Intelligent stacking industrial robot for packaging production line

Abstract

This invention relates to the field of intelligent robotic arm technology, and discloses an intelligent palletizing industrial robot for packaging production lines, comprising a mounting frame, a base rotation drive assembly, a first joint assembly, a second joint assembly, an end effector assembly, a palletizing fixture assembly, and control components. The control components include a pose acquisition module, a deviation analysis module, a trajectory planning module, a simulation verification module, an execution monitoring module, and a final confirmation module, all electrically connected in sequence. The pose acquisition module performs multi-source cross-comparison to output a preliminary target pose; the deviation analysis module performs corrections using anisotropic temperature compensation and vibration suppression; the trajectory planning module performs load feedforward and space avoidance compensation; the simulation verification module performs collision pre-verification; the execution monitoring module provides graded responses to tracking errors; and the final confirmation module updates the historical database. This invention improves palletizing accuracy and environmental adaptability.

Description

Technical Field

[0001] This invention relates to the field of intelligent robotic arm technology, specifically to an intelligent palletizing industrial robot for packaging production lines. Background Technology

[0002] In the palletizing operation of the packaging production line, multi-joint industrial robots are usually used to grasp and stack the packaged goods. Existing palletizing robots generally include a base, joint components, end effector and palletizing fixture. Their control method is mainly based on teach-and-playback or offline programming, and they are combined with vision sensors to complete target positioning.

[0003] However, existing palletizing industrial robots still have the following technical shortcomings:

[0004] First, pose acquisition relies on a single sensor and lacks cross-verification of multi-source information. When visual or ultrasonic data is interfered with by lighting, reflection, or occlusion, it is easy to produce incorrect positioning. The theoretical position calculated by the joint encoder has transmission backlash and flexible deformation error, which leads to the offset of the gripping point.

[0005] Second, the ambient temperature fluctuations and base vibrations are not properly compensated. Thermal expansion and contraction of the robotic arm links cause end-effector position drift. Existing technologies do not perform anisotropic temperature compensation for horizontal X, Y, vertical Z and orientation angles respectively. Vibration signals are ignored as noise or incorrectly used to directly correct the static target pose. There is a lack of vibration suppression coefficient and dynamic deceleration strategy.

[0006] Third, there is a lack of real-time compensation for load changes and dynamic obstacles. The mass and center of gravity of the packaged items being grasped change randomly. Existing controllers cannot compensate for the torque of each joint based on the Jacobian matrix feedforward. Pre-programmed trajectories cannot autonomously avoid temporary obstacles and are prone to collisions.

[0007] Fourth, the control process lacks closed-loop verification and self-learning, and lacks secondary data resampling and correction after pose fusion and trajectory planning; there is no graded response mechanism for tracking errors during execution; and the actual deviation is not fed back to the historical database after the operation is completed, making it impossible to optimize the subsequent positioning reference.

[0008] Therefore, there is a need to provide a palletizing industrial robot that can perform multi-source cross-validation, anisotropic temperature compensation, load feedforward compensation, double resampling correction, simulation verification, and self-learning update, in order to improve the palletizing accuracy and robustness in complex environments. Summary of the Invention

[0009] In order to solve the technical problems in the prior art, this application provides an intelligent palletizing industrial robot for packaging production lines.

[0010] The intelligent palletizing industrial robot for packaging production line provided in this application adopts the following technical solution: it includes a mounting frame, a base rotation drive assembly mounted on the mounting frame, a first joint assembly that is transmissionally connected to the base rotation drive assembly, a second joint assembly that is connected to the first joint assembly, an end effector assembly disposed at the end of the second joint assembly, and a palletizing fixture assembly mounted on the end effector assembly, and also includes a control component disposed on the mounting frame.

[0011] The control unit includes a pose acquisition module, a deviation analysis module, a trajectory planning module, a simulation verification module, an execution monitoring module, and a final confirmation module, which are connected in sequence.

[0012] The pose acquisition module is used to acquire real-time images of the packaging, the distance to the surface of the packaging, the rotation angle of the base rotation drive component, and the joint angle of the first joint component and the second joint component, and processes the acquired information to output a preliminary target pose.

[0013] The deviation analysis module is used to compare the preliminary target pose with the ideal reference pose in the historical palletizing parameter database, and to make corrections based on ambient temperature and vibration acceleration, and output the corrected target pose.

[0014] The trajectory planning module is used to generate a sequence of motion commands after load compensation and spatial avoidance compensation based on the corrected target pose and real-time load force data.

[0015] The simulation verification module is used to pre-verify the motion command sequence in a virtual simulation environment, and output an executable instruction set after the verification is successful;

[0016] The execution monitoring module is used to monitor and compare the robot's actual motion state with the expected motion state in real time during the execution of the executable instruction set, and make corrections accordingly.

