A robot motion reorientation method, apparatus, device, and medium

Abstract

A robot motion reorientation method, device, equipment and medium are disclosed. The method comprises: determining target rigid body pose information of a reference object in a first calibration posture and target chain link pose information of a robot in a second calibration posture; determining bias calibration information of each target matching item in the target rigid body pose information and the target chain link pose information, respectively; wherein the bias calibration information comprises translation bias calibration information and rotation bias calibration information; determining second motion pose information of the robot according to the first motion pose information of the target object and the bias calibration information of each target matching item; determining target joint configuration information of the robot according to the second motion pose information, so as to control the robot to perform a corresponding action based on the target joint configuration information. The present scheme can improve the calibration efficiency, accuracy and consistency by automatically calibrating the matching item bias of the imitation object rigid body pose and the robot chain link pose in the calibration posture.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a method, apparatus, device, and medium for redirecting robot motion. Background Technology

[0002] With the development of robot imitation learning, teleoperation, and motion reproduction, a common practice when redirecting robot motion is to use the motion capture or posture estimation results of the imitated object as a reference, and then obtain the robot joint configuration through inverse kinematics to drive the robot to execute.

[0003] The pose output of the imitated object is usually based on its own coordinate system (such as the sensor coordinate system, rigid body coordinate system, etc.), while robot control requires constructing pose constraints in the robot's link coordinate system. Because the imitated object's own coordinate system and the robot's link coordinate system have systematic differences in axial definitions, zero orientation, and initial alignment, in engineering practice, it is usually necessary to manually set corresponding offsets for each matching item to correctly convert the imitated object's pose into the robot's link pose, so as to achieve robot motion retargeting based on the robot's link pose.

[0004] However, when calibrating the offset of matching items by manually setting them, the calibration workload is large and the deployment cycle is long, resulting in low calibration efficiency. Moreover, it relies on human experience, making it difficult to guarantee the accuracy and consistency of the calibration, thus affecting the tracking accuracy and stability of the robot. Summary of the Invention

[0005] This invention provides a robot motion retargeting method, apparatus, device, and medium. By constructing a calibration posture, it automatically calibrates the offset of the matching item between the rigid body pose of the simulated object and the robot link pose, without relying on manual intervention. This effectively improves the efficiency, accuracy, and consistency of offset calibration, and helps to improve the accuracy and stability of robot motion retargeting.

[0006] According to one aspect of the present invention, a robot motion redirection method is provided, the method comprising: Determine the target rigid body pose information of the reference object in the first calibration posture and the target link pose information of the robot in the second calibration posture; wherein, the first calibration posture and the second calibration posture are adapted in overall shape, the target rigid body and the target link are matched, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. The offset calibration information for each target matching item in the target rigid body pose information and the target link pose information is determined respectively; wherein, the offset calibration information includes translation offset calibration information and rotation offset calibration information, the translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system; The robot's second motion pose information is determined based on the first motion pose information of the target object and the bias calibration information of each target matching item; The target joint configuration information of the robot is determined based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

[0007] According to another aspect of the present invention, a robot motion redirection device is provided, the device comprising: The calibration pose information determination module is used to determine the target rigid body pose information of the reference object in the first calibration pose and the target link pose information of the robot in the second calibration pose; wherein, the first calibration pose and the second calibration pose are adapted in overall shape, the target rigid body and the target link are matched, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. The offset calibration information determination module is used to determine the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively; wherein, the offset calibration information includes translation offset calibration information and rotation offset calibration information, the translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system; The motion pose information determination module is used to determine the second motion pose information of the robot based on the first motion pose information of the target object and the offset calibration information of each target matching item; The joint configuration information determination module is used to determine the target joint configuration information of the robot based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

[0008] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the robot motion reversal method according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the robot motion reversal method according to any embodiment of the present invention.

[0010] The technical solution of this invention first determines the target rigid body pose information of the reference object in a first calibration posture and the target link pose information of the robot in a second calibration posture. The first and second calibration postures are adapted in overall shape, the target rigid body matches the target link, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. Then, the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information are determined respectively. The offset calibration information includes translation offset calibration information and rotation offset calibration information. The translation offset calibration information describes the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information describes the rotation from the rigid body coordinate system to the link coordinate system. Then, the second motion pose information of the robot is determined based on the first motion pose information of the target object and the offset calibration information of each target matching item. Finally, the target joint configuration information of the robot is determined based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information. This technical solution automatically calibrates the offset of the matching item between the rigid body pose of the simulated object and the pose of the robot links by constructing a calibration posture. It does not rely on manual calibration and can effectively improve the efficiency, accuracy and consistency of offset calibration, which helps to improve the accuracy and stability of robot motion retargeting.

[0011] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a robot motion redirection method provided according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the overall process of a robot motion redirection method provided by an embodiment of the present invention; Figure 3This is a flowchart of another robot motion redirection method provided according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a robot motion redirection device according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device that implements a robot motion redirection method according to an embodiment of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] Example 1 Figure 1 This is a flowchart of a robot motion retargeting method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where the offset of a matching item is automatically calibrated by constructing a calibration posture. This method can be executed by a robot motion retargeting device, which can be implemented in hardware and / or software. The robot motion retargeting device can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes: S110, determine the target rigid body pose information of the reference object in the first calibration posture and the target link pose information of the robot in the second calibration posture.

