment: During normal positioning, the device records the strength of the response signal and base station location information in real time. This information is used to update and optimize the predicted path and positioning algorithm to adapt to different environments and conditions. In this way, autonomous mobile devices (robots) can self-correct and navigate using existing communication infrastructure and image data when the positioning system encounters interference or failure, improving the reliability and autonomy of the device in complex environments.

[0144] In one embodiment, the above-described dynamic environment adaptive navigation method may further include: During normal positioning, the electronic price tag unit records the strength value of the base station response signal and the base station location information in real time; The navigation control unit updates and optimizes the navigation path based on the strength of the base station response signal and the base station location information to adapt to different environments and conditions.

[0145] In specific implementation, the embodiments of the present invention also update and optimize the navigation path based on the strength value of the base station response signal and the base station location information recorded in real time by the ESL unit during normal positioning, so as to adapt to different environments and conditions and improve the accuracy and flexibility of robot positioning.

[0146] In summary, the key features and advantages of the dynamic environment adaptive navigation scheme provided by the embodiments of the present invention are: 1. Dual positioning system The key implementation of this invention involves two independent but complementary positioning systems: a lidar positioning system and a positioning system that fuses AOA (Automatic Aspect-Oriented Navigation) with a nine-axis sensor. These two systems provide multi-level data verification and positioning accuracy, ensuring the robot receives the most stable and reliable navigation support in complex or dynamically changing environments.

[0147] (1) LiDAR positioning provides high-precision spatial point cloud data, which is used to form detailed environmental maps and real-time obstacle recognition.

[0148] (2) The fusion positioning of AOA and nine-axis sensors provides precise position information about the robot in the environment through angle measurement and motion sensing data, especially when the lidar cannot provide accurate data due to line of sight obstruction or external interference.

[0149] In addition, the "AOA and nine-axis sensor fusion positioning" system in this embodiment of the invention can also be achieved through visual positioning.

[0150] 2. AOA and nine-axis fusion positioning assist lost robots in returning them to a safe and known location.

[0151] When a robot gets lost or encounters navigational anomalies while performing a task, a system that integrates AOA positioning and a nine-axis sensor can provide crucial positioning assistance. This fusion positioning technology can extract position information from two independent data sources, accurately locate the robot's current position through algorithmic optimization, and guide it back to the last known safe location.

[0152] During the assisted return process, the system ensures that the robot takes the optimal path through real-time data monitoring and analysis, and can continuously adjust to cope with possible new obstacles or environmental changes.

[0153] 3. Assist in reconstructing the map

[0154] After the robot returns to a known safe location and calibrates its position information, the LiDAR restarts scanning to update and rebuild the point cloud map. This process is crucial because it ensures that the robot's navigation map is always up-to-date, thereby improving the accuracy and reliability of navigation.

[0155] (1) Map updates include not only rescanning known areas, but also evaluating and integrating newly changed areas to ensure that any environmental changes can be identified by the system and incorporated into the navigation system.

[0156] (2) The accuracy and reliability of the map directly affect the robot’s performance in future tasks. By continuously optimizing and updating the map, the robot can better adapt to environmental changes and improve task efficiency.

[0157] By protecting and implementing these three key aspects, the embodiments of the present invention not only improve the robot's autonomy and flexibility in complex environments, but also greatly enhance the robot system's adaptability to environmental changes and overall operational safety. The integration and optimization of these technologies enable the robot to exhibit superior navigation and positioning performance in various environments.

[0158] In summary, the embodiments of this invention, combining Angular Arrival (AOA) positioning technology and a nine-axis sensor, can provide a solution. AOA positioning determines position by measuring the angle of arrival of signals, while the nine-axis sensor provides real-time data on the robot's motion state. The combination of these two technologies not only improves positional accuracy but also assists the robot in repositioning itself when it becomes lost, helping it return to its starting point. This allows the robot to restart the LiDAR mapping process in changing environments, effectively solving the problem of remapping caused by environmental changes.

[0159] This invention also provides a method for dynamic environment adaptive navigation applied to a central processing unit, as described in the following embodiments. Since the principle behind this method is similar to that of the method for dynamic environment adaptive navigation applied to robots, the implementation of this method can be found in the implementation of the method for dynamic environment adaptive navigation applied to robots; repeated details will not be elaborated further.

[0160] Figure 3 This is a flowchart illustrating the dynamic environment adaptive navigation method applied to the central processing unit in an embodiment of the present invention, as shown below. Figure 3 The method includes the following steps: Step 501: Compare the current point cloud data with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the LiDAR, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment; Step 502: When it is determined from the comparison results that the surrounding environment has changed, mark the abnormal areas on the map; Step 503: Evaluate the abnormal region and determine its status; Step 504: When the evaluation results determine that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, the robot's first current positioning position is obtained by fusing positioning based on the angle of arrival and nine-axis data; the angle of arrival is sent by the angle of arrival server, which is used to determine the robot's angle of arrival positioning position based on the angle of arrival, which is sent by the base station, which is used to determine the angle of arrival based on the measurement signal sent by the electronic price tag unit; the nine-axis data is sent by the nine-axis sensor, which is used to measure the robot's nine-axis data; Step 505: Retrieve the last confirmed safe location from the historical positioning data, determine the first optimal path from the first current positioning location back to the safe location, and send it to the navigation control unit; the navigation control unit is used to control the robot to return to the safe location according to the first optimal path; the lidar is also used to capture environmental changes that have occurred since the last scan from the abnormal area, and reconstruct a point cloud map according to the environmental changes; the step of regenerating the navigation path according to the reconstructed point cloud map; controlling the robot to continue to complete the task from the safe location according to the regenerated navigation path is also used to: regenerate the navigation path according to the reconstructed point cloud map; and control the robot to continue to complete the task from the safe location according to the regenerated navigation path.

