Robot automatic recharging control method and intelligent mowing robot
By employing multiple positioning algorithms and monitoring them in real time during the robot's recharging process, the division of the alignment positioning area and algorithm switching were realized. This solved the problems of inaccurate positioning and insufficient robustness of a single positioning algorithm in complex environments, and improved the stability and adaptability of the robot's recharging process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-15
- Publication Date
- 2026-05-05
AI Technical Summary
Existing robot automatic recharging and positioning algorithms are susceptible to environmental interference, have poor positioning reliability, are difficult to adapt to complex and ever-changing scenarios, and are difficult to balance accuracy and robustness, with weak fault recovery capabilities, which limits the widespread application of robots.
By employing multiple positioning algorithms and monitoring for anomalies in the target positioning algorithm in real time, and by switching algorithms and dividing the positioning area, the system can dynamically adapt to environmental changes, thereby improving fault recovery capabilities and positioning accuracy.
It improves the positioning accuracy and stability during the robot's recharging process, avoids frequent invalid adjustments caused by algorithm instability, and enhances fault recovery capability and environmental adaptability.
Smart Images
Figure CN121979208A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of robot recharging control technology, and in particular to automatic robot recharging control methods and intelligent lawn mowing robots. Background Technology
[0002] With the increasing popularity of service robots, automatic recharging has become a standard feature. Automatic recharging refers to the robot's ability to return to the charging station and recharge itself when its battery is low. This is a key factor in ensuring the robot's continuous and stable operation.
[0003] Currently, the automatic recharging function of robots mostly relies on a single positioning algorithm, which leads to poor positioning reliability.
[0004] Specifically, single-sensor localization algorithms are susceptible to environmental interference. For example, visual algorithms are greatly affected by lighting conditions; overexposure in strong light or blurring in weak light can lead to feature loss and inaccurate localization. Ultrasonic algorithms are prone to generating false signals due to reflections from obstacles. Furthermore, single-sensor localization algorithms struggle to adapt to complex and changing scenarios; changes in environmental layout and weather conditions can all degrade their performance. Simultaneously, it is difficult to balance accuracy and robustness; they are prone to localization drift when faced with anomalies such as noise and data loss, and their fault recovery capabilities are weak.
[0005] As the application fields of robots expand, the requirements for recharging and positioning are increasing. The limitations of single positioning algorithms restrict the widespread application of robots and urgently need to be improved. Summary of the Invention
[0006] To address the technical problems associated with a single positioning algorithm, this application provides an automatic recharging control method for a robot. The method includes: when moving to a guided range, determining a target positioning algorithm from multiple positioning algorithms; performing alignment and positioning based on the target positioning algorithm; adjusting the robot's pose within the guided range according to the positioning result to complete the docking of the charging interface; during the alignment and positioning process based on the target positioning algorithm, real-time monitoring to see if an algorithm switching trigger condition is met; when the condition is met, re-determining the target positioning algorithm from other positioning algorithms and performing alignment and positioning based on the new target positioning algorithm; wherein the trigger condition includes one or more of the following: failure to complete interface docking within a preset time threshold, an interface docking failure count exceeding a preset number threshold, or abnormal positioning data.
[0007] Based on the above technical solution, by applying multiple positioning algorithms and monitoring the abnormal situation of the target positioning algorithm in real time, the algorithm can be switched in time when an anomaly is detected, and a new target algorithm can be selected from the non-target positioning algorithms, thereby improving the fault recovery capability. Since different positioning algorithms have different requirements for input values, the algorithm can be specifically selected to adapt to the dynamic changes of the actual application scenario, thereby improving the environmental adaptability.
[0008] In one implementation, the method of aligning and positioning based on the target positioning algorithm and adjusting the pose within the guidance range according to the positioning result includes: real-time monitoring of whether a pre-set alignment and positioning area has been entered; if it is determined that the alignment and positioning area has been entered, re-determining the target positioning data based on the target positioning algorithm; and adjusting the pose based on the target positioning data.
