Palletizing robot facing the grasping pose adaptive decision method of damaged cigarette box

Abstract

The application discloses a stacking robot facing a damaged cigarette box grasping pose adaptive decision method, relates to the technical field of stacking robots, and comprises the following steps: in the stacking process of the stacking robot, continuously time-sequencing and calculating the difference of the visual data of the damaged cigarette box at each moment, generating a posture evolution trajectory sequence and a posture mutation indication sequence, and determining whether the visual result is updated in time in the case that the posture of the damaged cigarette box changes based on time matching judgment coding; in the case that the posture of the damaged cigarette box changes and the visual result is not updated in time, the posture mutation indication sequence is directionally accumulated and calculated, a posture driving change field is generated, and the posture offset at the current moment is obtained. The application solves the problem that the posture change of the damaged cigarette box and the visual update are not synchronized, leading to the mismatch of the grasping pose of the stacking robot, realizes the adaptive adjustment of the grasping pose of the damaged cigarette box, and improves the grasping stability and operation continuity of the stacking robot.

Description

Technical Field

[0001] This invention relates to the field of palletizing robot technology, and more specifically to an adaptive decision-making method for the gripping posture of a palletizing robot facing damaged cigarette boxes. Background Technology

[0002] The adaptive decision-making of gripping posture for palletizing robots facing damaged cigarette boxes refers to the technical process in tobacco logistics and warehousing automation scenarios where, for non-standard cigarette boxes with deformed, collapsed, or dimensional differences, the palletizing robot obtains the appearance and spatial information of the target object through a perception system, and dynamically generates the optimal gripping position and posture suitable for the current target based on the acquired data, thereby achieving stable and efficient automatic gripping and palletizing operations. Existing technologies typically use visual inspection systems (such as 2D image or 3D point cloud acquisition devices) to identify and locate cigarette boxes on a conveyor line, acquiring their boundary contours, posture angles, and surface feature information. Based on this, a geometric model or feature representation of the target cigarette box is constructed. Subsequently, multiple candidate grasping poses are generated through preset grasping rules or model-based computation methods. These candidate poses are then evaluated and selected based on the robot's kinematic constraints, end effector type, and grasping stability requirements. Furthermore, the optimal grasping pose is input into the robot control system to complete path planning and motion control, achieving the grasping operation of the target cigarette box. In some systems, real-time feedback information can be used to dynamically adjust the grasping process. Overall, this type of technology typically includes multiple stages such as perception acquisition, target recognition and modeling, grasping pose generation, pose evaluation and decision-making, and control execution. Through data interaction and collaboration between these stages, adaptive decision-making for grasping poses of damaged cigarette boxes can be achieved.

[0003] The existing technology has the following shortcomings: During palletizing, when a damaged cigarette box is slightly bumped during transport or contact with an adjacent cigarette box, its spatial posture changes instantaneously. Simultaneously, the vision system needs to go through image acquisition, data processing, and result transmission to complete a state update. Therefore, shortly after the collision, there may be a situation where the posture has changed but the visual result remains unchanged. Since the adaptive decision-making for the grasping posture relies on the current visual result as input, the posture information used for the decision may be inconsistent with the actual state. Existing technology cannot regenerate the grasping posture based on the state change caused by the change in the damaged cigarette box's posture and the failure to update the visual result in time. This results in the palletizing robot performing grasping according to an outdated posture, leading to grasping position deviation or insufficient contact, ultimately causing grasping failure, cigarette box slippage, or even palletizing interruption.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive decision-making method for the gripping pose of a palletizing robot facing damaged cigarette boxes, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive decision-making method for the gripping pose of a palletizing robot facing a damaged cigarette box, specifically including the following steps: S1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted by time and the differences are calculated to generate the posture evolution trajectory sequence and posture change indication sequence. Based on the time matching judgment code, it is determined whether there is a situation where the posture of the damaged cigarette box changes and the visual results are not updated in time. S2. In the case where the attitude of the damaged smoke box changes but the visual results are not updated in time, the direction accumulation calculation is performed on the attitude change indication sequence to generate the attitude driving change field and obtain the attitude offset at the current moment in order to determine the state change. S3. Based on the attitude-driven change field, perform spatial migration calculation on the visual coordinates to generate an attitude compensation coordinate set, and recalculate the grasping contact point and grasping direction according to the attitude compensation coordinate set to regenerate the grasping pose according to the state change. S4. Resample the damaged cigarette box to form an instantaneous attitude sampling sequence, and perform difference mapping calculation between the instantaneous attitude sampling sequence and the attitude compensation coordinate set to generate the pose offset convergence amount and perform pose correction. S5. During the palletizing robot's grasping execution, continuously update the posture evolution trajectory sequence and posture driving change field, and adjust the grasping trajectory in real time based on the updated posture driving change field.

[0007] Preferably, S1 includes the following sub-steps: S1-1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted according to the acquisition time sequence, and the spatial position information and attitude parameters corresponding to each moment are extracted. The spatial position information of adjacent consecutive moments is calculated by coordinate difference, the attitude parameters are calculated by angle difference, and the difference results of each moment are arranged in the order of time to form a difference data sequence consistent with the time sorting. S1-2. Connect the difference data sequence sequentially along the time sorting direction, map the difference results at each time point to a unified coordinate space in order to construct the attitude evolution trajectory sequence, determine the sign of the change direction of each difference result in the difference data sequence, divide the change amplitude into intervals, and extract the mutation features when the direction is reversed or the amplitude exceeds the preset interval. Generate a mark at the corresponding time position based on the mutation features to form an attitude mutation indication sequence that corresponds one-to-one with the attitude evolution trajectory sequence. S1-3. By segmenting and encoding the continuous change segments in the attitude evolution trajectory sequence and combining them with the marker positions in the attitude change indication sequence to construct a time matching judgment code, the attitude information in the current visual data is inserted into the time sorting position and aligned with the attitude evolution trajectory sequence. When the continuous change is interrupted or the change position is inconsistent during the alignment calculation, it is determined that there is a situation where the attitude of the defective smoke box has changed but the visual result has not been updated in time.

[0008] Preferably, S1-3 are as follows: The attitude evolution trajectory sequence is divided into continuous change segments according to the time order. Intervals with the same change direction and continuous change amplitude are divided into the same continuous change segment. Each continuous change segment is assigned a segment number and direction code to form a segmented coding sequence. The marker positions in the attitude mutation indication sequence are mapped to the corresponding segment number positions in the segmented coding sequence. A mutation identifier code is superimposed at the marker position. The segment number, direction code, and mutation identifier code are combined and arranged in chronological order to construct a time matching judgment code. The pose information in the current visual data is inserted into the corresponding position of the segmented coding sequence according to the time sorting position, and point-by-point alignment calculation is performed along the time sorting direction. The consistency of the directional coding and the consistency of the mutation marker coding position in the continuously changing segment are compared. When there is discontinuity in the directional coding or offset of the mutation marker coding position, it is determined that there is a situation where the pose of the damaged smoke box has changed and the visual result has not been updated in time.

[0009] Preferably, S2 includes the following sub-steps: S2-1. In the case of a change in the posture of a damaged smoke box without timely updating of the visual results, the posture change indication sequence is read point by point according to the time sorting direction. The change mark at each time position is combined with the corresponding direction information, and the direction information is continuously superimposed along the time sorting direction so that the change in the same direction forms an accumulated result at adjacent time positions, and the direction accumulation calculation sequence is obtained. S2-2. The cumulative results of each time position in the direction cumulative calculation sequence are mapped to the corresponding coordinate direction in the spatial coordinate system in the order of time. Starting from the spatial position in the initial visual data, the cumulative results of each time position are calculated point by point according to the time order, so that each time position forms a continuous spatial position node. The spatial position nodes are connected along the time direction to form a continuous change path, thereby constructing an attitude-driven change field covering a continuous time interval. S2-3. Extract the spatial change component corresponding to the current time position in the attitude-driven change field, combine the change components in each coordinate direction to obtain the attitude offset at the current time, and perform position superposition processing on the attitude offset at the current time and the attitude information in the visual data to determine the state change.

[0010] Preferably, S2-3 is as follows: In the attitude-driven change field, the spatial position node corresponding to the current time position is read according to the time order. The position difference between the spatial position node corresponding to the current time position and the spatial position node in the initial visual data is calculated. The position change amount in each coordinate direction is extracted. At the same time, the directional change information in the attitude-driven change field is separated according to the corresponding coordinate direction and combined to form the spatial change component corresponding to the current time position. The position change and direction change information in the spatial change components corresponding to the current time position are fused and calculated to generate the current attitude offset. The current attitude offset is then superimposed on the attitude information in the visual data to form the updated attitude information. The updated attitude information is compared with the attitude information in the original visual data item by item. When the attitude angle change or spatial position change exceeds the preset judgment range in the difference comparison, it is determined that the abnormally damaged smoke box has changed its state.