[0161] In one embodiment, the robot's first current location is obtained by fusing positioning based on the angle of arrival and nine-axis data, including: Obtain the robot's current position and velocity, as well as the covariance matrix of that position and velocity; Using acceleration and angular velocity data from a nine-axis sensor, along with the robot's motion equations, the robot's position and velocity at the next moment are predicted to obtain the current predicted state. Based on the prediction error and noise model, the covariance matrix is ​​updated to obtain the uncertainty of the current predicted state. Based on the current prediction state, calculate the predicted observation value of the nine-axis sensor; take the angle of arrival positioning position as the actual observation value, and calculate the difference between the actual observation value and the predicted observation value; dynamically adjust the weights of the angle of arrival positioning position and the nine-axis positioning position in the final fusion result according to the uncertainty of the current prediction state and the uncertainty of the actual measurement state. Based on the differences and the weights, the robot's current position and speed are corrected to obtain the first current positioning position.

[0162] In one embodiment, evaluating the anomalous region and determining its state includes: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold, or when the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold, and the navigation path fails, the abnormal area is identified as a state where navigation cannot be achieved normally and needs to be updated immediately.

[0163] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: When an abnormal area is determined to be in a state that requires updating but not immediate updating based on the evaluation results, the current point cloud data generated by the LiDAR scanning of the surrounding environment is acquired; based on the current point cloud data, the robot's second current location is determined; a second optimal path is generated based on the second current location and sent to the navigation control unit; the navigation control unit is also used to control the robot to continue executing the task from the second current location based on the second optimal path; after completing the task, the robot is controlled to return to the abnormal area; the LiDAR is also used to capture environmental changes that have occurred in the abnormal area since the last scan after completing the task, and reconstruct the point cloud map based on the environmental changes.

[0164] In one embodiment, evaluating the anomalous region and determining its state includes: If, under any of the following sub-conditions, a region changes beyond a change threshold but does not affect the overall navigation performance, the abnormal region is identified as needing to be updated but not immediately: Sub-case 1: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is less than the first overlap rate threshold, and the deviation between the robot's current location and the expected location is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0165] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: When the evaluation results determine that the abnormal area is in a state where normal navigation can be achieved without updating, a command to continue normal navigation is sent to the navigation control unit; the navigation control unit is also used to continue controlling the robot to complete the task when it receives the command to continue normal navigation.

[0166] In one embodiment, evaluating the anomalous region and determining its state includes: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than the second overlap rate threshold, and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold, and the changes in some areas do not exceed the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold, the abnormal area is determined to be in a state where normal navigation can be achieved without updating. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0167] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: When the robot is detected to have reached a safe position, the position is corrected using the angle of arrival and the nine-axis positioning to obtain the corrected safe position. The system then verifies whether the current environmental data is consistent with the preset data of the safe position. If the current environmental data is inconsistent with the preset data of the safe position, a map reconstruction command is sent to the LiDAR. The LiDAR is also used to rescan the area within the preset range of the safe position when it receives the map reconstruction command, and reconstruct the point cloud map according to the environmental changes.

[0168] In one embodiment, the navigation control unit controls the robot to return to the safe location according to the first optimal path, including: During the process of the navigation control unit controlling the robot to return to the safe position according to the first optimal path, the distance between the robot and the safe position is continuously monitored; Based on the distance between the robot and the safe location, control the robot to approach the safe location along a first optimal path using a preset process.

[0169] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: When a change in the surrounding environment is detected, the current visual image data is matched with a pre-stored supermarket image dataset to obtain target image information on the shelf; image recognition is performed on the target image information, and robot-assisted positioning is performed based on the preset location information of the identified goods in the supermarket; the current visual image data is obtained by an image acquisition device that collects the surrounding environment.

[0170] In one embodiment, the robot's first current location is obtained by fusing the angle of arrival positioning position and nine-axis data, including: outside the base station coverage area, positioning is performed using the positioning position obtained from the nine-axis data; within the base station coverage area, the nine-axis positioning position is corrected using the angle of arrival positioning position obtained from the measurement signal emitted by the electronic price tag unit.

[0171] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: when the nine-axis positioning position is located in an inaccessible shelf area, correcting the nine-axis positioning position according to pre-established map information to correct the nine-axis positioning position to the nearest accessible area.

