Target recognition methods, devices, robots, and storage media

Abstract

This invention discloses a target recognition method, apparatus, robot, and storage medium, applied to robots. The method includes: performing target recognition based on environmental data collected by the robot to obtain candidate targets that match a preset target, and recording the recognition parameters of the candidate targets; performing a corresponding parameter matching score between the recognition parameters and preset expected parameters to obtain the matching degree between the candidate targets and the preset target; and determining the recognition result based on the matching degree between the candidate targets and the preset target. This invention improves the robustness of robots in complex environments and enhances the accuracy of target recognition.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a target recognition method, apparatus, robot, and storage medium. Background Technology

[0002] Target identification and path planning are fundamental functions of robots. When identifying targets, robots often rely on specific shape features. For example, charging stations and smart vending machines are designed with V-shaped or circular outlines. Existing solutions typically assume that only one target in the environment conforms to that specific shape feature. However, in reality, multiple targets or objects similar to that shape feature often exist in the environment, interfering with the robot's work and leading to low recognition accuracy. Summary of the Invention

[0003] The main objective of this invention is to provide a target recognition method, device, robot, and storage medium, aiming to solve the technical problem of how to improve the accuracy of target recognition in robots.

[0004] To achieve the above objectives, the present invention provides a target recognition method applied to a robot, the method comprising:

[0005] Based on the environmental data collected by the robot, target recognition is performed to obtain candidate targets that match the preset target, and the recognition parameters of the candidate targets are recorded. The recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system.

[0006] The identification parameters are matched and scored with the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system.

[0007] The identification result is determined based on the matching degree between the candidate target and the preset target.

[0008] Optionally, the step of performing target recognition based on the environmental data collected by the robot to obtain candidate targets that match a preset target includes:

[0009] Based on the environmental data collected by the robot's sensing device, target recognition is performed to obtain an initial target that matches the preset shape of the target.

[0010] If the distance between the initially identified target and each of the identified candidate targets in the preset map coordinate system is greater than a preset threshold, then the initially identified target will be added as a candidate target.

[0011] Optionally, the step of determining the recognition result based on the matching degree between the candidate target and the preset target includes:

[0012] Based on the matching degree between at least one candidate target identified within the target recognition period and the preset target, an initial recognition result for the preset target within the target recognition period is obtained, wherein the initial recognition result includes the initial recognition position and initial recognition posture of the preset target relative to the robot coordinate system.

[0013] Based on the recognition results of the preset target in the previous recognition cycle and the robot's driving data in the target recognition cycle, a prediction result for the preset target in the target recognition cycle is calculated, wherein the prediction result includes the predicted position and predicted attitude of the preset target relative to the robot coordinate system.

[0014] The initial recognition result and the estimated result for the preset target within the target recognition period are fused according to a preset weight to obtain the recognition result for the preset target within the target recognition period.

[0015] Optionally, the step of obtaining the initial recognition result for the preset target within the target recognition period based on the matching degree between at least one candidate target identified within the target recognition period and the preset target includes:

[0016] Based on the matching degree between at least one candidate target identified within the target identification period and the preset target, a matching target that matches the preset target is selected from each candidate target;

[0017] The position and orientation of the identified matching target relative to the robot coordinate system are used as the initial recognition result;

[0018] The step of fusing the initial recognition result and the estimated result for the preset target within the target recognition period according to a preset weight to obtain the recognition result for the preset target within the target recognition period includes:

[0019] Based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension, the fusion weight corresponding to the initial recognition result is determined. The higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result.

[0020] The initial recognition result for the preset target within the target recognition period is fused with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

[0021] Optionally, the position of the candidate target relative to the preset coordinate system is represented by the first position of a first position point on a predetermined shape of the candidate target in the preset coordinate system, and the preset position is represented by the second position of a second position point on the preset target in the preset coordinate system;

[0022] The step of matching and scoring the identification parameters with the preset expected parameters includes:

[0023] Calculate the distance between the first position and the second position, and obtain a parameter matching score between the recognition parameters and the preset expected parameters in the position dimension based on the distance.

[0024] Optionally, the attitude of the candidate target relative to the preset coordinate system is represented by a direction line representing the orientation of the candidate target at a third position in the preset coordinate system, and the preset attitude is represented by a direction line representing the orientation of the preset target at a fourth position in the preset coordinate system.

[0025] The step of matching and scoring the identification parameters with the preset expected parameters includes:

[0026] Based on the third and fourth positions, the angle between the direction line representing the orientation of the candidate target and the direction line representing the orientation of the preset target is calculated, and the parameter matching score between the identification parameters and the preset expected parameters in the orientation dimension is obtained according to the angle.

[0027] Optionally, the step of matching and scoring the identification parameters with preset expected parameters includes:

[0028] Calculate the difference between the shape parameters of the candidate target and the preset shape parameters, and obtain a parameter matching score between the identification parameters and the preset expected parameters in the shape dimension based on the difference.

[0029] To achieve the above objectives, the present invention also provides a target recognition device, which is deployed on a robot, and the device includes:

[0030] The recognition module is used to perform target recognition based on the environmental data collected by the robot, obtain candidate targets that match the preset target, and record the recognition parameters of the candidate targets, wherein the recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system;

[0031] The scoring module is used to perform corresponding parameter matching scoring between the identification parameters and the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system.

[0032] The determination module is used to determine the recognition result based on the matching degree between the candidate target and the preset target.