[0011] Preferably, S3 includes the following sub-steps: S3-1. Based on the spatial position nodes and change direction information in the attitude-driven change field, obtain the initial spatial coordinates of the damaged smoke box in the visual coordinates, and perform direction-by-direction migration calculation on the initial spatial coordinates according to the change amount of each coordinate direction in the change direction information. Arrange the migrated spatial coordinates according to the time position to obtain the migrated visual coordinates corresponding to different time positions. S3-2. Associate the migrated visual coordinates corresponding to different time positions with the attitude direction information in the attitude-driven change field, combine the spatial coordinate information in the migrated visual coordinates with the attitude direction information at the corresponding time position, and sort them according to time position to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information. S3-3. Determine the gripping area on the surface of the damaged cigarette box based on the spatial coordinate information in the attitude compensation coordinate set, and calculate the contact direction between the end effector and the damaged cigarette box based on the attitude direction information in the attitude compensation coordinate set, and redetermine the gripping contact point and gripping direction. S3-4. Transform the newly determined gripping contact point and gripping direction into the execution coordinate system of the palletizing robot, and combine the posture information of the damaged cigarette box after the state change to perform spatial position adjustment and orientation posture matching, and generate the gripping pose after the corresponding state change, including gripping position and gripping direction.

[0012] Preferably, S3-2 is as follows: The migrated visual coordinates corresponding to different time positions are indexed according to time position, and the attitude direction information corresponding to the time position in the attitude-driven change field is read. The spatial coordinate position in the migrated visual coordinates is matched with the direction change parameter in the attitude direction information to form a combination of spatial coordinate information and attitude direction information containing time identifier. The combined data of spatial coordinate information and attitude direction information are arranged in chronological order. The spatial coordinate information and attitude direction information corresponding to the same time position are bound together, and the combined data corresponding to consecutive time positions are connected in sequence to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information.

[0013] Preferably, S4 is as follows: The damaged smoke box is visually sampled again. The spatial position data and attitude parameter data corresponding to the resampling time are obtained according to the sampling time sequence. The spatial position data and attitude parameter data obtained by continuous sampling are associated with time markers. The spatial position data and attitude parameter data corresponding to adjacent sampling time positions are combined according to the sampling order. The combined sampling data are arranged in sequence to form an instantaneous attitude sampling sequence. The spatial position data and attitude parameter data in the instant attitude sampling sequence are mapped to the corresponding time positions in the attitude compensation coordinate set according to the time position. The spatial coordinate information and attitude direction information at the same time position are calculated to calculate the difference. The difference data corresponding to each time position are arranged in chronological order to form the difference mapping data between the instant attitude sampling sequence and the attitude compensation coordinate set. The convergence calculation is performed on the continuous time position difference changes in the difference mapping data. The spatial position change difference and the attitude direction change difference corresponding to adjacent time positions are continuously compared. The trend of the difference change gradually decreasing within the continuous time position is extracted. The convergence amount of spatial position change and the convergence amount of attitude direction change are combined to generate the pose offset convergence amount. The spatial position and orientation of the grasping pose are adjusted based on the pose offset convergence amount, and the adjusted grasping position and orientation are replaced in the current grasping pose to complete the pose correction.

[0014] Preferably, S5 is as follows: During the palletizing robot's grasping and execution process, the current spatial position data and attitude parameter data of the damaged cigarette box are continuously acquired according to the grasping and execution time sequence. The spatial position data and attitude parameter data corresponding to the newly added time position are added to the attitude evolution trajectory sequence. The attitude evolution trajectory sequence is updated according to the spatial change nodes and attitude change direction information corresponding to the continuous time positions in the attitude evolution trajectory sequence. The attitude driving change field is regenerated based on the updated attitude evolution trajectory sequence. Based on the updated attitude-driven change field, the spatial change information corresponding to the current time position is read. The spatial change information is matched with the motion position and direction in the current grasping trajectory of the palletizing robot. The motion position and direction in the grasping trajectory are adjusted according to the spatial change information, and the palletizing robot is controlled to perform the grasping action according to the adjusted grasping trajectory.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a posture evolution trajectory sequence and a posture change indication sequence by performing temporal correlation analysis on the visual data of a damaged cigarette box at continuous time intervals. It then uses time-matching judgment coding to identify states where the posture of the damaged cigarette box changes but the visual results are not updated in time, achieving proactive perception of visual lag. Furthermore, by performing directional accumulation calculations on the posture change indication sequence, it generates a posture-driven change field covering a continuous time interval, obtaining the spatial change trend and posture offset of the damaged cigarette box during periods of visual non-update. This enables the palletizing robot to infer the current true state of the damaged cigarette box based on historical changes. On this basis, it performs spatial migration calculations on the visual coordinates through the posture-driven change field, generating a posture-compensated coordinate system containing both spatial coordinate information and posture direction information. The system uses a set of coordinates to re-determine the gripping area, contact point, and direction on the surface of the damaged cigarette box, forming a gripping pose adapted to the changing state. This prevents the palletizing robot from performing gripping based on outdated visual information. Simultaneously, a real-time pose sampling sequence is generated through re-visual sampling and compared with the pose compensation coordinate set to calculate the pose offset convergence. This provides closed-loop correction to the gripping pose, gradually bringing the compensated gripping pose closer to the actual state of the damaged cigarette box. During gripping, the pose evolution trajectory sequence and pose-driven change field are continuously updated, and the gripping trajectory is adjusted in real time based on the updated spatial change information. This allows the palletizing robot to maintain dynamic adaptation between the gripping position and direction as the damaged cigarette box continues to change. Therefore, this technical solution solves the problem of gripping pose mismatch caused by the asynchronous change in the posture of the damaged cigarette box and the update of the visual state, improving the palletizing robot's adaptive gripping capability and operational stability for non-standard cigarette boxes. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1The adaptive decision-making method for the palletizing robot's gripping pose when facing damaged cigarette boxes, as shown, specifically includes the following steps: S1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted by time and the differences are calculated to generate the posture evolution trajectory sequence and posture change indication sequence. Based on the time matching judgment code, it is determined whether there is a situation where the posture of the damaged cigarette box changes and the visual results are not updated in time. In this embodiment, S1 includes the following sub-steps: S1-1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted according to the acquisition time sequence, and the spatial position information and attitude parameters corresponding to each moment are extracted. The spatial position information of adjacent consecutive moments is calculated by coordinate difference, the attitude parameters are calculated by angle difference, and the difference results of each moment are arranged in the order of time to form a difference data sequence consistent with the time sorting. During the palletizing process of the palletizing robot, the vision acquisition device continuously collects data on the damaged cigarette boxes on the conveyor line at a fixed sampling frequency, forming a continuous sequence of visual data arranged in ascending order of time. By performing target localization and attitude analysis on each frame of visual data, the spatial position information and attitude parameters at the corresponding moment can be obtained. The spatial position information can be represented as three-dimensional coordinate values, and the attitude parameters can be represented as rotation angles around each coordinate axis. In practical implementation, the data from two adjacent acquisition moments can be matched one-to-one. Coordinate difference calculation is performed on the spatial position information, that is, the values ​​in the three coordinate directions are subtracted respectively. At the same time, angle difference calculation is performed on the attitude parameters, that is, the angle changes in each rotation direction are calculated item by item. For example, in a continuous acquisition process, the spatial position information at the first moment is (100, 200, 300), and the attitude parameters are (0°, 0°, 0°). The spatial position information at the second moment is (102, 198, 305), and the attitude parameters are (2°, -1°, 1°). Through differential calculation, the corresponding difference data can be obtained as (2, -2, 5) and (2°, -1°, 1°). The same processing is performed on subsequent continuous moments, and all difference results are arranged in chronological order to form a difference data sequence consistent with the time order. This processing can transform discrete visual observation data into a continuous description of changes, thus providing a unified data foundation for subsequent attitude change analysis.

[0020] Palletizing robots refer to automated equipment used to perform cigarette box grasping and stacking operations. The palletizing process of a palletizing robot refers to the continuous operation process from the entry of the cigarette box into the work area to the completion of grasping and stacking. Damaged cigarette boxes refer to cigarette box objects that have deformed, collapsed, or exhibited abnormal postures during transportation or stacking. Continuous-time visual data refers to multi-frame images or point cloud data acquired by a visual acquisition device at different time points. Spatial position information refers to the position coordinates of the cigarette box in three-dimensional space. Attitude parameters refer to the orientation angles of the cigarette box relative to a reference coordinate system. Coordinate difference calculation refers to the process of performing axis-by-axis numerical difference processing on the spatial position information of adjacent time points to obtain the displacement change. Angle difference calculation refers to the calculation of the angle change of the attitude parameters of adjacent time points to obtain the attitude change. The difference data sequence consistent with time order refers to the set of displacement and attitude change quantities arranged in the order of acquisition time, used to describe the motion and attitude changes of damaged cigarette boxes over continuous time.