[0172] In one embodiment, the dynamic environment adaptive navigation method applied to the central processing unit may further include: The areas surrounding the abnormal areas that affect navigation are reassessed and verified, and the point cloud map is reconstructed based on the reassessment and verification results.

[0173] This invention also provides a dynamic environment adaptive navigation robot, as described in the following embodiments. Since the principle by which this robot solves the problem is similar to the dynamic environment adaptive navigation method applied to robots, the implementation of this robot can refer to the implementation of the dynamic environment adaptive navigation method applied to robots; repeated details will not be elaborated further.

[0174] Figure 4 This is a schematic diagram of the structure of the dynamic environment adaptive navigation robot in an embodiment of the present invention, as shown below. Figure 4 As shown, the robot includes: LiDAR 01 is used to scan the surrounding environment, generate current point cloud data of the surrounding environment and send it to the central processing unit; it captures environmental changes that have occurred since the last scan from abnormal areas and reconstructs the point cloud map based on the environmental changes. Electronic price tag unit 03 is used to send measurement signals to base station 06; the base station is used to determine the angle of arrival based on the measurement signals and send the angle of arrival to angle of arrival server 07; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. Nine-axis sensor 04 is used to measure nine-axis data of the robot and send it to the central processing unit. The central processing unit 05 is used to compare the current point cloud data with the pre-generated supermarket point cloud map data. When it is determined that the surrounding environment has changed based on the comparison results, it marks the abnormal area on the map, evaluates the abnormal area, and determines the state of the abnormal area. When it is determined that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately based on the evaluation results, it performs fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position. It calls the last confirmed safe position in the historical positioning data and determines the first optimal path from the first current positioning position back to the safe position. The navigation control unit 02 is used to control the robot to return to the safe position according to the first optimal path; to regenerate the navigation path according to the reconstructed point cloud map; and to control the robot to continue to perform the task from the safe position according to the regenerated navigation path.

[0175] In one embodiment, fusing positioning based on the angle-of-arrival location and nine-axis data to obtain the robot's first current location may include: Obtain the robot's current position and velocity, as well as the covariance matrix of that position and velocity; Using acceleration and angular velocity data from a nine-axis sensor, along with the robot's motion equations, the robot's position and velocity at the next moment are predicted to obtain the current predicted state. Based on the prediction error and noise model, the covariance matrix is ​​updated to obtain the uncertainty of the current predicted state. Based on the current prediction state, calculate the predicted observation value of the nine-axis sensor; take the angle of arrival positioning position as the actual observation value, and calculate the difference between the actual observation value and the predicted observation value; dynamically adjust the weights of the angle of arrival positioning position and the nine-axis positioning position in the final fusion result according to the uncertainty of the current prediction state and the uncertainty of the actual measurement state. Based on the differences and the weights, the robot's current position and speed are corrected to obtain the first current positioning position.

[0176] In one embodiment, evaluating the anomalous region and determining its state may include: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold, or when the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold, and the navigation path fails, the abnormal area is identified as a state where navigation cannot be achieved normally and needs to be updated immediately.

[0177] In one embodiment, the central processing unit is further configured to, when determining that an abnormal area is in a state that needs to be updated but does not need to be updated immediately based on the evaluation results, acquire current point cloud data generated by the LiDAR scanning of the surrounding environment; determine the robot's second current location based on the current point cloud data; and generate a second optimal path based on the second current location. The aforementioned navigation control unit is also used to control the robot to continue executing the task from the second current location according to the second optimal path; after completing the task, control the robot to return to the abnormal area; The aforementioned lidar is also used to capture environmental changes that have occurred since the last scan in abnormal areas after the mission is completed, and to reconstruct point cloud maps based on these environmental changes.

[0178] In one embodiment, evaluating the anomalous region and determining its state may include: If, under any of the following sub-conditions, a region changes beyond a change threshold but does not affect the overall navigation performance, the abnormal region is identified as needing to be updated but not immediately: Sub-case 1: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is less than the first overlap rate threshold, and the deviation between the robot's current location and the expected location is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0179] In one embodiment, the central processing unit is further configured to send a command to the navigation control unit to continue normal navigation when the abnormal area is determined to be in a state where normal navigation can be achieved without updating based on the evaluation results. The navigation control unit is also used to continue controlling the robot to complete the task when it receives an instruction to continue normal navigation.

[0180] In one embodiment, evaluating the anomalous region and determining its state may include: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than the second overlap rate threshold, and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold, and the changes in some areas do not exceed the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold, the abnormal area is determined to be in a state where normal navigation can be achieved without updating. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0181] In one embodiment, the central processing unit is further configured to, when detecting that the robot has reached a safe position, use the angle of arrival positioning position and the nine-axis positioning position to correct the safe position to obtain the corrected safe position; verify whether the current environmental data is consistent with the preset data of the safe position; and when verifying that the current environmental data is inconsistent with the preset data of the safe position, initiate a map reconstruction command to the LiDAR. In one embodiment, the lidar is also used to rescan the area within a preset range of the safe location when a map reconstruction instruction is received, and reconstruct the point cloud map according to environmental changes.