[0017] The final confirmation module is used to compare the actual palletizing position with the target pose after the palletizing operation is completed, and update the historical palletizing parameter database based on the comparison result.

[0018] In summary, this application includes at least one of the following beneficial technical effects:

[0019] 1. This invention achieves closed-loop control and multiple redundancy checks throughout the entire process from target recognition to palletizing by setting up a closed-loop control system that includes a pose acquisition module, a deviation analysis module, a trajectory planning module, a simulation verification module, an execution monitoring module, and a final confirmation module, and by introducing a first data resampling correction and a second data resampling correction. In particular, the pose acquisition module cross-compares the theoretical pose calculated from vision, ultrasonic ranging, and joint angles, effectively avoiding erroneous positioning caused by errors from a single sensor. The two data resampling corrections perform secondary confirmation and weighted fusion of key data after pose fusion and trajectory planning, respectively, which significantly improves the robustness and fault tolerance of the system and reduces the risk of operation failure due to instantaneous interference or sudden environmental changes.

[0020] 2. This invention employs anisotropic temperature compensation correction in the deviation analysis module. Based on the linear expansion coefficient of each link material of the robotic arm and the equivalent thermal expansion length in different directions, independent compensation is performed for the horizontal X direction, horizontal Y direction, vertical Z direction, and orientation angle, overcoming the problem of large deviations between the traditional uniform temperature compensation model and the actual situation. At the same time, the vibration signal is low-pass filtered and the vibration suppression coefficient is calculated. The maximum motion speed and acceleration are dynamically reduced in the execution monitoring module. Combined with the Kalman filter in the pose acquisition module, vibration measurement errors are suppressed, avoiding the unreasonable correction of directly superimposing vibration interference onto the static target pose. This significantly improves the palletizing accuracy of the robot in complex industrial environments such as temperature changes and vibrations.

[0021] 3. In the trajectory planning module, this invention employs a feedforward load compensation method based on a six-dimensional force / torque sensor and a Jacobian matrix. By measuring the load force vector and calculating the compensation torque of each joint, the compensation amount is superimposed on the torque command of the servo controller without modifying the target joint angle, thus achieving precise adaptive compensation for dynamic loads. Simultaneously, it automatically inserts safe intermediate points for spatial avoidance by combining a three-dimensional model of the surrounding environment, and superimposes the expected maximum error vector in the simulation verification module for collision prediction and clamping reliability analysis, effectively avoiding collisions and grasping failures in actual operations, greatly improving operational safety and adaptability to unstructured environments.

[0022] 4. This invention achieves refined real-time intervention in trajectory tracking errors by implementing a hierarchical response mechanism in the monitoring module and a closed-loop dynamic adjustment of the clamping force; the final confirmation module updates the historical palletizing parameter database after comparing the actual palletizing pose with the target pose, enabling the system to have self-learning capabilities and continuously optimize the ideal reference pose based on historical successful experiences, thereby forming an intelligent closed loop of perception-decision-execution-feedback-learning. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the overall three-dimensional structure of the present invention;

[0024] Figure 2 This is a schematic diagram of the structure of the base rotation drive assembly and the palletizing fixture assembly of the present invention;

[0025] Figure 3 This is a block diagram of the architecture of the control component of the present invention;

[0026] Figure 4 This is a flowchart illustrating the workflow of the pose acquisition module of the present invention.

[0027] Figure 5 This is a flowchart illustrating the workflow of the deviation analysis module of the present invention.

[0028] Figure 6 This is a flowchart illustrating the workflow of the trajectory planning module of the present invention.

[0029] Figure 7 This is a flowchart illustrating the workflow of the simulation verification module of the present invention.

[0030] Figure 8 This is a flowchart illustrating the workflow of the monitoring module in this invention.

[0031] Figure 9 This is a flowchart illustrating the workflow of the final confirmation module of this invention.

[0032] Explanation of reference numerals in the attached drawings: 1. Mounting frame; 2. Base rotation drive assembly; 3. First joint assembly; 301. Second joint assembly; 4. End effector assembly; 5. Palletizing fixture assembly; 6. Control unit. Detailed Implementation

[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The intelligent palletizing industrial robot for packaging production lines involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figures 1-9The diagram illustrates an intelligent palletizing industrial robot for a packaging production line, comprising a mounting frame 1, which serves as the supporting foundation for the entire robot, used to fix and support other functional components. A base rotation drive assembly 2 is mounted on the mounting frame 1, providing rotational motion around a vertical axis to achieve overall steering of the robot. The output end of the base rotation drive assembly 2 is connected to a first joint assembly 3 via a splined shaft or servo flange. The first joint assembly 3 includes one or more swing arms and corresponding reducers and servo motors, used to achieve a wide range of height changes. The end of the first joint assembly 3 is further connected to a second joint assembly 301 via a joint pin, the second joint assembly 301 being used for... The robot arm allows for more precise posture adjustments, including a multi-degree-of-freedom fine-tuning mechanism. An end effector 4 is located at the end of the second joint assembly 301. This end effector 4 serves as an interface connecting the palletizing fixture and the robot arm, and integrates power lines and signal cables. A palletizing fixture assembly 5 is mounted on the end effector 4 via bolts or a quick-change mechanism. This palletizing fixture assembly 5 is used to directly grasp, hold, or suck up packaging boxes or bags to complete the palletizing operation. The palletizing industrial robot also includes a control component 6 mounted on the mounting frame 1. The control component 6 is located at an appropriate position on the side or inside the mounting frame 1 and is connected to each motion component via cables or wireless means. It is used to control the coordinated movements of each component according to a preset program or external commands.