[0033] To achieve the above objectives, the present invention also provides a robot, the robot comprising: a memory, a processor, and a target recognition program stored in the memory and executable on the processor, wherein the target recognition program, when executed by the processor, implements the steps of the target recognition method as described above.

[0034] Furthermore, to achieve the above objectives, the present invention also proposes a computer-readable storage medium storing a target recognition program, which, when executed by a processor, implements the steps of the target recognition method as described above.

[0035] In this invention, a robot performs target recognition based on collected environmental data to obtain candidate targets that match a preset target. The recognition parameters of these candidate targets are recorded, including shape parameters and position and orientation relative to a preset coordinate system. The recognition parameters are then matched and scored against preset expected parameters to obtain the matching degree between the candidate targets and the preset target. The preset expected parameters include preset shape parameters and preset position and orientation relative to the preset coordinate system. Based on the matching degree between the candidate targets and the preset target, the recognition result is determined. This invention implements a robot target recognition scheme that scores candidate targets from three dimensions: shape, distance, and orientation, thereby obtaining the matching degree between the candidate targets and the preset target. The recognition result for the preset target is derived based on the matching degree of each candidate target, improving the robot's robustness in complex environments and increasing the accuracy of target recognition. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the hardware operating environment involved in the embodiments of the present invention;

[0037] Figure 2 This is a flowchart illustrating the first embodiment of the target recognition method of the present invention;

[0038] Figure 3 This is a schematic diagram of a charging pile with a V-shape according to an embodiment of the present invention;

[0039] Figure 4 This is a schematic diagram of a charging pile with an arc shape according to an embodiment of the present invention;

[0040] Figure 5 This is a schematic diagram of the functional modules of a preferred embodiment of the target recognition device of the present invention.

[0041] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0042] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0043] like Figure 1 As shown, Figure 1 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of the present invention.

[0044] It should be noted that the device in the embodiments of the present invention can be a smartphone, a personal computer or a server or other device with data processing capabilities. The device can be deployed in a mobile robot, and no specific limitation is made here.

[0045] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU, a memory 1002, and a communication bus 1003. The communication bus 1003 is used to enable communication between these components. The memory 1002 may be a high-speed RAM or a stable, non-volatile memory, such as a disk drive. Optionally, the memory 1002 may also be a storage device independent of the aforementioned processor 1001.

[0046] Those skilled in the art will understand that Figure 1 The device structure shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0047] like Figure 1 As shown, the memory 1002, as a computer storage medium, may include an operating system and a target recognition program. The operating system is a program that manages and controls the device's hardware and software resources, supporting the operation of the target recognition program and other software or programs. Figure 1 In the device shown, the processor 1001 can be used to call the target identification program stored in the memory 1002 and perform the following operations:

[0048] Based on the environmental data collected by the robot, target recognition is performed to obtain candidate targets that match the preset target, and the recognition parameters of the candidate targets are recorded. The recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system.

[0049] The identification parameters are matched and scored with the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system.

[0050] The identification result is determined based on the matching degree between the candidate target and the preset target.

[0051] Furthermore, the operation of identifying candidate targets that match preset targets based on the environmental data collected by the robot includes:

[0052] Based on the environmental data collected by the robot's sensing device, target recognition is performed to obtain an initial target that matches the preset shape of the target.

[0053] If the distance between the initially identified target and each of the identified candidate targets in the preset map coordinate system is greater than a preset threshold, then the initially identified target will be added as a candidate target.

[0054] Furthermore, the operation of determining the recognition result based on the matching degree between the candidate target and the preset target includes:

[0055] Based on the matching degree between at least one candidate target identified within the target recognition period and the preset target, an initial recognition result for the preset target within the target recognition period is obtained, wherein the initial recognition result includes the initial recognition position and initial recognition posture of the preset target relative to the robot coordinate system.

[0056] Based on the recognition results of the preset target in the previous recognition cycle and the robot's driving data in the target recognition cycle, a prediction result for the preset target in the target recognition cycle is calculated, wherein the prediction result includes the predicted position and predicted attitude of the preset target relative to the robot coordinate system.

[0057] The initial recognition result and the estimated result for the preset target within the target recognition period are fused according to a preset weight to obtain the recognition result for the preset target within the target recognition period.

[0058] Further, the operation of obtaining the initial recognition result for the preset target within the target recognition period based on the matching degree between at least one candidate target identified within the target recognition period and the preset target includes:

[0059] Based on the matching degree between at least one candidate target identified within the target identification period and the preset target, a matching target that matches the preset target is selected from each candidate target;

[0060] The position and orientation of the identified matching target relative to the robot coordinate system are used as the initial recognition result;

[0061] The operation of fusing the initial recognition result and the estimated result for the preset target within the target recognition period according to a preset weight to obtain the recognition result for the preset target within the target recognition period includes:

[0062] Based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension, the fusion weight corresponding to the initial recognition result is determined. The higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result.

[0063] The initial recognition result for the preset target within the target recognition period is fused with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

[0064] Furthermore, the position of the candidate target relative to the preset coordinate system is represented by the first position of a first position point on a predetermined shape of the candidate target in the preset coordinate system, and the preset position is represented by the second position of a second position point on the preset target in the preset coordinate system;

[0065] The operation of matching and scoring the identified parameters with preset expected parameters includes:

[0066] Calculate the distance between the first position and the second position, and obtain a parameter matching score between the recognition parameters and the preset expected parameters in the position dimension based on the distance.