[0182] In one embodiment, the navigation control unit controlling the robot to return to the safe location based on the first optimal path may include: During the process of the navigation control unit controlling the robot to return to the safe position according to the first optimal path, the distance between the robot and the safe position is continuously monitored; Based on the distance between the robot and the safe location, control the robot to approach the safe location along a first optimal path using a preset process.

[0183] In one embodiment, the image acquisition device is also used to acquire current visual image data of the surrounding environment; The central processing unit is also used to match the current visual image data with a pre-stored supermarket image dataset when a change in the surrounding environment is detected, to obtain target image information on the shelf; to perform image recognition on the target image information, and to perform robot-assisted positioning based on the preset position information of the identified goods in the supermarket.

[0184] In one embodiment, the electronic price tag unit is also used to record the strength value of the base station response signal and the base station location information in real time during normal positioning. The navigation control unit is also used to update and optimize the navigation path based on the strength value of the base station response signal and the base station location information to adapt to different environments and conditions.

[0185] In one embodiment, the electronic price tag unit is also used to interact with smart electronic tags in the supermarket to obtain the location information of the current product bound to the smart electronic tag, and to perform robot-assisted positioning based on the location information of the current product bound to the smart electronic tag.

[0186] In one embodiment, fusing the angle of arrival (AHA) position and nine-axis data to obtain the robot's first current position may include: outside the base station coverage area, using the position obtained from the nine-axis data for positioning; and within the base station coverage area, using the AHA position obtained from the measurement signal emitted by the electronic price tag unit to correct the nine-axis position.

[0187] In one embodiment, the central processing unit is further configured to correct the nine-axis positioning position to the nearest accessible area based on pre-established map information when the nine-axis positioning position is located in an inaccessible shelf area.

[0188] In one embodiment, the central processing unit is also used to reassess and verify the area around the abnormal area that affects navigation, and reconstruct the point cloud map based on the reassessment and verification results.

[0189] This invention also provides a central processing unit for dynamic environment adaptive navigation, as described in the following embodiments. Since the principle by which this central processing unit solves the problem is similar to that of the dynamic environment adaptive navigation method applied to robots, the implementation of this device can refer to the implementation of the dynamic environment adaptive navigation method applied to robots; repeated details will not be elaborated further.

[0190] Figure 5 This is a schematic diagram of the central processing unit for dynamic environment adaptive navigation in an embodiment of the present invention, as shown below. Figure 5 As shown, the central processing unit includes: The comparison module 051 is used to compare the current point cloud data with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the lidar, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment. The marking module 052 is used to mark abnormal areas on the map when it is determined from the comparison results that the surrounding environment has changed; Evaluation module 053 is used to evaluate the abnormal region and determine the state of the abnormal region; The fusion positioning module 054 is used to perform fusion positioning based on the angle of arrival positioning position and nine-axis data when the evaluation results determine that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, to obtain the robot's first current positioning position; the angle of arrival positioning position is sent by the angle of arrival server, which is used to determine the robot's angle of arrival positioning position based on the angle of arrival, which is sent by the base station, which is used to determine the angle of arrival based on the measurement signal sent by the electronic price tag unit; the nine-axis data is sent by the nine-axis sensor, which is used to measure the robot's nine-axis data; The optimal path determination module 055 is used to call the last confirmed safe location in the historical positioning data, determine the first optimal path from the first current positioning location back to the safe location, and send it to the navigation control unit; the navigation control unit is used to control the robot to return to the safe location according to the first optimal path; the lidar is also used to capture environmental changes that have occurred since the last scan from the abnormal area, and reconstruct a point cloud map according to the environmental changes; the regeneration of the navigation path according to the reconstructed point cloud map; and controlling the robot to continue to complete the task from the safe location according to the regenerated navigation path are further configured to: regenerate the navigation path according to the reconstructed point cloud map; and control the robot to continue to complete the task from the safe location according to the regenerated navigation path.

[0191] In one embodiment, the above-mentioned fusion positioning module is specifically used for: Obtain the robot's current position, velocity, and its uncertainties; Using the aforementioned nine-axis data and robot motion equations, predict the robot's position, velocity, and uncertainties at the next moment; The position, velocity, and uncertainties of the robot at the next moment are corrected using the angle of arrival positioning. The corrected position of the robot at the next moment is taken as the first current positioning position.

[0192] In one embodiment, the evaluation module is further configured to: determine the abnormal area as a state where navigation cannot be achieved normally and needs to be updated immediately when the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than a first overlap rate threshold, or the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold and the navigation path fails.

[0193] In one embodiment, the optimal path determination module is further configured to: when the abnormal area is determined to be in a state that needs to be updated but does not need to be updated immediately based on the evaluation result, acquire the current point cloud data generated by the LiDAR scanning the surrounding environment; determine the robot's second current location based on the current point cloud data; generate a second optimal path based on the second current location and send it to the navigation control unit; the navigation control unit is further configured to control the robot to continue executing the task from the second current location based on the second optimal path; after completing the task, control the robot to return to the abnormal area; the LiDAR is further configured to capture environmental changes that have occurred since the last scan from the abnormal area after completing the task, and reconstruct the point cloud map based on the environmental changes.

