Dual-robot pose error iterative compensation control method and system, and storage medium

Abstract

The application discloses a dual-robot pose error iterative compensation control method and system and a storage medium, and belongs to the field of dual-robot cooperative control and medical imaging technology. In view of the problems that it is difficult for a ball tube and a detector end target point to be real-time synchronized in a dual-robot CT system, and existing methods depend on dynamic control authority, the application collects the end target point pose through a motion capture system and quantifies the synchronization pose error, adopts a synchronization pose error distribution method to decompose the error to the expected trajectory compensation amount of the two robots according to a preset coefficient, then superimposes the error to the expected trajectory for iterative updating after zero-phase filtering processing, and only position control instructions are used to realize compensation. The application does not need torque or speed control authority, can converge the synchronization error to millimeter level, and effectively improves the CT reconstruction image quality.

Description

Technical Field

[0001] This embodiment belongs to the field of dual-robot cooperative control and medical imaging technology. Specifically, it relates to a dual-robot pose error iterative compensation control method, system, and storage medium. Background Technology

[0002] Traditional medical imaging equipment (such as O-type CT scanners and C-type CT scanners) relies on a rigid physical connection design integrating the X-ray tube and detector. When executing a CT scan trajectory based on an arc, the rigid structure effectively ensures the relative positional accuracy of the two components, thus meeting the basic requirements of CT scanning for imaging geometric accuracy. However, rigid structures have significant limitations in adapting to complex environments and flexibly adjusting scanning angles.

[0003] To overcome these limitations, researchers proposed a dual-robot CT system with a non-rigid connection between the X-ray tube and the detector, where two robots independently carry the X-ray tube and detector. The core technical challenge of this system lies in achieving proper alignment between the X-ray tube's light exit port and the detector center (i.e., the X-ray tube robot target point) throughout the entire CT scan process, following a pre-set circular trajectory. With the detector robot target Real-time and accurate pose synchronization.

[0004] Existing dual-robot systems face the following key challenges: First, the actual motion trajectory of the robot end effector is affected by multiple factors such as motion error, load deformation, and environmental disturbances, making it difficult to maintain dynamic and precise synchronization with the target point. This leads to X-ray beam propagation path deviation, causing problems such as blurred imaging and reduced spatial resolution, which seriously affects the accuracy of clinical diagnosis. Second, there is a risk of collision between the robot's workspace and the physical dimensions of the X-ray tube and detector. Existing methods cannot achieve precise synchronization of large-scale circular trajectories while meeting collision avoidance constraints. Third, most existing control methods rely on dynamic control authority such as robot torque or speed, which is not practical for industrial robots with only position control modes. Fourth, there is a lack of an effective mechanism for calculating the overall synchronization error of the two robots, decoupling it, and systematically distributing it to a single robot, making it difficult for each robot to perform precise compensation independently.

[0005] Therefore, there is an urgent need to develop an efficient method that can achieve precise synchronization of the end-effector target points of two robots by combining error modeling, error allocation and iterative learning compensation under only position control mode constraints. Summary of the Invention

[0006] The purpose of this application is to provide a dual-robot pose error iterative compensation control method, which has the advantages of being able to achieve decoupled allocation and iterative learning feedforward compensation of the synchronous pose error of the dual-robot end-point target based solely on the position control mode, thereby improving the engineering practicality, synchronous control accuracy and CT reconstruction image quality of the control method.

[0007] This application provides a dual-robot pose error iterative compensation control method, including:

[0008] Acquire real-time pose data of the end-effector targets of the X-ray tube robot and the detector robot;

[0009] Based on the real-time pose data, the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system is determined, and the synchronization pose error between the two robots is quantified based on the homogeneous transformation matrix.

[0010] A synchronous pose error allocation method is adopted to decompose the synchronous pose error into the expected trajectory compensation amount of each of the two robots according to a preset allocation coefficient.

[0011] The expected trajectory compensation values ​​of the two robots are processed by zero-phase filtering and then superimposed onto the corresponding expected trajectory. The motion control inputs of the two robots are iteratively updated until the convergence termination condition is met.

[0012] Furthermore, acquiring the real-time pose data of the end-effector target points of the X-ray tube robot and the detector robot includes:

[0013] Fluorescent ball marker connection devices are installed at the ends of the X-ray tube robot and the detector robot, respectively. Each connection device is equipped with multiple fluorescent ball markers to form a rigid body coordinate system.

[0014] The position vectors and attitude data of the two rigid body coordinate systems relative to the world coordinate system are collected in real time by the motion capture system.

[0015] Based on the fixed structural transformation relationship between each rigid body coordinate system and the corresponding end target coordinate system, the real-time pose data of the end target is determined.

[0016] Furthermore, determining the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system, and quantifying the synchronization pose error between the two robots based on the homogeneous transformation matrix, includes:

[0017] Based on the real-time pose of the X-ray tube side rigid body coordinate system and the detector side rigid body coordinate system relative to the world coordinate system, and the fixed transformation matrix between the pre-calibrated rigid body coordinate system and the target coordinate system, the homogeneous transformation matrix of the X-ray tube target coordinate system relative to the detector target coordinate system is derived.

[0018] The synchronization attitude error is determined based on the deviation between the rotation matrix in the homogeneous transformation matrix and the ideal unit rotation matrix.

[0019] The origin of the target coordinate system of the sphere is projected onto the target plane of the coupled coordinate system determined by the target coordinate system of the detector and the synchronization attitude error. The synchronization position error is determined based on the deviation between the projection point and the origin of the target coordinate system of the detector.

[0020] Furthermore, the synchronous pose error allocation method, which decomposes the synchronous pose error into the desired trajectory compensation amounts of the two robots according to a preset allocation coefficient, includes:

[0021] Based on the rotation vector of the synchronization attitude error and the preset allocation coefficient, the attitude calibration amounts of the ball tube robot and the detector robot are determined respectively.

[0022] The synchronization position error is projected onto the target plane of the coupled coordinate system, and the position calibration amount of the X-ray tube robot and the detector robot is determined according to the preset allocation coefficient.

[0023] Based on the attitude calibration, position calibration, and deviation between the actual and desired poses of the two robots' ends, the desired trajectory compensation for each robot is determined.

[0024] Furthermore, the step of superimposing the expected trajectory compensation amounts of the two robots after zero-phase filtering onto the corresponding expected trajectories, and iteratively updating the motion control inputs of the two robots, includes:

[0025] In the current iteration, the expected trajectory compensation amounts of the two robots are filtered by a zero-phase filter to obtain the filtered error correction amount.

[0026] The filtered error correction is weighted by the learning gain coefficient and then added to the expected trajectory of the current round to generate the expected trajectory of the next round.

[0027] The desired trajectory for the next round is sent as motion control input to the controllers of the two robots. After execution, the synchronization pose error quantization step is returned until the desired trajectory compensation amount meets the convergence termination condition.