[0009] Based on the above technical solution, by dividing the alignment and positioning area and controlling the re-alignment and positioning within this area, it is possible to achieve pose adjustment only after a stable and accurate positioning result is determined, avoiding frequent and invalid adjustments caused by algorithm instability during the positioning and adjustment process. Furthermore, it is possible to divide multiple alignment and positioning areas according to the actual application scenario, enabling phased positioning and pose adjustment, thereby improving positioning accuracy and fault tolerance.
[0010] In one implementation, the retargeting data based on the target positioning algorithm includes: acquiring multiple sets of positioning data within a preset positioning time period based on the target positioning algorithm; maintaining a sliding time window and calculating the deviation between each set of positioning data and the mean of the time window to which it belongs; if the deviation is greater than a first preset deviation threshold, then determining the positioning data as an outlier; removing the outlier; and performing a weighted average on the retained positioning data to obtain the target positioning data.
[0011] In one implementation, obtaining multiple sets of positioning data within a preset positioning time period based on the target positioning algorithm includes: within the preset positioning time period, when obtaining the valid input value required by the target positioning algorithm, remaining stationary, and obtaining the multiple sets of positioning data according to a preset time interval.
[0012] In one implementation, multiple alignment and positioning areas are defined within the guidance range, and each alignment and positioning area is divided according to its distance from the charging pile.
[0013] In one implementation, the method for re-determining the target positioning algorithm from the other positioning algorithms includes: selecting the one with the highest preset priority among the other positioning algorithms as the new target positioning algorithm; or, acquiring real-time positioning data determined based on each positioning algorithm, determining the similarity of the positioning results of each positioning algorithm based on the real-time positioning data, and determining the one with the highest similarity of the positioning results as the new target positioning algorithm; wherein, the real-time positioning data is obtained by executing each positioning algorithm in real time after moving to the guidance range, and the similarity of the positioning results is determined based on intra-group similarity or inter-group similarity.
[0014] Based on the above technical solution, a more stable positioning algorithm can be selected based on the similarity of real-time positioning data, and used as a new target positioning algorithm, thereby ensuring the accuracy and stability of the new target positioning algorithm.
[0015] In one implementation, the positioning algorithms vary in their dependence on the image recognition results of the charging interface positioning marker.
[0016] In one implementation, the method for alignment and positioning based on one of the positioning algorithms includes: acquiring an RGB image of the surrounding environment; calling a deep learning segmentation model to identify the charging interface positioning mark region in the RGB image to obtain the pixel coordinates of the charging interface positioning mark region; generating point cloud data of the charging pile; extracting a subset of the point cloud data of the charging interface positioning mark region from the point cloud data according to the pixel coordinates; preprocessing the point cloud subset, performing plane fitting, and extracting the center point of the fitted plane and the long and short side directions of the charging interface positioning mark; using the center point, the long side direction, and the short side direction, determining the pose and position of the charging interface positioning mark in the camera coordinate system, and converting the information of the charging interface positioning mark in the camera coordinate system into relative pose information with the charging interface of the charging pile through coordinate transformation.
[0017] Based on the above technical solution, the "deep learning segmentation + point cloud" positioning algorithm effectively solves the pain point of "stopping upon seeing the code" in traditional vision solutions. Even if the QR code is obscured, worn, or the image is unclear due to extremely poor lighting conditions, as long as the geometric plane of the charging pile is visible, it can still accurately enter the charging pile through point cloud fitting.
[0018] In one implementation, the abnormal positioning data includes one or more of the following: the number of positioning jumps exceeds a preset jump threshold, a valid input value cannot be obtained, or the device is positioned outside the guidance range.
[0019] Based on the same inventive concept, this application also provides an intelligent lawn mowing robot, which automatically recharges based on the above-described control method. Attached Figure Description
[0020] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0021] Figure 1 This is a flowchart of the robot automatic recharging control method provided in an embodiment of this application.
[0022] Figure 2 This is a flowchart of the alignment and positioning method based on the target positioning algorithm in the embodiments of this application.
[0023] Figure 3 This is a schematic diagram of the alignment and positioning area in an embodiment of this application.