[0194] In one embodiment, the evaluation module is further configured to: determine the abnormal region as needing to be updated but not immediately updated when any of the following sub-conditions are met, where changes in some areas exceed a change threshold but do not affect the overall navigation effect: Sub-case 1: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is less than the first overlap rate threshold, and the deviation between the robot's current location and the expected location is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0195] In one embodiment, the central processing unit further includes: The "Continue Normal Navigation" instruction unit is used to send a "Continue Normal Navigation" instruction to the navigation control unit when the abnormal area is determined to be in a state where normal navigation can be achieved without updating, based on the evaluation results. The navigation control unit is also used to continue controlling the robot to complete the task upon receiving the "Continue Normal Navigation" instruction.

[0196] In one embodiment, the evaluation module is further configured to: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than the second overlap rate threshold, and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold, and the changes in some areas do not exceed the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold, the abnormal area is determined to be in a state where normal navigation can be achieved without updating. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

[0197] In one embodiment, the central processing unit may further include: a correction unit, configured to correct the safe position using the angle of arrival and nine-axis positioning when the robot is detected to have reached a safe position, to obtain a corrected safe position; and a verification unit, configured to verify whether the current environmental data is consistent with the preset data of the safe position, and to initiate a map reconstruction command to the lidar when the current environmental data is inconsistent with the preset data of the safe position; the lidar is further configured to rescan the area within the preset range of the safe position when it receives the map reconstruction command, and reconstruct a point cloud map according to the environmental changes.

[0198] In one embodiment, the navigation control unit is specifically used for: During the process of the navigation control unit controlling the robot to return to the safe position according to the first optimal path, the distance between the robot and the safe position is continuously monitored; Based on the distance between the robot and the safe location, control the robot to approach the safe location along a first optimal path using a preset process.

[0199] In one embodiment, the central processing unit further includes a visual-assisted positioning unit, which, when a change in the surrounding environment is detected, matches the current visual image data with a pre-stored supermarket image dataset to obtain target image information on the shelf; performs image recognition on the target image information, and performs robot-assisted positioning based on the preset location information of the identified goods in the supermarket; wherein the current visual image data is obtained by an image acquisition device collecting the surrounding environment.

[0200] In one embodiment, the electronic price tag unit is further configured to record the strength value of the base station response signal and the base station location information in real time during normal positioning; the navigation control unit is further configured to update and optimize the navigation path based on the strength value of the base station response signal and the base station location information to adapt to different environments and conditions.

[0201] In one embodiment, the electronic price tag unit is also used to interact with smart electronic tags in the supermarket to obtain the location information of the current product bound to the smart electronic tag, and to perform robot-assisted positioning based on the location information of the current product bound to the smart electronic tag.

[0202] In one embodiment, the fusion positioning module is specifically used to: perform positioning using the positioning position obtained from the nine-axis data outside the base station coverage area; and correct the nine-axis positioning position using the angle of arrival positioning position obtained from the measurement signal emitted by the electronic price tag unit within the base station coverage area.

[0203] In one embodiment, the central processing unit further includes a correction unit, configured to correct the nine-axis positioning position to the nearest accessible area based on pre-established map information when the nine-axis positioning position is located in an inaccessible shelf area.

[0204] In one embodiment, the central processing unit further includes a re-evaluation and verification unit, used to re-evaluate and verify the area around the abnormal area that affects navigation, and reconstruct the point cloud map based on the re-evaluation and verification results.

[0205] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for dynamic environment adaptive navigation.

[0206] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for dynamic environment adaptive navigation.

[0207] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for dynamic environment adaptive navigation.

[0208] In this embodiment of the invention, the dynamic environment adaptive navigation scheme involves: a lidar scanning the surrounding environment to generate current point cloud data; a navigation control unit generating a navigation path based on the current point cloud data, and controlling the robot to perform tasks according to the navigation path; an ESL unit sending measurement signals to a base station; the base station determining the angle of arrival based on the measurement signals and sending the angle of arrival to an AOA server; the AOA server determining the robot's AOA positioning position based on the angle of arrival and returning it to the central processing unit; a nine-axis sensor measuring the robot's nine-axis data and sending it to the central processing unit; the central processing unit marking abnormal areas on the map when it detects changes in the surrounding environment; comparing the current point cloud data with pre-generated supermarket point cloud map data; evaluating the abnormal areas based on the comparison results and determining the state of the abnormal areas; and determining the abnormal areas as [missing information - likely related to navigation]. When navigation cannot be achieved normally and an immediate update is required, the robot's nine-axis positioning position is determined based on the nine-axis data. The AOA positioning position and the nine-axis positioning position are fused to obtain the robot's first current positioning position. The last confirmed safe position in the historical positioning data is retrieved to determine the first optimal path from the first current positioning position back to that safe position. The navigation control unit controls the robot to return to the safe position according to the first optimal path. The LiDAR rescans the abnormal area, capturing environmental changes that have occurred since the last scan, and reconstructs a point cloud map based on these changes. The navigation control unit regenerates the navigation path based on the reconstructed point cloud map. Based on the regenerated navigation path, the robot is controlled to continue performing the task from the safe position, achieving adaptive navigation in a dynamic environment and improving the robot's navigation efficiency and accuracy.