[0017] The reference object can be an object used for bias calibration, which the robot can learn and imitate, and which has similar attributes to the robot (such as morphology and functional behavior). For example, the reference object can be a human body or other entities with movement capabilities (such as animals), and correspondingly, the robot can be a humanoid robot or an animal robot. The first calibration posture and the second calibration posture can refer to the reference object posture and the robot posture pre-set according to the actual application scenario, respectively. It should be noted that the first and second calibration postures are adapted in overall shape (e.g., when the reference object is in a standing posture, the robot also selects a standing initial configuration) to reduce the absolute value of subsequent bias calculations and improve the smoothness of robot motion retargeting.

[0018] In this context, the target rigid body and target link can refer to multiple rigid bodies of the robot's emulation object and multiple links of the robot, respectively, pre-defined according to actual needs, and it is essential to ensure that the target rigid body matches the target link. The robot's emulation object serves as a reference object during the calibration phase and as the target object during the operational phase. It should be noted that the selection of the target rigid body must be combined with the robot's link configuration, prioritizing coverage of the emulation object parts corresponding to the robot's movable links. For example, taking the human body as the robot's emulation object, when the robot possesses upper limb, lower limb, and torso movement capabilities, the human side needs to simultaneously select key rigid bodies such as the head, neck, torso, left and right upper arms, left and right wrists, left and right thighs, left and right lower legs, and left and right feet as target rigid bodies.

[0019] The target rigid body pose information can be used to describe the position and orientation of the target rigid body. The target link pose information can be used to describe the position and orientation of the target links. It should be noted that since the data source of the imitated object (such as different types of motion capture devices) may output pose information in different coordinate systems (such as the sensor local coordinate system and a custom coordinate system), coordinate system unification processing needs to be introduced during the calibration and operation phases to transform the pose information of the imitated object and the robot to a unified coordinate system (such as the world coordinate system). This ensures that the target rigid body pose information and the target link pose information are described based on the same coordinate system, thereby ensuring the coordinate system consistency of the pose information of the imitated object and the robot.

[0020] Specifically, coordinate system processing can include the following: (1) Axis replacement: For the differences in coordinate axis definitions of different devices (e.g., the X-axis of some devices is horizontal forward and the Z-axis of some devices is horizontal forward), the coordinate axis direction of the pose data is adjusted by the axis replacement matrix to ensure consistency; (2) Left-hand coordinate system conversion: If the pose data of the imitation object adopts the left-hand coordinate system and the pose data of the robot adopts the right-hand coordinate system (or vice versa), the left-hand coordinate system conversion is achieved by the coordinate transformation matrix to avoid the orientation direction being reversed; (3) Unit conversion: The length units of the pose data of the imitation object and the robot are unified, and units such as centimeters (cm) and millimeters (mm) are converted to meters (m) to keep the pose data units of the imitation object and the robot consistent; (4) Time synchronization: Since there may be differences in the acquisition frequency between the pose data of the imitation object and the pose data of the robot (e.g., the acquisition frequency of the imitation object data is 60Hz and the control frequency of the robot is 100Hz), the imitation object data can be interpolated to the unified frequency of the robot control by linear interpolation and other methods to ensure the synchronization of frame-by-frame processing.

[0021] In this embodiment, to clarify the mathematical expressions for the robot's imitation object, the robot, and the biases of the matching items, and to unify the technical context, the core data and symbols are strictly defined in advance to ensure the consistency of calculations between the calibration and operation phases. Specifically, for the target rigid body, each rigid body is denoted as symbol s, and the global pose of rigid body s in the world coordinate system is represented as a tuple. Among them, the position component , represents the three-dimensional coordinates of key feature points of rigid body s (such as the center of mass or center point of the rigid body) in the world coordinate system, in meters (m); attitude components (Special orthogonal group) represents the attitude of a rigid body s relative to the world coordinate system. It can be represented by rotation matrix, quaternion or axis-angle equivalent form according to the calculation requirements, ensuring the orthogonality and singularity of the attitude description.

[0022] For each target link of the robot (such as the movable links of the robot arm, legs, torso, head, etc.), each link is denoted by the symbol r, and the global pose of link r in the world coordinate system is represented as a tuple. Among them, the position component This represents the three-dimensional coordinates of the reference point of robot link r (such as the joint center or the end effector center point) in the world coordinate system, with the unit consistent with its simulated object (meters); attitude components. , represents the posture of the robot link r relative to the world coordinate system. Its representation should be consistent with the posture components of the imitated object (e.g., both use rotation matrices or both use quaternions) to facilitate subsequent offset calculation.

[0023] To address the matching term bias and establish the correspondence between the target rigid body and the target link, the matching term is defined as a bidirectional correspondence between the rigid body s and the robot link r (denoted as ). Each matching item corresponds to a unique set of translation and rotation offsets, used to describe the relative pose relationship between the two in the calibration attitude. Among them, the translation offset... This represents the translation of rigid body s relative to link r in the local coordinate system (i.e., the link coordinate system) of robot link r, with dimensions […]. The core function of translational offset is to eliminate the inherent positional deviation between the target rigid body and the target links under the calibration attitude. Because it is based on the local coordinate system of the links, it possesses coordinate system invariance and can be transferred to robots with different initial positions. Rotational offset... Let represent the relative rotation matrix from the local coordinate system (i.e., the rigid body coordinate system) of rigid body s to the local coordinate system (i.e., the link coordinate system) of robot link r. Its dimensions are 3×3 (4-dimensional if quaternions are used). Space. Rotational offset is used to align the emulated object with the robot's pose reference, ensuring directional consistency when the emulated object's pose is converted to the robot's pose.