[0028] Furthermore, the motion control input is a position control command, and the iterative update is performed without obtaining torque or speed control permissions from the two robots.

[0029] Furthermore, it also includes: decomposing the circular arc scanning trajectory of the two robots into at least two adjacent path segments, wherein each path segment corresponds to a different scanning radius or arc span; and independently performing the synchronization pose error quantization, error allocation, and iterative update for each path segment until each path segment meets the corresponding convergence termination condition.

[0030] Furthermore, the convergence termination condition includes:

[0031] The laser beam emitted by the laser emitter located at the end of the X-ray tube robot passes through the corresponding through-holes on the receiving plate located at the end of the detector robot at all waypoints of the scanning trajectory; or

[0032] The position synchronization error between the target point of the X-ray tube robot and the target point of the detector robot is less than a preset position threshold, and the attitude synchronization error is less than a preset attitude threshold.

[0033] Secondly, this embodiment also proposes a dual-robot pose error iterative compensation control system, including:

[0034] The acquisition unit is used to acquire real-time pose data of the end-effector target points of the ball tube robot and the detector robot.

[0035] The error calculation unit is used to determine the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system based on the real-time pose data, and to quantify the synchronization pose error between the two robots based on the homogeneous transformation matrix.

[0036] An error allocation unit is used to decompose the synchronous pose error into the expected trajectory compensation amount of each of the two robots according to a preset allocation coefficient using a synchronous pose error allocation method.

[0037] The iterative compensation unit is used to superimpose the expected trajectory compensation amounts of the two robots after zero-phase filtering onto the corresponding expected trajectory, and iteratively update the motion control inputs of the two robots until the convergence termination condition is met.

[0038] Thirdly, this embodiment also proposes a computer-readable storage medium storing a computer program; the program is loaded and executed by a processor to implement the steps of the dual-robot pose error iterative compensation control method as described in the first aspect.

[0039] This application has the following technical advantages:

[0040] First, the proposed Synchronous Pose Error Allocation (SPEAA) method integrates a bidirectional projection method with an error allocation mechanism to achieve coordinated allocation of attitude and position errors between the two robots. Compared with the traditional single-robot tracking control method, it is more suitable for multi-robot collaborative scenarios and the error compensation direction is more targeted.

[0041] Second, the ILFCC algorithm used only relies on the robot's position control mode and does not require dynamic control permissions such as torque and speed, which has stronger engineering applicability and can be directly adapted to industrial collaborative robots.

[0042] Third, a multi-stage iterative convergence mechanism is adopted, and the pose deviation between the two robot targets is gradually eliminated by using the principle of "incremental optimization". This effectively avoids system oscillation caused by large adjustments to the control input and ensures the stability of the synchronization error correction process.

[0043] Fourth, after ILFCC iterative compensation, key quality indicators such as MSE, SSIM, and PSNR of CT reconstructed images are improved, reducing geometric artifacts caused by projection data misalignment due to synchronization errors and improving the imaging quality of medical CT images. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of this embodiment, the accompanying drawings used in the embodiment will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this embodiment and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart illustrating the steps of the dual-robot pose error iterative compensation control method disclosed in this embodiment;

[0046] Figure 2 This is a schematic diagram of the coordinate system distribution of the dual-robot CT system disclosed in this embodiment;

[0047] Figure 3 This is a schematic diagram of the dual-robot CT system disclosed in this embodiment, showing the arc scanning trajectory and target point distribution;

[0048] Figure 4 This is a schematic diagram of the laser synchronization alignment detection experimental platform disclosed in this embodiment regarding the synchronous verification experiment of the motion layer;

[0049] Figure 5 This is a schematic diagram showing the configuration relationship between the three laser pointer array and the double-layer receiving aperture plate in the laser detection system disclosed in this embodiment;

[0050] Figure 6 This is a convergence curve of the first segment trajectory synchronization error as a function of the number of iterations in Experiment 1 disclosed in this embodiment;

[0051] Figure 7 This is a convergence curve of the second segment trajectory synchronization error as a function of the number of iterations in Experiment 1 disclosed in this embodiment;

[0052] Figure 8 This is a convergence curve of the first segment trajectory synchronization error as a function of the number of iterations in Experiment 2 disclosed in this embodiment;

[0053] Figure 9 This is a convergence curve of the second segment trajectory synchronization error as a function of the number of iterations in Experiment 2 disclosed in this embodiment;

[0054] Figure 10 This embodiment presents a comparison of CT reconstruction images of the turntable phantom before and after compensation, as well as a magnified view of a portion thereof.

[0055] Figure 11 This embodiment presents a comparison of CT reconstruction images of the school emblem resin phantom before and after compensation, as well as a magnified view of a portion thereof.

[0056] Figure 12 This is a schematic diagram of the dual-robot pose error iterative compensation control system disclosed in this embodiment.

[0057] The component numbers and names are shown in the attached diagram:

[0058] 1. X-ray tube robot; 2. Detector robot; 3. X-ray tube; 4. X-ray detector; 5. Fluorescent ball marker connection device (X-ray tube end); 6. Fluorescent ball marker connection device (detector end); 7. Three laser pointer connection device; 8. Double-layer laser spot receiving plate. Detailed Implementation

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which these embodiments belong; the terminology used herein and in the specification of the application is for the purpose of describing particular embodiments only and is not intended to limit these embodiments; the terms "comprising" and "having," and any variations thereof, in the specification of these embodiments and the foregoing description of the accompanying drawings, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification of these embodiments and the foregoing description of the accompanying drawings are used to distinguish different objects, not to describe a particular order.

[0060] The implementation details of the technical solution in this embodiment are described in detail below:

[0061] Firstly, this embodiment proposes an iterative compensation control method for dual-robot pose error, such as... Figure 1 As shown, the method includes:

[0062] S101, acquire real-time pose data of the end-effector target points of the ball tube robot and the detector robot.

[0063] Specifically, acquiring the real-time pose data of the end-effector target points of the X-ray tube robot and the detector robot includes: installing fluorescent ball marker connection devices at the ends of the X-ray tube robot and the detector robot respectively, with multiple fluorescent ball markers set on each connection device to form a rigid body coordinate system; real-time acquisition of the position vectors and attitude data of the two rigid body coordinate systems relative to the world coordinate system by a motion capture system; and determining the real-time pose data of the end-effector target points based on the fixed structural transformation relationship between each rigid body coordinate system and the corresponding end-effector target point coordinate system.

[0064] Specifically, such as Figure 2 The diagram shown illustrates the coordinate system distribution of the dual-robot CT system in this embodiment. Figure 3 The diagram shown illustrates the circular scanning trajectory and target distribution of the dual-robot CT system in this embodiment. The system hardware includes: two UR20 collaborative robots, an X-ray tube 3 (i.e., an X-ray emission tube), an X-ray detector 4, and a high-precision motion capture camera. The X-ray tube robot 1 has the X-ray emission tube 3 mounted at its end, and the detector robot 2 has the X-ray detector 4 mounted at its end.