[0024] Figure 4 This is a flowchart of a method for re-determining target location data based on a target location algorithm in an embodiment of this application.
[0025] Figure 5 This is a flowchart of a method for real-time monitoring of a target localization algorithm in an embodiment of this application.
[0026] Figure 6 This is a flowchart illustrating a method for alignment and positioning based on a second positioning algorithm, as an application example of an embodiment of this application. Detailed Implementation
[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this invention.
[0028] In the description of this application, it should be noted that the terms "first" and "second" are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0029] Please refer to Figure 1 The automatic recharging control method for robots provided in this application embodiment is applied to mobile robots and specifically includes the following steps.
[0030] S100 responds to recharge commands, plans a recharge route based on the current location, moves along the recharge route, and determines in real time whether it has reached the guidance range.
[0031] In practice, when the battery level is lower than the charging threshold, the robot control system automatically generates a return-to-charge command and plans a return-to-charge route based on the current location and the pre-determined location or guidance range of the charging station.
[0032] The guidance range is the short-range activity range used to achieve alignment and positioning with the charging interface of the charging pile. It is determined based on the positioning of the charging pile. Within the guidance range, in the absence of interference, the robot can obtain effective input values for various positioning algorithms.
[0033] In one example, a sector-shaped area with a radius of 1 meter and a central angle of 120°, centered on the charging port of the charging station, is used as the guiding range. It is understood that the radius and the size of the central angle can be adjusted according to actual needs, and this application does not impose any limitations on them.
[0034] In one implementation, during the movement along the charging cable, different types of positioning algorithms can be used for positioning. When the robot has not moved into the guidance range, positioning algorithms suitable for long-distance positioning can be used, such as positioning algorithms based on RTK (Real-time Kinematic); when the robot moves into the guidance range, positioning algorithms more suitable for close-range alignment can be used, such as positioning algorithms based on QR codes.
[0035] S101, when moving into the guidance range, the target positioning algorithm is determined from multiple positioning algorithms.
[0036] During implementation, the system pre-deploys multiple positioning algorithms. When it is determined that the system has entered the guidance range based on real-time positioning information, the most suitable algorithm can be selected from the multiple positioning algorithms as the target positioning algorithm.
[0037] In one example, the target positioning algorithm can be determined based on the priority set pre-defined for each positioning algorithm. The priority of each positioning algorithm can be determined based on the theoretical reliability value determined by the algorithm principle, or it can be dynamically determined periodically based on the historical performance of each positioning algorithm.
[0038] S102 performs alignment and positioning based on a target positioning algorithm, and adjusts its posture within the guidance range according to the positioning result to complete the docking of the charging interface.
[0039] During implementation, please refer to Figure 2 The method for alignment and positioning based on target localization algorithm specifically includes the following steps.
[0040] S201 monitors in real time whether the system has entered the pre-set alignment and positioning area.
[0041] Specifically, the alignment and positioning area is used to trigger a re-alignment calculation. In other words, when the robot enters the alignment and positioning area, it needs to be re-aligned and positioned, and after the positioning result is clear, the robot body is controlled to move or adjust its posture.
[0042] In one example, please refer to Figure 3 Within the guidance range D, there are three alignment and positioning areas A1, A2, and A3. Each alignment and positioning area is divided according to its distance from the charging pile, from farthest to closest, and there is a certain interval between them.
[0043] It is understood that the number, size, and spacing of the alignment and positioning areas can be adjusted according to actual needs, and this application is not limited thereto.
[0044] S202, if it is determined that the target has entered the alignment and positioning area, the target positioning data is re-determined based on the target positioning algorithm.
[0045] In implementation, please refer to Figure 4 The method for re-determining target location data based on the target location algorithm includes the following steps.
[0046] S401, based on the target positioning algorithm, obtains multiple sets of positioning data within a preset positioning time period.
[0047] The preset positioning time period is a pre-set duration used to re-determine the relative position information with the charging interface. The starting point of the time period can be the current time or a historical time before the current time. In other words, the current time can be the starting point, intermediate point or end point of the preset positioning time period, and the length of the time period can be fixed.