[0209] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0210] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0211] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0212] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0213] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive navigation in a dynamic environment, characterized in that, This method is applied to robots, and the method includes: The lidar scans the surrounding environment and generates current point cloud data of the surrounding environment, which is then sent to the central processing unit. The electronic price tag unit sends a measurement signal to the base station; the base station is used to determine the angle of arrival based on the measurement signal, and sends the angle of arrival to the angle of arrival server; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. The nine-axis sensor measures the robot's nine-axis data and sends it to the central processing unit. The central processing unit compares the current point cloud data with pre-generated supermarket point cloud map data. When it determines that the surrounding environment has changed based on the comparison results, it marks abnormal areas on the map, evaluates the abnormal areas, and determines their state. When the evaluation results determine that the abnormal area is in a state where navigation cannot be achieved normally and needs to be updated immediately, it performs fusion positioning based on the angle of arrival positioning position and nine-axis data to obtain the robot's first current positioning position. This includes: acquiring the robot's current position and velocity, and the covariance matrix of the position and velocity; using the acceleration and angular velocity data of the nine-axis sensor, and the robot's motion equations, predicting the robot's position and velocity at the next moment to obtain the current predicted state; updating the covariance matrix based on the prediction error and noise model to obtain the uncertainty of the current predicted state; and based on the current prediction... The system measures the state and calculates the predicted observations from the nine-axis sensor; it uses the angle of arrival (AHA) position as the actual observation and calculates the difference between the actual observation and the predicted observation; based on the uncertainty of the current predicted state and the uncertainty of the actual measured state, it dynamically adjusts the weights of the AHA position and the nine-axis position in the final fusion result; based on the difference and the weights, it corrects the robot's current position and velocity to obtain the first current positioning position; or, outside the base station coverage area, it uses the positioning position obtained from the nine-axis data for positioning; within the base station coverage area, it uses the AHA position obtained from the measurement signal emitted by the electronic price tag unit to correct the nine-axis positioning position; it calls the last confirmed safe position in the historical positioning data and determines the first optimal path from the first current positioning position back to that safe position; The navigation control unit controls the robot to return to the safe position according to the first optimal path; LiDAR captures environmental changes that have occurred since the last scan in anomaly areas and reconstructs point cloud maps based on these changes. The navigation control unit regenerates the navigation path based on the reconstructed point cloud map; based on the regenerated navigation path, the robot is controlled to continue performing the task from that safe position.

2. The method as described in claim 1, characterized in that, Assess the abnormal region and determine its state, including: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is lower than the first overlap rate threshold, or when the deviation between the robot's current location and the expected location exceeds the navigation tolerance threshold, and the navigation path fails, the abnormal area is identified as a state where navigation cannot be achieved normally and needs to be updated immediately.

3. The method as described in claim 2, characterized in that, Also includes: When the central processing unit determines that an abnormal area needs to be updated but not immediately based on the evaluation results, it acquires the current point cloud data generated by the LiDAR scanning of the surrounding environment; determines the robot's second current location based on the current point cloud data; and generates a second optimal path based on the second current location. The navigation control unit controls the robot to continue executing the task from the second current location according to the second optimal path; after completing the task, the robot is controlled to return to the abnormal area; After completing its mission, the lidar captures environmental changes that have occurred in the abnormal area since the last scan and reconstructs a point cloud map based on these environmental changes.

4. The method as described in claim 3, characterized in that, Assess the abnormal region and determine its state, including: If, under any of the following sub-conditions, a region changes beyond a change threshold but does not affect the overall navigation performance, the abnormal region is identified as needing to be updated but not immediately: Sub-case 1: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 2: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is less than the first overlap rate threshold, and the deviation between the robot's current location and the expected location is less than or equal to the navigation tolerance threshold and greater than or equal to the allowable boundary threshold. Sub-case 3: The overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than or equal to the first overlap rate threshold and less than or equal to the second overlap rate threshold, and the deviation between the robot's current positioning position and the expected position is less than the allowable boundary threshold. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

5. The method as described in claim 2, characterized in that, Also includes: When the central processing unit determines, based on the evaluation results, that the abnormal area is in a state where normal navigation can be achieved without updating, it sends a command to the navigation control unit to continue normal navigation. When the navigation control unit receives an instruction to continue normal navigation, it continues to control the robot to complete the task.

6. The method as described in claim 5, characterized in that, Assess the abnormal region and determine its state, including: When the overlap rate between the current point cloud data and the pre-established supermarket point cloud map data is greater than the second overlap rate threshold, and the deviation between the robot's current location and the expected location is less than the allowable boundary threshold, and the changes in some areas do not exceed the change threshold, and the impact rate on the navigation path is lower than the preset impact rate threshold, the abnormal area is determined to be in a state where normal navigation can be achieved without updating. Wherein, the first overlap rate threshold is less than the second overlap rate threshold, and the allowable boundary threshold is less than the navigation tolerance threshold.