[0024] In this embodiment, it is first necessary to determine the global pose set of the target rigid body under the first calibration posture {( The target rigid body pose information serves as a reference object, where the selection of the first calibration pose must balance stability, ease of acquisition, and noise resistance. Specifically, this can be achieved in the following three ways: (1) Acquisition based on static standard pose: Taking the human body as a reference object, a static pose that is easy for the human body to maintain and has no redundant movement can be selected as the first calibration pose, such as the static standing pose (feet shoulder-width apart, arms hanging naturally, torso upright, head looking straight ahead), T-Pose (arms horizontally spread at 90° to the torso, legs straight and together), or A-Pose (arms slightly bent at 120° to the torso, feet straight and together). The advantage of this type of pose is that the human body is rigid and the force is uniform, and the posture is stable. It can be quickly acquired by motion capture equipment (such as optical motion capture system, inertial motion capture suit, etc.) and can effectively reduce the posture deviation caused by human muscle tension. In actual operation, the reference object can be required to maintain the first calibration pose for 3-5 seconds to ensure that the motion capture system acquires enough effective frames for subsequent pose extraction.

[0025] (2) Constructing calibration poses based on offline files: When it is impossible to acquire the pose of the reference object in real time, the first calibration pose can be constructed using preset poses in offline files. For example, taking the human body as the reference object, the default pose, zero-rotation pose, or reference frame pose of the human body model stored offline can be called. These poses are usually provided by the human body model manufacturer or are predefined based on a general human skeleton model, and their pose parameters have been standardized, and can be directly used as ( This implementation paradigm is suitable for scenarios such as system pre-deployment and individual robot debugging, without relying on real-time human data. (Among them, ( ) represents the pose information of the target rigid body of the reference object.

[0026] (3) Robust estimation based on multi-frame data: To reduce the impact of measurement noise from motion capture equipment and slight jitter of the reference object on the calibration posture, robust estimation can be performed using multi-frame reference object posture data within a short time window to construct the first calibration posture. The specific process is as follows: Collect multi-frame (usually 30-50 frames) posture data of the reference object within 1-2 seconds of maintaining the first calibration posture (such as a static standing posture). The position component of each target rigid body is processed by median filtering, and the posture component is processed by least squares fitting or quaternion interpolation averaging. Finally, a set of global poses of the rigid body under the first calibration posture with stronger robustness is obtained {( This implementation method can effectively suppress random noise, thereby improving the calculation accuracy of bias calibration, and is especially suitable for sensor systems with relatively high noise, such as inertial motion capture.

[0027] In this embodiment, the global pose set of the target link of the robot in the second calibration posture is also required {( As the target link pose information of the robot, the specific process is as follows: (1) Determine the initial joint configuration of the robot. The initial joint configuration consists of a preset set of joint angles for the robot (each joint corresponds to an angle value, with dimensions consistent with the number of robot joints). Typically, the robot's zero-position posture, standing posture, or initial calibration posture is selected as the second calibration posture. Initial Joint Configuration The preset parameters can be called through robot control, or the robot joints can be manually adjusted to the second calibration posture by the user and then saved. (2) Calculate the global pose set of the links: based on the robot's initial joint configuration Using the Forward Kinematics (FK) algorithm, the global pose of each target link r of the robot in the world coordinate system is calculated. ), ultimately forming the global pose set of the links under the second calibration attitude {( The target link pose information is used. The Freeform Kinematics (FK) calculation relies on the robot's Denavit-Hartenberg (DH) parameter table or the robot's kinematic model to ensure the accuracy of the pose calculation. If the robot has a built-in FK solver interface, the link pose data can be directly obtained by calling the interface, eliminating the need to manually implement the FK algorithm. It is important to note that the global pose calculation of the robot links must be based on the same coordinate system as the rigid body pose calculation of the reference object to ensure the comparability of the pose data and lay the foundation for subsequent offset calculations.

[0028] S120, respectively determine the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information.

[0029] The offset calibration information includes translational offset calibration information and rotational offset calibration information. Translational offset calibration information describes the translation of the rigid body relative to the link in the link coordinate system, while rotational offset calibration information describes the rotation from the rigid body coordinate system to the link coordinate system. Target matching terms can refer to any matching terms between the target rigid body pose information and the target link pose information. It should be noted that, to better achieve robot motion retargeting, multiple matching rigid bodies and links are typically set as target rigid bodies and target links; therefore, the target rigid body pose information and target link pose information include multiple target matching terms. It should be noted that the calibration phase (corresponding to steps S110-S120) is a one-time execution process, which is only completed during system initialization. It does not need to be calculated repeatedly during the running phase, which can greatly reduce the real-time processing pressure. Its core objective is to obtain the global pose set of the reference object in the first calibration pose and the robot in the second calibration pose, calculate the translation offset and rotation offset of each target matching item as translation offset calibration information and rotation offset calibration information, and cache them for use in the running phase.