[0021] S1-2. Connect the difference data sequence sequentially along the time sorting direction, map the difference results at each time point to a unified coordinate space in order to construct the attitude evolution trajectory sequence, determine the sign of the change direction of each difference result in the difference data sequence, divide the change amplitude into intervals, and extract the mutation features when the direction is reversed or the amplitude exceeds the preset interval. Generate a mark at the corresponding time position based on the mutation features to form an attitude mutation indication sequence that corresponds one-to-one with the attitude evolution trajectory sequence. When processing the differential data sequence, the difference results at each time step are first concatenated sequentially according to the chronological order. The displacement and attitude changes at each time step are then mapped to the same reference coordinate system, ensuring a consistent spatial reference for data acquired at different times, thus forming a continuously changing attitude evolution trajectory sequence. In specific implementation, the difference results at each time step can be accumulated sequentially to the initial position. For example, if the initial spatial position information is (100, 200, 300), and the subsequent difference results at consecutive time steps are (2, -2, 5), (1, -1, 3), and (-3, 2, -4), the accumulated trajectory points are (102, 198, 305), (103, 197, 308), and (100, 199, 304), forming a continuous attitude evolution trajectory. Based on this, the sign of the change direction of each difference result is determined; for example, a positive value indicates movement in the positive direction, and a negative value indicates movement in the negative direction. Simultaneously, the magnitude of the change is... Interval division, for example, dividing the amplitude into small change zone, medium change zone, and large change zone, when the sign of adjacent difference results on the same coordinate axis changes from positive to negative or from negative to positive, it is determined that the direction has reversed. When the change amplitude jumps from the small change zone to the large change zone, it is determined that the amplitude exceeds the preset interval. In these cases, the abrupt change features at the corresponding time are extracted, and a mark is generated at the corresponding position of the trajectory sequence. For example, when the change from positive to negative occurs at the third time, the abrupt change mark is recorded at the position of that time, thus obtaining an attitude abrupt change indication sequence that corresponds one-to-one with the attitude evolution trajectory sequence. In this way, the abnormal changes of the damaged smoke box during continuous movement can be accurately reflected.

[0022] The difference results at each time point refer to the set of displacement and attitude changes obtained through coordinate and angle difference calculations between consecutive time points; the unified coordinate space refers to the spatial environment in which data collected at different times are transformed into the same reference coordinate system for representation, to ensure trajectory continuity; the attitude evolution trajectory sequence refers to the spatial position and attitude change path constructed by chronological order and accumulation; each difference result refers to the displacement and attitude changes corresponding to each time point in the difference data sequence; the direction of change of each difference result refers to the positive or negative trend of each difference result on each coordinate axis or in each rotation direction; sign determination refers to the process of marking the direction of change as positive or negative; variation The amplitude refers to the magnitude of the difference result; the interval division refers to classifying the amplitude of change according to a preset range; the reversal of direction refers to the opposite trend in the direction of change of the difference result at adjacent time points; the amplitude exceeding the preset interval refers to the amplitude of change crossing the preset range boundary; the preset interval refers to the range of amplitude of change set according to actual operating experience; the mutation feature refers to the abnormal change information reflected when the direction is reversed or the amplitude exceeds the preset interval; the attitude mutation indication sequence corresponding one-to-one with the attitude evolution trajectory sequence refers to the set of markers that record whether mutation features exist at each time position of the trajectory sequence, which is used to describe the distribution of abnormal nodes in the attitude change process of the damaged smoke box.

[0023] S1-3. By segmenting and encoding the continuous change segments in the attitude evolution trajectory sequence and combining them with the marker positions in the attitude change indication sequence to construct a time matching judgment code, the attitude information in the current visual data is inserted into the time sorting position and aligned with the attitude evolution trajectory sequence. When the continuous change is interrupted or the change position is inconsistent during the alignment calculation, it is determined that there is a situation where the attitude of the defective smoke box has changed but the visual result has not been updated in time.

[0024] In this embodiment, S1-3 specifically refers to: The attitude evolution trajectory sequence is divided into continuous change segments according to the time order. Intervals with the same change direction and continuous change amplitude are divided into the same continuous change segment. Each continuous change segment is assigned a segment number and direction code to form a segmented coding sequence. The marker positions in the attitude mutation indication sequence are mapped to the corresponding segment number positions in the segmented coding sequence. A mutation identifier code is superimposed at the marker position. The segment number, direction code, and mutation identifier code are combined and arranged in chronological order to construct a time matching judgment code. The pose information in the current visual data is inserted into the corresponding position of the segmented coding sequence according to the time sorting position, and point-by-point alignment calculation is performed along the time sorting direction. The consistency of the directional coding and the consistency of the mutation marker coding position in the continuously changing segment are compared. When there is discontinuity in the directional coding or offset of the mutation marker coding position, it is determined that there is a situation where the pose of the damaged smoke box has changed and the visual result has not been updated in time.

[0025] When dividing the attitude evolution trajectory sequence into continuous change segments, the difference results at each time point can be read point by point along the time sorting direction. The change trend of each difference result in each coordinate direction is marked with a direction, and the continuity of the change amplitude within a preset range is judged. When the change trend of adjacent time points in the same direction is consistent and the change amplitude does not cross the set range, the time period is divided into the same continuous change segment, and all difference results in the segment are grouped into the same group. When the change trend changes or the change amplitude crosses the set range, the current position is used as the segment boundary to re-divide the segment. By numbering each segment in sequence and recording the dominant direction in the segment, a segmented coding sequence containing segment number and direction code is formed. This process can split the continuous change process into several intervals with consistent characteristics, which is convenient for subsequent unified processing.

[0026] After completing the division of continuously changing segments, the marked positions in the attitude change indication sequence are mapped to the corresponding segment positions in the segmented coding sequence in chronological order. A change identifier code is superimposed at each marked position, and the segment number, direction code, and change identifier code are combined and arranged in chronological order to construct a time matching judgment code. In specific implementation, the segment number can be used to identify the change stage, the direction code can be used to identify the change trend, and the change identifier code can be used to identify the change node, so that each time position has stage attribute, direction attribute, and change attribute at the same time. A complete time matching judgment code sequence is formed by the combination and arrangement of the three types of codes.

[0027] During alignment calculation, the pose information in the current visual data is inserted into the corresponding position of the segmented coding sequence according to the time order. The direction codes of adjacent segments before and after the insertion position are compared point by point, and the position of the mutation marker code in the time series is checked synchronously. In specific implementation, the segmented coding sequence can be scanned point by point forward and backward from the insertion position to determine whether the change trend remains continuous. When the direction code is inconsistent with the segment direction, it is determined that the direction code is discontinuous. When the relative position of the mutation marker code in the time series is offset, it is determined that the mutation marker code position is offset. Through this point-by-point alignment calculation method, the accurate judgment of the time matching state is achieved.

[0028] The continuous change segment division refers to dividing time periods with consistent change characteristics into the same group based on the changing trend and magnitude of the difference results at each time step in the attitude evolution trajectory sequence; the interval with consistent change direction and continuous change magnitude refers to the interval in which the changing trend of each difference result in the same coordinate direction is consistent and the change magnitude is within a continuous range; the segment number and direction code are used to identify the sequential position and dominant change direction of each segment; the segmented coding sequence refers to the sequence structure formed by arranging the segment number and direction code in chronological order; the marked position in the attitude change indication sequence indicates the time position of the change; the change identification code is used to mark whether the change feature exists at that position; the time matching judgment code refers to the comprehensive coding sequence formed by combining the segment number, direction code, and change identification code.

[0029] The pose information in the current visual data refers to the spatial pose parameters corresponding to the current acquisition time; point-by-point alignment calculation refers to the process of comparing the consistency of the encoding at each position along the time sorting direction; discontinuous orientation encoding refers to the inconsistency between the direction of change of the insertion position and the orientation encoding of the segment; shift in the position of the mutation marker encoding refers to the change in the relative position of the mutation location in the time series; change in the pose of the damaged cigarette box refers to the change in position or orientation of the cigarette box in a short period of time; and failure to update the visual results in a timely manner refers to the visual data not reflecting the latest pose state. By comprehensively judging the discontinuity of orientation encoding and the shift in the position of the mutation marker encoding, the asynchronous situation between pose changes and visual updates can be identified.