[0048] The location data is the relative position information of the charging pile's charging interface, calculated using a target positioning algorithm based on the collected input values. The input values are related to the target positioning algorithm, such as environmental images and RTK differential data.
[0049] In one example, during a preset positioning period, the device first rotates, and once it obtains the valid input value required by the target positioning algorithm, it remains stationary, obtaining multiple sets of positioning data according to a preset time interval.
[0050] S402, maintain a sliding time window and calculate the deviation of each set of positioning data from the mean of its respective time window.
[0051] Among them, the window length is less than the preset positioning time period, and it can be an overlapping window or a non-overlapping window.
[0052] The mean of the time window to which the location data belongs is calculated by averaging all the location data within that time window.
[0053] If the deviation is greater than the first preset deviation threshold, then step S403 is executed. The first preset deviation threshold can be determined based on the historical performance of the target positioning algorithm, or it can be determined based on the theoretical error of the target positioning algorithm. In one example, the theoretical error can be evaluated based on data such as the acquisition error and calculation error of the input value.
[0054] S403, the location data is identified as an outlier and removed.
[0055] S404, the retained positioning data is weighted and averaged to serve as the target positioning data.
[0056] S203, adjust pose based on target positioning data.
[0057] In practice, adjusting the posture includes moving towards the charging port of the charging station and aligning the charging port.
[0058] Based on the above steps, alignment and positioning based on the target localization algorithm can be achieved. However, in actual application, the target localization algorithm may fail, such as inaccurate input value acquisition due to environmental changes, leading to incorrect calculation results and preventing successful interface integration. To address this issue, this embodiment of the application, while executing the above method steps, simultaneously executes... Figure 5 The method and steps shown are used to monitor the alignment and positioning execution of the target positioning algorithm in real time, and to switch algorithms in a timely manner when an anomaly is detected.
[0059] Please refer to Figure 5 The method for real-time monitoring of the target localization algorithm in this application embodiment includes the following steps.
[0060] S501 monitors in real time whether the algorithm switching trigger conditions are met during the alignment and positioning process based on the target positioning algorithm.
[0061] The triggering conditions include one or more of the following: failure to complete interface connection within a preset time threshold, failure to complete interface connection more than a preset number of times, or abnormal location data.
[0062] In one implementation, when a target positioning algorithm is determined, a timer is started to record the alignment and positioning time based on the target positioning algorithm. If the alignment and positioning time is longer than a preset time threshold, the triggering condition is determined to be met.
[0063] In one implementation, after interface docking is performed at the target location determined by the target positioning algorithm, if a successful interface docking signal is not detected, it is recorded as docking failure. After docking failure, the system moves to the nearest positioning alignment area for repositioning. If the cumulative number of docking failures exceeds a preset threshold, the triggering condition is determined to be met.
[0064] In one implementation, abnormal positioning data includes the number of positioning jumps exceeding a preset jump threshold or the inability to obtain valid input values.
[0065] In one example, determining whether a location jump has occurred can be based on the deviation between the aforementioned location data and the window average. If the deviation is greater than a second preset deviation threshold, a jump is determined to have occurred, where the second preset deviation threshold is greater than or equal to a first preset deviation threshold. If, during a single relocation of the target location, the cumulative number of jumps exceeds a preset jump threshold, the triggering condition is determined to be met.
[0066] The inability to obtain valid input values includes, but is not limited to, the inability to acquire image data, the inability to extract valid input values from the acquired image data, and the robot being located outside the guidance range (i.e., the robot has moved out of the guidance range). Based on this, it can promptly identify situations where effective calculations cannot be performed due to factors such as environmental changes or algorithm malfunctions; by monitoring whether the robot is within the guidance range in real time, it can determine that the positioning algorithm is malfunctioning when the robot moves out of the guidance range, and thus switch algorithms in a timely manner.
[0067] When the triggering conditions are met, step S502 is executed; if the algorithm switching triggering conditions are not met, monitoring continues until the charging interface docking is completed, i.e., the charging interface docking is successful.