[0030] In this embodiment, for each preset target matching item Based on the target rigid body pose information of the reference object ( ) and the robot's target link pose information ( ), calculate translation offset With rotational bias This serves as bias calibration information. For example, a formula can be used. Calculate translation offset As translation offset calibration information, among which... Rotation matrix under the second calibration attitude The transpose of the rotation matrix (since the rotation matrix is ​​orthogonal, its transpose is equal to its inverse) is the core operation of converting the position difference between the target rigid body and the target link in the world coordinate system. Transform to the robot's link coordinate system to obtain a translational offset that is invariant in that coordinate system. For example, a formula can be used... The rotation offset is calculated as rotation offset calibration information. This formula uses the transpose of the rotation matrix to represent the attitude of the target rigid body. The process involves transforming from the world coordinate system to the robot's link coordinate system to obtain the relative rotational relationship between the target rigid body and the target links. It should be noted that if the attitude components are represented using quaternions, the quaternions must first be converted into rotation matrices for calculation, or the corresponding calculation formula should be derived based on the quaternion operation rules to ensure consistency of the calculation results.

[0031] After determining the bias calibration information for each target match, it can be written into a matching bias table. This matching bias table can be stored as one or more task tables / IK (Inverse Kinematics) tables. For example, it can be divided into multiple matching bias tables based on body parts (upper limbs, lower limbs, trunk), or into matching bias tables with different weights based on task requirements (high-precision tasks, fast-response tasks). Furthermore, the matching bias table needs to be cached in the robot controller's high-speed memory to avoid disk read latency and ensure rapid access during runtime, thereby reducing data read latency.

[0032] To further reduce the noise impact during the calibration stage and improve the accuracy of bias calculation, a multi-frame refinement and optimization step can be introduced after obtaining the closed-form solution of the above bias calibration information. This step is suitable for scenarios with extremely high requirements for motion reversal accuracy (such as precision operation and anthropomorphic interaction scenarios). The specific process is as follows: Collect multiple frames (e.g., 50-100 frames) of reference object pose data and robot pose data under the calibration posture, determine the bias calibration information of each frame of data, and obtain multiple sets of bias data; use the least squares algorithm to fit and optimize the multiple sets of bias data to obtain the refined bias calibration information, which is used to replace the bias data in the original matching bias table.

[0033] S130, determine the robot's second motion pose information based on the first motion pose information of the target object and the offset calibration information of each target matching item.

[0034] The target object can refer to the object to be imitated that requires robot motion retargeting. This object can also be learned and imitated by the robot and has similar attributes to the robot. The first motion pose information can be used to reflect the position and orientation of the target rigid body in multiple consecutive motion frames. The second motion pose information can refer to the position and orientation information of the target links obtained by retargeting the first motion pose information using bias calibration information. It should be noted that the running phase (corresponding to steps S130-S140) is a real-time execution process. It needs to process the first motion pose information of the target object frame by frame, quickly convert the bias calibration information stored in the calibration phase into the second motion pose information required by the robot IK, and finally obtain the robot joint configuration through IK solution and output it. The core objective of this phase is to ensure real-time performance and ensure that the motion retargeting delay is lower than the system requirements (usually lower than 10ms).

[0035] In this embodiment, the rigid body pose of the target object in continuous motion frames can be acquired in real time using a motion capture system (such as optical capture, inertial capture, or visual capture) or a motion sensor, resulting in a set of global rigid body poses of the target object across multiple frames {( This serves as the first motion pose information. It should be noted that the source of this first motion pose information can be flexibly switched (e.g., from optical capture to inertial capture) without needing to re-execute the calibration stage. It is only necessary to ensure that the data has undergone preprocessing such as coordinate system first-order preprocessing to adapt to existing offset calibration information. For each target matching item of the first motion pose information in each frame of the motion image... It can access the bias calibration information cached during the calibration phase (including translation bias). With rotational bias ), to obtain the target rigid body pose information of the target object ( Converted into target link pose information in link coordinate system. From this, the robot's second motion pose information can be obtained {( )}.

[0036] Specifically, a formula can be used. To achieve motion posture transitions. Among them, For rotational bias The inverse matrix (since the rotation matrix is ​​an orthogonal matrix, its inverse matrix is ​​equal to its transpose). This formula, through the inverse operation of rotation offset, converts the real-time pose of the target rigid body into the pose of the robot's target links, ensuring pose consistency between the two. Alternatively, the formula can be used... This achieves motion position transformation. The formula combines the transformed robot target link posture. With translation offset This method converts the real-time position of the target rigid body into the position of the robot's target links, eliminating the inherent positional deviation between the target object and the robot, while ensuring the coordination of position and attitude transformations. The conversion process involves only basic operations such as matrix multiplication and addition, without complex iterative processes. The processing time per frame can be controlled within 1ms, meeting the requirements for low-latency real-time motion retargeting. Furthermore, since the offset is defined based on a local coordinate system, this conversion method has good portability and can be adapted to robots with different initial configurations and scales without requiring recalibration.