[0065] Four symmetrically distributed fluorescent ball markers are precisely assembled on fluorescent ball marker connection devices 5 and 6. A spatial square configuration is formed. In the motion capture system software, dedicated tracking rigid bodies are established for both the X-ray tube robot 1 and the detector robot 2: X-ray tube robot rigid body coordinate system. With the rigid body coordinate system of the detector robot The geometric centers of the four fluorescent spheres are defined as the target points of the X-ray tube robot. With the detector robot target The transformation relationship between the coordinate system of the robot's rigid body center of mass and the target coordinate system is uniquely determined by the mechanical structural parameters of the marker connection device, and is a fixed constant.

[0066] The motion capture system acquires the position vectors and attitude data of the two rigid body coordinate systems relative to the world coordinate system in real time. The motion capture camera used in this embodiment is a Swift 30, with a data update frequency of 120 Hz. The motion capture system outputs the position vectors and quaternion attitudes of the two rigid body centers of mass in the world coordinate system in real time at a sampling rate of 120 Hz. Based on the spatial distribution of each marker point, the motion capture system establishes rigid body coordinate systems on the X-ray tube side. and the rigid body coordinate system on the detector side It outputs its pose relative to the world coordinate system {w} in real time.

[0067] Based on the fixed structural transformation relationship between each rigid body coordinate system and the corresponding end-target coordinate system, the real-time pose data of the end-target is determined. Specifically, a quaternion-rotation matrix transformation algorithm is used to map the rigid body pose quaternions into a three-dimensional orthogonal rotation matrix. This is combined with the position vector of the corresponding rigid body's center of mass in the world coordinate system, and the fixed transformation matrix uniquely determined by the mechanical structure parameters. , Derive the coordinate system of the X-ray tube target point Relative to the detector target coordinate system The homogeneous transformation matrix. For the X-ray tube side-mounted robot, define... The base coordinate system of the X-ray tube robot is used. Let X-ray tube side target point coordinate system be defined; for detector side robot, define The detector-side robot base coordinate system, Let be the coordinate system of the target point on the detector side. Through the above coordinate system definition and transformation relationships, the calculation process of synchronous pose error in a dual-robot CT system can be uniformly expressed.

[0068] S102, based on the real-time pose data, determine the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system, and quantify the synchronization pose error between the two robots based on the homogeneous transformation matrix.

[0069] Specifically, based on the real-time pose data of the rigid body coordinate systems of the X-ray tube side and the detector side relative to the world coordinate system obtained in step S101, this step aims to establish the relative pose mapping relationship between the end-effector target points of the two robots, thereby quantifying the synchronization pose error at the system level and providing accurate input for subsequent error allocation and compensation control.

[0070] Furthermore, in this embodiment, the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system is determined, and the synchronization pose error between the two robots is quantified based on the homogeneous transformation matrix. This includes: deriving the homogeneous transformation matrix of the X-ray tube target coordinate system relative to the detector target coordinate system based on the real-time pose of the X-ray tube side rigid body coordinate system and the detector side rigid body coordinate system relative to the world coordinate system, and the fixed transformation matrix between the pre-calibrated rigid body coordinate system and the target coordinate system; determining the synchronization pose error based on the deviation between the rotation matrix in the homogeneous transformation matrix and the ideal unit rotation matrix; projecting the origin of the X-ray tube target coordinate system onto the target plane of the coupled coordinate system determined by the detector target coordinate system and the synchronization pose error; and determining the synchronization position error based on the deviation between the projection point and the origin of the detector target coordinate system.

[0071] Specifically, the X-ray tube target coordinate system Relative to the detector target coordinate system The homogeneous transformation matrix is ​​calculated as follows:

[0072]

[0073] in, and Obtained through online real-time measurements by a motion capture system. and To obtain a fixed transformation matrix in advance through structural calibration, This achieves the transformation of the coordinate system from the world coordinate system to the detector-side rigid body coordinate system. Through the above matrix multiplication operations, the real-time relative pose of the X-ray tube target point with respect to the detector target point can be uniformly expressed.

[0074] Under ideal synchronization conditions, synchronization constraints include two conditions: attitude constraints require that the coordinate systems of the two target points be in the same orientation, i.e., the rotation matrix from the X-ray tube target coordinate system to the detector target coordinate system. Position constraints require that the origin of the X-ray tube target coordinate system be in the detector target coupled coordinate system. - The projection point of the plane (a coupled coordinate system combining the actual position of the target with the desired synchronization attitude) coincides with the origin of the detector target. Synchronization attitude error. By rotation vector Characterization, synchronization position error Characterized by projection deviation, both together constitute the overall synchronization error of the system.

[0075] Specifically, regarding the positional constraints, let's assume... The coupled coordinate system formed by the origin of the detector target point and the synchronous attitude is represented by the origin of the tube target point coordinate system. Projected onto this coupled coordinate system - Plane, to obtain the projection point ,Require relative to the origin of the detector target Overlap, that is:

[0076]

[0077] In actual system operation, due to factors such as calibration, mechanical assembly errors, and environmental interference, the target coordinate system's attitude is difficult to achieve absolute synchronization. The rotation matrix of the tube target coordinate system relative to the detector target coordinate system is... Using rotation vector This indicates that the synchronization attitude error is obtained:

[0078]

[0079] To decouple the calculation of attitude error from position error, a coupled coordinate system is constructed: the attitude rotation vector is converted into a rotation matrix, and the transformation matrix of the current target point coordinate system in the coupled coordinate system is constructed.

[0080]

[0081] in, This represents the rotation matrix from the X-ray tube target coordinate system to the X-ray tube coupled coordinate system. This represents the rotation matrix from the detector target coordinate system to the detector coupled coordinate system.

[0082] Furthermore, the target point position of the ball tube Projected onto the detector target coupled coordinate system - Plane, to obtain the projection point This leads to the synchronization position error. Synchronization position error is defined as the deviation between the current position of the projection point and the current position of the detector target point:

[0083]

[0084] Therefore, the system-level synchronization pose error is composed of the synchronization attitude error. Synchronization position error Together, they represent the degree of mismatch between the end-point targets of the two robots in terms of attitude angle and planar position, respectively, providing a quantitative basis for decomposing the system error into the expected trajectory compensation of each individual robot through the synchronous pose error allocation method.

[0085] S103, The synchronous pose error allocation method is adopted to decompose the synchronous pose error into the expected trajectory compensation amount of each of the two robots according to the preset allocation coefficient.