[0068] S502, from other positioning algorithms, a new target positioning algorithm is determined.
[0069] In practice, when the triggering condition is met, it indicates that the current target localization algorithm is abnormal and the localization result is unreliable. Therefore, it is necessary to select a new target localization algorithm.
[0070] In one example, the algorithm with the highest priority among other localization algorithms can be selected as the new target localization algorithm. The method for setting the priority has been explained above and will not be repeated here.
[0071] In another example, real-time positioning data determined by each positioning algorithm is obtained, the similarity of the positioning results of each positioning algorithm is determined based on the real-time positioning data, and the algorithm with the highest similarity is selected as the new target positioning algorithm.
[0072] Specifically, the real-time positioning data is calculated by executing various positioning algorithms in real time after the user moves into the guided area.
[0073] In one example, the similarity of the positioning results can be determined based on intra-group similarity and / or inter-group similarity. Intra-group similarity can be determined based on the range of differences between all real-time positioning data obtained by the same positioning algorithm; inter-group similarity can be determined based on the range of differences between real-time positioning data and those of the target positioning algorithm or other positioning algorithms. The larger the range of differences, the lower the similarity.
[0074] Based on this, by calculating horizontal and vertical similarity, the algorithm with the highest similarity among other positioning algorithms can be selected as the new target positioning algorithm. This can largely guarantee the reliability of the new target positioning algorithm, and this reliability is based on real-time positioning data, which is more reliable than simply relying on historical performance.
[0075] S503 performs alignment and positioning based on a new target localization algorithm.
[0076] In practice, once the target positioning algorithm is redefined, the system first moves to the nearest positioning alignment area, then redefines the target position of the charging interface based on the new target positioning algorithm, and performs movement control or pose adjustment based on the target position.
[0077] It is worth noting that after determining the new target location, the process returns to step S501 above to monitor the alignment and positioning process of the new target location algorithm. In this way, by switching the target location algorithm promptly upon detecting an anomaly, the problems of insufficient robustness and weak fault recovery capability under a single location algorithm can be effectively avoided.
[0078] Therefore, the robot automatic recharging control method provided in this application embodiment monitors the alignment and positioning process of the target positioning algorithm in real time, and selects a new target positioning algorithm from other positioning algorithms when the algorithm switching trigger condition is met, so as to realize timely recovery of positioning failure and improve the reliability and stability of the control system.
[0079] Furthermore, by dividing the positioning alignment area, close-range positioning alignment is divided into multiple stages, and realignment is performed at each stage. This facilitates the timely detection of anomalies in the positioning algorithm and stabilizes the robot's movement during close-range positioning, avoiding frequent movements and rotations caused by inaccurate positioning. With multiple positioning alignment areas, segmented alignment can effectively solve the problem of alignment failure caused by environmental factors during movement, thereby ensuring the success rate of final interface docking.
[0080] Specifically, in the alignment and positioning process based on the technical solution provided in this application, the robot's movement involves adjusting its posture by remaining stationary or rotating within a small range within the alignment area, and then moving directly to the next alignment area or performing interface docking. When a target algorithm switch occurs, the robot retreats to the nearest alignment area to re-align and adjust its posture. Therefore, the overall movement range and mode are regular, and it can promptly retreat to the alignment area in case of an anomaly. The overall movement is regular and efficient. In contrast, existing implementations involve the robot moving or adjusting its posture directly based on the positioning result for one-time alignment. After detecting docking failure or an incorrect movement direction, the robot repositions itself in place and then moves or adjusts its posture based on the new positioning result. However, when the positioning algorithm malfunctions, due to a lack of fault recovery capability, frequent movement and posture adjustments occur, and the direction of movement and posture adjustment are chaotic. Especially when real-time positioning changes abruptly, large movement deviations are likely to occur, thereby reducing charging docking efficiency.
[0081] The following examples illustrate the application of the control method provided in the embodiments of this application in an intelligent lawnmower robot.