[0067] Furthermore, the attitude of the candidate target relative to the preset coordinate system is represented by a direction line representing the orientation of the candidate target at a third position in the preset coordinate system, and the preset attitude is represented by a direction line representing the orientation of the preset target at a fourth position in the preset coordinate system.

[0068] The operation of matching and scoring the identified parameters with preset expected parameters includes:

[0069] Based on the third and fourth positions, the angle between the direction line representing the orientation of the candidate target and the direction line representing the orientation of the preset target is calculated, and the parameter matching score between the identification parameters and the preset expected parameters in the orientation dimension is obtained according to the angle.

[0070] Furthermore, the operation of matching and scoring the identified parameters with preset expected parameters includes:

[0071] Calculate the difference between the shape parameters of the candidate target and the preset shape parameters, and obtain a parameter matching score between the identification parameters and the preset expected parameters in the shape dimension based on the difference.

[0072] Based on the above structure, various embodiments of the target recognition method of the present invention are proposed.

[0073] Reference Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the target recognition method of the present invention.

[0074] This invention provides embodiments of a target recognition method. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order. The executing entity of each embodiment of the target recognition method of this invention can be a robot. The robot can be a conventional robot controlled by an automatic control program. The embodiments do not limit the type or style of the robot or its specific implementation details. In this embodiment, the target recognition method includes the following steps S10-S30:

[0075] Step S10: Based on the environmental data collected by the robot, target recognition is performed to obtain candidate targets that match the preset target, and the recognition parameters of the candidate targets are recorded. The recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system.

[0076] The robot can be equipped with sensing devices, and the data collected by the sensing devices is used for target recognition. In this embodiment, the type of sensing device is not limited, and any sensing device that can achieve target recognition can be used, such as LiDAR, depth camera and other sensing devices. The robot can perform target recognition based on LiDAR data.

[0077] The preset target can be a physical object that the robot needs to identify, such as a charging pile or a smart vending machine. The preset target can have a specific shape outline structure, so that the preset target has a specific shape (i.e., a set shape). This set shape is used to distinguish other interfering objects. For example, a V-shaped outline structure can be set on the preset target to make the preset target have a V shape, or an arc-shaped outline structure can be set on the preset target to make the preset target have an arc shape.

[0078] The set shape enables the target's attitude relative to the robot coordinate system to be determined based on the position of the set shape in the robot coordinate system. In this embodiment, the set shape is not limited and any set shape that can meet the requirements can be used.

[0079] For example, in one feasible embodiment, the set shape can be set to any shape that satisfies the following condition: based on the position of the set shape in the robot coordinate system of the identified target, a direction line representing the orientation of the target in the robot coordinate system can be uniquely determined. When the set shape is determined, this embodiment does not limit the contour structure that gives the preset target the set shape. For example, in one feasible embodiment, taking a charging pile as an example, the charging pile can be equipped with... Figure 3 The V-shaped outline structure shown gives the charging station a V-shape. When the robot identifies the charging station, it can determine the position of the V-shape in the robot coordinate system, such as the positions of the two sides of the V-shape. Based on these positions, the position of the angle bisector of the V-shape in the robot coordinate system can be determined. This angle bisector can be used to indicate the orientation of the charging station in the robot coordinate system. For example, in one feasible embodiment, taking the charging station as an example, the charging station can be equipped with... Figure 4 The arc-shaped contour structure shown gives the charging station an arc shape. When the robot identifies the charging station, it can identify the position of the arc shape in the robot coordinate system, such as the position of the two ends of the arc shape in the robot coordinate system. Based on this position, the position of the perpendicular bisector of the two ends of the arc shape in the robot coordinate system can be determined. The direction of the perpendicular bisector from the midpoint of the arc to the center of the circle can be used to indicate the orientation of the charging station in the robot coordinate system.

[0080] The robot performs target recognition based on data collected by sensing devices. It can identify targets with a set shape that conforms to a preset target as candidate targets. The robot can identify at least one candidate target based on single-frame or multi-frame data collected by sensing devices as a candidate target group, and determine the recognition result for the preset target based on the candidate target group.

[0081] For each identified candidate target, the robot can record its identification parameters. These parameters may include shape parameters characterizing the candidate target, as well as its position and orientation relative to a preset coordinate system. The preset coordinate system can be a pre-defined coordinate system, such as a robot coordinate system, a LiDAR coordinate system, or a map coordinate system; this embodiment is not limited to any particular system. In one feasible implementation, the orientation of the candidate target relative to a certain coordinate system can be represented by attitude angles.

[0082] This embodiment does not limit the specific implementation method of the robot to obtain candidate targets by identifying targets based on the data collected by the sensing device; conventional target recognition algorithms can be used to achieve this.

[0083] In a specific implementation, the target recognition algorithm used by the robot first obtains the position and orientation of the candidate target relative to the robot coordinate system. When the preset coordinate system is the map coordinate system, the robot can combine the robot's current position and orientation in the map coordinate system to transform the position and orientation of the candidate target relative to the robot coordinate system into the map coordinate system.

[0084] In one feasible implementation, step S10 may include S101-S102:

[0085] Step S101: Based on the environmental data collected by the robot's sensing device, target recognition is performed to obtain an initial target that matches the preset shape of the target.

[0086] Step S102: If the distance between the initially identified target and each of the identified candidate targets in the preset map coordinate system is greater than a preset threshold, then the initially identified target is added as a candidate target.