[0086] Specifically, in this embodiment, the Synchronous Pose Error Allocation (SPEAA) method is used to decompose the system-level synchronous pose error quantified in step S102 into the expected trajectory compensation amounts of the pneumatic tube robot and the detector robot according to a preset allocation coefficient, thereby realizing the mapping from system-level error to single-robot-level control objectives. The implementation process of this allocation method includes steps S3001 to S3003.

[0087] Furthermore, a synchronization pose error allocation method is adopted to decompose the synchronization pose error into the expected trajectory compensation amounts of the two robots according to a preset allocation coefficient, including:

[0088] Based on the rotation vector of the synchronous attitude error and the preset allocation coefficient, the attitude calibration amounts of the ball tube robot and the detector robot are determined respectively.

[0089] Specifically, the synchronization attitude error According to allocation coefficient Assignment: The rotation vector from the X-ray tube target coordinate system to the synchronous attitude is... The rotation vector from the detector target coordinate system to the synchronous attitude is: Therefore, the tube robot undertakes... The attitude calibration is performed by the detector robot. The attitude calibration amount is borne by the two robots according to the allocation coefficient, and the attitude mismatch of the system is shared by both robots.

[0090] The synchronization position error is projected onto the target plane of the coupled coordinate system, and the position calibration amount of the ball tube robot and the detector robot is determined according to the preset allocation coefficient.

[0091] Specifically, the position error projection and allocation are completed on the xy plane of the coupled coordinate system. Combined with the aforementioned synchronization target pose allocation results, the expected end-effector poses of the two robots in their respective base coordinate systems can be obtained. In this embodiment, α=0.5 is taken in the experiment, and the synchronization error is evenly distributed between the two robots, that is, each robot bears half of the synchronization position error correction amount, so as to ensure the symmetry and convergence stability of error compensation.

[0092] Based on the attitude calibration, position calibration, and deviation between the actual and desired poses of the two robots' ends, the desired trajectory compensation for each robot is determined.

[0093] Specifically, based on the first The actual pose data of the target point coordinate system collected in each iteration are used to calculate the expected trajectory compensation amount of the single robot in the pneumatic tube. Compensation amount for the expected trajectory of the detector robot alone Based on the real-time output and calibration results of the motion capture system, the actual pose of the robot's end effector in the base coordinate system is determined by the following transformation relationship:

[0094]

[0095] in, , Let represent the homogeneous transformation matrices of the robot's actual end-effector coordinate system relative to the robot's base coordinate system on the X-ray tube side and the detector side, respectively. Let represent the homogeneous transformation matrices of the world coordinate system relative to the robot base coordinate systems on the X-ray tube side and the detector side, respectively. , Let represent the homogeneous transformation matrices of the rigid body coordinate systems on the X-ray tube side and the detector side relative to the world coordinate system, respectively. , These represent the homogeneous transformation matrices of the robot's actual end-effector coordinate system relative to their respective rigid body coordinate systems on the X-ray tube side and the detector side, respectively. , These represent the rotation matrices of the actual end-effector coordinate systems of the robot on the X-ray tube side and the detector side relative to their respective robot base coordinate systems. , These represent the translation vectors of the actual end-effector coordinate systems of the X-ray tube side and the detector side relative to their respective robot base coordinate systems.

[0096] Accordingly, combining the aforementioned synchronization target pose allocation results, the expected synchronization end poses of the two robots in their respective base coordinate systems can be obtained as follows:

[0097] ,

[0098] .

[0099] in, , Let represent the homogeneous transformation matrices of the desired end-effector coordinate system relative to the robot's base coordinate system on the X-ray tube side and the detector side, respectively. Let represent the homogeneous transformation matrices of the robot target coordinate system on the X-ray tube side and the detector side relative to their respective rigid body coordinate systems. , Let represent the homogeneous transformation matrices of the desired target coordinate system relative to the target coordinate system for robot synchronization on the X-ray tube side and the detector side, respectively. , Let represent the homogeneous transformation matrices of the robot's rigid body coordinate system on the X-ray tube side and the detector side relative to their respective synchronous desired target point coordinate systems. , These represent the homogeneous transformation matrices of the desired end-effector coordinate system of the robot on the X-ray tube side and the detector side, respectively, relative to their respective rigid body coordinate systems.

[0100] Based on this, the synchronization error at the system level can be reconstructed into the desired trajectory compensation at the robot level. Specifically, the position error is defined as the translation difference between the synchronous desired end-effector pose and the actual end-effector pose in the corresponding base coordinate system, i.e.:

[0101] ,

[0102] .

[0103] in, This represents the expected trajectory compensation amount for the robot's position on the X-ray tube side. This represents the expected trajectory compensation amount for the robot's position on the detector side. , Let represent the translation vectors of the desired end-effector coordinate system relative to their respective base coordinate systems on the X-ray tube side and the detector side, respectively. , These represent the translation vectors of the robot's actual end-effector coordinate system relative to their respective base coordinate systems on the X-ray tube side and the detector side, respectively.

[0104] Attitude error is defined as the residual rotation required to compensate for the actual end-effector attitude to the synchronized desired end-effector attitude, i.e.:

[0105] ,

[0106] .

[0107] in, , These represent the attitude errors of the robot on the X-ray tube side and the detector side, respectively. , These represent the rotation matrices of the robot's actual end-effector coordinate system relative to its respective base coordinate system on the X-ray tube side and the detector side, respectively. , These represent the rotation matrices of the desired end-effector coordinate system of the robot on the X-ray tube side and the detector side relative to their respective base coordinate systems.

[0108] To obtain a unified robot-level error representation, the above posture errors are further converted into rotation vectors, denoted as follows: and Therefore, the expected trajectory compensation for the two robots can be uniformly expressed as:

[0109] ,

[0110] .

[0111] in, This represents the expected trajectory compensation amount for the uniform pose of the X-ray tube-side robot. This represents the expected trajectory compensation amount for the uniform pose of the robot on the detector side.

[0112] At this point, the system-level synchronization pose error has been completely decomposed into the independent expected trajectory compensation amounts for each of the two robots through the synchronization pose error allocation method, providing a precise single-robot level input for subsequent iterative learning feedforward compensation control.

[0113] S104: The expected trajectory compensation amounts of the two robots are processed by zero-phase filtering and then superimposed onto the corresponding expected trajectory. The motion control inputs of the two robots are iteratively updated until the convergence termination condition is met.

[0114] In this embodiment, the iterative learning feedforward compensation control process is based on the expected trajectory compensation amount of the unified pose of the X-ray tube side and the detector side obtained in step S103. Without obtaining the torque or speed control authority of the two robots, it achieves the gradual convergence of the synchronization error of the end target point of the two robots by only correcting the position control command round by round.

[0115] Furthermore, the desired trajectory compensation values ​​of the two robots are processed by zero-phase filtering and then superimposed onto the corresponding desired trajectories. The motion control inputs of the two robots are iteratively updated, including:

[0116] In the current iteration, the expected trajectory compensation amounts of the two robots are filtered by a zero-phase filter to obtain the filtered error correction amount.