[0082] Intelligent lawnmowers are typically used in outdoor lawns where environmental conditions are variable and require adaptability to different lighting conditions. However, most existing intelligent lawnmowers rely on a single positioning algorithm based on a charging port location marker for alignment. Therefore, they have high requirements for image acquisition, and when lighting changes, such as sudden overcast skies, sunsets, or when the charging port location marker is obscured or damaged, they may be unable to achieve positioning based on the acquired images. In this example, the charging port location marker is an ArUco (Augmented Reality University of Cordoba) QR code.
[0083] To address this issue, in one example, multiple positioning methods can be deployed simultaneously on the intelligent lawnmower robot, including a first positioning algorithm based on QR code positioning principle, a second positioning algorithm based on deep learning and point cloud data, and a positioning algorithm based on global coordinates, such as RTK-based positioning algorithm and adaptive Monte Carlo positioning algorithm.
[0084] The three positioning algorithms depend on the image recognition results of the charging port positioning mark to varying degrees. The first positioning algorithm is completely dependent on the image recognition results of the QR code, the second positioning algorithm is partially dependent, and the third positioning algorithm is completely independent. The priority of the three positioning algorithms is set as follows: the first positioning algorithm has the highest priority, followed by the second positioning algorithm.
[0085] When it is determined that the robot needs to recharge, its control system, after planning the recharge path, will execute a third positioning algorithm if it determines that the robot is not currently within the guidance range, thus achieving long-distance positioning. If the robot detects that it has moved into the guidance range, it will stop moving, identify the first positioning algorithm as the target positioning algorithm, and begin alignment and positioning based on the first positioning algorithm.
[0086] During this process, a monitoring process is initiated to monitor the alignment and positioning process of the first positioning algorithm in real time. When the algorithm switching conditions are met, a new target positioning algorithm is selected from the second and third positioning algorithms. The alignment and positioning process of the new target positioning algorithm is then monitored until the docking is successful or fails. The methods for determining docking failure include, but are not limited to, the cumulative docking time exceeding a preset total time threshold, the cumulative docking number exceeding a preset total number threshold, and remaining battery power alarm.
[0087] In this example, refer to Figure 6 The method for alignment and positioning based on the second positioning algorithm specifically includes the following steps.
[0088] S601, acquires an RGB image of the surrounding environment.
[0089] In practice, the built-in camera can acquire RGB images of the surrounding environment as input for subsequent processing.
[0090] S602, calls a deep learning segmentation model to identify the charging interface positioning marker region in the RGB image, so as to obtain the pixel coordinates of the charging interface positioning marker region.
[0091] In practice, pre-trained deep learning segmentation models (such as U-Net, Mask R-CNN, etc.) can be used to process RGB images and extract the pixel coordinates of the QR code region from the model's output to determine the specific location of the QR code in the image.
[0092] The advantage of this model is that it does not require modeling and recognizing the entire charging pile, and it can be effectively recognized even if the QR code is partially obscured.
[0093] S603 generates point cloud data for charging piles.
[0094] In practice, the intelligent lawnmower robot is equipped with a binocular camera, which can generate point cloud data of the environment using the principle of binocular vision. The point cloud data includes the three-dimensional spatial information of the charging pile.
[0095] S604 extracts a subset of the point cloud of the charging interface location marker area from the point cloud data based on the pixel coordinates.
[0096] S605 preprocesses the point cloud subset, performs plane fitting, and extracts the center point of the fitted plane and the long and short sides of the charging interface positioning mark.
[0097] S606 uses the center point, long side direction, and short side direction to determine the attitude and position of the charging interface positioning mark in the camera coordinate system, and converts the information of the charging interface positioning mark in the camera coordinate system into relative pose information with the charging pile interface through coordinate transformation.
[0098] Based on the above positioning algorithm, accurate positioning can be achieved in scenarios where only a portion of the QR code image can be recognized.
[0099] The above description is merely a preferred embodiment of this application and is not intended to limit this application in any way. Although this application has been disclosed above with reference to preferred embodiments, it is not intended to limit this application. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the technical solution of this application. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of this application without departing from the scope of the technical solution of this application shall still fall within the scope of the technical solution of this application.