[0030] S2. In the case where the attitude of the damaged smoke box changes but the visual results are not updated in time, the direction accumulation calculation is performed on the attitude change indication sequence to generate the attitude driving change field and obtain the attitude offset at the current moment in order to determine the state change. In this embodiment, S2 includes the following sub-steps: S2-1. In the case of a change in the posture of a damaged smoke box without timely updating of the visual results, the posture change indication sequence is read point by point according to the time sorting direction. The change mark at each time position is combined with the corresponding direction information, and the direction information is continuously superimposed along the time sorting direction so that the change in the same direction forms an accumulated result at adjacent time positions, and the direction accumulation calculation sequence is obtained. In cases where the attitude of a damaged cigarette box changes but the visual results are not updated in a timely manner, the system first reads the mutation markers corresponding to each time position in the attitude mutation indication sequence and simultaneously acquires the direction information corresponding to each time position. The mutation markers and direction information are combined to form the attitude change description information for a single time position. Then, the attitude change description information for consecutive time positions is superimposed according to the time order. In the specific implementation, the direction information can be divided into the change trend in different coordinate directions. For example, when the damaged cigarette box moves forward along the spatial coordinate direction, the positive direction information is recorded; when it moves in the opposite direction, the negative direction information is recorded. Simultaneously, the mutation markers are used as the basis for adjusting the change intensity, and changes in the same direction in consecutive time positions are cumulatively recorded. For example, if the direction information corresponding to consecutive time positions in the attitude mutation indication sequence is positive, positive, positive, negative, and the mutation markers are normal change, normal change, mutation change, normal change, then the positive direction changes in the previous consecutive time positions are first superimposed to form a cumulative result of continuous changes in the same direction. When a change in the opposite direction occurs, the change trend is re-recorded, thus forming a direction accumulation calculation sequence that reflects the cumulative process of the attitude change of the damaged cigarette box. By continuously superimposing directional information, the problem that single-time position change information cannot reflect the overall attitude change trend can be avoided, so that the subsequent attitude-driven change field generation has a continuous change basis.

[0031] The attitude change indicator sequence refers to the sequence information recording the positions of abrupt change nodes during the attitude change of the damaged cigarette box in chronological order, used to identify the locations of significant attitude changes in continuous visual data. The change marker and corresponding direction information at each time point refer to the change description formed by combining information on whether an attitude change occurred and spatial change direction information at the same time node. The change marker indicates the change state, and the direction information indicates the change trend. Continuous overlay processing refers to the cumulative association processing of direction information from multiple consecutive time points according to the chronological order, creating a continuous change process between adjacent time points. The cumulative result refers to the cumulative state of direction change formed after continuous overlay processing, used to represent the attitude change trend of the damaged cigarette box over continuous time. The direction accumulation calculation sequence refers to the set of multiple accumulation results arranged in chronological order, used to describe the continuous evolution of direction change during the attitude change of the damaged cigarette box. Through the data association between the attitude change indicator sequence, change marker, direction information, and direction accumulation calculation sequence, discrete attitude changes during the visual acquisition process can be converted into continuous change information, providing a data foundation for the subsequent generation of attitude-driven change fields.

[0032] S2-2. The cumulative results of each time position in the direction cumulative calculation sequence are mapped to the corresponding coordinate direction in the spatial coordinate system in the order of time. Starting from the spatial position in the initial visual data, the cumulative results of each time position are calculated point by point according to the time order, so that each time position forms a continuous spatial position node. The spatial position nodes are connected along the time direction to form a continuous change path, thereby constructing an attitude-driven change field covering a continuous time interval. When constructing the attitude-driven change field based on the directional cumulative calculation sequence, the cumulative results corresponding to each time position in the directional cumulative calculation sequence are first read. Then, the cumulative results at each time position are mapped according to the corresponding coordinate direction in the spatial coordinate system, so that the change trend in the directional cumulative calculation sequence is converted into spatial position change information. In specific implementation, the spatial position of the aberrant smoke box when the initial visual data is acquired can be used as the starting point. Then, the cumulative results at each time position are read sequentially according to the time sorting order. The change corresponding to each time position is superimposed on the position state formed by the previous time position to complete the point-by-point position migration calculation. For example, if the initial visual data acquisition location of the damaged smoke box is (100, 200, 300), and the cumulative results corresponding to consecutive time positions in the directional cumulative calculation sequence represent spatial movement changes as (2, 1, 0), (4, 2, 1), and (6, 3, 2), then, starting from the initial spatial position, the spatial position nodes (102, 201, 300), (104, 202, 301), and (106, 203, 302) are obtained sequentially. Connecting these spatial position nodes in chronological order forms a continuous change path. This continuous change path records the spatial change process of the damaged smoke box from its initial visual state to its current state. By aggregating the multiple spatial position nodes formed within the continuous time interval, an attitude-driven change field covering the continuous time interval can be constructed. Through this process, the temporal change information in the directional cumulative calculation sequence can be converted into a spatial change trajectory, providing a spatial basis for subsequently acquiring the current attitude change state of the damaged smoke box.

[0033] The cumulative results at each time point refer to the cumulative change information corresponding to each time node formed by sorting by time in the direction cumulative calculation sequence, which is used to represent the attitude change trend of the damaged cigarette box over a continuous time period; the corresponding coordinate direction in the spatial coordinate system refers to the directional component that maps the change information in the cumulative results to the spatial position coordinates, including the spatial movement direction and the attitude change direction; the initial visual data refers to the initial visual information collected by the palletizing robot before the attitude change of the damaged cigarette box; the spatial position in the initial visual data refers to the position coordinates of the damaged cigarette box in the spatial coordinate system recorded in the initial visual data; the starting point refers to the initial spatial position used as a reference for position change when performing point-by-point position migration calculation; Point-by-point position migration calculation refers to the process of applying the accumulated results of each time position sequentially to the previous spatial position node according to the time order, thereby obtaining the calculation process of continuous spatial position change. Continuous spatial position nodes refer to multiple spatial position data points arranged in time order, used to describe the spatial movement trajectory during the attitude change process of the damaged smoke box. Continuous change path refers to the spatial change route formed by connecting the spatial position nodes in time order. The attitude-driven change field covering the continuous time interval refers to the set of spatial change information composed of multiple continuous spatial position nodes and the change trends between nodes, used to represent the attitude change process of the damaged smoke box when the visual results are not updated in time. Through the accumulated results of each time position, spatial coordinate system mapping, point-by-point position migration calculation, and the connection process of continuous spatial position nodes, discrete time change information can be converted into continuous spatial change information.

[0034] S2-3. Extract the spatial change component corresponding to the current time position in the attitude-driven change field, combine the change components in each coordinate direction to obtain the attitude offset at the current time, and perform position superposition processing on the attitude offset at the current time and the attitude information in the visual data to determine the state change.

[0035] In this embodiment, S2-3 specifically refers to: In the attitude-driven change field, the spatial position node corresponding to the current time position is read according to the time order. The position difference between the spatial position node corresponding to the current time position and the spatial position node in the initial visual data is calculated. The position change amount in each coordinate direction is extracted. At the same time, the directional change information in the attitude-driven change field is separated according to the corresponding coordinate direction and combined to form the spatial change component corresponding to the current time position. The position change and direction change information in the spatial change components corresponding to the current time position are fused and calculated to generate the current attitude offset. The current attitude offset is then superimposed on the attitude information in the visual data to form the updated attitude information. The updated attitude information is compared with the attitude information in the original visual data item by item. When the attitude angle change or spatial position change exceeds the preset judgment range in the difference comparison, it is determined that the abnormally damaged smoke box has changed its state.

[0036] If the damaged cigarette box exhibits an attitude change but the visual results are not updated in a timely manner, the spatial position node corresponding to the current time position is first retrieved from the attitude-driven change field, ordered by time. The attitude-driven change field consists of multiple spatial position nodes arranged in chronological order, each containing spatial movement information of the damaged cigarette box at that time position. In the specific implementation, the spatial position node corresponding to the current time position can be obtained through a time index, and the spatial position node in the initial visual data is used as a reference. The positional difference between the two spatial position nodes in terms of spatial coordinate directions is calculated. During the positional difference calculation, the position parameters in each coordinate direction of the spatial position node are compared to obtain the positional change in different coordinate directions. Simultaneously, the directional change information recorded in the attitude-driven change field is read and decomposed according to the spatial coordinate direction, establishing a correspondence between the spatial movement direction and the attitude change direction. For example, the spatial location nodes in the initial visual data record the initial spatial position of the damaged cigarette box within the transport area. The spatial location nodes corresponding to the current time position in the attitude-driven change field record the spatial positional change of the damaged cigarette box after a minor collision. The positional changes of the damaged cigarette box along different spatial directions are calculated through positional differences. Simultaneously, combined with the directional change information in the attitude-driven change field, a spatial change component corresponding to the current time position is formed, containing both the direction of spatial movement and the degree of change. This process converts the continuous trajectory in the attitude-driven change field into specific change information corresponding to the current time position, providing a data source for attitude offset calculation.