7. The method as described in claim 1, characterized in that, Also includes: When the central processing unit detects that the robot has reached a safe position, it uses the angle of arrival and nine-axis positioning to correct the safe position and obtain the corrected safe position; it verifies whether the current environmental data is consistent with the preset data of the safe position. If the current environmental data is inconsistent with the preset data of the safe position, it initiates a map reconstruction command to the LiDAR. When the lidar receives a map reconstruction command, it rescans the area within the preset range of the safe location and reconstructs the point cloud map according to the environmental changes.

8. The method as described in claim 1, characterized in that, The navigation control unit controls the robot to return to the safe location according to the first optimal path, including: During the process of the navigation control unit controlling the robot to return to the safe position according to the first optimal path, the distance between the robot and the safe position is continuously monitored; Based on the distance between the robot and the safe location, control the robot to approach the safe location along a first optimal path using a preset process.

9. The method as described in claim 1, characterized in that, Also includes: The image acquisition device collects current visual image data of the surrounding environment; When the central processing unit detects a change in the surrounding environment, it matches the current visual image data with a pre-stored supermarket image dataset to obtain target image information on the shelf; it then performs image recognition on the target image information and performs robot-assisted positioning based on the preset location information of the identified goods in the supermarket.

10. The method as described in claim 1, characterized in that, Also includes: During normal positioning, the electronic price tag unit records the strength value of the base station response signal and the base station location information in real time; The navigation control unit updates and optimizes the navigation path based on the strength of the base station response signal and the base station location information to adapt to different environments and conditions.

11. The method as described in claim 1, characterized in that, Also includes: The electronic price tag unit interacts with the smart electronic tags in the supermarket to obtain the location information of the product currently bound to the smart electronic tag. Based on the location information of the product currently bound to the smart electronic tag, the robot performs assisted positioning.

12. The method as described in claim 1, characterized in that, Also includes: When the nine-axis positioning position is located in an unreachable shelf area, the nine-axis positioning position is corrected based on pre-established map information, and the nine-axis positioning position is corrected to the nearest reachable area.

13. The method as described in claim 1, characterized in that, Also includes: The central processing unit reassesses and verifies the areas surrounding the abnormal region that affect navigation, and reconstructs the point cloud map based on the reassessment and verification results.

14. A method for adaptive navigation in a dynamic environment, characterized in that, This method is applied to a central processing unit, and the method includes: The current point cloud data is compared with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the lidar, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment. When changes in the surrounding environment are determined based on the comparison results, abnormal areas on the map are marked. Assess the abnormal region and determine its status; When an abnormal area is determined to be a state where navigation cannot be achieved normally and an immediate update is required based on the evaluation results, a fusion positioning is performed based on the angle of arrival (AHA) and nine-axis data to obtain the robot's first current positioning position. This includes: acquiring the robot's current position and velocity, and the covariance matrix of that position and velocity; using the acceleration and angular velocity data from the nine-axis sensors, and the robot's motion equations, predicting the robot's position and velocity at the next moment to obtain the current predicted state; updating the covariance matrix based on the prediction error and noise model to obtain the uncertainty of the current predicted state; calculating the predicted observations from the nine-axis sensors based on the current predicted state; using the AHA as the actual observation, calculating the difference between the actual observation and the predicted observation; and considering the uncertainty of the current predicted state and the uncertainty of the actual measurement state... The system dynamically adjusts the weights of the angle-of-arrival (AOA) and nine-axis (NJA) positioning positions in the final fusion result. Based on the differences and weights, it corrects the robot's current position and velocity to obtain the first current positioning position. Alternatively, outside the base station coverage area, it uses the positioning position obtained from the NJA data for positioning. Within the base station coverage area, it uses the AOA positioning position obtained from the measurement signal emitted by the electronic price tag unit to correct the NJA positioning position. The AOA positioning position is sent by an AOA server, which determines the robot's AOA positioning position based on the AOA angle. The AOA angle is sent by a base station, which determines the AOA angle based on the measurement signal sent by the electronic price tag unit. The NJA data is sent by a nine-axis sensor, which measures the robot's NJA data. The system retrieves the last confirmed safe location from historical positioning data, determines a first optimal path from the first current positioning location back to that safe location, and sends it to the navigation control unit. The navigation control unit controls the robot to return to the safe location based on the first optimal path. The lidar is also used to capture environmental changes that have occurred since the last scan from abnormal areas, reconstruct a point cloud map based on the environmental changes, regenerate a navigation path based on the reconstructed point cloud map, and control the robot to continue performing the task from the safe location based on the regenerated navigation path. The system further includes: regenerating a navigation path based on the reconstructed point cloud map; and controlling the robot to continue performing the task from the safe location based on the regenerated navigation path.

