Robot charging methods, apparatus, equipment and computer-readable storage media

Abstract

This application provides a robot charging method, apparatus, device, and computer-readable storage medium, belonging to the field of radar communication technology. The method includes: controlling a robot to move to a target area where the entrance to a target charging station is located; determining the contour information of the target charging station based on a radar device installed on the robot; determining the position offset information of the target charging station based on the contour information; inputting the position offset information of the target charging station into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging station and the current position of the robot; and controlling the robot to move to the charging position of the target charging station to charge. This application can reduce the complexity of determining the charging position of a charging station, improve computational efficiency, and thus quickly determine the specific location of the target charging station.

Description

Technical Field

[0001] This application relates to the field of radar communication technology, and more specifically, to a robot charging method, apparatus, device, and computer-readable storage medium. Background Technology

[0002] To enable the robot to automatically recharge, it is usually necessary to determine the charging location based on the robot's infrared, radar, or radio frequency detection methods.

[0003] In existing technologies, a brute-force matching method is usually used to determine multiple valid offset values ​​involved in the target charging pile, and then the actual location of the target charging pile is calculated based on the multiple valid offset values.

[0004] However, the brute-force matching algorithm requires a lot of processing. Even if a certain threshold range is set in order to save computing power, it is not easy to calculate the size of the threshold range. Therefore, the complexity of determining the location of the target charging station is too high and the efficiency is low. Summary of the Invention

[0005] The purpose of this application is to provide a robot charging method, apparatus, device, and computer-readable storage medium, which can reduce the complexity of determining the charging location of a charging pile, improve computational efficiency, and thus quickly determine the specific location of the target charging pile.

[0006] The embodiments of this application are implemented as follows:

[0007] One aspect of this application provides a robot charging method, including:

[0008] Control the robot to move to the target area where the entrance to the target charging station is located;

[0009] The outline information of the target charging station is determined based on the radar device installed on the robot;

[0010] The position offset information of the target charging pile is determined based on the contour information. The position offset information includes: relative position offset, target depth offset and target width offset. The relative position offset is the relative position between the robot and the target charging pile. The target depth offset is the longitudinal depth of the charging position of the target charging pile. The target width offset is the lateral width of the charging position of the target charging pile.

[0011] The position offset information of the target charging pile is input into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot.

[0012] Control the robot to move to the target charging station and charge.

[0013] Optionally, the location offset information of the target charging station is determined based on the contour information, including:

[0014] Determine the location of the reference point based on the contour information;

[0015] Determine the offset between the reference point and the robot's position;

[0016] The offset between the reference point and the robot's position is used as the relative position offset.

[0017] Optionally, the location offset information of the target charging station is determined based on the contour information, including:

[0018] The target depth offset of the target charging pile is determined based on the positional deviation between the upper and lower edges in the contour information.

[0019] Optionally, the location offset information of the target charging station is determined based on the contour information, including:

[0020] The target width offset of the target charging pile is determined based on the width information of the upper edge in the contour information.

[0021] Optionally, before inputting the location offset information of the target charging station into the pre-trained deep learning model, the method further includes:

[0022] Obtain initial sample offset information, which includes: sample relative position offset, sample target depth offset, and sample target width offset;

[0023] The initial sample offset information is processed by sample augmentation to obtain a set of sample offset information.

[0024] A deep learning model is obtained by training the initial model based on the sample offset information set.

[0025] Optionally, the initial sample offset information is subjected to sample augmentation processing to obtain a set of sample offset information, including:

[0026] Determine the offset ranges for the relative position offset of the sample, the depth offset of the sample target, and the width offset of the sample target, respectively;

[0027] Random number generation is performed based on the offset range to obtain multiple sets of sample offset information;

[0028] The initial sample offset information and the set of multiple sets of sample offset information are used as the sample offset information set.