[0087] In this embodiment, the robot identifies targets based on data collected by the sensing device. Targets with a predetermined shape that conform to a preset target are initially identified. Then, the distance between the initial identified target and the identified candidate targets in a preset map coordinate system is calculated, and this distance is compared with a preset threshold. There may be one or more identified candidate targets, such as those identified by the robot within a recognition cycle. The recognition cycle can be defined as the time span corresponding to multiple frames of data collected by the sensing device, for example, 3 seconds per recognition cycle. Since single-frame data collected by the robot's sensing device may contain noise or be accidentally occluded, resulting in the failure to identify candidate targets based on a single frame, in this specific embodiment, multiple frames of data within the recognition cycle are identified separately, and the recognition result for the preset target is obtained based on the candidate targets identified in each frame, thereby improving the recognition accuracy of the preset target.

[0088] When performing recognition based on multi-frame data, the same target may appear in each frame. To avoid generating too many duplicate candidate targets, in specific implementations, duplicate candidate targets can be deduplicated. There are various deduplication methods. For example, in this implementation, the distance between the initially identified target and the already identified candidate targets in a preset map coordinate system is calculated and compared with a preset threshold. For example, if the distance between the initially identified target and each of the already identified candidate targets in the preset map coordinate system is greater than the preset threshold, the initially identified target is added as a candidate target; otherwise, it is determined to be a duplicate and discarded. This process is used to determine whether the initially identified target is a candidate target that has already been identified. Only when it is determined that there is no duplicate is the initially identified target included as one of the candidate targets, thus achieving the purpose of deduplication.

[0089] It is understandable that, since the robot may move or adjust its posture during the process of recognizing the preset target, this embodiment calculates the distance between the initially recognized target and the identified candidate target in the preset map coordinate system. If the distance between the initially recognized target and the identified candidate target in the preset map coordinate system is too close (less than a preset threshold), it indicates that the initially recognized target and the candidate target are very likely the same target. The specific value of the preset threshold can be set as needed and is not limited in this embodiment. In a feasible embodiment, the preset threshold can be set according to the size of the preset target; for example, half the width of the preset target can be used as the preset threshold.

[0090] Step S20: Match the identification parameters with the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system.

[0091] The desired parameters (referred to as preset desired parameters) corresponding to the preset target can be set in advance or obtained from other devices. The preset desired parameters may include the shape parameters of the preset target (referred to as preset shape parameters for distinction), as well as the position and orientation of the preset target relative to the preset coordinate system (referred to as preset position and preset orientation for distinction).

[0092] It is understandable that the position and orientation parameters of the candidate target are relative to the preset position and orientation of the preset target, which are in the same coordinate system, i.e., the preset coordinate system. If the candidate target is the preset target, then the matching degree between the shape parameter of the candidate target and the preset shape parameter of the preset target is high, the matching degree between the position of the candidate target relative to the preset coordinate system and the preset position of the preset target is high, and the matching degree between the orientation of the candidate target relative to the preset coordinate system and the preset orientation of the preset target is also high. Therefore, in this embodiment, the matching degree between the candidate target and the preset target can be calculated based on the parameter matching scores between the identification parameters and the preset expected parameters in the three dimensions of position, orientation, and shape. That is, the parameter matching score between the shape parameter of the candidate target and the shape parameter of the preset expected parameters is calculated, the parameter matching score between the position of the candidate target and the preset position of the preset expected parameters is calculated, and the parameter matching score between the orientation of the candidate target and the preset orientation of the preset expected parameters is calculated. Then, the matching degree between the candidate target and the preset target is calculated based on the three parameter matching scores. There are many ways to calculate parameter matching scores, and this embodiment does not impose any restrictions. For example, it can be to calculate the differences between parameters, or it can be to normalize after calculating the differences to obtain the score.

[0093] In one feasible implementation, after the robot travels to the target location, it can perform target recognition at that location to obtain candidate targets. The target location can be a position within a certain range around the preset target. Further, in another feasible implementation, when the preset coordinate system is the robot coordinate system, the robot can further adjust to the target posture after reaching the target location. At this location and posture, target recognition is performed to obtain candidate targets. In this case, the position and posture in the candidate target's recognition parameters are relative to the robot coordinate system when the robot is at the target location and posture. Correspondingly, the preset position and preset posture are also set as the position and posture of the preset target relative to the robot coordinate system when the robot is at the target location and posture.

[0094] Step S30: Determine the recognition result based on the matching degree between the candidate target and the preset target.

[0095] The recognition result of the preset target may include a result indicating whether the preset target has been recognized or not. Alternatively, if the preset target has been recognized, it may further include a result indicating the location and orientation of the preset target. The location and orientation may be relative to the robot coordinate system, map coordinate system or other coordinate system. The specific settings may be based on the robot's purpose for recognizing the preset target, and are not limited in this embodiment.

[0096] In this embodiment, the robot can obtain the recognition result of the preset target based on the matching degree between at least one candidate target and the preset target. For example, it can select the candidate target with the highest matching degree with the preset target, take the candidate target as the preset target, and take the position and posture of the candidate target as the position and posture of the recognized preset target. Alternatively, it can compare the matching degree of the candidate target with a threshold. Only if the matching degree is greater than the threshold is it determined that the preset target has been recognized. That is, only if the matching degree is greater than the threshold is the candidate target taken as the preset target, which indicates that the preset target has been recognized.