[0035] In this embodiment, it should be specifically noted that the mounting frame 1 preferably adopts a welded steel structure frame, the surface of which is sandblasted and rust-removed and coated with an anti-rust coating to ensure structural stability and corrosion resistance during long-term use in industrial sites; the base rotation drive assembly 2 preferably adopts a structure in which a harmonic reducer is directly connected to a servo motor to achieve high-precision rotation positioning; the first joint assembly 3 and the second joint assembly 301 are both driven by servo motors, and work with precision planetary reducers to achieve high-rigidity power transmission; therefore, the mounting frame 1, base rotation drive assembly 2, first joint assembly 3, second joint assembly 301, end effector assembly 4, and palletizing fixture assembly 5 mentioned in this embodiment are all mature components widely used in industrial robots and automated palletizing equipment in this field, and the specific structural forms, driving methods, and transmission connection methods of these components are common knowledge or conventional technical means known to those skilled in the art.

[0036] It should be further explained that the control unit 6 includes a pose acquisition module, a deviation analysis module, a trajectory planning module, a simulation verification module, an execution monitoring module, and a final confirmation module. The output terminal of the pose acquisition module is electrically connected to the input terminal of the deviation analysis module, the output terminal of the deviation analysis module is electrically connected to the input terminal of the trajectory planning module, the output terminal of the trajectory planning module is electrically connected to the input terminal of the simulation verification module, the output terminal of the simulation verification module is electrically connected to the input terminal of the execution monitoring module, and the output terminal of the execution monitoring module is electrically connected to the input terminal of the final confirmation module, forming a complete data processing and control link.

[0037] The information acquired by the pose acquisition module includes: real-time images of packaging boxes or bags taken by an industrial camera mounted on the mounting frame 1; the distance to the surface of the packaged item measured by an ultrasonic ranging sensor mounted on the end effector 4; the current rotation angle fed back by the encoder built into the base rotation drive assembly 2; and the joint angle fed back by the Hall sensors of the servo motors of the first joint assembly 3 and the second joint assembly 301.

[0038] In this embodiment, it should be specifically noted that the industrial camera is preferably an industrial-grade GigE camera with a resolution of not less than 5 million pixels, which, together with the light source assembly, enables stable target image acquisition; the ultrasonic ranging sensor is a model with a measurement accuracy better than ±1 mm to meet the requirements of high-precision pose measurement; and each encoder is an absolute encoder with more than 17 bits to ensure the resolution and reliability of angle feedback.

[0039] The processing logic of the pose acquisition module is as follows: First, edge detection and template matching are performed on the real-time image to obtain the approximate center coordinates and orientation angle of the packaging in the horizontal plane; simultaneously, the vertical height value is calculated using the distance value measured by the ultrasonic ranging sensor; the above position information is cross-compared for the first time with the theoretical position of the robot end effector calculated by the joint encoder angle; if the deviation exceeds a preset threshold, the image and ranging data are re-acquired, up to three times consecutively; if it is still unqualified, an alarm is triggered and the machine is stopped; the output result is the preliminary target pose, including horizontal coordinates. Orientation angle and height ;

[0040] It should be further noted that the edge detection adopts the Canny algorithm or an improved morphological edge detection method to improve adaptability to different packaging surfaces; the template matching adopts a matching algorithm based on normalized correlation coefficient, which performs fast matching through a variety of pre-stored packaging specification templates;

[0041] The formula for calculating the preliminary target pose output by the pose acquisition module is as follows:

[0042] Preliminary target pose vector ,in In the horizontal plane Direction coordinates In the horizontal plane Direction coordinates The angle at which the packaging is facing. The height of the packaging;

[0043] First cross-comparison bias The calculation formula is: ,in This is the theoretical position of the robot's end effector calculated using the angles of the joint encoder. The encoder angle of the base rotation drive assembly 2. The encoder angle of the first joint assembly 3. For the encoder angle of the second joint assembly 301, Denotes the inverse kinematic transformation function, which is... Based on the inherent DH parameters of the robot (including the length of each link, joint offset, etc.), these parameters are known quantities that can be obtained by those skilled in the art during robot design, and based on them, the function can be uniquely determined. Specific form;

[0044] The preset threshold setting range is: ,in Determined based on sensor resolution, typically set to a value of [value to be filled in]. millimeters The value is determined based on the repeatability of the robot's end effector, and is typically set to a value of [value missing]. millimeters, if This will trigger a re-collection;

[0045] The deviation analysis module is used to dynamically correct and quantitatively evaluate the initial target pose.