[0029] Optionally, a deep learning model is obtained by training the initial model based on the sample offset information set, including:

[0030] Each set of sample offset information in the sample offset information set is input into the initial model in sequence to obtain the output result, and the loss information of the initial model is determined based on the actual charging position of the sample corresponding to the output result and the sample offset information set.

[0031] The parameters of the initial model are corrected based on the loss information of the initial model, and the corrected initial model is retrained using the sample offset information set until the initial model meets the convergence condition. The initial model is then used as the deep learning model.

[0032] In another aspect of this application, a robot charging device is provided, comprising: a moving module, a detection module, a determining module, and a calculating module;

[0033] The mobile module is used to control the robot to move to the target area where the target charging station entrance is located;

[0034] The detection module is used to determine the outline information of the target charging station based on the radar device installed on the robot;

[0035] The determination module is used to determine the position offset information of the target charging pile based on the contour information. The position offset information includes: relative position offset, target depth offset and target width offset. The relative position offset is the relative position between the robot and the target charging pile, the target depth offset is the longitudinal depth of the charging position of the target charging pile, and the target width offset is the lateral width of the charging position of the target charging pile.

[0036] The calculation module is used to input the position offset information of the target charging pile into the pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot.

[0037] The mobile module is also used to control the robot to move to the charging location of the target charging station for charging.

[0038] Optionally, the determining module is specifically used to determine the position of the reference point based on the contour information; determine the offset between the position of the reference point and the position of the robot; and use the offset between the position of the reference point and the position of the robot as the relative position offset.

[0039] Optionally, the determining module is further used to determine the target depth offset of the target charging pile based on the positional deviation between the upper and lower edges in the contour information.

[0040] Optionally, the determining module is further configured to determine the target width offset of the target charging pile based on the width information of the upper edge in the contour information.

[0041] Optionally, the calculation module is also used to obtain initial sample offset information, which includes: sample relative position offset, sample target depth offset, and sample target width offset; to perform sample addition processing on the initial sample offset information to obtain a sample offset information set; and to train the initial model based on the sample offset information set to obtain a deep learning model.

[0042] Optionally, the calculation module is specifically used to determine the offset range of the sample relative position offset, the sample target depth offset, and the sample target width offset, respectively; perform random number generation processing based on the offset range to obtain multiple sets of sample offset information; and use the initial sample offset information and the set of multiple sets of sample offset information as the sample offset information set.

[0043] Optionally, the calculation module is specifically used to input each set of sample offset information in the sample offset information set into the initial model in sequence to obtain the output result, and determine the loss information of the initial model based on the actual charging position of the sample corresponding to the output result and the set of sample offset information; correct the parameters of the initial model based on the loss information of the initial model, and retrain the corrected initial model using the sample offset information set until the initial model meets the convergence condition, and use the initial model as a deep learning model.

[0044] In another aspect of this application, a computer device is provided, including: a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of a robot charging method.

[0045] In another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of a robot charging method.

[0046] The beneficial effects of the embodiments of this application include:

[0047] This application provides a robot charging method, apparatus, device, and computer-readable storage medium. The method allows control of a robot to move to a target area where the entrance to a target charging station is located. It determines the contour information of the target charging station based on a radar device mounted on the robot; determines the position offset information of the target charging station based on the contour information; inputs the position offset information of the target charging station into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging station and the robot's current position; and controls the robot to move to the charging position of the target charging station for charging. By controlling the robot to move to the target area where the entrance to the target charging station is located, the robot can detect more accurate and clear contour information of the target charging station. Based on this contour information, more accurate position offset information can be obtained. Furthermore, the pre-trained deep learning model reduces the amount of position offset information required, allowing the model to directly obtain the relative positional relationship between the charging position of the target charging station and the robot's current position. This enables the robot to move more accurately to the charging position of the target charging station, reducing the complexity of determining the charging position and improving computational efficiency. This allows for rapid determination of the specific location of the target charging station, enabling the robot to reach the charging position more quickly and accurately. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A schematic diagram of a robot working scenario provided in an embodiment of this application;