[0037] S140, determine the target joint configuration information of the robot based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

[0038] In this embodiment, after obtaining the robot's second motion pose information, this second motion pose information can be input as the IK task target into the IK solver for IK solving, thereby obtaining the robot's joint configuration as the target joint configuration information. For example, an IK solver based on the Minkowski (Mink) Generalized Minimum Residual (GMR) algorithm can be used. The GMR algorithm has advantages such as fast convergence speed, strong robustness, and the ability to handle multi-constraint tasks. Under the premise of satisfying preset constraints, it can complete the multi-constraint IK solution within milliseconds to obtain the robot's target joint configuration information, adapting to high-frequency real-time control requirements, so that the robot can be controlled to perform corresponding actions based on the target joint configuration information under different scenario requirements. The preset constraints can be flexibly set based on the actual application scenario, such as including joint angle restrictions and obstacle avoidance constraints. For example, after the IK solution is completed, the target joint configuration information can be output in real time to a robot simulator (for virtual simulation scenarios), a controller (for actual robot control scenarios), or an actuator (such as a servo motor) to drive the robot to perform the corresponding motion, realizing real-time redirection from target object motion to robot motion.

[0039] For example, see Figure 2 Taking the human body as the model and a humanoid robot as an example, the overall process of robot motion retargeting is as follows: In the calibration phase, through a motion capture system (i.e., Figure 2 The robot acquires the global pose of the target rigid body in the human calibration pose (i.e., the first calibration pose) and the global pose of the target links in the robot's initial joint configuration (i.e., the second calibration pose) using motion capture systems or sensors. Through data preprocessing, both are transformed into a unified coordinate system. This allows for the calculation of offset calibration information for each target matching item, forming a matching offset table which is stored in the robot's high-speed memory. During operation, the motion capture system (i.e., ...) acquires the global pose of the target rigid body in the human calibration pose (i.e., the first calibration pose) and the global pose of the target links in the robot's initial joint configuration (i.e., the second calibration pose). Then, through data preprocessing, both are transformed into a unified coordinate system. This allows for the calculation of offset calibration information for each target matching item and the formation of a matching offset table, which is then stored in the robot's high-speed memory. Figure 2The robot acquires real-time human pose information (i.e., first motion pose information) from motion capture or sensors, and then calls the offset calibration information in the matching offset table stored in the robot to convert the real-time human pose information to the robot link coordinate system to obtain the second motion pose information. The second motion pose information is then input into the IK solver, and the target joint configuration information of the robot is obtained through IK solving, and finally the robot joint commands are generated.

[0040] The technical solution of this invention first determines the target rigid body pose information of the reference object in a first calibration posture and the target link pose information of the robot in a second calibration posture. The first and second calibration postures are adapted in overall shape, the target rigid body matches the target link, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. Then, the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information are determined respectively. The offset calibration information includes translation offset calibration information and rotation offset calibration information. The translation offset calibration information describes the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information describes the rotation from the rigid body coordinate system to the link coordinate system. Then, the second motion pose information of the robot is determined based on the first motion pose information of the target object and the offset calibration information of each target matching item. Finally, the target joint configuration information of the robot is determined based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information. This technical solution automatically calibrates the offset of the matching item between the rigid body pose of the simulated object and the pose of the robot links by constructing a calibration posture. It does not rely on manual calibration and can effectively improve the efficiency, accuracy and consistency of offset calibration, which helps to improve the accuracy and stability of robot motion retargeting.

[0041] In this embodiment, optionally, before determining the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively, the method further includes: determining a reference scale factor based on the size difference between the robot and the reference object; and updating the position information in the target rigid body pose information based on the reference scale factor.

[0042] It should be noted that when there is a difference between the size of the reference object and the size of the robot, directly performing pose conversion will lead to a misalignment of the robot's motion proportions (e.g., the swing amplitude of a human arm is too large when adapted to a small robot). Therefore, this embodiment introduces a scale factor to adjust the position component in the target rigid body's pose information. Normalization is performed, and the specific process is as follows: (1) Calculate the reference scale factor: In the calibration stage, the reference scale factor can be determined based on the ratio of the robot size in the second calibration posture to the reference object size in the first calibration posture. If the height of the reference object and the height of the robot are known, the height parameter can be directly called to calculate the reference scale factor, that is, the reference scale factor = robot height / reference object height; if the height of the reference object and / or the height of the robot are unknown, the reference scale factor can be calculated based on the target rigid body distance of the reference object (such as the length of the human arm and leg) and the length of the target link corresponding to the robot (such as the length of the robot arm and leg), that is, the reference scale factor = target link length / target rigid body distance. (2) Application of scale normalization: Before calculating the translation offset in the calibration stage, or before performing motion pose conversion in the running stage, the position component in the target rigid body pose information is normalized. ( or Multiplying the normalized position component by the reference scale factor k yields the normalized position component. This allows for the updating of location information.

[0043] This solution, through this setup, can effectively improve the compatibility between the robot and the emulated objects of different sizes, and enhance the proportional coordination and accuracy of motion redirection.

[0044] In this embodiment, optionally, after determining the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information, the method further includes: obtaining the foot rigid body position information from the target rigid body pose information and obtaining the foot link position information from the target link pose information; determining the height difference between the reference object and the robot in the vertical direction based on the foot rigid body position information and the foot link position information; and adjusting the robot's base height or root node translation amount based on the height difference so that the foot rigid body and the foot link are at the same height.