[0117] The filtered error correction is weighted by the learning gain coefficient and then added to the expected trajectory of the current round to generate the expected trajectory of the next round.

[0118] The desired trajectory for the next round is sent as motion control input to the controllers of the two robots. After execution, the synchronization pose error quantization step is returned until the desired trajectory compensation amount meets the convergence termination condition.

[0119] Specifically, in the current iteration, the expected trajectory compensation amounts for both robots are filtered using a zero-phase filter to obtain the filtered error correction. The zero-phase filter can filter out high-frequency disturbances while avoiding phase distortion, ensuring the temporal consistency of the trajectory. Specifically, the ILFCC algorithm updates the expected trajectories of the two robots according to the following iterative learning update law:

[0120]

[0121] in, Indicates the first Learning gain in each iteration Indicates the first The expected trajectory compensation amount for the X-ray tube side robot in the next iteration. Indicates the first The expected trajectory compensation amount for the robot on the detector side in the next iteration.

[0122] Due to the desired pose trajectory of the dual-robot CT system , A zero-phase filter with sufficient smoothness (satisfying the condition of continuous differentiability). The filtering effect on smooth trajectories is negligible, and the iterative update formula simplifies to:

[0123]

[0124] Where δ is the zero-phase filter and σ is the positive learning gain coefficient. Desired trajectory , This can be viewed as the control experience accumulated before the nth iteration, and , The updated trajectory, which is the effective correction information obtained in the current iteration, is stored in the memory system as the control input for the next iteration.

[0125] The filtered error correction is weighted by the learning gain coefficient and then added to the desired trajectory of the current round to generate the desired trajectory of the next round. This desired trajectory of the next round is sent as a motion control input to the controllers of both robots. This desired trajectory is then used as a position control command to be issued to the robot controllers, driving the tube robot and the detector robot to execute the new trajectory.

[0126] After the drive execution returns to the quantization step of the synchronization pose error, the real-time pose data of the end-point target of the two robots are reacquired, and the quantization, allocation, and iterative update of the synchronization pose error are repeated to form a closed-loop compensation circuit. The above ILFCC compensation process is applied independently to all scanning trajectories until the synchronization accuracy of the dual robot scanning trajectories meets the convergence termination condition.

[0127] This embodiment employs a multi-stage iterative convergence mechanism, utilizing the principle of "incremental optimization" to gradually eliminate pose deviations between the two robot target points. This effectively avoids system oscillations caused by significant adjustments to control inputs, ensuring the stability of the synchronization error correction process. After ILFCC iterative compensation, key quality indicators such as MSE, SSIM, and PSNR of the CT reconstructed images are improved, reducing geometric artifacts caused by projection data misalignment due to synchronization errors and improving the imaging quality of medical CT images.

[0128] Furthermore, the motion control input is a position control command, and the iterative update is performed without obtaining torque or speed control permissions from the two robots.

[0129] In this embodiment, the two UR20 collaborative robots only have external interfaces for position control mode, and the robot controllers do not accept externally input torque or velocity loop control commands. Therefore, the desired trajectory for the next round generated by iterative learning compensation control is directly sent to the respective controllers of the two robots in the form of position control commands. The controllers drive the servo motors of each joint to perform positioning movements according to the received desired position trajectory.

[0130] Under the aforementioned control architecture, the entire synchronization error compensation process is achieved entirely through position-level correction of the desired trajectory, without needing to access the robot's internal dynamic model or obtain joint torque feedback or velocity loop adjustment permissions. The desired trajectory compensation amounts for both the ball tube robot and the detector robot are processed by zero-phase filtering and then compensated solely through iterative superposition of position commands. Each robot's underlying servo system independently performs trajectory tracking control.

[0131] Compared to traditional synchronous control methods that rely on dynamic models and torque / velocity control authority, the position control mode compatibility of this embodiment allows the synchronous pose error allocation and iterative learning compensation control method to be directly deployed on conventional industrial collaborative robot platforms without requiring underlying modifications to the robot control system or opening of dynamic control interfaces. This significantly reduces the engineering implementation difficulty of the dual-robot CT system and has stronger field applicability and promotion value.

[0132] Furthermore, the method in this embodiment also includes: decomposing the circular arc scanning trajectory of the two robots into at least two adjacent path segments, wherein each path segment corresponds to a different scanning radius or arc span; and independently performing the synchronization pose error quantization, error allocation, and iterative update for each path segment until each path segment meets the corresponding convergence termination condition.

[0133] In this embodiment, the complete scanning angle in the actual scanning task of the dual-robot CT system is 190°. Since the X-ray tube robot and the detector robot are respectively equipped with an X-ray tube and an X-ray detector at their ends, their physical dimensions are relatively large, and the working range and posture constraints of the two robots in the confined space differ. If a single 190° circular arc trajectory is directly executed, mechanical collisions are likely to occur between the robot body or the end-effector load during the scanning process. Therefore, to avoid the risk of collisions during the scanning process and to consider the accessible working space of each robot, the complete scanning trajectory is segmented.

[0134] The original 190° circular trajectory was decomposed into two adjacent path segments. The first path segment has an arc span of 100°, with a scanning radius of 55 mm for the X-ray tube robot and 45 mm for the detector robot. The second path segment has an arc span of 90° and smoothly connects to the first path segment. Both trajectories have the same scanning center and nominal X-ray tube-detector geometric relationship. Through this segmentation method, the attitude angles and joint configurations of both robots within each trajectory segment are within the safe and feasible region, effectively avoiding the risk of collision during movement.

[0135] The quantization, allocation, and iterative update of the synchronization pose error are performed independently for each path segment. Specifically, for the first trajectory segment, steps S102 to S104 are executed sequentially: the synchronization pose error of this trajectory segment is quantified based on the pose data collected in real time by the motion capture system; the synchronization pose error allocation method is used to decompose the error into the expected trajectory compensation amounts of the two robots; after zero-phase filtering, the expected trajectory is iteratively updated until the synchronization accuracy of this trajectory segment meets the corresponding convergence termination condition. Subsequently, the same process is used to independently iteratively compensate the second trajectory segment until the second trajectory segment also meets its convergence termination condition. The compensation processes of the two trajectory segments are independent of each other, and each has its own independent expected trajectory memory and error convergence determination.

[0136] Experimental results show that the segmented iterative compensation strategy has a significant effect on suppressing synchronization errors in both scanning trajectories. For example... Figure 6 The figure shown is a convergence curve of the synchronization error of the first segment of the trajectory in Experiment 1 of this embodiment, which varies with the number of iterations; as shown Figure 7 The figure shown is a convergence curve of the synchronization error of the second segment of the trajectory in Experiment 1 of this embodiment as a function of the number of iterations.