[0050] Figure 2 A schematic flowchart illustrating the robot charging method provided in this application embodiment;

[0051] Figure 3 A schematic diagram of the target area provided in the embodiments of this application;

[0052] Figure 4 A schematic diagram illustrating the contour information provided in the embodiments of this application;

[0053] Figure 5 This is another schematic diagram of the robot charging method provided in the embodiments of this application;

[0054] Figure 6 This is another schematic diagram of the robot charging method provided in the embodiments of this application;

[0055] Figure 7 This is another schematic diagram of the robot charging method provided in the embodiments of this application;

[0056] Figure 8 This is another schematic diagram of the robot charging method provided in the embodiments of this application;

[0057] Figure 9 This is a schematic diagram of the structure of the robot charging device provided in the embodiments of this application;

[0058] Figure 10 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, 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. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0060] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0061] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0062] In the description of this application, it should be noted that the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0063] In existing technologies, when calculating the charging location of a target charging pile, it is usually necessary to determine multiple valid offset values ​​involved in the target charging pile through brute-force matching. That is, it is necessary to obtain multiple sets of valid offset values, and then perform calculations such as averaging based on multiple sets of valid offset values. This results in high computational complexity and requires a large amount of computational processing. In order to solve this problem of high computational complexity in existing technologies, this application proposes a robot charging method. The specific application scenario of the robot in the implementation of this application will be explained below.

[0064] Figure 1 Please refer to the schematic diagram of the robot working scenario provided in the embodiments of this application. Figure 1 In this scenario, the following components may be included: robot 100 and target charging station 200.

[0065] The robot 100 can be a humanoid robot, a wheeled robot, or any type of robot that can move in the environment. No specific form is limited here. The robot 100 can be equipped with a radar device for radar detection to determine the approximate location of the target charging station 200.

[0066] The target charging station 200 can be any type of charging station available on the market. The target charging station 200 is used to charge the robot 100. The target charging station 200 can usually be set near a wall, with a groove at the front end. The groove can have a charging interface, and this position can be the charging position for the robot.

[0067] The following section will explain the specific implementation process of the robot charging method provided in the embodiments of this application based on the above application scenarios.

[0068] Figure 2 Please refer to the flowchart illustrating the robot charging method provided in this application embodiment. Figure 2 Robot charging methods include:

[0069] S210: Control the robot to move to the target area where the entrance to the target charging station is located.

[0070] Optionally, the subject executing this method can be a robot, specifically a robot controller on the robot.

[0071] First, after the robot detects that it needs to be charged or receives a charging command, it can be controlled to move to the target area where the entrance to the target charging station is located.

[0072] It should be noted that the target area can be any shape, such as rectangle, sector, ring, etc. There are no specific restrictions here. Specifically, the area can be at the entrance of the target charging station, that is, in front of the target charging station (it can be directly in front, or slightly off).

[0073] To provide a more intuitive explanation of the target area, the scope of the target area will be explained in detail below.

[0074] Figure 3 Please refer to the schematic diagram of the target area provided in the embodiments of this application. Figure 3 , Figure 3In the diagram, target area 310 is represented by a rectangle. 311 is the entrance to the target charging station. Starting from the exit, the area in the direction facing the entrance can be considered target area 310. Figure 3 The example used here is a rectangle, but in actual implementation, it can also be other shapes such as a sector or a ring, and there are no restrictions here.

[0075] Once it is determined that the robot needs to be charged, the robot can be controlled to move to the target area, specifically from its current position to a closer position within the target area 310.

[0076] S220: Determine the outline information of the target charging station based on the radar device installed on the robot.

[0077] Optionally, once the robot reaches the target area, the outline information of the target charging station can be determined based on the radar device installed on the robot.

[0078] It should be noted that once the robot reaches the target area, the determined outline information of the target charging station is relatively accurate with a small error.