[0045] In this embodiment, in order to improve the stability of the robot's foot contact and the rationality of the IK solution during the subsequent motion reorientation process, the height alignment of the reference object's foot and the robot's foot can be completed in the calibration stage to eliminate the height deviation between the two in the calibration posture. The specific process is as follows: (1) Select the corresponding rigid body / link: Select the foot rigid body of the reference object (such as the rigid body of the two feet, the rigid body of the toes) and the foot link of the robot (such as the end link of the robot's two feet) as the height alignment object; (2) Calculate the height difference: Extract the foot rigid body position information from the target rigid body pose information in the first calibration posture, extract the foot link position information from the target link pose information in the second calibration posture, and calculate the height difference between the two on the Z-axis (i.e., the vertical direction) of the world coordinate system. (3) Height compensation: based on the height difference Compensation adjustments are made to the robot's base height or root node translation. For example, if the height of the human foot is higher than the robot's foot height (…), If the value is greater than 0, then the robot base will be translated upwards along the Z-axis. Alternatively, adjust the Z-axis coordinate of the robot's root node so that the robot's foot links are at the same height plane as the rigid body of the human foot in the calibrated posture.

[0046] This solution, through its configuration, can effectively prevent the robot from exhibiting unreasonable postures such as tiptoeing or missing steps during operation, improve the consistency of initial landing, reduce IK oscillations caused by penetration, suspension, and contact instability, and enhance the naturalness and stability of motion redirection. It is especially suitable for motion scenarios involving foot contact, such as walking and standing.

[0047] Example 2 Figure 3 This is a flowchart of a robot motion redirection method provided in Embodiment 2 of the present invention. This embodiment is an optimization based on the above embodiment. Figure 3 As shown, the method in this embodiment specifically includes the following steps: S210, determine the target rigid body pose information of the reference object in the first calibration posture and the target link pose information of the robot in the second calibration posture.

[0048] Among them, the first calibration posture and the second calibration posture are adapted in overall shape, the target rigid body and the target link are matched, and the pose information of the target rigid body and the pose information of the target link are described based on the same coordinate system.

[0049] S220, determine the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information respectively.

[0050] The offset calibration information includes translation offset calibration information and rotation offset calibration information. The translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system.

[0051] S230, determine the weight information of each target matching item under different motion scenarios, and construct a matching bias table under different motion scenarios based on the bias calibration information and weight information of each target matching item.

[0052] In this embodiment, to adapt to different motion scenario requirements, multiple matching bias tables can be constructed based on bias calibration information during the calibration phase. Each matching bias table corresponds to a different task priority or motion scenario. Each matching bias table includes the bias calibration information of the target matching item and the corresponding weight information (the weight is determined based on different motion scenarios), and each matching bias table is stored independently and does not interfere with each other. For example, motion scenarios may include upper limb fine manipulation scenarios and lower limb stable movement scenarios (such as walking scenarios). For example, the matching bias tables may include upper limb priority tables, lower limb priority tables, and stability enhancement tables. The target matching items in each matching bias table are the same, but the weights are different. Specifically, upper limb matching items (such as head, upper arm, forearm, etc.) in the upper limb priority table have a larger weight, lower limb matching items (such as thigh, calf, foot, etc.) in the lower limb priority table have a larger weight, and trunk matching items in the stability enhancement table have a larger weight.

[0053] S240, determine the target matching table from the matching bias table according to the current motion scene, and determine the robot's second motion pose information according to the first motion pose information of the target object at each motion moment and the bias calibration information in the target matching table.

[0054] In this embodiment, during the operation phase, one or more target matching tables can be automatically determined from the matching bias table based on the current motion scenario. For example, the upper limb fine manipulation scenario switches to the upper limb priority table, while the walking scenario switches to the lower limb priority table. Then, the bias calibration information in the target matching table is used to perform pose transformation on the first motion pose information of the target object at each motion moment, thereby obtaining the robot's second motion pose information in the current motion scenario. It should be noted that if multiple target matching tables exist simultaneously, weights can be assigned to each target matching table (e.g., upper limb priority table weight 0.6, lower limb priority table weight 0.4). Then, based on the weights of each target matching table and the bias calibration information in the target matching table, pose transformation is performed on the first motion pose information of the target object at each motion moment, thereby obtaining the second motion pose information integrating multiple target matching tables, thus optimizing the motion redirection effect.

[0055] S250, if there are insufficient confidence data in the first motion pose information, then the weight information corresponding to the insufficient confidence data in the target matching table is reduced in weight.

[0056] It should be noted that in actual motion capture scenarios, the first motion pose information may have low confidence due to sensor failure or performance degradation (such as sensor drift). Low confidence data will affect the IK solution results, leading to improper robot joint configuration and thus reducing the motion retargeting effect.

[0057] To address the aforementioned issues, this embodiment, after obtaining the first motion pose information of the target object, can perform data confidence assessment. If the confidence level of a certain data is lower than a first threshold (which can be set according to accuracy requirements), the data is considered to have insufficient confidence. If there are data with insufficient confidence in the first motion pose information, the target matching items corresponding to the data with insufficient confidence in the target matching table are weighted down (e.g., the weight is reduced from 1.0 to 0.3-0.5), that is, a smaller weight is set during IK solving to reduce the impact of low-confidence data on the IK solution results.

[0058] Furthermore, if the confidence level of a certain data is lower than the second threshold, where the second threshold is less than the first threshold, then the confidence level of the data can be considered extremely low. In this case, the weight can be set to 0, that is, the IK task of the target matching item corresponding to the data is temporarily disabled in the current frame, and the robot joint configuration is solved only based on other valid target matching items. The task is re-enabled after the data is recovered to ensure the continuity of motion redirection.

[0059] S260, based on the weight information in the target matching table, the inverse kinematics solution of the second motion pose information is performed to obtain the target joint configuration information of the robot, so as to control the robot to perform corresponding actions based on the target joint configuration information.