[0097] In this embodiment, the robot performs target recognition based on collected environmental data to obtain candidate targets that match a preset target, and records the recognition parameters of the candidate targets. These parameters include the shape parameters of the candidate targets, and their position and orientation relative to a preset coordinate system. The recognition parameters are then matched and scored against preset expected parameters to obtain the matching degree between the candidate targets and the preset target. The preset expected parameters include preset shape parameters of the preset target, and preset position and orientation relative to the preset coordinate system. Based on the matching degree between the candidate targets and the preset target, the recognition result is determined. This invention implements a robot target recognition scheme that scores candidate targets from three dimensions: shape, distance, and direction, thereby obtaining the matching degree between the candidate targets and the preset target. The recognition result for the preset target is derived based on the matching degree corresponding to each candidate target, improving the robot's robustness in complex environments and increasing the accuracy of target recognition.

[0098] Furthermore, based on the first embodiment described above, a second embodiment of the target recognition method of the present invention is proposed. In this embodiment, step S30, which determines the recognition result based on the matching degree between the candidate target and the preset target, includes steps S301-S303:

[0099] Step S301: Based on the matching degree between at least one candidate target identified within the target recognition period and the preset target, an initial recognition result for the preset target within the target recognition period is obtained, wherein the initial recognition result includes the initial recognition position and initial recognition posture of the preset target relative to the robot coordinate system.

[0100] Because the robot may need to continuously identify the preset target while approaching it—for example, during the docking process—the sensor installation location and the charging port location may be on the front or back of the robot, or the robot may need to rotate significantly due to motion planning strategies. This necessitates the robot continuously identifying the charging pile's location during the docking process to ensure accurate placement. During this period, due to noise or motion-induced distortion, the preset target's position and orientation may be missed for several frames, have errors, or fluctuate in value, resulting in the robot failing to dock with the preset target or experiencing excessively long docking times.

[0101] To address the aforementioned issues, in this embodiment, the robot's driving data can be combined to estimate the current position and orientation of the preset target based on the previously identified position and orientation. The estimated position and orientation are then fused with the currently identified position and orientation of the preset target according to a certain weight, thereby achieving a smoothing effect and preventing the robot from failing to dock with the preset target or from docking for too long.

[0102] In a specific implementation, the robot can identify a preset target according to a recognition cycle. That is, each recognition cycle yields one recognition result for the preset target. This recognition result includes the position and orientation of the preset target relative to the robot's coordinate system within that recognition cycle (hereinafter referred to as recognition position and recognition orientation for distinction). The division of the recognition cycle is not limited here. The following description uses one recognition cycle as an example, and for distinction, this recognition cycle is referred to as the target recognition cycle.

[0103] The robot identifies at least one candidate target within a target recognition cycle and calculates the matching degree between each candidate target and a preset target. Based on this matching degree, it obtains an initial recognition result for the preset target within the target recognition cycle. The initial recognition result includes the position and orientation of the preset target relative to the robot's coordinate system (hereinafter referred to as initial recognition position and initial recognition orientation for distinction). The specific implementation method for obtaining the initial recognition result based on the matching degree is not limited. For example, the candidate target with the highest matching degree to the preset target can be used as the preset target, and the position and orientation of this identified candidate target relative to the robot's coordinate system can be used as the initial recognition result for the preset target within the target recognition cycle.

[0104] Step S302: Based on the recognition result of the preset target in the previous recognition cycle of the target recognition cycle and the driving data of the robot in the target recognition cycle, calculate the estimated result of the preset target in the target recognition cycle, wherein the estimated result includes the estimated position and estimated attitude of the preset target relative to the robot coordinate system.

[0105] The robot can acquire the recognition results for a preset target in the previous recognition cycle. It's understood that if the previous recognition cycle is not the first, then the recognition results for the preset target in the previous cycle are obtained by smoothing the initial recognition results for the preset target in the previous cycle. The robot can acquire driving data within the target recognition cycle. In a specific implementation, the driving data can include the robot's displacement and attitude changes in various directions. Based on the recognition results in the previous recognition cycle and the driving data in the target recognition cycle, the robot can calculate the expected position and attitude of the preset target relative to the robot's coordinate system after the robot's movement within the target recognition cycle (hereinafter referred to as estimated position and estimated attitude for distinction).

[0106] Step S303: The initial recognition result and the estimated result for the preset target within the target recognition period are fused according to a preset weight to obtain the recognition result for the preset target within the target recognition period.

[0107] The robot can fuse the initial recognition result and the estimated result for a preset target within a target recognition period according to preset weights. These preset weights can be set in advance as needed and may include the weights corresponding to the initial recognition result and the estimated result; the sum of these two weights is 1. During fusion, the initial recognition position and the estimated position are weighted and summed according to the preset weights to obtain the recognition position of the preset target within the target recognition period, and the initial recognition posture and the estimated posture are weighted and summed according to the preset weights to obtain the recognition posture of the preset target within the target recognition period.

[0108] In one feasible embodiment, step S301, obtaining the initial recognition result for the preset target within the target recognition period based on the matching degree between at least one candidate target identified within the target recognition period and the preset target, includes steps S3011-S3012. Step S303, fusing the initial recognition result and the estimated result for the preset target within the target recognition period according to a preset weight, to obtain the recognition result for the preset target within the target recognition period, includes steps S3031-S3032:

[0109] Step S3011: Based on the matching degree between at least one candidate target identified within the target identification period and the preset target, select a matching target that matches the preset target from among the candidate targets.