[0037] After obtaining the spatial change components corresponding to the current time position, the position change and direction change information in the spatial change components are fused and calculated to generate the current attitude offset. Specifically, the position change quantities in different coordinate directions are combined according to their corresponding directions, and the attitude change trend in the direction change information is added to the spatial change calculation process, so that the generated current attitude offset includes both spatial position offset information and attitude direction offset information. For example, if a damaged cigarette box moves along the conveying direction after being hit by a collision, and simultaneously rotates around the spatial direction, the movement and rotation changes are combined according to the position change direction and attitude change direction recorded in the spatial change components to generate an attitude offset reflecting the current state change. Subsequently, the current attitude offset is superimposed on the attitude information in the visual data to update the attitude state in the original visual data, forming updated attitude information. When comparing the updated attitude information with the attitude information in the original visual data item by item, the spatial position change and attitude angle change can be compared separately. When the updated attitude information contains position changes or attitude angle changes exceeding a preset judgment range, it is determined that the damaged cigarette box has undergone a state change. For example, if the original visual data records that the posture of the damaged cigarette box remains stable, but the updated posture information shows that the position of the box has moved significantly or the orientation of the box has changed significantly, then it is determined that the current state of the damaged cigarette box has changed.

[0038] The spatial position node corresponding to the current time position refers to the spatial position data point formed by sorting by time in the attitude-driven change field, used to record the spatial position of the damaged cigarette box in the current time state; the spatial position node in the initial visual data refers to the position data formed when the spatial state of the damaged cigarette box is first acquired during the visual acquisition process, used as a reference benchmark for spatial change calculation; the position difference calculation refers to comparing the spatial coordinates of the spatial position node corresponding to the current time position with the spatial position node in the initial visual data to obtain the position change in different coordinate directions; the position change in each coordinate direction refers to the distance the damaged cigarette box moves in different spatial directions, used to describe the spatial position change; the directional change information in the attitude-driven change field refers to the attitude change direction information of the damaged cigarette box recorded in the attitude-driven change field, used to describe the change trend; the spatial change component corresponding to the current time position refers to the comprehensive change information formed by the combination of position change and directional change information, used to represent the spatial change state of the damaged cigarette box at the current time position.

[0039] Fusion computing refers to the process of combining the positional change and orientation change information in the spatial change components corresponding to the current time position to generate the current attitude offset. The current attitude offset refers to the set of information reflecting the positional and attitude changes of the damaged cigarette box compared to its initial visual state. Visual data refers to the image information or spatial perception information of the damaged cigarette box acquired by the palletizing robot through visual acquisition equipment. The attitude information in the visual data refers to the spatial orientation and attitude state of the damaged cigarette box obtained from the visual data. Updated attitude information refers to the new attitude state formed after applying the current attitude offset to the attitude information in the visual data. Original visual data refers to the visual acquisition information acquired before the attitude change of the damaged cigarette box. The attitude information in the original visual data refers to the original... The initial posture state of the damaged cigarette box recorded in the initial visual data; item-by-item difference comparison refers to comparing the updated posture information with the posture information in the original visual data according to spatial position parameters and posture parameters respectively; the presence of posture angle change means that the updated posture information has a change in direction and angle compared with the posture information in the original visual data; spatial position change exceeds the preset judgment range means that the amount of spatial position change after the update exceeds the preset position change allowable range; preset judgment range refers to the position change range and posture change range set according to the requirements of grasping the damaged cigarette box; the state change of the damaged cigarette box means that the current spatial state of the damaged cigarette box has changed compared with the initial state at the time of visual acquisition. Through the above data processing process, the continuous calculation process from posture-driven change field to state change determination can be completed.

[0040] S3. Based on the attitude-driven change field, perform spatial migration calculation on the visual coordinates to generate an attitude compensation coordinate set, and recalculate the grasping contact point and grasping direction according to the attitude compensation coordinate set to regenerate the grasping pose according to the state change. In this embodiment, S3 includes the following sub-steps: S3-1. Based on the spatial position nodes and change direction information in the attitude-driven change field, obtain the initial spatial coordinates of the damaged smoke box in the visual coordinates, and perform direction-by-direction migration calculation on the initial spatial coordinates according to the change amount of each coordinate direction in the change direction information. Arrange the migrated spatial coordinates according to the time position to obtain the migrated visual coordinates corresponding to different time positions. When performing spatial migration calculations on visual coordinates based on the attitude-driven change field, the spatial position nodes and change direction information in the attitude-driven change field are first read. The spatial position nodes are used to represent the spatial movement state of the damaged cigarette box in continuous time, and the change direction information is used to represent the trend of spatial position change. Then, the initial spatial coordinates of the damaged cigarette box in the visual coordinates are obtained. The initial spatial coordinates are used as the starting position for spatial migration calculation, and the initial spatial coordinates are migrated in one direction according to the change amount of each coordinate direction in the change direction information. In the specific implementation process, the positional changes of the damaged cigarette box can be recorded separately in different directions of the spatial coordinate system. For example, the initial spatial coordinates of the damaged cigarette box in the visual coordinate system are (500, 300, 200). The change direction information corresponding to the continuous time positions in the attitude-driven change field shows that the damaged cigarette box moves along the length direction, shifts in the width direction, and changes in the height direction, with corresponding coordinate direction changes of (8, -3, 2). Then, the initial spatial coordinates are migrated and calculated according to each coordinate direction, resulting in migrated spatial coordinates of (508, 297, 202). As the time position continues to change, the change direction information corresponding to the subsequent time position is applied to the migration result of the previous time position. For example, if the change in the next time position is (5, -2, 1), new migrated spatial coordinates are obtained, and they are arranged sequentially according to the time position to form migrated visual coordinates corresponding to different time positions. Through this process, the spatial change trend in the attitude-driven change field can be transformed into the visual coordinate space, so that the visual coordinates can reflect the spatial state of the damaged cigarette box after the attitude change.

[0041] The spatial position nodes and change direction information in the attitude-driven change field refer to the spatial position change points and corresponding movement direction information recorded in the attitude-driven change field in chronological order. The spatial position nodes are used to describe the spatial state of the damaged cigarette box at different time positions, and the change direction information is used to describe the trend of spatial state change. The initial spatial coordinates of the damaged cigarette box in the visual coordinates refer to the initial spatial position data of the damaged cigarette box acquired during the visual acquisition process, which is used as the reference starting point for spatial migration calculation. The coordinate direction change amount refers to the position change data generated along different spatial directions in the attitude-driven change field, which is used to describe the movement of the damaged cigarette box in each spatial direction. The direction-by-direction migration calculation refers to the calculation process of reading the coordinate direction change amount according to the spatial coordinate direction and gradually applying the coordinate direction change amount to the initial spatial coordinates, so that the spatial position changes continuously with time. The post-migration visual coordinates corresponding to different time positions refer to the set of visual coordinate positions formed at different time nodes after the completion of the spatial migration calculation, which is used to record the spatial position change process of the damaged cigarette box from the initial visual state to the current state. By linking spatial location nodes, change direction information, initial spatial coordinates, and direction-by-direction migration calculations, spatial transformation between attitude-driven change fields and visual coordinates can be achieved, providing basic data for the subsequent generation of attitude compensation coordinate sets.

[0042] S3-2. Associate the migrated visual coordinates corresponding to different time positions with the attitude direction information in the attitude-driven change field, combine the spatial coordinate information in the migrated visual coordinates with the attitude direction information at the corresponding time position, and sort them according to time position to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information. S3-3. Determine the gripping area on the surface of the damaged cigarette box based on the spatial coordinate information in the attitude compensation coordinate set, and calculate the contact direction between the end effector and the damaged cigarette box based on the attitude direction information in the attitude compensation coordinate set, and redetermine the gripping contact point and gripping direction. When redetermining the gripping contact point and gripping direction based on the attitude compensation coordinate set, the spatial coordinate information in the attitude compensation coordinate set is first read to spatially locate the external contour position of the damaged cigarette box in its current state, and then the contactable area on the surface of the damaged cigarette box is determined based on the spatial coordinate information. In the specific implementation process, the continuous spatial coordinate information in the attitude compensation coordinate set can be converted to the visual coordinate space of the palletizing robot. The position area that can meet the contact conditions of the end effector is selected according to the spatial distribution state of different positions on the surface of the damaged cigarette box. Then, the motion direction of the end effector when approaching the damaged cigarette box is determined by combining the attitude direction information in the attitude compensation coordinate set, so that the contact direction of the end effector matches the current attitude of the damaged cigarette box. For example, in the original visual state of the damaged cigarette box, the top area is used as the gripping position. However, after a slight impact, the damaged cigarette box tilts, and the attitude compensation coordinate set shows that the spatial position of the top of the box has shifted. At the same time, the attitude direction information shows that the orientation of the box has changed. At this time, the accessible area on the surface of the box is repositioned based on the updated spatial coordinate information. The stable area on the side of the box after the shift is determined as the new gripping contact point. The entry direction of the end effector is adjusted according to the tilt direction of the box, so that the gripping contact point and gripping direction are adapted to the current state of the damaged cigarette box. Through this process, the gripping parameters can be re-determined based on the spatial change information and attitude change information in the attitude compensation coordinate set.