[0046] The processing logic of the deviation analysis module is as follows: the initial target pose is compared with the ideal reference pose of the same packaging specification in the historical palletizing parameter database to calculate the deviation value in each direction; then, the deviation value is dynamically weighted and corrected by combining the current ambient temperature (measured by the thermistor on the mounting frame 1) and the vibration acceleration of the robot base (collected by the accelerometer); based on the corrected deviation value, a normal allowable range and an emergency intervention range are constructed; if the deviation falls within the emergency intervention range, an abnormal label is directly output and an audible and visual alarm is triggered to lock subsequent modules; if it is within the normal range, the corrected target pose and the corresponding confidence interval boundary are output.

[0047] In this embodiment, it should be specifically noted that the historical palletizing parameter database is stored in non-volatile memory, pre-stores ideal reference pose data for different packaging specifications, and supports online updates; the ambient temperature measurement uses a digital thermistor sensor with an accuracy of ±0.5℃; and the vibration acceleration acquisition uses a MEMS accelerometer with a sampling frequency of not less than [missing information]. Hz;

[0048] It should be further explained that the specific calculation formula of the deviation analysis module is as follows:

[0049] Step 1: Calculate the deviation values ​​in each direction. , , ,in This serves as an ideal reference pose for the same packaging specifications in the historical database.

[0050] Step 2: Calculate the temperature compensation correction amount based on the linear expansion coefficient of each link material (steel or aluminum alloy) of the robotic arm. (Unit: 1 / ℃) and the geometric relationship of thermal deformation in each direction are determined using the following anisotropic compensation formula:

[0051]

[0052]

[0053]

[0054] ;

[0055] in, These represent the distance from the base to the end effector in the base coordinate system. Equivalent thermal expansion length in the direction (unit: mm). The equivalent linear expansion coefficient in the corresponding direction (obtained through factory calibration, typical value). ), The angle drift coefficient (unit: rad / ℃, typically less than) If detailed calibration data is lacking, it can be simplified to:

[0056] ;

[0057] in These are the temperature coefficients (which can be positive or negative) measured experimentally in each direction. ;

[0058] Step 3: Process the vibration acceleration signal without directly correcting the static target pose. Specifically, process the effective value of the vibration acceleration collected by the accelerometer. The input is fed into a second-order low-pass filter to filter out measurement noise above 20Hz; simultaneously, a vibration suppression coefficient is introduced into the monitoring module. :

[0059] ;

[0060] in ,when At that time, multiply both the current planned maximum speed and acceleration by [missing information]. Speed ​​reduction measures were implemented to minimize the impact of vibration on tracking accuracy.

[0061] Step 4: The corrected target pose is:

[0062] ,

[0063] ,

[0064] ,

[0065] ;

[0066] The sign of the temperature compensation coefficient is determined by the actual thermal deformation direction (usually, a positive temperature rise causes the robotic arm to extend, thus increasing the end-effector coordinates, hence the plus sign). The position measurement error caused by vibration is suppressed by the Kalman filter in the pose acquisition module (which dynamically adjusts the measurement noise covariance matrix based on accelerometer data), and no static vibration compensation term is added.

[0067] Step 5: Construct the confidence interval boundaries, with the standard allowable range set as follows: ,in The standard deviation is calculated based on the most recent 100 successful palletizing data, and is typically taken as a value of [value missing]. Millimeters; the emergency intervention range is set as follows: If the deviation exceeds this range, an anomaly label will be triggered.

[0068] The first data resampling correction is set between the deviation analysis module and the trajectory planning module, and is used to confirm the initial target pose a second time.

[0069] The processing logic for the first data re-sampling correction is as follows: First, the controller 6 instructs the industrial camera and ultrasonic sensor to perform a second data acquisition on the current packaging, while recording new ambient temperature and vibration data. Then, the cross-comparison logic of the pose acquisition module is re-run to obtain a new preliminary target pose. Then execute the correction logic: change the newly acquired pose. The original corrected target pose in the deviation analysis module A second comparison is performed. If the deviation between the two is within the allowable range (less than 1 mm in the horizontal direction)... Millimeters, angle direction less than If the deviation is not large enough, the original corrected pose is retained and the process continues; if the deviation is too large, a weighted fusion algorithm is used to update the original corrected pose to obtain the updated corrected target pose. Its confidence interval is recalculated based on the fused pose;

[0070] In this embodiment, it should be specifically explained that the secondary comparison formula for the first data resampling correction is as follows:

[0071] ,in express Norm, if Then the original corrected pose is retained. ,in As the allowable range threshold, the horizontal value is [value to be filled in]. millimeters; if Then for the position component ( Linear weighting is used:

[0072] ,

[0073] ,

[0074] ;

[0075] Regarding the orientation angle To avoid erroneous averaging caused by angular periodicity, a circular weighted average is used:

[0076] , ;

[0077] in This is a weighting coefficient, usually taken as 0.7. If the result angle... If the value is negative, add 360° to transform it to the [0°, 360°] interval, and finally obtain the updated corrected target pose. ;

[0078] It should be further explained that the above weighted fusion algorithm can effectively balance the real-time nature of newly collected data with the reliability of existing corrected data, through weighting coefficients. The proper settings can not only respond quickly to changes in on-site working conditions, but also avoid control jitter caused by a single abnormal data.

[0079] The trajectory planning module is used to generate a sequence of motion commands after load compensation and spatial avoidance compensation.

[0080] The inputs to the trajectory planning module are the target pose and confidence interval boundary after the first correction, the real-time load force data of each joint component, and the current clamping force of the palletizing fixture component 5.

[0081] The processing logic of the trajectory planning module is as follows: Based on the target pose and the current joint angle, the target joint angle sequence is calculated through inverse kinematics; however, it is not executed directly, but rather a two-step progressive compensation is performed: First, load compensation: Based on the static and dynamic load measured by the force sensor, the control parameters of the servo motor of each joint are adjusted in real time; Second, spatial avoidance compensation: The 3D model of the surrounding environment stored on the mounting frame 1 is read, and two safe intermediate points are automatically inserted in the trajectory to avoid collisions; Finally, a compensated discrete motion command sequence is output, including the angle, angular velocity, angular acceleration of each joint on the time axis and the rotation angle of the end effector 4, and an expected maximum error vector is also output.

[0082] In this embodiment, it should be specifically noted that the force sensor is installed at the output end of each joint and is used to collect the force situation of the robot end effector 4 during the movement in real time; the three-dimensional model of the surrounding environment is established through offline teaching or three-dimensional scanning and stored in the model database of the control component 6, which includes the spatial position information of obstacles such as conveying equipment, pallets, and other robots;

[0083] It should be further explained that the specific calculation formula of the trajectory planning module is as follows:

[0084] Step 1: Calculate the target joint angle sequence using inverse kinematics. ,in Denotes the inverse kinematic function, which is... Based on the determination of the robot's DH parameters, those skilled in the art can calculate them according to the robot's actual geometric dimensions. The target angle sequence for five joints;

[0085] Step 2: Load compensation, measuring the current load force vector using a six-dimensional force / torque sensor mounted on the end effector 4. (Units are N and N·m), according to the static model of the robotic arm, the additional gravity compensation torque required by each joint to balance the load. , of which The compensating torque for each joint is:

[0086] ;

[0087] here It is the first under the current configuration The row vectors of the Jacobian matrix of forces / torques from the joint to the end effector will... It is added as a feedforward term to the torque command of the servo controller without modifying the target joint angle;

[0088] If the robotic arm is not equipped with a six-dimensional force sensor, a simplified compensation method based on the end-effector load mass is used: The load mass m (in kg) is measured by the force sensor, and the gravity compensation amount for each joint is:

[0089] ;

[0090] in For the first The equivalent force arm from the joint's center of mass to the end-load center. Given gravitational acceleration, the corrected joint command torque is:

[0091]

[0092] The final output sequence of motion commands is still based on the original target joint angles. However, in actual execution, tracking is performed by a servo driver with feedforward torque compensation;

[0093] Step 3: Spatial avoidance compensation, inserting two safe intermediate points in the trajectory. and , and Calculations are performed by reading a 3D model of the surrounding environment to ensure that the distance between the robot's end effector 4 and obstacles is greater than a safe distance threshold, typically set to [value missing]. millimeters;

[0094] Step 4: Output the expected maximum error vector Based on historical data, the values ​​of each component are typically taken as follows: millimeters;

[0095] The second data resampling and correction mechanism is located between the trajectory planning module and the simulation verification module, and is used to perform real-time status confirmation of the motion command sequence.

[0096] The processing logic for the second data re-acquisition correction is as follows: First, the controller 6 instructs all joint encoders, force sensors, and vision sensors on the end effector 4 to re-acquire a frame of data in the current state, including the robot's current posture, load distribution, and the positions of environmental obstacles; then, the correction logic is executed: the newly acquired real-time state data is input into the trajectory planning module, the trajectory compensation logic is run again, and a new motion command sequence is generated. Then the new sequence is compared with the original sequence. A full path comparison is performed. If the difference in angle of any joint exceeds a preset safety threshold (usually 2 degrees), it is considered that the current environment has changed significantly (such as accidental displacement of the packaging box or the appearance of temporary obstacles), and the original trajectory is no longer reliable. Therefore, a new sequence is selected as the basis for execution; otherwise, the original sequence is retained. Under normal operation and a stable environment, the difference between the new and old trajectories usually will not exceed the threshold, so unnecessary frequent switching will not occur.