[0137] Experiment 1: Verification Experiment of Synchronous Motion Control

[0138] To verify the effectiveness of the dual-robot CT synchronization control method based on synchronous pose error allocation and iterative learning feedforward compensation proposed in this embodiment, a laser synchronization alignment experiment was conducted on a self-built dual-robot collaborative CT platform.

[0139] The experimental platform comprises two UR20 collaborative robots, an X-ray source, an X-ray detector, a high-precision motion capture system, a laser synchronous detection device, and a local recording device. The two robots move along a preset circular arc scanning trajectory, with a complete scanning angle of 190°. Considering the working range and attitude constraints of the two robots in the confined space, to avoid collisions during scanning, the complete scanning trajectory is divided into two adjacent trajectories for separate execution. The first trajectory corresponds to a 100° circular arc scan, and the second trajectory corresponds to a 90° circular arc scan. Both trajectories share the same scanning center and nominal X-ray tube-detector geometric relationship.

[0140] To visually verify the synchronization alignment effect, three laser emitters were installed at the end of the X-ray tube-side robot, arranged in an equilateral triangle with a spacing of 30 mm. A double-layer receiving plate was installed at the end of the detector-side robot. The first layer had circular through-holes corresponding to the three laser beams, with a diameter of 6 mm. The second layer was used to receive the laser spots after passing through the through-holes. Under ideal synchronization conditions, the three laser beams should be able to continuously pass through the corresponding through-holes throughout the scanning process. When there is a positional or orientational deviation between the X-ray tube-side robot and the detector-side robot, the laser beams will experience partial obstruction or even complete failure to pass through the through-holes. The perforation status of the three laser spots (A, B, C) was recorded in each iteration. The perforation status was divided into four categories: Status 0 – the spot is blocked by the first layer of the perforated plate; Status 1 – the spot is briefly perforated; Status 2 – the spot is intermittently perforated; Status 3 – the spot is continuously and stably perforated.

[0141] During the experiment, rigid body markers were installed on the laser device and the receiving plate, respectively. The pose information of the rigid bodies on the X-ray tube side and the detector side was acquired in real time using a motion capture system, and the synchronization position error and synchronization attitude error of the two robots during the current scanning process were calculated accordingly. Then, using the synchronization pose error allocation method proposed in this embodiment, the system-level synchronization error was decomposed into the expected trajectory compensation amount for each of the two robots, and the reference trajectory for the next scan was updated using an iterative learning feedforward compensation method. In this embodiment, the error allocation coefficient is set to 0.5, meaning the synchronization error is symmetrically distributed between the two robots. Each trajectory segment is repeatedly scanned and compensated starting from the initial trajectory until the laser beam stably penetrates the hole.

[0142] Synchronous motion control experimental results

[0143] Experimental results show that the method in this embodiment has a significant effect on suppressing synchronization error for both scanning trajectories, and exhibits good repeatability and stability.

[0144] For the first trajectory segment, without compensation, the average position synchronization error was 7.10 mm, the root mean square error was 7.22 mm, and the maximum error was 10.50 mm. After 8 iterations of compensation, the average position synchronization error decreased to 0.89 mm, the root mean square error decreased to 0.94 mm, and the maximum error decreased to 2.77 mm. That is, the average error was reduced to about 1 / 8 of the original, the root mean square error was reduced to about 1 / 7.7 of the original, and the maximum error was reduced to about 1 / 3.8 of the original.

[0145] For the second trajectory segment, without compensation, the average position synchronization error was 6.12 mm, the root mean square error was 6.18 mm, and the maximum error was 7.77 mm. After 8 iterations of compensation, the average position synchronization error decreased to 0.76 mm, the root mean square error decreased to 0.79 mm, and the maximum error decreased to 1.51 mm. That is, the average error decreased to approximately 1 / 8.1 of its original value, the root mean square error decreased to approximately 1 / 7.8 of its original value, and the maximum error decreased to approximately 1 / 5.1 of its original value.

[0146] From the perspective of laser perforation status, in the initial stages of both the first and second trajectory segments, three laser points could not be fully perforated. With each compensation iteration, the laser points gradually transitioned from being completely unable to perforate to momentary perforation and then intermittent perforation. Finally, after the 8th iteration, all three laser points were able to stably pass through the corresponding through-holes throughout the entire scan, achieving a stable perforation state throughout, as shown in Table 1. This result demonstrates that the method in this embodiment not only numerically reduces synchronization errors but also physically restores the X-ray tube-detector geometric alignment required by a dual-robot CT system.

[0147] Table 1. Evolution of laser spot penetration state

[0148]

[0149] Therefore, the method proposed in this embodiment can effectively improve the motion layer synchronization accuracy of the dual-robot CT system, providing a stable and reliable motion basis for subsequent CT imaging.

[0150] like Figure 8 The figure shown is a convergence curve of the synchronization error of the first segment of the trajectory in Experiment 2 of this embodiment, which varies with the number of iterations; as shown... Figure 9 The figure shown is a convergence curve of the synchronization error of the second segment of the trajectory in Experiment 2 of this embodiment as a function of the number of iterations.

[0151] Experiment 2: CT Radiographic Imaging Verification Experiment

[0152] After completing the above-mentioned synchronous compensation, in order to further verify the effect of the method of this embodiment on improving CT imaging quality, a CT imaging experiment was carried out using the compensated dual-robot scanning trajectory.

[0153] In this embodiment, the same synchronization compensation process as in the aforementioned motion layer experiment is first employed to perform eight iterations of learning on the dual-robot CT system, enabling the X-ray tube-side robot and the detector-side robot to achieve high-precision synchronization. After the synchronization error converges, CT projection data is acquired using the compensated scan trajectory, and image reconstruction is performed.

[0154] To comprehensively evaluate the impact of the method in this embodiment on image quality, two types of phantoms were selected as test objects. The first type was a cylindrical phantom containing holes of different diameters, namely 1 mm, 2 mm, 3 mm, 5 mm, and 7 mm, used to verify the system's ability to recover multi-scale geometric structures. The second type was a resin 3D-printed school badge phantom with a minimum stroke width of 2 mm, used to verify the system's ability to recover fine structural edges and local details. Reference images acquired by professional CT equipment were used as the comparison benchmark. Image quality evaluation metrics included mean squared error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM).

[0155] CT radiographic imaging experimental results

[0156] During the compensated CT scan, the two scan trajectories maintained a low level of synchronization error. Specifically, after eight compensations, the average positional synchronization error of the first trajectory in the actual CT scan decreased from 8.38 mm to 1.14 mm, and the root mean square error decreased from 8.5 mm to 1.23 mm; the average positional synchronization error of the second trajectory in the actual CT scan decreased from 5.7 mm to 0.89 mm, and the root mean square error decreased from 5.72 mm to 0.97 mm. This demonstrates that the improved synchronization accuracy obtained in the aforementioned motion layer experiment can be maintained during actual CT image acquisition.