[0110] Matching with a preset target can mean that the matching degree between the candidate target and the preset target meets certain preset conditions. For example, the matching target that matches the preset target can be the candidate target with the highest matching degree. However, it should be noted that in some feasible implementations, it is not necessary to use the candidate target with the highest matching degree as the matching target. For example, it is possible to select candidate targets with a higher matching degree and a matching score of the position dimension parameter greater than a certain threshold as the matching target as needed.

[0111] Step S3012: The position and orientation of the identified matching target relative to the robot coordinate system are taken as the initial recognition result.

[0112] The robot can use the position and orientation of the matching target relative to the robot's coordinate system, which are obtained when performing target recognition based on data from sensing devices, as the initial recognition result.

[0113] Step S3031: Determine the fusion weight corresponding to the initial recognition result based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension. The higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result.

[0114] Step S3032: The initial recognition result for the preset target within the target recognition period is fused with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

[0115] When the matching score between the target's identification parameters and the preset expected parameters in the shape dimension is low, it indicates that the target has undergone severe distortion. The initial identification result determined based on this target is unsuitable as the identification result for the preset target. Conversely, when the matching score between the target's identification parameters and the preset expected parameters in the shape dimension is high, it indicates that the target has not undergone distortion or has not undergone severe distortion. The initial identification result determined based on this target is suitable as the identification result for the preset target. Therefore, in this embodiment, the fusion weight corresponding to the initial identification result is determined based on the matching score of the target in the shape dimension. Furthermore, the higher the matching score of the target in the shape dimension, the higher the fusion weight corresponding to the initial identification result, thereby improving the identification accuracy of the preset target. The fusion weight corresponding to the predicted result is 1 minus the fusion weight corresponding to the initial identification result.

[0116] Furthermore, based on the first and / or second embodiments described above, a third embodiment of the target recognition method of the present invention is proposed. In this embodiment, the recognition parameters are matched and scored with preset expected parameters. For example, the robot can calculate the difference between the shape parameter in the recognition parameters of the candidate target and the preset shape parameter in the preset expected parameters, and obtain a parameter matching score for the shape dimension based on the difference (either directly as a parameter or through mapping). The robot can also calculate the difference between the pose in the recognition parameters of the candidate target and the preset pose in the preset expected parameters, and obtain a parameter matching score for the pose dimension based on the difference (either directly as a parameter or through mapping). Finally, the robot can calculate the difference between the position in the recognition parameters of the candidate target and the preset position in the preset expected parameters, and obtain a parameter matching score for the position dimension based on the difference (either directly as a parameter or through mapping).

[0117] In one feasible implementation, the position of a candidate target relative to a preset coordinate system can be represented by the position of a first position point on a defined shape within the candidate target in the preset coordinate system (hereinafter referred to as the first position for distinction). The preset position of the preset target relative to the preset coordinate system can be represented by the position of a second position point on the preset target in the preset coordinate system (hereinafter referred to as the second position for distinction). It should be noted that when performing target recognition based on data collected by a sensing device, the positions of each point on the defined shape within the candidate target in the preset coordinate system can be identified. It is precisely because the shape features formed by connecting the identified points have a certain similarity to the defined shape that the candidate target can be recognized by the target recognition algorithm. A specific point (the first position point) on the defined shape within the candidate target can be preset to represent the candidate target, but the specific point is not limited; a point with a uniform position can be used for all candidate targets. For example, when the defined shape is a V-shape, the vertex of the V can be used to represent the candidate target; similarly, when the defined shape is an arc shape, the center of the arc or the center of the circle can be used to represent the candidate target. A pre-defined point (second position point) on a preset target can be used to represent the preset target. The second position point and the first position point correspond to the same point on a predetermined shape, such as the vertex of a V-shape. The robot can calculate the distance between the first and second positions and obtain a parameter matching score in the position dimension between the recognition parameters and the preset expected parameters based on this distance. In a specific implementation, this distance can be directly used as the parameter matching score in the position dimension; alternatively, the distance can be normalized, and the normalized result can be used as the parameter matching score. Normalization ensures that small defects in a single dimension do not have a significant absolute impact on the final calculated matching degree. The normalization method is not limited here; for example, an exponential function exp(-x) can be used to transform the data (e.g., distance) to be normalized to the (0,1] interval to achieve normalization.

[0118] In one feasible implementation, the attitude of the candidate target relative to a preset coordinate system can be represented by the position of a direction line representing the orientation of the candidate target in the preset coordinate system (hereinafter referred to as the third position for distinction). The preset attitude can be represented by the fourth position of the direction line representing the orientation of the preset target in the preset coordinate system. In a specific implementation, the attitude of the candidate target can be represented by the attitude of a predetermined shape within the candidate target. In this case, the direction line representing the orientation of the candidate target can be represented by the direction line representing the orientation of the predetermined shape within the candidate target, and the third position can be calculated by identifying the position of the predetermined shape within the candidate target in the preset coordinate system. For example, when the predetermined shape is a V-shape, the direction of the angle bisector of the V-shape in the candidate target in the preset coordinate system can be used as the third position; similarly, when the predetermined shape is an arc shape, the direction of the ray from the midpoint of the arc to the center of the arc in the candidate target in the preset coordinate system can be used as the third position. The orientation of the preset target can be represented by the orientation of the predetermined shape within the preset target, or it can be represented by the orientation of other objects within the preset target, such as the orientation of an electrode sheet. Understandably, the fourth position is obtained through pre-calibration. The robot can calculate the angle between the direction line representing the orientation of the candidate target and the direction line representing the orientation of the preset target based on the third and fourth positions. Based on this angle, it obtains a parameter matching score in the pose dimension between the identified parameters and the preset expected parameters. In specific implementations, this angle can be directly used as the parameter matching score in the pose dimension; alternatively, the angle can be normalized, and the normalized result can be used as the parameter matching score. Normalization ensures that small defects in a single dimension do not have a significant absolute impact on the final calculated matching degree. The method of normalization is not limited here.