[0079] Optionally, the contour information may specifically include the top-view contour of the target charging pile and the relative positional relationship between the target charging pile and the robot. The top-view contour is the approximate shape of the target charging pile obtained from a top-view perspective. For example, if the charging pile is a quadrangular prism and has two small quadrangular prism grooves on the front, then for the top-view contour, the upper edge will have two short lines (representing the position behind the two grooves, specifically the recessed position of the two grooves), and the lower edge will have three short lines (representing the position in front of the grooves, specifically the left and right sides of the grooves and the position between the two grooves where there is no recess).

[0080] To facilitate further explanation of the contour information, the contour information of the target charging pile involved in this application is shown in detail below.

[0081] Figure 4 Please refer to the schematic diagram illustrating the contour information provided in the embodiments of this application. Figure 4 The contour information may include: the top view contour 410 of the target charging pile and the relative positional relationship between the target charging pile and the robot 100. Since the information obtained by the radar is planar information, the relative positional relationship between the target charging pile and the robot 100 can be represented in a planar rectangular coordinate system.

[0082] In a Cartesian coordinate system, robot 100 can be at the origin of the coordinate system, and the top view profile 410 of the target charging pile can be at the corresponding position in the coordinate system and can occupy a certain range. Thus, the relative positional relationship between the target charging pile and robot 100 can be determined based on the coordinate set of the top view profile.

[0083] Optionally, the points that form the outline can be radar points, that is, points obtained based on radar detection.

[0084] S230: Determine the location offset information of the target charging pile based on the contour information.

[0085] The position offset information includes: relative position offset, target depth offset, and target width offset. The relative position offset is the relative position between the robot and the target charging station. The target depth offset is the longitudinal depth of the charging position of the target charging station. The target width offset is the lateral width of the charging position of the target charging station.

[0086] Optionally, after obtaining the contour information based on the radar device, the position offset information of the target charging pile can be further calculated based on the contour information.

[0087] The relative position offset can be the relative position of the robot and the target charging station as a whole. For example, the target charging station can be equivalent to a point, and the distance between the robot and the equivalent point can be determined. Alternatively, the relative position of the robot and each point of the target charging station as a whole can be determined. No specific restrictions are imposed here.

[0088] The target depth offset is specifically the longitudinal depth of the charging position of the target charging pile, which is the depth of the groove in the actual structure of the target charging pile (this depth is the longitudinal depth of the front-to-back distance of the groove, not the depth of the vertical distance).

[0089] The target width offset is specifically the lateral width of the charging position of the target charging pile, which is also the width of the groove in the actual structure of the target charging pile (this width is the lateral width of the left and right distance of the groove).

[0090] The target depth offset and target width offset can both be represented in a Cartesian coordinate system based on the range of depth or width. When calculating the actual depth or width, it can be determined by converting the scale in the coordinate system to a scale in reality.

[0091] S240: Input the position offset information of the target charging pile into the pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot.

[0092] Optionally, the pre-trained deep learning model can be a model trained using deep learning methods. The input of this model can be the three types of position offset information mentioned above, and the output can be the relative positional relationship between the charging position of the target charging pile and the current position of the robot.

[0093] After inputting the location offset information of the target charging pile into the model, the relative positional relationship between the charging position of the target charging pile and the current position of the robot can be obtained.

[0094] S250: Controls the robot to move to the charging location of the target charging station for charging.

[0095] Optionally, after obtaining the relative positional relationship, the robot can be navigated to move from its current position in the target area to the charging position for charging. Specifically, this can be done using methods such as LiDAR navigation or map planning. No specific restrictions are imposed here, as long as the charging position can be reached.