[0043] The gripping area on the surface of the damaged cigarette box refers to the range of positions that the end effector can contact after the attitude compensation coordinate set is mapped onto the surface of the damaged cigarette box. It describes the spatial area in which gripping operations can be performed in the current state of the damaged cigarette box. The contact direction between the end effector and the damaged cigarette box refers to the spatial direction in which the end effector approaches the surface of the damaged cigarette box and establishes contact. It describes the directional state during the gripping action. The gripping contact point refers to the coordinates of the actual contact position between the end effector and the surface of the damaged cigarette box. It determines the effective position of the robot when performing the gripping action. The gripping direction refers to the direction of movement and contact of the end effector when it moves from its current position to the gripping contact point. It describes the spatial directional state of the gripping action. The spatial coordinate information in the attitude compensation coordinate set provides the surface position data of the damaged cigarette box after state changes, and the attitude direction information provides the current attitude change trend data of the damaged cigarette box. By jointly processing the spatial coordinate information and the attitude direction information, the gripping contact point and gripping direction that conform to the current state can be re-determined, enabling the palletizing robot to complete the gripping pose adjustment based on the spatial position and attitude information of the damaged cigarette box after state changes.

[0044] S3-4. Transform the newly determined gripping contact point and gripping direction into the execution coordinate system of the palletizing robot, and combine the posture information of the damaged cigarette box after the state change to perform spatial position adjustment and orientation posture matching, and generate the gripping pose after the corresponding state change, including gripping position and gripping direction.

[0045] When generating the gripping pose after the state change, the newly determined gripping contact point and gripping direction are first transformed into the palletizing robot's execution coordinate system to ensure that the gripping position and gripping direction are consistent with the motion space of the palletizing robot's end effector. Specifically, based on the spatial transformation relationship between the visual coordinate system and the palletizing robot's execution coordinate system, the coordinates of the gripping contact point determined in the posture compensation coordinate set are converted into executable spatial coordinates for the palletizing robot, while the gripping direction is converted into a direction vector in the execution coordinate system. Subsequently, combined with the posture information of the damaged cigarette box after the state change, the transformed gripping position is spatially adjusted, and the gripping direction is matched with the orientation to ensure that the motion direction of the end effector is consistent with the current posture of the damaged cigarette box. For example, the posture compensation coordinate set determines the new gripping contact point of the damaged cigarette box in the spatial coordinate system, with the gripping direction being tilted. After this spatial position is transformed into the palletizing robot's execution coordinate system, the corresponding execution coordinate position is obtained based on the robot's installation position, visual acquisition position, and coordinate transformation parameters. Simultaneously, the approach direction of the end effector is adjusted according to the posture information of the damaged cigarette box after the state change, enabling the end effector to approach the damaged cigarette box according to the new spatial position and direction. This ultimately forms a gripping pose that includes the current gripping position and direction after the state change. Through this process, the spatial information of the damaged cigarette box after the state change can be converted into motion parameters executable by the palletizing robot.

[0046] The palletizing robot's execution coordinate system refers to the spatial coordinate environment established with reference to the movement range of the palletizing robot itself, used to describe the position and orientation state that the end effector can reach; the attitude information of the damaged cigarette box after state change refers to the current spatial orientation and position state of the damaged cigarette box after the attitude drive change field calculation, used to reflect the true attitude of the damaged cigarette box after the change; spatial position adjustment refers to the position correction of the transformed gripping position according to the attitude information of the damaged cigarette box after state change, so that the gripping position is consistent with the current spatial state of the damaged cigarette box; orientation and attitude matching refers to the association and adjustment of the gripping direction with the current orientation direction of the damaged cigarette box, so that the entry direction of the end effector conforms to the current tilt direction and spatial orientation of the damaged cigarette box; the gripping pose after state change, including gripping position and gripping direction, refers to the robot execution parameters formed by the combination of the current gripping contact point and the current gripping direction, used to describe the spatial position and orientation state of the palletizing robot when performing gripping on the damaged cigarette box after state change. By changing the gripping contact point, changing the gripping direction, adjusting the spatial position, and matching the orientation and posture, a gripping posture that is adapted to the damaged cigarette box after the state change can be formed, enabling the palletizing robot to perform subsequent gripping actions based on the updated spatial state.

[0047] In this embodiment, S3-2 specifically refers to: The migrated visual coordinates corresponding to different time positions are indexed according to time position, and the attitude direction information corresponding to the time position in the attitude-driven change field is read. The spatial coordinate position in the migrated visual coordinates is matched with the direction change parameter in the attitude direction information to form a combination of spatial coordinate information and attitude direction information containing time identifier. The combined data of spatial coordinate information and attitude direction information are arranged in chronological order. The spatial coordinate information and attitude direction information corresponding to the same time position are bound together, and the combined data corresponding to consecutive time positions are connected in sequence to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information.

[0048] When generating the attitude compensation coordinate set, firstly, time index information is established for the migrated visual coordinates corresponding to different time positions, so that each migrated visual coordinate has a unique time position identifier. Simultaneously, the attitude direction information arranged according to time position in the attitude-driven change field is read, and the correspondence between the migrated visual coordinates and the attitude direction information is determined through time index matching. In the specific implementation, spatial coordinate positions can be read one by one according to the time sequence of the formation of the migrated visual coordinates, and the corresponding attitude direction information can be extracted from the attitude-driven change field based on the same time position. The spatial coordinate position is used as a description of the current spatial state of the damaged smoke box, and the attitude direction information is used as a description of the current directional change of the damaged smoke box. The two types of information are then associated and combined to form combined data of spatial coordinate information and attitude direction information with time identifiers. For example, the migrated visual coordinates record the spatial coordinate change state of the damaged smoke box at continuous time positions, and the attitude-driven change field records the directional change trend at the same time position. After time identifier matching, each spatial coordinate position can be associated with the corresponding directional change information, thereby avoiding time misalignment between spatial position change information and attitude direction change information.

[0049] After generating a combination of spatial coordinate and attitude direction information containing time markers, the combined data is arranged according to time position order, and the spatial coordinate and attitude direction information at the same time position are bound together, so that each time position forms a complete attitude compensation data node. Subsequently, the attitude compensation data nodes corresponding to consecutive time positions are connected in chronological order, forming a continuously changing data set, thus generating an attitude compensation coordinate set containing both spatial coordinate and attitude direction information. In specific implementation, the first attitude compensation data node in a consecutive time position can be determined as the initial node, and then the data nodes corresponding to subsequent time positions can be connected sequentially, ensuring that the attitude compensation coordinate set remains consistent with the change process in the attitude-driven change field. For example, after a damaged cigarette box is slightly impacted during transport, its visual coordinates continuously change spatially after migration, while the attitude-driven change field records the box's orientation change trend. The attitude compensation coordinate set formed by arranging consecutive time positions can simultaneously record the box's position movement trajectory and orientation change trajectory, providing a complete data foundation for subsequent recalculation of the grasping contact point and grasping direction.

[0050] The attitude direction information corresponding to the time position in the attitude-driven change field refers to the directional change data of the damaged cigarette box stored in the attitude-driven change field in chronological order, used to describe the attitude change trend of the damaged cigarette box under different time states; the spatial coordinate position in the migrated visual coordinates refers to the spatial position data of the damaged cigarette box after spatial migration calculation, used to represent the current position of the visual coordinates after attitude change correction; the direction change parameter in the attitude direction information refers to the data content describing the directional change state of the damaged cigarette box, used to represent the directional change when the box body deflects or moves; time position matching refers to the process of associating the migrated visual coordinates with the attitude direction information in the attitude-driven change field according to the time identifier, so that the spatial position state and the directional change state are kept synchronized; the combined data of spatial coordinate information and attitude direction information containing time identifier refers to the data unit composed of time identifier, spatial coordinate position and attitude direction information, used to record the position state and orientation state of the damaged cigarette box at the same time position.

[0051] The attitude compensation coordinate set, containing both spatial coordinate and orientation information, is a continuous data set arranged chronologically. Each data node contains the spatial position and orientation of the damaged cigarette box, describing the actual state change of the damaged cigarette box when the visual results are not updated in a timely manner. Spatial coordinate information represents the position of the damaged cigarette box in the spatial coordinate system; orientation information represents the orientation change of the damaged cigarette box relative to the spatial coordinate system. Combined data connection refers to the continuous arrangement of data nodes corresponding to multiple time positions in chronological order, forming a continuous change trajectory between each time position. The attitude compensation coordinate set is formed by temporally associating, combining, and continuously arranging the post-migration visual coordinates with the orientation information in the attitude-driven change field, providing a data foundation for recalculating the grasping contact point and grasping direction based on the attitude compensation coordinate set. Through the above processing, a complete generation process from spatial migration results to the attitude compensation coordinate set can be achieved, enabling a unified description of the spatial position and orientation information of the damaged cigarette box after its state change.