[0119] In one feasible implementation, the robot can calculate the difference between the shape parameters of the candidate target and preset shape parameters, and obtain a parameter matching score in the shape dimension between the identification parameters and the preset expected parameters based on this difference. In a specific implementation, this difference can be directly used as the parameter matching score in the shape dimension; alternatively, the difference can be normalized, and the normalized result can be used as the parameter matching score. Normalization ensures that small defects in a single dimension do not have a significant absolute impact on the final calculated matching degree. The normalization method is not limited here. In a specific implementation, when there are multiple shape parameters, the difference between each shape parameter of the candidate target and each preset shape parameter can be calculated separately, and the differences can be added together or averaged to obtain the parameter matching score in the shape dimension.

[0120] For example, when the set shape is a V shape, the shape parameters of the candidate target can include: the lengths of the two sides of the V shape in the candidate target and the included angle in the middle. Correspondingly, the preset shape parameters can include the lengths of the two sides of the V shape in the preset target and the included angle in the middle. The difference between the lengths of the two sides in the shape parameters of the candidate target and the lengths of the two sides in the preset shape parameters can be calculated, and the difference between the included angle in the shape parameters of the candidate target and the included angle in the preset shape parameters can be calculated. The result of the two differences can be used as the parameter matching score in the shape dimension, or the result can be normalized and used as the parameter matching score in the shape dimension.

[0121] For example, when the set shape is an arc shape, the shape parameters of the candidate target can include: the arc length and the radius of the arc shape in the candidate target. Correspondingly, the preset shape parameters can include the arc length and the radius of the arc shape in the preset target. The difference between the arc length in the shape parameters of the candidate target and the arc length in the preset shape parameters can be calculated, and the difference between the arc radius in the shape parameters of the candidate target and the arc radius in the preset shape parameters can be calculated. The result of the two differences can be used as the parameter matching score in the shape dimension, or the result can be normalized and used as the parameter matching score in the shape dimension.

[0122] Furthermore, embodiments of the present invention also propose a target recognition device, which is deployed on a robot, with reference to... Figure 5 The device includes:

[0123] The recognition module 10 is used to perform target recognition based on the environmental data collected by the robot, obtain candidate targets that match the preset target, and record the recognition parameters of the candidate targets, wherein the recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system;

[0124] The scoring module 20 is used to perform corresponding parameter matching scoring between the identification parameters and the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system.

[0125] The determining module 30 is used to determine the recognition result based on the matching degree between the candidate target and the preset target.

[0126] In one feasible embodiment, the identification module 10 is further configured to:

[0127] Based on the environmental data collected by the robot's sensing device, target recognition is performed to obtain an initial target that matches the preset shape of the target.

[0128] If the distance between the initially identified target and each of the identified candidate targets in the preset map coordinate system is greater than a preset threshold, then the initially identified target will be added as a candidate target.

[0129] In one feasible implementation, the determining module 30 is further configured to:

[0130] Based on the matching degree between at least one candidate target identified within the target recognition period and the preset target, an initial recognition result for the preset target within the target recognition period is obtained, wherein the initial recognition result includes the initial recognition position and initial recognition posture of the preset target relative to the robot coordinate system.

[0131] Based on the recognition results of the preset target in the previous recognition cycle and the robot's driving data in the target recognition cycle, a prediction result for the preset target in the target recognition cycle is calculated, wherein the prediction result includes the predicted position and predicted attitude of the preset target relative to the robot coordinate system.

[0132] The initial recognition result and the estimated result for the preset target within the target recognition period are fused according to a preset weight to obtain the recognition result for the preset target within the target recognition period.

[0133] In one feasible implementation, the determining module 30 is further configured to:

[0134] Based on the matching degree between at least one candidate target identified within the target identification period and the preset target, a matching target that matches the preset target is selected from each candidate target;

[0135] The position and orientation of the identified matching target relative to the robot coordinate system are used as the initial recognition result;

[0136] Based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension, the fusion weight corresponding to the initial recognition result is determined. The higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result.

[0137] The initial recognition result for the preset target within the target recognition period is fused with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

[0138] In one feasible implementation, the position of the candidate target relative to the preset coordinate system is represented by the first position of a first position point on a predetermined shape of the candidate target in the preset coordinate system, and the preset position is represented by the second position of a second position point on the preset target in the preset coordinate system;

[0139] The scoring module 20 is also used for:

[0140] Calculate the distance between the first position and the second position, and obtain a parameter matching score between the recognition parameters and the preset expected parameters in the position dimension based on the distance.

[0141] In one feasible implementation, the attitude of the candidate target relative to the preset coordinate system is represented by a direction line representing the orientation of the candidate target at a third position in the preset coordinate system, and the preset attitude is represented by a direction line representing the orientation of the preset target at a fourth position in the preset coordinate system.