15. A robot with dynamic environment adaptive navigation, characterized in that, include: LiDAR is used to scan the surrounding environment, generate current point cloud data of the surrounding environment and send it to the central processing unit; it captures environmental changes that have occurred since the last scan from abnormal areas and reconstructs the point cloud map based on the environmental changes. An electronic price tag unit is used to send measurement signals to a base station; the base station is used to determine the angle of arrival based on the measurement signals and send the angle of arrival to an angle of arrival server; the angle of arrival server is used to determine the robot's angle of arrival positioning position based on the angle of arrival and return it to the central processing unit. A nine-axis sensor is used to measure data from the robot's nine axes and send it to the central processing unit. The central processing unit compares the current point cloud data with pre-generated supermarket point cloud map data. When the comparison results indicate a change in the surrounding environment, it marks abnormal areas on the map, evaluates these abnormal areas, and determines their state. If the evaluation results determine that an abnormal area is in a state where navigation cannot be achieved normally and requires immediate updating, it performs fusion positioning based on the angle of arrival and nine-axis data to obtain the robot's first current positioning position. This includes: acquiring the robot's current position and velocity, and the covariance matrix of that position and velocity; using the acceleration and angular velocity data from the nine-axis sensors, and the robot's motion equations, predicting the robot's position and velocity at the next moment to obtain the current predicted state; updating the covariance matrix based on the prediction error and noise model to obtain the uncertainty of the current predicted state; and based on the current... The system predicts the state and calculates the predicted observations from the nine-axis sensor. Using the angle-of-arrival (AOA) position as the actual observation, it calculates the difference between the actual observation and the predicted observation. Based on the uncertainty of the current predicted state and the uncertainty of the actual measurement state, it dynamically adjusts the weights of the AOA position and the nine-axis position in the final fusion result. Based on the difference and the weights, it corrects the robot's current position and velocity to obtain the first current positioning position. Alternatively, outside the base station coverage area, it uses the positioning position obtained from the nine-axis data for positioning. Within the base station coverage area, it uses the AOA position obtained from the measurement signal emitted by the electronic price tag unit to correct the nine-axis positioning position. It then calls the last confirmed safe position from the historical positioning data to determine the first optimal path from the first current positioning position back to that safe position. The navigation control unit is used to control the robot to return to the safe position according to the first optimal path; to regenerate the navigation path according to the reconstructed point cloud map; and to control the robot to continue to perform the task from the safe position according to the regenerated navigation path.

16. A central processing unit for dynamic environment adaptive navigation, characterized in that, include: The comparison module is used to compare the current point cloud data with the pre-generated supermarket point cloud map data; the current point cloud data is sent by the lidar, which is used to scan the surrounding environment and generate the current point cloud data of the surrounding environment. The marking module is used to mark abnormal areas on the map when changes in the surrounding environment are determined based on comparison results; The evaluation module is used to evaluate the abnormal region and determine the state of the abnormal region; The fusion localization module is used to perform fusion localization based on the angle of arrival (AHA) and nine-axis data when an abnormal area is determined to be a state where navigation cannot be achieved normally and immediate updates are required, based on the evaluation results. This results in the robot's first current localization position, which includes: acquiring the robot's current position and velocity, and the covariance matrix of that position and velocity; using the acceleration and angular velocity data from the nine-axis sensors and the robot's motion equations to predict the robot's position and velocity at the next moment, obtaining the current predicted state; updating the covariance matrix based on the prediction error and noise model to obtain the uncertainty of the current predicted state; calculating the predicted observations from the nine-axis sensors based on the current predicted state; using the AHA as the actual observation, calculating the difference between the actual observation and the predicted observation; and calculating the difference based on the uncertainty of the current predicted state and the actual measurement status. To address the uncertainty of the state, the weights of the angle of arrival (AHA) and nine-axis positioning positions in the final fusion result are dynamically adjusted. Based on the differences and weights, the robot's current position and velocity are corrected to obtain the first current positioning position. Alternatively, outside the base station coverage area, the positioning position obtained from the nine-axis data is used for positioning. Within the base station coverage area, the nine-axis positioning position is corrected using the AHA positioning position obtained from the measurement signal emitted by the electronic price tag unit. The AHA positioning position is sent by an AHA server, which determines the robot's AHA positioning position based on the AHA angle. The AHA angle is sent by a base station, which determines the AHA angle based on the measurement signal emitted by the electronic price tag unit. The nine-axis data is sent by a nine-axis sensor, which measures the robot's nine-axis data. The optimal path determination module is used to call the last confirmed safe location in the historical positioning data, determine the first optimal path from the first current positioning location back to the safe location, and send it to the navigation control unit; the navigation control unit is used to control the robot to return to the safe location according to the first optimal path; the lidar is also used to capture environmental changes that have occurred since the last scan from the abnormal area, reconstruct a point cloud map according to the environmental changes; regenerate a navigation path according to the reconstructed point cloud map; and control the robot to continue to perform the task from the safe location according to the regenerated navigation path. It is also used to: regenerate a navigation path according to the reconstructed point cloud map; and control the robot to continue to perform the task from the safe location according to the regenerated navigation path.

17. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 14.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 14.

19. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 14.

Images