[0052] S4. Resample the damaged cigarette box to form an instantaneous attitude sampling sequence, and perform difference mapping calculation between the instantaneous attitude sampling sequence and the attitude compensation coordinate set to generate the pose offset convergence amount and perform pose correction. In this embodiment, S4 specifically refers to: The damaged smoke box is visually sampled again. The spatial position data and attitude parameter data corresponding to the resampling time are obtained according to the sampling time sequence. The spatial position data and attitude parameter data obtained by continuous sampling are associated with time markers. The spatial position data and attitude parameter data corresponding to adjacent sampling time positions are combined according to the sampling order. The combined sampling data are arranged in sequence to form an instantaneous attitude sampling sequence. The spatial position data and attitude parameter data in the instant attitude sampling sequence are mapped to the corresponding time positions in the attitude compensation coordinate set according to the time position. The spatial coordinate information and attitude direction information at the same time position are calculated to calculate the difference. The difference data corresponding to each time position are arranged in chronological order to form the difference mapping data between the instant attitude sampling sequence and the attitude compensation coordinate set. The convergence calculation is performed on the continuous time position difference changes in the difference mapping data. The spatial position change difference and the attitude direction change difference corresponding to adjacent time positions are continuously compared. The trend of the difference change gradually decreasing within the continuous time position is extracted. The convergence amount of spatial position change and the convergence amount of attitude direction change are combined to generate the pose offset convergence amount. The spatial position and orientation of the grasping pose are adjusted based on the pose offset convergence amount, and the adjusted grasping position and orientation are replaced in the current grasping pose to complete the pose correction.

[0053] When re-sampling the damaged cigarette box to form an instantaneous attitude sampling sequence, the visual sampling time is first determined based on the current execution state of the palletizing robot, and the visual acquisition device is controlled to re-acquire the current spatial state of the damaged cigarette box. During the re-visual sampling process, not only is the current spatial position of the damaged cigarette box acquired, but also its current attitude direction, spatial orientation, and attitude change parameters are acquired simultaneously, so that the acquired data can completely describe the actual state of the damaged cigarette box after attitude changes. After acquiring continuous sampling data, a time stamp is configured for each sampled data. The time stamp is used to distinguish the spatial position data and attitude parameter data corresponding to different sampling times, and a continuous data arrangement relationship is established according to the sampling order. When forming the instantaneous attitude sampling sequence, the spatial position data and attitude parameter data corresponding to adjacent time positions are combined so that each time position contains complete information about the state of the damaged cigarette box. Then, the data combinations of each time position are connected in chronological order to form a continuously changing data sequence. For example, while the palletizing robot is waiting to perform the grasping process, the damaged cigarette box undergoes a change in posture due to external contact. The vision acquisition device re-acquires the state of the damaged cigarette box, and records the changes in the spatial position and orientation of the box at multiple sampling moments before and after. After being associated with time markers, the data at each sampling moment are arranged in the order of occurrence, thus forming an instantaneous posture sampling sequence that can reflect the real-time posture change process of the damaged cigarette box.

[0054] When performing difference mapping calculations between the instantaneous attitude sampling sequence and the attitude compensation coordinate set, the time stamp in the instantaneous attitude sampling sequence is first read, and data nodes with the same time position are searched in the attitude compensation coordinate set based on the time stamp, ensuring that the real-time state data in the instantaneous attitude sampling sequence and the predicted change data in the attitude compensation coordinate set are on the same time reference. Subsequently, positional differences are calculated for the spatial coordinate information at the same time position, and directional differences are calculated for the attitude direction information. The positional and directional differences are recorded separately and arranged in chronological order to form difference mapping data. In specific implementation, the spatial coordinate information in the attitude compensation coordinate set can be used as the expected change state, and the spatial position data in the instantaneous attitude sampling sequence can be used as the actual change state. By comparing the spatial offset of the two types of data at the same time position, the change deviation between the actual state and the compensated state is obtained; simultaneously, by comparing the attitude direction information in the two types of data, the deviation between the actual attitude direction and the predicted attitude direction of the damaged smoke box is obtained. For example, the attitude compensation coordinate set records the expected movement trajectory of the damaged cigarette box after the collision, while the instantaneous attitude sampling sequence obtained by visual sampling records the actual movement trajectory. After time-position matching, the spatial position change and direction change at each time position are compared, and the difference data at continuous time positions are arranged to form difference mapping data.

[0055] When generating pose offset convergence based on difference mapping data, the difference data corresponding to consecutive time positions is first read according to the time sorting direction, and the changing trend between adjacent time positions is analyzed. Specifically, the spatial position changes within consecutive time positions are continuously compared to determine if the spatial position deviation is gradually decreasing. Simultaneously, the attitude direction changes are also continuously compared to determine if the attitude direction deviation is gradually decreasing. When the differences within consecutive time positions show a gradually decreasing trend, the reduction process of both spatial and attitude direction changes is recorded, and spatial position change convergence and attitude direction change convergence are generated respectively. Subsequently, the spatial position change convergence and attitude direction change convergence are combined to form the pose offset convergence. Through this process, the pose offset convergence is not directly generated but calculated based on the actual changing trend at consecutive time positions. For example, after a damaged cigarette box is impacted, there is a deviation between the attitude compensation coordinate set and the instantaneous attitude sampling sequence. As resampling continues, the actual state gradually approaches the compensation state, the spatial position deviation gradually decreases, and the attitude direction deviation decreases synchronously. Through continuous change analysis, a pose offset convergence representing the current degree of offset adjustment is generated.

[0056] When performing pose correction based on the pose offset convergence amount, the pose offset convergence amount is decomposed into gripping position adjustment information and gripping direction adjustment information, which are then applied to the spatial position parameters and directional attitude parameters in the current gripping pose, respectively. Specifically, the current gripping position is adjusted based on the spatial position change convergence amount, ensuring the gripping contact point adapts to the actual spatial position of the damaged cigarette box; the current gripping direction is adjusted based on the orientation direction change convergence amount, ensuring the end effector's entry direction aligns with the current orientation direction of the damaged cigarette box. After adjustment, the corrected gripping position and orientation replace the original parameters in the current gripping pose, forming an updated gripping pose. For example, if the current gripping pose has already been generated based on the attitude compensation coordinate set, but re-visual sampling reveals a slight spatial offset in the damaged cigarette box, the gripping position is moved and adjusted using the pose offset convergence amount, while the end effector's attitude is corrected based on the orientation change convergence amount, so that the palletizing robot uses the corrected gripping pose when performing subsequent gripping actions.

[0057] Re-visual sampling refers to the process by which the palletizing robot re-visually acquires the current state of the damaged cigarette box based on existing posture compensation information, in order to obtain real-time state information after visual updates. Spatial position data and posture parameter data refer to the spatial position and posture state information of the damaged cigarette box extracted during re-visual sampling. Spatial position data describes the positional changes of the damaged cigarette box in spatial coordinates, while posture parameter data describes the current orientation and posture changes of the damaged cigarette box. Time stamp association refers to the process of establishing the arrangement relationship between different sampled data through sampling time information, so that data from different time positions can be organized in the order of occurrence. Instantaneous posture sampling sequence refers to the data set formed by combining spatial position data and posture parameter data obtained from continuous sampling in chronological order, used to describe the real-time posture change process of the damaged cigarette box. Corresponding time position in the posture compensation coordinate set refers to the data position in the posture compensation coordinate set that has the same time stamp as the instantaneous posture sampling sequence, used to ensure accurate comparison between real-time state data and compensated state data.

[0058] The difference mapping calculation refers to the data processing process of comparing the instantaneous attitude sampling sequence with the spatial coordinate information and attitude direction information in the attitude compensation coordinate set to obtain the deviation between the actual state and the compensated state. The difference mapping data refers to the set of spatial position difference data and attitude direction difference data arranged according to time position, used to record the deviation of the state change of the damaged cigarette box within a continuous time period. The continuous time position difference change in the difference mapping data refers to the change process of spatial position deviation and attitude direction deviation over time in multiple continuous time positions. The convergence calculation refers to the calculation process of analyzing the gradual decrease of deviation and extracting the degree of change based on the difference change trend at continuous time positions. The spatial position change difference and attitude direction change difference are used to represent the deviation between the actual position state and the compensated position state of the damaged cigarette box, and the deviation between the actual attitude direction and the compensated attitude direction, respectively. The gradual decrease in the difference change within continuous time positions refers to the change law of the gradual decrease in the deviation between the actual state and the compensated state during continuous sampling.