[0142] The scoring module 20 is also used for:

[0143] Based on the third and fourth positions, the angle between the direction line representing the orientation of the candidate target and the direction line representing the orientation of the preset target is calculated, and the parameter matching score between the identification parameters and the preset expected parameters in the orientation dimension is obtained according to the angle.

[0144] In one feasible implementation, the scoring module 20 is further configured to:

[0145] Calculate the difference between the shape parameters of the candidate target and the preset shape parameters, and obtain a parameter matching score between the identification parameters and the preset expected parameters in the shape dimension based on the difference.

[0146] The specific implementation details of the target recognition device of the present invention are basically the same as the embodiments of the target recognition method described above, and will not be repeated here.

[0147] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a target recognition program, wherein the target recognition program, when executed by a processor, implements the steps of the target recognition method described below.

[0148] The various embodiments of the robot and computer-readable storage medium of the present invention can be referred to the various embodiments of the target recognition method of the present invention, and will not be repeated here.

[0149] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0150] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0151] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0152] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A target recognition method applied to a robot, characterized in that, The method includes: Based on the environmental data collected by the robot, target recognition is performed to obtain candidate targets that match the preset target, and the recognition parameters of the candidate targets are recorded. The recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system. The identification parameters are matched and scored with the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system. Based on the matching degree between at least one candidate target identified within the target identification period and the preset target, a matching target that matches the preset target is selected from each candidate target; The position and orientation of the identified matching target relative to the robot coordinate system are used as the initial recognition result; Based on the recognition results of the preset target in the previous recognition cycle and the robot's driving data in the target recognition cycle, a prediction result for the preset target in the target recognition cycle is calculated, wherein the prediction result includes the predicted position and predicted attitude of the preset target relative to the robot coordinate system. Based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension, the fusion weight corresponding to the initial recognition result is determined. The higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result. The initial recognition result for the preset target within the target recognition period is fused with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

2. The target identification method of claim 1, wherein, The steps for identifying candidate targets that match preset targets based on environmental data collected by the robot include: Based on the environmental data collected by the robot's sensing device, target recognition is performed to obtain an initial target that matches the preset shape of the target. If the distance between the initially identified target and each of the identified candidate targets in the preset map coordinate system is greater than a preset threshold, then the initially identified target will be added as a candidate target.

3. The target recognition method as described in claim 1, characterized in that, The position of the candidate target relative to the preset coordinate system is represented by the first position of a first position point on a predetermined shape of the candidate target in the preset coordinate system, and the preset position is represented by the second position of a second position point on the preset target in the preset coordinate system; The step of matching and scoring the identification parameters with the preset expected parameters includes: Calculate the distance between the first position and the second position, and obtain a parameter matching score between the recognition parameters and the preset expected parameters in the position dimension based on the distance.

4. The target recognition method as described in claim 1, characterized in that, The attitude of the candidate target relative to the preset coordinate system is represented by the direction line representing the orientation of the candidate target at the third position in the preset coordinate system, and the preset attitude is represented by the direction line representing the orientation of the preset target at the fourth position in the preset coordinate system. The step of matching and scoring the identification parameters with the preset expected parameters includes: Based on the third and fourth positions, the angle between the direction line representing the orientation of the candidate target and the direction line representing the orientation of the preset target is calculated, and the parameter matching score between the identification parameters and the preset expected parameters in the orientation dimension is obtained according to the angle.

5. The target recognition method as described in claim 1, characterized in that, The step of matching and scoring the identification parameters with the preset expected parameters includes: Calculate the difference between the shape parameters of the candidate target and the preset shape parameters, and obtain a parameter matching score between the identification parameters and the preset expected parameters in the shape dimension based on the difference.

6. A target recognition device, said device being deployed on a robot, characterized in that, The device includes: The recognition module is used to perform target recognition based on the environmental data collected by the robot, obtain candidate targets that match the preset target, and record the recognition parameters of the candidate targets, wherein the recognition parameters include the shape parameters of the candidate targets and the position and orientation of the candidate targets relative to the preset coordinate system; The scoring module is used to perform corresponding parameter matching scoring between the identification parameters and the preset expected parameters to obtain the matching degree between the candidate target and the preset target. The preset expected parameters include the preset shape parameters of the preset target and the preset position and preset posture of the preset target relative to the preset coordinate system. A determining module is configured to: select a matching target that matches the preset target from among the candidate targets based on the matching degree between at least one candidate target identified within the target recognition period and the preset target; use the position and orientation of the identified matching target relative to the robot coordinate system as an initial recognition result; calculate an estimated result for the preset target within the target recognition period based on the recognition result for the preset target in the previous recognition period and the robot's driving data within the target recognition period, wherein the estimated result includes the estimated position and estimated orientation of the preset target relative to the robot coordinate system; determine the fusion weight corresponding to the initial recognition result based on the parameter matching score between the recognition parameters of the matching target and the preset expected parameters in the shape dimension, wherein the higher the parameter matching score of the matching target in the shape dimension, the higher the fusion weight corresponding to the initial recognition result; and fuse the initial recognition result for the preset target within the target recognition period with the estimated result according to the fusion weight to obtain the recognition result for the preset target within the target recognition period.

7. A robot, characterized in that, The robot includes: a memory, a processor, and a target recognition program stored in the memory and executable on the processor, wherein the target recognition program, when executed by the processor, implements the steps of the target recognition method as described in any one of claims 1 to 5.

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

Images