[0096] This application provides a robot charging method that allows the robot to move to a target area where the entrance to a target charging station is located. The robot's radar device determines the outline information of the target charging station. Based on the outline information, the robot's position offset information is determined. This position offset information is then input into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging station and the robot's current position. Finally, the robot is controlled to move to the charging position of the target charging station. By controlling the robot to move to the target area where the entrance to the target charging station is located, the robot can detect more accurate and clear outline information of the target charging station. Based on this outline information, more accurate position offset information can be obtained. Furthermore, the pre-trained deep learning model reduces the amount of position offset information required, allowing the robot to directly obtain the relative positional relationship between the charging position of the target charging station and the robot's current position. This enables the robot to move more accurately to the charging position of the target charging station, reducing the complexity of determining the charging position and improving computational efficiency. This allows for rapid determination of the target charging station's specific location, enabling the robot to reach the charging position more quickly and accurately.

[0097] The following section will explain in detail the specific implementation process of determining position offset information in the robot charging method provided in the embodiments of this application.

[0098] Figure 5 For another schematic diagram of the robot charging method provided in the embodiments of this application, please refer to... Figure 5 If the location information is a relative position offset, the location offset information of the target charging station is determined based on the contour information, including:

[0099] S510: Determine the position of the reference point based on the contour information.

[0100] Optionally, a reference point can be determined based on the top view profile in the outline information of the target charging pile. The reference point can be any edge point in the top view profile, or it can be the overall center point of the top view profile, etc. The reference point can be determined according to actual needs, and the position of the reference point can be obtained.

[0101] S520: Determine the offset between the reference point and the robot's position.

[0102] Optionally, after obtaining the position of the reference point, the offset between the reference point and the robot's position can be determined based on the positions of both points in a Cartesian coordinate system. For example, when the robot's position is the origin, the offset between the reference point and the robot's position can be determined based on the coordinates of the reference point.

[0103] S530: The offset between the reference point's position and the robot's position is used as the relative position offset.

[0104] Optionally, after obtaining the offset between the position of the reference point and the position of the robot, this offset can be used as the relative position offset.

[0105] Optionally, if the location information is a target depth offset, the location offset information of the target charging pile is determined based on the contour information, including: determining the target depth offset of the target charging pile based on the positional deviation between the upper edge and the lower edge in the contour information.

[0106] You can refer to Figure 4 The specific shape of the top-down profile is calculated based on the positional difference between the upper and lower edges to obtain the target depth offset of the target charging pile.

[0107] For example, if the vertical coordinate of the upper edge is 100 (unit length) and the vertical coordinate of the lower edge is 95 (unit length), then the difference of 5 (unit length) is the target depth offset of the target charging pile in the Cartesian coordinate system, which can then be converted into the actual target depth offset of the target charging pile according to the scale.

[0108] Optionally, if the location information is a target width offset, the location offset information of the target charging pile is determined based on the contour information, including: determining the target width offset of the target charging pile based on the width information of the upper edge in the contour information.

[0109] You can refer to Figure 4 The specific shape of the top-down profile is determined, and the target width offset of the target charging pile is determined based on the width information of the upper edge in the profile information.

[0110] For example, if the horizontal coordinate of the left side of the top edge is 1 (unit length) and the horizontal coordinate of the right side of the top edge is 3 (unit length), then the difference between the two, 2 (unit length), is the target width offset of the target charging pile in the Cartesian coordinate system. This can then be converted into the actual target width offset of the target charging pile according to the ratio.

[0111] The following section will explain in detail the specific implementation process of training the deep learning model in the robot charging method provided in this application embodiment.

[0112] Figure 6 For another schematic diagram of the robot charging method provided in the embodiments of this application, please refer to... Figure 6 Before inputting the location offset information of the target charging station into the pre-trained deep learning model, the method further includes:

[0113] S610: Obtain initial sample offset information.

[0114] The sample offset information includes: sample relative position offset, sample target depth offset, and sample target width offset.

[0115] Optionally, the initial sample offset information can be a single sample offset information, wherein each sample offset information can include the sample relative position offset, the sample target depth offset, and the sample target width offset. These offsets are similar to the aforementioned target charging pile position offset information, and will not be elaborated here.