[0060] In this embodiment, after reducing the weight of data with insufficient confidence, the IK solution can be performed on the second motion pose information based on the weight information in the target matching table, thereby obtaining the target joint configuration information of the robot adapted to the current motion scenario. It is understood that the larger the weight of the matching item in the target matching table, the greater its contribution to the IK solution. Furthermore, if there is data with insufficient confidence in the first motion pose information, the constraints on the target matching items corresponding to the data with insufficient confidence can be relaxed during the IK solution.

[0061] The technical solution of this invention constructs a matching bias table for different motion scenarios based on the weight information and bias calibration information of each target matching item under different motion scenarios. It determines the robot's second motion pose information based on the first motion pose information of the target object at each motion moment and the bias calibration information in the target matching table adapted to the current motion scenario. When there are data with insufficient confidence in the first motion pose information, the weight information corresponding to the data with insufficient confidence in the target matching table is reduced in weight. Then, based on the weight information in the target matching table, inverse kinematics is performed on the second motion pose information to obtain the robot's target joint configuration information. This technical solution can better adapt to the motion relocalization requirements of different motion scenarios and further improve the accuracy and robustness of robot motion retargeting by reducing the weight of low-confidence data.

[0062] In this embodiment, optionally, before determining the robot's second motion pose information based on the first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table, the method further includes: determining whether there is data anomaly in the first motion pose information; wherein, data anomaly includes data missing and insufficient data confidence; if so, then using a nearby rigid body to infer the predicted motion pose information corresponding to the abnormal data; updating the first motion pose information based on the predicted motion pose information; correspondingly, determining the robot's second motion pose information based on the first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table includes: determining the robot's second motion pose information based on the updated first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table.

[0063] It should be noted that in actual motion capture scenarios, the first motion pose information may be missing or have low confidence due to occlusion, sensor failure, or excessively fast movement. This can lead to interruption of pose conversion or excessive error, affecting the accuracy and continuity of motion redirection.

[0064] To address the aforementioned issues, this embodiment first determines whether the first motion pose information contains data anomalies (including missing data and insufficient data confidence) before determining the robot's second motion pose information. If data anomalies are found, the motion pose data corresponding to the anomaly data is inferred from neighboring rigid bodies as predicted motion pose information to update the first motion pose information. The second motion pose information is then determined based on the updated first motion pose information and the bias calibration information in the target matching table. For example, the parent-child joint relationship of human rigid bodies can be utilized (e.g., when a forearm bone segment is missing, the forearm bone segment pose can be extrapolated from the poses of the upper arm and wrist bone segments), or the motion trend of adjacent rigid bodies can be used (e.g., inferring the right upper arm pose from the left upper arm pose, applicable to symmetrical motion scenarios) to infer the pose data of the missing rigid body, ensuring the integrity of the pose set. Furthermore, if the pose data of a certain rigid body is missing and cannot be inferred from neighboring rigid bodies, the IK task for the target matching item corresponding to the missing data can be temporarily disabled in the current frame, and the task can be re-enabled after the data is recovered.

[0065] This solution enhances the robustness of robot motion redirection by employing mechanisms such as weight reduction, constraint relaxation, and disabling IK tasks when data anomalies exist, such as missing data or low confidence levels.

[0066] Example 3 Figure 4This is a schematic diagram of a robot motion redirection device provided in Embodiment 3 of the present invention. This device can execute the robot motion redirection method provided in any embodiment of the present invention, and possesses the corresponding functional modules and beneficial effects of the method. For example... Figure 4 As shown, the device includes: The calibration pose information determination module 310 is used to determine the target rigid body pose information of the reference object in the first calibration pose and the target link pose information of the robot in the second calibration pose; wherein, the first calibration pose and the second calibration pose are adapted in overall shape, the target rigid body and the target link are matched, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. The offset calibration information determination module 320 is used to determine the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively; wherein, the offset calibration information includes translation offset calibration information and rotation offset calibration information, the translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system; The motion pose information determination module 330 is used to determine the second motion pose information of the robot based on the first motion pose information of the target object and the offset calibration information of each target matching item; The joint configuration information determination module 340 is used to determine the target joint configuration information of the robot based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

[0067] Optionally, the device further includes: a rigid body position correction module, used for: Before determining the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively, a reference scale factor is determined based on the size difference between the robot and the reference object. The position information in the pose information of the target rigid body is updated according to the reference scale factor.

[0068] Optionally, the device further includes: a robot height compensation module, used for: After determining the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively, the foot rigid body position information is obtained from the target rigid body pose information, and the foot link position information is obtained from the target link pose information. The height difference between the reference object and the robot in the vertical direction is determined based on the foot rigid body position information and the foot link position information; The robot's base height or root node translation is adjusted based on the height difference to ensure that the foot rigid body and foot link are at the same height.

[0069] Optionally, the apparatus further includes: a matching bias table construction module, used for: After determining the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively, the weight information of each target matching item in different motion scenarios is determined. The matching bias table for different motion scenarios is constructed based on the bias calibration information and weight information of each target matching item.

[0070] Optionally, the motion pose information determination module 330 is used for: Determine the target matching table from the matching bias table based on the current motion scenario; The robot's second motion pose information is determined based on the target object's first motion pose information at each motion moment and the offset calibration information in the target matching table.