[0097] In this embodiment, it should be specifically explained that the full path point comparison formula for the second data resampling correction is as follows:

[0098] ,in If for any have ,but ,otherwise ,in As a safety threshold, it is usually set to a value of Spend;

[0099] It should be further explained that the above-mentioned real-time status confirmation mechanism can effectively cope with dynamic changes in the field environment, such as the positional deviation of the conveying equipment and the appearance of temporary obstacles, to ensure the reliability of motion commands; and the re-sampling and re-planning calculations added by this module can be completed in milliseconds on modern industrial controllers (such as multi-core DSP+FPGA architecture), without affecting the production line cycle time requirements.

[0100] The simulation verification module is used to pre-verify motion command sequences in a virtual simulation environment.

[0101] The inputs to the simulation verification module are the motion command sequence after the second correction and the expected maximum error vector, the confidence interval boundary output by the deviation analysis module, the current timestamp of the internal clock of the control unit 6, and the expected action duration.

[0102] The processing logic of the simulation verification module is as follows: The module quickly simulates an action process in a virtual simulation environment; during the simulation, the expected maximum error vector is superimposed on each critical path point of the end-effector 4 to check whether there will be interference with surrounding objects; if any collision risk is detected, a secondary correction loop is triggered: the coordinates of the conflict position are fed back to the trajectory planning module, requiring replanning, with a maximum of three loops; at the same time, the gripping action of the palletizing fixture component 5 is simulated to calculate whether the package can be effectively gripped under the expected error; if the gripping is unreliable, an instruction to adjust the opening and closing stroke parameters of the fixture is output and passed back to the trajectory planning module; when the simulation passes, the confirmed executable instruction set and a unique action verification code are output;

[0103] In this embodiment, it should be specifically noted that the virtual simulation environment is built based on a multi-rigid-body dynamics model, which can accurately simulate the kinematics and dynamics characteristics of the robot; the action verification code is generated using a hash algorithm and is used for subsequent verification by the execution monitoring module.

[0104] It should be further explained that the collision detection formula for the simulation verification module is as follows:

[0105] For each path point ,in Given the total number of path points, calculate the end position of the final execution component 4. ,in Represents the positive kinematics function; if obstacle coordinates exist... satisfy ,in If the safe distance threshold is reached, a collision risk is determined, and trajectory replanning needs to be triggered.

[0106] The execution monitoring module is used to track, monitor, and precisely control the execution process of motion commands in real time.

[0107] The inputs to the execution monitoring module are the confirmed executable instruction set and action check code output by the simulation verification module, the real-time angles fed back by the encoders of each joint during the actual movement, the real-time load data collected by each force sensor, and the clamping force sensor data on the end effector 4.

[0108] The processing logic of the execution monitoring module is as follows: Before the motion command is executed, the motion verification code is first used to perform a handshake verification with the simulation verification module to confirm that the command set has not been tampered with and is consistent with the current simulation verification result; if the verification fails, the simulation verification module is re-triggered for re-verification.

[0109] During the execution of motion commands, the actual position, velocity, and acceleration data of each joint are collected in real time at a sampling frequency of no less than 100Hz. These data are compared with the expected values ​​in the command sequence to calculate the tracking error. If the tracking error exceeds a preset threshold (usually set to a value of...), the tracking error is considered to be... If the force exceeds the normal range (too loose or too tight), local trajectory correction or pause processing is triggered; at the same time, the clamping force data is monitored in real time. If the clamping force exceeds the normal range (too loose or too tight), the clamping action parameters of the fixture are dynamically adjusted; key data in the entire execution process, including timestamps, joint angles, tracking errors, clamping forces, etc., are recorded in real time to the log file for the final confirmation module to call.

[0110] In this embodiment, it should be specifically noted that the handling of tracking error exceeding the limit adopts a graded response mechanism: slight exceeding the limit (less than...) Local smoothing correction was performed (millimeters), and moderate over-limit correction was applied. to (millimeters) trigger trajectory fine-tuning, severely exceeding limits (greater than) (mm) Immediately pause and alarm; the clamping force adjustment adopts closed-loop feedback control, dynamically optimizing the clamping force setting value according to the actual gripping feedback force;

[0111] It should be further noted that the execution monitoring module also has the ability to fuse multi-source data, and comprehensively analyze joint position feedback, force feedback and visual feedback to improve the accuracy and reliability of state perception.

[0112] The final confirmation module is used to accurately compare and record the actual results after the palletizing operation is completed:

[0113] The inputs to the final confirmation module are the execution process log output by the execution monitoring module, the actual palletizing position image collected by the end vision sensor after palletizing is completed, and the update request for the historical palletizing database.