[0116] It should be noted that the initial sample offset information can be obtained by the robot based on the actual working conditions, or it can be manually input by the user based on the actual data, or it can be randomly generated.

[0117] Taking random generation as an example, radar devices typically acquire contour information by scanning radar points. For example, if a radar scan covers 360 radar points, a subset of radar points can be selected as the contour of a sample charging station. For instance, if the contour occupies 10 radar points, the remaining 350 radar points can be randomly generated within the range. Then, the initial sample offset information can be determined based on the selected contour of the sample charging station.

[0118] S620: Perform sample addition processing on the initial sample offset information to obtain a set of sample offset information.

[0119] Optionally, after obtaining the initial sample offset information, sample addition processing can be performed based on the information, thereby obtaining multiple sample offset information from one initial sample offset information. These sample offset information together form the above-mentioned sample offset information set, which includes multiple sample offset information, each of which includes the above three types of offset.

[0120] S630: The initial model is trained based on the sample offset information set to obtain a deep learning model.

[0121] Optionally, after obtaining the sample offset information set, the initial model can be trained based on the sample offset information. Specifically, a deep learning algorithm can be used for training to obtain a deep learning model.

[0122] The following is a detailed explanation of the specific implementation process of the sample addition process in the robot charging method provided in the embodiments of this application.

[0123] Figure 7 For another schematic diagram of the robot charging method provided in the embodiments of this application, please refer to... Figure 7 The initial sample offset information is processed by sample augmentation to obtain a set of sample offset information, including:

[0124] S710: Determine the offset range of sample relative position offset, sample target depth offset, and sample target width offset respectively.

[0125] Optionally, after obtaining the initial sample offset information, the offset ranges of the sample relative position offset, sample target depth offset, and sample target width offset can be determined respectively. Specifically, these ranges can be manually set, determined based on contour information, or randomly generated, etc., without specific restrictions.

[0126] S720: Performs random number generation based on the offset range to obtain multiple sets of sample offset information.

[0127] Optionally, after determining the offset range, random numbers can be generated within the offset range to obtain multiple sets of sample offset information.

[0128] For example, the relative position offset of the sample is in the range of 0-10, the target depth offset of the sample is in the range of 0.2-0.8, and the target width offset of the sample is in the range of 2-4. A set of sample offset information can be randomly generated from the corresponding range, such as (1, 0.3, 2.5). Multiple sets of sample offset information can be obtained in this way.

[0129] S730: The initial sample offset information and the set of multiple sets of sample offset information are used as the sample offset information set.

[0130] Optionally, after obtaining multiple sets of sample offset information, the set of these sample offset information and the initial sample offset information can be used as the sample offset information set.

[0131] In the robot charging method provided in this application embodiment, the offset ranges of sample relative position offset, sample target depth offset, and sample target width offset can be determined respectively; random number generation processing is performed based on the offset range to obtain multiple sets of sample offset information; the initial sample offset information and the set of multiple sets of sample offset information are used as the sample offset information set. By obtaining multiple sample offset information in the above manner, the number of samples can be increased, improving sample saturation. This enhances the diversity of training during subsequent training, thereby obtaining a model that better meets the requirements.

[0132] The following section will explain in detail the specific implementation process of model training in the robot charging method provided in the embodiments of this application.

[0133] Figure 8 For another schematic diagram of the robot charging method provided in the embodiments of this application, please refer to... Figure 8 A deep learning model is obtained by training the initial model based on the sample offset information set, including:

[0134] S810: Input each set of sample offset information in the sample offset information set into the initial model in sequence to obtain the output result, and determine the loss information of the initial model based on the actual charging position of the sample corresponding to the output result and the sample offset information set.

[0135] Optionally, during model training, each set of sample offset information from the sample offset information set can be sequentially input into the initial model to obtain the output result, which can represent the calculated actual charging position. The loss information of the initial model can be calculated by comparing the output result with the actual charging position of the sample corresponding to the set of sample offset information. Specifically, the calculation can be based on a pre-configured loss function, which is not limited here.