[0071] Optionally, the device further includes: a motion posture abnormality correction module, used for: Before determining the robot's second motion pose information based on the target object's first motion pose information at each motion moment and the bias calibration information in the target matching table, it is determined whether there are any data anomalies in the first motion pose information; wherein, the data anomalies include data missing and insufficient data confidence. If so, the predicted motion pose information corresponding to the abnormal data is inferred using the neighboring rigid body; The first motion pose information is updated based on the predicted motion pose information; Accordingly, the motion pose information determination module 330 is used for: The robot's second motion pose information is determined based on the updated first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table.

[0072] Optionally, the joint configuration information determination module 340 is used for: If there are data with insufficient confidence in the first motion pose information, then the weight information corresponding to the data with insufficient confidence in the target matching table is reduced in weight. Based on the weight information in the target matching table, the inverse kinematics solution of the second motion pose information is performed to obtain the target joint configuration information of the robot.

[0073] The robot motion redirection device provided in this embodiment of the invention can execute a robot motion redirection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0074] Example 4 Figure 5 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0075] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0076] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0077] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as robot motion reversal methods.

[0078] In some embodiments, the robot motion redirection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the robot motion redirection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the robot motion redirection method by any other suitable means (e.g., by means of firmware).

[0079] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0080] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0081] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0082] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0083] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0084] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0085] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0086] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for redirecting robot motion, characterized in that, The method includes: Determine the target rigid body pose information of the reference object in the first calibration posture and the target link pose information of the robot in the second calibration posture; wherein, the first calibration posture and the second calibration posture are adapted in overall shape, the target rigid body and the target link are matched, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. The offset calibration information for each target matching item in the target rigid body pose information and the target link pose information is determined respectively; wherein, the offset calibration information includes translation offset calibration information and rotation offset calibration information, the translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system; The robot's second motion pose information is determined based on the first motion pose information of the target object and the bias calibration information of each target matching item; The target joint configuration information of the robot is determined based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

2. The method according to claim 1, characterized in that, Before determining the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information, the method further includes: The reference scale factor is determined based on the size difference between the robot and the reference object; The position information in the pose information of the target rigid body is updated according to the reference scale factor.

3. The method according to claim 1, characterized in that, After determining the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information, the method further includes: Obtain the foot rigid body position information from the target rigid body pose information, and obtain the foot link position information from the target link pose information; The height difference between the reference object and the robot in the vertical direction is determined based on the foot rigid body position information and the foot link position information; The robot's base height or root node translation is adjusted based on the height difference to ensure that the foot rigid body and foot link are at the same height.

4. The method according to claim 1, characterized in that, After determining the offset calibration information for each target matching item in the target rigid body pose information and the target link pose information, the method further includes: Determine the weight information of each target matching item under different motion scenarios; The matching bias table for different motion scenarios is constructed based on the bias calibration information and weight information of each target matching item.

5. The method according to claim 4, characterized in that, The robot's second motion pose information is determined based on the first motion pose information of the target object and the bias calibration information of each target matching item, including: Determine the target matching table from the matching bias table based on the current motion scenario; The robot's second motion pose information is determined based on the target object's first motion pose information at each motion moment and the offset calibration information in the target matching table.

6. The method according to claim 5, characterized in that, Before determining the robot's second motion pose information based on the target object's first motion pose information at each motion moment and the offset calibration information in the target matching table, the method further includes: Determine whether there are any data anomalies in the first motion pose information; wherein, the data anomalies include missing data and insufficient data confidence. If so, the predicted motion pose information corresponding to the abnormal data is inferred using the neighboring rigid body; The first motion pose information is updated based on the predicted motion pose information; Accordingly, the second motion pose information of the robot is determined based on the first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table, including: The robot's second motion pose information is determined based on the updated first motion pose information of the target object at each motion moment and the offset calibration information in the target matching table.

7. The method according to claim 5 or 6, characterized in that, The target joint configuration information of the robot is determined based on the second motion pose information, including: If there are data with insufficient confidence in the first motion pose information, then the weight information corresponding to the data with insufficient confidence in the target matching table is downweighted. Based on the weight information in the target matching table, the inverse kinematics solution of the second motion pose information is performed to obtain the target joint configuration information of the robot.

8. A robot motion redirection device, characterized in that, The device includes: The calibration pose information determination module is used to determine the target rigid body pose information of the reference object in the first calibration pose and the target link pose information of the robot in the second calibration pose; wherein, the first calibration pose and the second calibration pose are adapted in overall shape, the target rigid body and the target link are matched, and the target rigid body pose information and the target link pose information are described based on the same coordinate system. The offset calibration information determination module is used to determine the offset calibration information of each target matching item in the target rigid body pose information and the target link pose information respectively; wherein, the offset calibration information includes translation offset calibration information and rotation offset calibration information, the translation offset calibration information is used to describe the translation of the rigid body relative to the link in the link coordinate system, and the rotation offset calibration information is used to describe the rotation from the rigid body coordinate system to the link coordinate system; The motion pose information determination module is used to determine the second motion pose information of the robot based on the first motion pose information of the target object and the offset calibration information of each target matching item; The joint configuration information determination module is used to determine the target joint configuration information of the robot based on the second motion pose information, so as to control the robot to perform corresponding actions based on the target joint configuration information.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the robot motion reversal method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the robot motion reversal method according to any one of claims 1-7.

Images