[0114] The processing logic of the final confirmation module is as follows: First, the execution process log is checked for integrity to confirm that there are no abnormal interruptions or missing data; then, images of the actual palletizing position are acquired through a vision sensor, and the coordinates and angle information of the actual palletizing position are extracted using image processing algorithms; the actual pose is compared with the target pose to calculate the final deviation value; if the final deviation is within the allowable range (usually taken as a certain value), the final confirmation module will be considered as a complete verification of the data. If the deviation is within millimeters, the palletizing is considered successful, the relevant data is updated to the historical palletizing database, and a work completion signal is output; if the final deviation exceeds the limit, a deviation alarm signal is output and the deviation data is recorded for subsequent process optimization analysis.

[0115] It should be specifically noted in this embodiment that the final confirmation module also has a statistical analysis function, which can automatically count key indicators such as palletizing success rate, average deviation value, and distribution of abnormal types within a certain period, and present them to the user through a display panel or network interface.

[0116] It should be further explained that the specific calculation formula for the final confirmation module is as follows:

[0117] Calculation of the deviation between the actual palletizing position and the target position: ,in This is the actual palletizing position vector extracted by the vision sensor. The target pose output by the trajectory planning module; if ,in The allowable deviation threshold is typically set to a value of [value to be filled in]. If the deviation is within millimeters, the palletizing is considered successful; otherwise, an alarm is output and the deviation data is recorded for subsequent analysis.

[0118] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0119] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent palletizing industrial robot for a packaging production line, comprising a mounting frame (1), a base rotation drive assembly (2) mounted on the mounting frame (1), a first joint assembly (3) connected to the base rotation drive assembly (2), a second joint assembly (301) connected to the first joint assembly (3), an end effector assembly (4) disposed at the end of the second joint assembly (301), and a palletizing fixture assembly (5) mounted on the end effector assembly (4), characterized in that: It also includes a control unit (6) set on the mounting frame (1); The control unit (6) includes a pose acquisition module, a deviation analysis module, a trajectory planning module, a simulation verification module, an execution monitoring module, and a final confirmation module, which are connected in sequence. The pose acquisition module is used to acquire real-time images of the package, the distance to the surface of the package, the rotation angle of the base rotation drive component (2), and the joint angle of the first joint component (3) and the second joint component (301), and to process the acquired information to output the preliminary target pose. The deviation analysis module is used to compare the preliminary target pose with the ideal reference pose in the historical palletizing parameter database, and to make corrections based on ambient temperature and vibration acceleration, and output the corrected target pose. The trajectory planning module is used to generate a sequence of motion commands after load compensation and spatial avoidance compensation based on the corrected target pose and real-time load force data. The simulation verification module is used to pre-verify the motion command sequence in a virtual simulation environment, and output an executable instruction set after the verification is successful; The execution monitoring module is used to monitor and compare the robot's actual motion state with the expected motion state in real time during the execution of the executable instruction set, and make corrections accordingly. The final confirmation module is used to compare the actual palletizing position with the target pose after the palletizing operation is completed, and update the historical palletizing parameter database based on the comparison result.

2. The intelligent palletizing industrial robot for a packaging production line according to claim 1, characterized in that, The specific logic of the pose acquisition module in processing the acquired information to output the preliminary target pose includes: performing edge detection and template matching on the acquired real-time image to obtain the horizontal center coordinates and orientation angle of the packaging; calculating the vertical height value using the acquired surface distance of the packaging; cross-comparing the obtained horizontal center coordinates, orientation angle, and height value with the theoretical position of the robot end effector calculated by the joint angle; if the deviation exceeds a preset threshold, re-acquisition is triggered.

3. The intelligent palletizing industrial robot for a packaging production line according to claim 2, characterized in that, The deviation analysis module compares the preliminary target pose with the ideal reference pose in the historical palletizing parameter database, and makes corrections based on ambient temperature and vibration acceleration. The specific logic includes: The initial target pose is compared with the ideal reference pose of the same packaging specification in the historical palletizing parameter database, and the deviation value in each direction is calculated. Based on the ambient temperature, the linear expansion coefficient of each link material of the robotic arm and the equivalent thermal expansion length in each direction are used to calculate the temperature compensation correction for anisotropy, and independent compensation is performed for the horizontal X direction, horizontal Y direction, vertical Z direction and orientation angle respectively. The vibration acceleration signal is low-pass filtered, and the vibration suppression coefficient is calculated based on the effective value of the vibration acceleration. This coefficient is output to the execution monitoring module to reduce the maximum motion speed and acceleration when the vibration exceeds the threshold. The position measurement error caused by vibration is suppressed by the Kalman filter in the pose acquisition module, and the static target pose is not directly corrected. The deviation value is corrected using the temperature compensation correction amount to obtain the corrected target pose; Based on the corrected target pose, a normal allowable range and an emergency intervention range are constructed. If the deviation falls within the emergency intervention range, an anomaly label is output.