Claims
1. A method for controlling automatic recharging of a robot, characterized in that, Applied to robots, the method includes: When the device moves into the guidance range, a target positioning algorithm is determined from multiple positioning algorithms; alignment and positioning are performed based on the target positioning algorithm; and pose adjustment is performed within the guidance range according to the positioning result to complete the docking of the charging interface. During the alignment and positioning process based on the target positioning algorithm, it is monitored in real time whether the algorithm switching trigger condition is met; when it is determined that the condition is met, a new target positioning algorithm is determined from the other positioning algorithms, and alignment and positioning are performed based on the new target positioning algorithm; wherein, The triggering conditions include one or more of the following: failure to complete interface connection within a preset time threshold, failure to complete interface connection more than a preset number of times, or abnormal location data.
2. The control method according to claim 1, characterized in that, The method for alignment and positioning based on the target localization algorithm, and for adjusting the pose within the guidance range according to the positioning result, includes: Real-time monitoring to see if it enters the pre-set alignment and positioning area; If it is determined that the target has entered the alignment and positioning area, the target positioning data is re-determined based on the target positioning algorithm; Adjust pose based on target positioning data.
3. The control method according to claim 2, characterized in that, The target relocation data based on the target localization algorithm includes: Based on the target positioning algorithm, multiple sets of positioning data within a preset positioning time period are obtained; Maintain a sliding time window and calculate the deviation of each set of positioning data from the mean of its respective time window; If the deviation is greater than a first preset deviation threshold, the positioning data is determined to be an abnormal value; Remove the outliers; The retained positioning data is weighted and averaged to obtain the target positioning data.
4. The control method according to claim 3, characterized in that, The step of obtaining multiple sets of positioning data within a preset positioning time period based on the target positioning algorithm includes: During the preset positioning time period, when the valid input value required by the target positioning algorithm is obtained, the system remains stationary and obtains the multiple sets of positioning data according to the preset time interval.
5. The control method according to claim 2, characterized in that, The guidance range is defined by multiple alignment and positioning areas, each of which is divided according to its distance from the charging pile.
6. The control method according to claim 1, characterized in that, The method for re-determining the target localization algorithm from the other localization algorithms includes: Select the algorithm with the highest preset priority among the other positioning algorithms as the new target positioning algorithm; or... The system acquires real-time positioning data determined by each of the aforementioned positioning algorithms, determines the similarity of the positioning results of each of the aforementioned positioning algorithms based on the real-time positioning data, and identifies the algorithm with the highest similarity as the new target positioning algorithm. The real-time positioning data is obtained by executing each of the aforementioned positioning algorithms in real time after the system moves into the guidance range, and the similarity of the positioning results is determined based on intra-group similarity or inter-group similarity.
7. The control method according to claim 1, characterized in that, The positioning algorithms described above depend on the image recognition results of the charging interface positioning marker to varying degrees.
8. The control method according to claim 1, characterized in that, The method for alignment and positioning based on one of the positioning algorithms includes: Obtain an RGB image of the surrounding environment; A deep learning segmentation model is invoked to identify the charging interface location marker region in the RGB image, so as to obtain the pixel coordinates of the charging interface location marker region; Generate point cloud data for charging piles; Based on the pixel coordinates, a subset of the point cloud of the charging interface positioning identifier region is extracted from the point cloud data; The point cloud subset is preprocessed, a plane is fitted, and the center point of the fitted plane and the long and short sides of the charging interface positioning mark are extracted. Using the center point, the long side direction, and the short side direction, the attitude and position of the charging interface positioning mark in the camera coordinate system are determined, and the information of the charging interface positioning mark in the camera coordinate system is converted into relative pose information with the charging interface of the charging pile through coordinate transformation.
9. The control method according to claim 1, characterized in that, The abnormal positioning data includes one or more of the following: the number of positioning jumps exceeds a preset jump threshold, a valid input value cannot be obtained, or the device is positioned outside the guidance range.
10. An intelligent lawnmower robot, characterized in that, The intelligent lawnmower robot achieves automatic recharging based on the control method described in any one of claims 1 to 9.