[0136] S820: Based on the loss information of the initial model, the parameters of the initial model are corrected, and the corrected initial model is retrained using the sample offset information set until the initial model meets the convergence condition, and the initial model is used as a deep learning model.

[0137] Optionally, after obtaining the loss information of the initial model, the relevant parameters of the initial model can be corrected, and the corrected initial model can be retrained using the sample offset information set until the initial model meets the convergence condition.

[0138] After obtaining the initial model that satisfies the convergence condition, the initial model is used as the deep learning model.

[0139] The following describes the apparatus, equipment, and storage medium used to implement the robot charging method provided in this application. The specific implementation process and technical effects are described above and will not be repeated below.

[0140] Figure 9 Please refer to the structural schematic diagram of the robot charging device provided in the embodiments of this application. Figure 9 The robot charging device includes: a moving module 910, a detection module 920, a determination module 930, and a calculation module 940;

[0141] The mobile module 910 is used to control the robot to move to the target area where the target charging pile entrance is located;

[0142] The detection module 920 is used to determine the outline information of the target charging pile based on the radar device installed on the robot;

[0143] The determination module 930 is used to determine the position offset information of the target charging pile based on the contour information. The position offset information includes: relative position offset, target depth offset and target width offset. The relative position offset is the relative position between the robot and the target charging pile. The target depth offset is the longitudinal depth of the charging position of the target charging pile. The target width offset is the lateral width of the charging position of the target charging pile.

[0144] The calculation module 940 is used to input the position offset information of the target charging pile into the pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot.

[0145] The mobile module 910 is also used to control the robot to move to the charging location of the target charging station for charging.

[0146] Optionally, the determining module 930 is specifically used to determine the position of the reference point based on the contour information; determine the offset between the position of the reference point and the position of the robot; and use the offset between the position of the reference point and the position of the robot as the relative position offset.

[0147] Optionally, the determining module 930 is further used to determine the target depth offset of the target charging pile based on the positional deviation between the upper edge and the lower edge in the contour information.

[0148] Optionally, the determining module 930 is further used to determine the target width offset of the target charging pile based on the width information of the upper edge in the contour information.

[0149] Optionally, the calculation module 940 is also used to obtain initial sample offset information, which includes: sample relative position offset, sample target depth offset, and sample target width offset; to perform sample addition processing on the initial sample offset information to obtain a sample offset information set; and to train the initial model based on the sample offset information set to obtain a deep learning model.

[0150] Optionally, the calculation module 940 is specifically used to determine the offset range of the sample relative position offset, the sample target depth offset, and the sample target width offset, respectively; perform random number generation processing based on the offset range to obtain multiple sets of sample offset information; and use the initial sample offset information and the set of multiple sets of sample offset information as the sample offset information set.

[0151] Optionally, the calculation module 940 is specifically used to input each set of sample offset information in the sample offset information set into the initial model in sequence to obtain the output result, and determine the loss information of the initial model based on the actual charging position of the sample corresponding to the output result and the set of sample offset information; correct the parameters of the initial model based on the loss information of the initial model, and retrain the corrected initial model using the sample offset information set until the initial model meets the convergence condition, and use the initial model as a deep learning model.

[0152] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0153] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0154] Figure 10 Please refer to the schematic diagram of the computer device provided in the embodiments of this application. Figure 10A computer device is provided, including: a memory 950 and a processor 960. The memory 950 stores a computer program that can run on the processor 960. When the processor 960 executes the computer program, it implements the steps of a robot charging method.

[0155] Alternatively, the computer device may specifically be a robot controller for a robot.

[0156] Another aspect of this application embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a robot charging method.