[0157] For the cylindrical phantom, as shown in Table 2, the average MSE of the five scan results before compensation was 163.93, the average SSIM was 0.7915, and the average PSNR was 25.99 dB. After compensation, the average MSE of the five scan results decreased to 120.72, the average SSIM increased to 0.7988, and the average PSNR increased to 27.31 dB. That is, the MSE decreased by approximately 35.8% and the PSNR increased by 1.32 dB after compensation, indicating a reduction in overall image distortion and clearer restoration of the boundaries of holes at different scales. Particularly for the 2 mm and 7 mm hole regions, the edge contours were more defined after compensation, and local artifacts and blurring were significantly reduced.

[0158] Table 2 Comparison of CT image quality indicators before and after turntable phantom compensation

[0159]

[0160] Table 3 Comparison of CT image quality indicators before and after compensation for the school emblem resin phantom

[0161]

[0162] For the school emblem phantom, as shown in Table 3, the average MSE of the five scan results before compensation was 111.56, the average SSIM was 0.9779, and the average PSNR was 27.65 dB. After compensation, the average MSE of the five scan results decreased to 74.62, the average SSIM increased to 0.9854, and the average PSNR increased to 29.42 dB. That is, the MSE decreased by approximately 49.5% and the PSNR increased by 1.77 dB after compensation. This result indicates that for finely structured phantoms containing small strokes, letter and number regions, the method of this embodiment can more significantly improve edge sharpness and local detail resolution, resulting in higher separation between adjacent fine lines and a smoother and clearer outline.

[0163] For the first trajectory segment, the average position synchronization error was 7.10 mm without compensation, which decreased to 0.89 mm after 8 iterations of compensation. For the second trajectory segment, the average position synchronization error was 6.12 mm without compensation, which decreased to 0.76 mm after 8 iterations of compensation. The laser perforation state of both trajectory segments gradually improved with the increase of the number of iterations, and finally, stable laser beam perforation was achieved during the complete scanning of each trajectory segment, verifying the effectiveness and reliability of the segmented independent compensation strategy.

[0164] Furthermore, the convergence termination conditions include: (1) the laser beam emitted by the laser emitter set at the end of the ball tube robot passes through the corresponding through hole on the receiving plate set at the end of the detector robot at all waypoints of the scanning trajectory; or, (2) the position synchronization error between the target point of the ball tube robot and the target point of the detector robot is less than a preset position threshold, and the attitude synchronization error is less than a preset attitude threshold.

[0165] Specifically, regarding the convergence termination condition (1). For example... Figure 4 The diagram shown is a schematic of the laser synchronization alignment and detection experimental platform used in the motion layer synchronization verification experiment of this embodiment. Figure 5 The diagram shown is a schematic representation of the configuration relationship between the three laser pointer array and the double-layer receiving aperture plate in the laser detection system of this embodiment.

[0166] In this embodiment, to visually verify the synchronization alignment effect, three laser emitters are installed at the end of the X-ray tube-side robot. These three laser emitters are arranged in an equilateral triangle with a spacing of 30 mm. A double-layer receiving plate is installed at the end of the detector-side robot. The first layer has circular through-holes corresponding to the three laser beams, with a diameter of 6 mm. The second layer receives the laser beams after they pass through the through-holes. Under ideal synchronization conditions, the three laser beams should be able to continuously pass through the corresponding through-holes throughout the entire scanning process. When there is a positional or orientational deviation between the X-ray tube-side robot and the detector-side robot, the laser beams will experience partial obstruction or even complete failure to pass through the through-holes. The penetration status of the three laser spots (A, B, C) in each iteration is recorded. The penetration status is divided into four categories: State 0 indicates that the spot is blocked by the first layer of the perforated plate; State 1 indicates that the spot is briefly penetrated; State 2 indicates that the spot is intermittently penetrated; and State 3 indicates that the spot is continuously and stably penetrated. The convergence termination condition is set to state 3 at all discrete waypoints of the scanning trajectory for all three laser beams, that is, the laser beams pass through the corresponding through holes without obstruction during the entire scanning process. This is used as the physical layer criterion for determining that the target point of the dual robots has achieved high-precision spatial alignment.

[0167] For the convergence termination condition (2), the numerical error index is used as the convergence termination condition, that is, the position synchronization error between the X-ray tube robot target point and the detector robot target point is less than a preset position threshold, and the attitude synchronization error is less than a preset attitude threshold. In this embodiment, based on the imaging accuracy requirements of the dual-robot CT system and the iterative convergence experimental data, the preset position threshold is set to 2 mm, and the preset attitude threshold is set to 1°. When the position synchronization error and attitude synchronization error between the end targets of the two robots simultaneously meet the above threshold requirements, it is determined that the iterative compensation of the trajectory segment has reached the convergence standard, and the iterative learning process of the current path segment can be terminated.

[0168] The two convergence termination conditions mentioned above can be used individually or in combination in practical applications. In Experiment 1, the laser perforation state was used as the main criterion. After eight iterations of compensation, the three laser points of the first and second trajectory segments were able to stably pass through the corresponding through-holes throughout the entire scan. Simultaneously, the average position synchronization error of the first trajectory segment decreased from 7.10 mm to 0.89 mm, and the average position synchronization error of the second trajectory segment decreased from 6.12 mm to 0.76 mm, both meeting the accuracy requirements of the numerical threshold. In the CT radiographic verification of Experiment 2, the average position synchronization error of the first trajectory segment in the actual CT scan decreased to 1.14 mm after compensation, and that of the second trajectory segment decreased to 0.89 mm, further verifying the effectiveness of the convergence termination conditions under real imaging loads.

[0169] To verify the improvement in CT reconstructed image quality after compensation, imaging experiments were conducted using cylindrical turntable phantoms containing holes of different diameters and resin 3D-printed school badge phantoms, with reference images acquired by professional CT equipment used as a comparison benchmark. Figure 10 The image shown is a comparison of CT reconstruction images of the turntable phantom before and after compensation in this embodiment, along with a magnified view of a local area. It can be seen that after compensation, the boundaries of holes at different scales are more clearly restored, and local artifacts are significantly reduced. Figure 11 The image shown is a comparison of the CT reconstruction images of the school emblem resin phantom before and after compensation in this embodiment, as well as a magnified view of a part of it; it can be seen that the sharpness of the fine structural edges and the ability to resolve local details are significantly improved after compensation.

[0170] Secondly, this embodiment also proposes a dual-robot pose error iterative compensation control system, such as... Figure 12 As shown, it includes:

[0171] The acquisition unit 1201 is used to acquire real-time pose data of the end-effector target points of the ball tube robot and the detector robot.