4. The intelligent palletizing industrial robot for a packaging production line according to claim 3, characterized in that, The control unit (6) further includes a first data resampling correction disposed between the deviation analysis module and the trajectory planning module, the first data resampling correction being used for: The instruction states that the pose acquisition module should perform secondary acquisition to obtain a new preliminary target pose; The new preliminary target pose is compared a second time with the corrected target pose output by the deviation analysis module. If the deviation between the two is within the allowable range, the original corrected pose is retained; If the deviation is too large, a weighted fusion algorithm is used to update the original corrected pose: linear weighting is used for the position component, and circular weighted averaging is used for the orientation angle. That is, the angle is converted into a vector on a unit circle, weighted and summed, and then restored to the angle to obtain the updated corrected target pose.

5. The intelligent palletizing industrial robot for a packaging production line according to claim 4, characterized in that, The specific logic of the trajectory planning module in generating a motion command sequence after load compensation and spatial avoidance compensation includes: Based on the updated corrected target pose and the current joint angles, the target joint angle sequence is calculated using inverse kinematics. Load compensation is performed by measuring the load force vector using a six-dimensional force / torque sensor mounted on the end effector. The compensation torque for each joint is calculated based on the static model of the robotic arm and the Jacobian matrix. This compensation torque is then added as a feedforward term to the torque command of the servo controller without modifying the target joint angle. If a six-dimensional force sensor is not configured, a simplified compensation based on load mass is used. The load mass is measured by the force sensor and then combined with the equivalent arm to calculate the gravity compensation torque. Read the pre-stored 3D model of the surrounding environment and automatically insert safe intermediate points in the trajectory for spatial avoidance compensation; The output is a compensated discrete motion command sequence, which is actually tracked by a servo driver with feedforward torque compensation during execution.

6. The intelligent palletizing industrial robot for a packaging production line according to claim 5, characterized in that, The control unit (6) also includes a second data resampling correction set between the trajectory planning module and the simulation verification module. The second data resampling correction is used to: instruct all joint encoders, force sensors and vision sensors on the end effector (4) to re-collect current state data; input the newly collected real-time state data into the trajectory planning module to generate a new motion command sequence; compare the new motion command sequence with the original motion command sequence at all path points, and if the difference in any joint angle exceeds a preset safety threshold, select the new motion command sequence as the basis for execution.

7. The intelligent palletizing industrial robot for a packaging production line according to claim 6, characterized in that, The specific logic of the simulation verification module in pre-verifying the motion command sequence in the virtual simulation environment includes: simulating the execution process of the motion command sequence in the virtual simulation environment; superimposing the expected maximum error vector onto each critical path point of the end-effector (4) and checking whether there is interference with surrounding objects; if a collision risk is detected, triggering a secondary correction loop and feeding back the coordinates of the conflict position to the trajectory planning module to request replanning; simultaneously simulating the gripping action of the palletizing fixture component (5) and calculating the reliability of gripping under the expected error; after the simulation passes, outputting the confirmed executable instruction set and a unique action verification code.

8. The intelligent palletizing industrial robot for a packaging production line according to claim 7, characterized in that, The specific logic of the execution monitoring module in real-time monitoring and correction during the execution of the executable instruction set includes: before executing the executable instruction set, performing a handshake verification with the simulation verification module using the action check code; during execution, collecting the actual position, velocity, and acceleration data of each joint in real time at a set sampling frequency, and comparing them with the expected values ​​in the instruction sequence to calculate the tracking error; if the tracking error exceeds a preset threshold, triggering local trajectory correction or pausing processing; simultaneously monitoring the clamping force data in real time and dynamically adjusting the clamping action parameters of the fixture; and recording key data during the execution process to a log file in real time.

9. The intelligent palletizing industrial robot for a packaging production line according to claim 8, characterized in that, The execution monitoring module adopts a hierarchical response mechanism to handle tracking error exceeding the limit: for minor exceedances, local smoothing correction is performed; for moderate exceedances, trajectory fine-tuning is triggered; and for severe exceedances, the system is immediately paused and an alarm is triggered.

10. The intelligent palletizing industrial robot for a packaging production line according to claim 9, characterized in that, The specific logic of the final confirmation module in comparing the actual palletizing position with the target pose and updating the historical palletizing parameter database based on the comparison result includes: performing integrity verification on the execution process log; acquiring an image of the actual palletizing position through a visual sensor and extracting the coordinates and angle information of the actual palletizing position; performing a final comparison between the actual pose and the target pose and calculating the final deviation value; if the final deviation is within the allowable range, the palletizing is determined to be successful, and the relevant data is updated to the historical palletizing database; if the final deviation exceeds the limit, a deviation alarm signal is output and the deviation data is recorded.

Images