[0157] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0158] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0159] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0160] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute partial steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0161] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0162] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for charging a robot, characterized in that, include: Control the robot to move to the target area where the entrance to the target charging station is located; The outline information of the target charging station is determined based on the radar device installed on the robot; Based on the contour information, the position offset information of the target charging pile is determined. The position offset information includes: relative position offset, target depth offset, and target width offset. The relative position offset is the relative position between the robot and the target charging pile. The target depth offset is the longitudinal depth of the charging position of the target charging pile. The target width offset is the lateral width of the charging position of the target charging pile. The position offset information of the target charging pile is input into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot. Control the robot to move to the charging location of the target charging station to charge; Before inputting the location offset information of the target charging pile into the pre-trained deep learning model, the method further includes: Obtain initial sample offset information, which includes: sample relative position offset, sample target depth offset, and sample target width offset; The initial sample offset information is subjected to sample increment processing to obtain a sample offset information set; The deep learning model is obtained by training the initial model based on the sample offset information set; The initial sample offset information is processed by sample augmentation to obtain a set of sample offset information, including: Determine the offset ranges for the relative position offset of the sample, the depth offset of the sample target, and the width offset of the sample target, respectively; Random number generation is performed based on the offset range to obtain multiple sets of sample offset information; The initial sample offset information and the set of multiple sets of sample offset information are used as the sample offset information set.

2. The robot charging method as described in claim 1, characterized in that, Determining the position offset information of the target charging pile based on the contour information includes: The position of the reference point is determined based on the contour information; Determine the offset between the position of the reference point and the position of the robot; The offset between the position of the reference point and the position of the robot is used as the relative position offset.

3. The robot charging method as described in claim 1, characterized in that, Determining the position offset information of the target charging pile based on the contour information includes: Based on the positional deviation between the upper and lower edges in the contour information, the target depth offset of the target charging pile is determined.

4. The robot charging method as described in claim 1, characterized in that, Determining the position offset information of the target charging pile based on the contour information includes: The target width offset of the target charging pile is determined based on the width information of the upper edge in the contour information.

5. The robot charging method as described in claim 1, characterized in that, The process of training the initial model based on the sample offset information set to obtain the deep learning model includes: Each set of sample offset information in the sample offset information set is sequentially input into the initial model to obtain the output result, and the loss information of the initial model is determined based on the output result and the actual charging position of the sample corresponding to the set of sample offset information. Based on the loss information of the initial model, the parameters of the initial model are corrected, and the corrected initial model is retrained using the sample offset information set until the initial model meets the convergence condition, and the initial model is used as the deep learning model.

6. A robot charging device, characterized in that, include: Movement module, detection module, determination module, calculation module; The mobile module is used to control the robot to move to the target area where the entrance to the target charging station is located; The detection module is used to determine the outline information of the target charging pile based on the radar device installed on the robot; The determining module is used to determine the position offset information of the target charging pile based on the contour information. The position offset information includes: relative position offset, target depth offset, and target width offset. The relative position offset is the relative position of the robot and the target charging pile. The target depth offset is the longitudinal depth of the charging position of the target charging pile. The target width offset is the lateral width of the charging position of the target charging pile. The calculation module is used to input the position offset information of the target charging pile into a pre-trained deep learning model to obtain the relative positional relationship between the charging position of the target charging pile and the current position of the robot. The mobile module is also used to control the robot to move to the charging position of the target charging pile for charging; The computing module is also used for: Obtain initial sample offset information, which includes: sample relative position offset, sample target depth offset, and sample target width offset; The initial sample offset information is subjected to sample increment processing to obtain a sample offset information set; The deep learning model is obtained by training the initial model based on the sample offset information set; The computing module is also used for: Determine the offset ranges for the relative position offset of the sample, the depth offset of the sample target, and the width offset of the sample target, respectively; Random number generation is performed based on the offset range to obtain multiple sets of sample offset information; The initial sample offset information and the set of multiple sets of sample offset information are used as the sample offset information set.

7. A computer device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program that can run on the processor, and when the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1 to 5.

Images