[0172] Error calculation unit 1202 is used to determine the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system based on the real-time pose data, and to quantify the synchronization pose error between the two robots based on the homogeneous transformation matrix.

[0173] Error allocation unit 1203 is used to decompose the synchronous pose error into the expected trajectory compensation amount of each of the two robots according to a preset allocation coefficient by using a synchronous pose error allocation method.

[0174] The iterative compensation unit 1204 is used to superimpose the expected trajectory compensation amounts of the two robots after zero-phase filtering onto the corresponding expected trajectory, and iteratively update the motion control inputs of the two robots until the convergence termination condition is met.

[0175] This system can be used to execute the dual-robot pose error iterative compensation control method described in the first aspect, which will not be elaborated further here.

[0176] Thirdly, this embodiment also proposes a computer-readable storage medium storing a computer program loaded and executed by a processor to implement the steps of the dual-robot pose error iterative compensation control method as described in the first aspect.

[0177] The computer-readable storage medium may be non-volatile or volatile. Optionally, the computer-readable storage medium includes: read-only memory (ROM), random access memory (RAM), magnetic disk, optical disk, solid-state drive, USB flash drive, or other removable storage device, as well as any combination of the above storage media.

[0178] Furthermore, this embodiment also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the steps of the dual-robot pose error iterative compensation control method as described in the first aspect.

[0179] Furthermore, this embodiment also provides a computer device, which includes a processor and a memory. The memory stores a computer program, and the processor calls and executes the computer program stored in the memory, so that the computer device performs the steps of the dual-robot pose error iterative compensation control method as described in the first aspect. The computer device may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices; the processor may be a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices.

[0180] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A dual-robot pose error iterative compensation control method, characterized in that, include: Acquire real-time pose data of the end-effector targets of the X-ray tube robot and the detector robot; Based on the real-time pose data, a homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system is determined, and the synchronization pose error between the two robots is quantified based on the homogeneous transformation matrix. This includes: deriving the homogeneous transformation matrix of the X-ray tube target coordinate system relative to the detector target coordinate system based on the real-time pose of the X-ray tube side rigid body coordinate system and the detector side rigid body coordinate system relative to the world coordinate system, and a pre-calibrated fixed transformation matrix between the rigid body coordinate system and the target coordinate system; determining the synchronization pose error based on the deviation between the rotation matrix in the homogeneous transformation matrix and the ideal unit rotation matrix; projecting the origin of the X-ray tube target coordinate system onto the target plane of the coupled coordinate system determined by the detector target coordinate system and the synchronization pose error, and determining the synchronization position error based on the deviation between the projection point and the origin of the detector target coordinate system. A synchronous pose error allocation method is adopted, which decomposes the synchronous pose error into the expected trajectory compensation amounts of the two robots according to a preset allocation coefficient. This includes: determining the pose calibration amounts of the X-ray tube robot and the detector robot respectively based on the rotation vector of the synchronous pose error and the preset allocation coefficient; projecting the synchronous position error onto the target plane of the coupled coordinate system, and determining the position calibration amounts of the X-ray tube robot and the detector robot respectively based on the preset allocation coefficient; and determining the expected trajectory compensation amounts of the two robots respectively based on their pose calibration amounts, position calibration amounts, and the deviation between their actual end-effector poses and expected poses. The expected trajectory compensation values ​​of the two robots are processed by zero-phase filtering and then superimposed onto the corresponding expected trajectory. The motion control inputs of the two robots are iteratively updated until the convergence termination condition is met.

2. The dual-robot pose error iterative compensation control method according to claim 1, characterized in that, The acquisition of real-time pose data of the end-effector target points of the X-ray tube robot and the detector robot includes: Fluorescent ball marker connection devices are installed at the ends of the X-ray tube robot and the detector robot, respectively. Each connection device is equipped with multiple fluorescent ball markers to form a rigid body coordinate system. The position vectors and attitude data of the two rigid body coordinate systems relative to the world coordinate system are collected in real time by the motion capture system. Based on the fixed structural transformation relationship between each rigid body coordinate system and the corresponding end target coordinate system, the real-time pose data of the end target is determined.

3. The dual-robot pose error iterative compensation control method according to claim 2, characterized in that, The step of superimposing the expected trajectory compensation values ​​of the two robots after zero-phase filtering onto the corresponding expected trajectory, and iteratively updating the motion control inputs of the two robots, includes: In the current iteration, the expected trajectory compensation amounts of the two robots are filtered by a zero-phase filter to obtain the filtered error correction amount. The filtered error correction is weighted by the learning gain coefficient and then added to the expected trajectory of the current round to generate the expected trajectory of the next round. The desired trajectory for the next round is sent as motion control input to the controllers of the two robots. After execution, the synchronization pose error quantization step is returned until the desired trajectory compensation amount meets the convergence termination condition.

4. The dual-robot pose error iterative compensation control method according to claim 1, characterized in that, The motion control input is a position control command, and the iterative update is performed without obtaining torque or speed control permissions from the two robots.

5. The dual-robot pose error iterative compensation control method according to claim 1, characterized in that, Also includes: The circular scanning trajectories of the two robots are decomposed into at least two adjacent path segments, each path segment corresponding to a different scanning radius or arc span; For each path segment, the synchronization pose error quantization, error allocation, and iterative update are performed independently until each path segment meets the corresponding convergence termination condition.

6. The dual-robot pose error iterative compensation control method according to claim 5, characterized in that, The convergence termination conditions include: The laser beam emitted by the laser emitter located at the end of the X-ray tube robot passes through the corresponding through-holes on the receiving plate located at the end of the detector robot at all waypoints of the scanning trajectory; or The position synchronization error between the target point of the X-ray tube robot and the target point of the detector robot is less than a preset position threshold, and the attitude synchronization error is less than a preset attitude threshold.

7. A dual-robot pose error iterative compensation control system, used to execute the method according to any one of claims 1-6, characterized in that, include: The acquisition unit is used to acquire real-time pose data of the end-effector target points of the ball tube robot and the detector robot. The error calculation unit is used to determine the homogeneous transformation matrix between the X-ray tube target coordinate system and the detector target coordinate system based on the real-time pose data, and to quantify the synchronization pose error between the two robots based on the homogeneous transformation matrix. An error allocation unit is used to decompose the synchronous pose error into the expected trajectory compensation amount of each of the two robots according to a preset allocation coefficient using a synchronous pose error allocation method. The iterative compensation unit is used to superimpose the expected trajectory compensation amounts of the two robots after zero-phase filtering onto the corresponding expected trajectory, and iteratively update the motion control inputs of the two robots until the convergence termination condition is met.

8. A computer-readable storage medium storing a computer program; characterized in that, The program is loaded and executed by a processor to implement the steps of the dual-robot pose error iterative compensation control method as described in any one of claims 1-6.

Images