[0059] Spatial position change convergence refers to the offset adjustment information calculated based on the differences in spatial position changes over continuous time, used to represent the degree of spatial position deviation. Attitude direction change convergence refers to the orientation adjustment information calculated based on the differences in attitude direction changes over continuous time, used to represent the degree of attitude direction deviation. Pose offset convergence refers to the comprehensive adjustment information formed by combining spatial position change convergence and attitude direction change convergence, used to describe the position and orientation changes that need to be corrected in the grasping pose. Pose correction refers to adjusting the grasping position and grasping direction in the grasping pose based on the pose offset convergence, so that the current grasping pose adapts to the real state of the damaged cigarette box after resampling. Through real-time attitude sampling sequence generation, difference mapping calculation, pose offset convergence generation, and pose correction processing, a continuous processing procedure from real-time state acquisition to grasping pose adjustment can be formed.

[0060] S5. During the palletizing robot's grasping execution, continuously update the posture evolution trajectory sequence and posture driving change field, and adjust the grasping trajectory in real time based on the updated posture driving change field.

[0061] In this embodiment, S5 specifically refers to: During the palletizing robot's grasping and execution process, the current spatial position data and attitude parameter data of the damaged cigarette box are continuously acquired according to the grasping and execution time sequence. The spatial position data and attitude parameter data corresponding to the newly added time position are added to the attitude evolution trajectory sequence. The attitude evolution trajectory sequence is updated according to the spatial change nodes and attitude change direction information corresponding to the continuous time positions in the attitude evolution trajectory sequence. The attitude driving change field is regenerated based on the updated attitude evolution trajectory sequence. Based on the updated attitude-driven change field, the spatial change information corresponding to the current time position is read. The spatial change information is matched with the motion position and direction in the current grasping trajectory of the palletizing robot. The motion position and direction in the grasping trajectory are adjusted according to the spatial change information, and the palletizing robot is controlled to perform the grasping action according to the adjusted grasping trajectory.

[0062] During the palletizing robot's grasping process, to continuously update the attitude evolution trajectory sequence and attitude-driven change field, the current spatial position data and attitude parameter data of the damaged cigarette box are continuously acquired according to the grasping execution time sequence. The data corresponding to newly added time positions are then added to the existing attitude evolution trajectory sequence. Specifically, visual acquisition can be continuously run while the palletizing robot performs the grasping action. Whenever new spatial state information of the damaged cigarette box is acquired, a new time position identifier is generated based on the acquisition time, and the newly added spatial position data and attitude parameter data are inserted into the corresponding time position in the attitude evolution trajectory sequence. Subsequently, the spatial change nodes and attitude change direction information between the newly added time position and adjacent time positions are analyzed. The newly added change information is connected with the continuous change trend in the original attitude evolution trajectory sequence to form an updated attitude evolution trajectory sequence. The attitude-driven change field is then regenerated based on the updated spatial change nodes and attitude change direction information. For example, when a palletizing robot is approaching a damaged cigarette box to perform a grasping action, the damaged cigarette box may experience a slight disturbance during transport, causing a spatial shift. Continuous visual acquisition captures this new spatial position change data. If the change path recorded in the original attitude evolution trajectory sequence deviates from the newly added spatial position, the new spatial position node is added to the original trajectory, and the attitude-driven change field is reconstructed based on the new direction of change. This ensures that the attitude-driven change field continuously reflects the current motion state of the damaged cigarette box. Through continuous updating, the attitude evolution trajectory sequence and the attitude-driven change field can change synchronously with the state changes during the grasping process.

[0063] When adjusting the grasping trajectory in real time based on the updated attitude-driven change field, the spatial change information corresponding to the current time position is first read from the updated attitude-driven change field, and then matched with the movement position and direction in the current grasping trajectory of the palletizing robot. Specifically, the spatial change information determines the positional offset of the damaged cigarette box relative to the original grasping trajectory, and the direction of change information determines the direction trend that the end effector needs to adjust. Then, the spatial position offset information and direction change information are applied to the corresponding position nodes in the current grasping trajectory to correct the movement position and direction in real time. For example, if the palletizing robot originally planned to move along a predetermined trajectory to the top grasping area of ​​the damaged cigarette box, but the updated attitude-driven change field shows that the damaged cigarette box has shifted laterally and changed direction, then after reading the current spatial change information, the lateral offset is applied to the target movement position in the grasping trajectory. Simultaneously, the entry direction of the end effector is adjusted according to the direction change information, so that the adjusted grasping trajectory matches the changed spatial state of the damaged cigarette box, and the palletizing robot is controlled to perform the grasping action according to the adjusted grasping trajectory. By adjusting the grasping trajectory in real time, the movement position and direction during the grasping process can be continuously adapted to the changing state of the damaged cigarette box.

[0064] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0065] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0066] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0067] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0068] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0069] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0070] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An adaptive decision-making method for the gripping pose of a palletizing robot facing damaged cigarette boxes, characterized in that, Specifically, the following steps are included: S1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted by time and the differences are calculated to generate the posture evolution trajectory sequence and posture change indication sequence. Based on the time matching judgment code, it is determined whether there is a situation where the posture of the damaged cigarette box changes and the visual results are not updated in time. S2. In the case where the attitude of the damaged smoke box changes but the visual results are not updated in time, the direction accumulation calculation is performed on the attitude change indication sequence to generate the attitude driving change field and obtain the attitude offset at the current moment in order to determine the state change. S3. Based on the attitude-driven change field, perform spatial migration calculation on the visual coordinates to generate an attitude compensation coordinate set, and recalculate the grasping contact point and grasping direction according to the attitude compensation coordinate set to regenerate the grasping pose according to the state change. S4. Resample the damaged cigarette box to form an instantaneous attitude sampling sequence, and perform difference mapping calculation between the instantaneous attitude sampling sequence and the attitude compensation coordinate set to generate the pose offset convergence amount and perform pose correction. S5. During the palletizing robot's grasping execution, continuously update the posture evolution trajectory sequence and posture driving change field, and adjust the grasping trajectory in real time based on the updated posture driving change field.

2. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 1, characterized in that, S1 includes the following sub-steps: S1-1. During the palletizing process of the palletizing robot, the visual data of the damaged cigarette box at continuous moments are sorted according to the acquisition time sequence, and the spatial position information and attitude parameters corresponding to each moment are extracted. The spatial position information of adjacent consecutive moments is calculated by coordinate difference, the attitude parameters are calculated by angle difference, and the difference results of each moment are arranged in the order of time to form a difference data sequence consistent with the time sorting. S1-2. Connect the difference data sequence sequentially along the time sorting direction, map the difference results at each time point to a unified coordinate space in order to construct the attitude evolution trajectory sequence, determine the sign of the change direction of each difference result in the difference data sequence, divide the change amplitude into intervals, and extract the mutation features when the direction is reversed or the amplitude exceeds the preset interval. Generate a mark at the corresponding time position based on the mutation features to form an attitude mutation indication sequence that corresponds one-to-one with the attitude evolution trajectory sequence. S1-3. By segmenting and encoding the continuous change segments in the attitude evolution trajectory sequence and combining them with the marker positions in the attitude change indication sequence to construct a time matching judgment code, the attitude information in the current visual data is inserted into the time sorting position and aligned with the attitude evolution trajectory sequence. When the continuous change is interrupted or the change position is inconsistent during the alignment calculation, it is determined that there is a situation where the attitude of the defective smoke box has changed but the visual result has not been updated in time.

3. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 2, characterized in that, S1-3 specifically refers to: The attitude evolution trajectory sequence is divided into continuous change segments according to the time order. Intervals with the same change direction and continuous change amplitude are divided into the same continuous change segment. Each continuous change segment is assigned a segment number and direction code to form a segmented coding sequence. The marker positions in the attitude mutation indication sequence are mapped to the corresponding segment number positions in the segmented coding sequence. A mutation identifier code is superimposed at the marker position. The segment number, direction code, and mutation identifier code are combined and arranged in chronological order to construct a time matching judgment code. The pose information in the current visual data is inserted into the corresponding position of the segmented coding sequence according to the time sorting position, and point-by-point alignment calculation is performed along the time sorting direction. The consistency of the directional coding and the consistency of the mutation marker coding position in the continuously changing segment are compared. When there is discontinuity in the directional coding or offset of the mutation marker coding position, it is determined that there is a situation where the pose of the damaged smoke box has changed and the visual result has not been updated in time.

4. The adaptive decision-making method for gripping posture of a palletizing robot facing damaged cigarette boxes according to claim 1, characterized in that, S2 includes the following sub-steps: S2-1. In the case of a change in the posture of a damaged smoke box without timely updating of the visual results, the posture change indication sequence is read point by point according to the time sorting direction. The change mark at each time position is combined with the corresponding direction information, and the direction information is continuously superimposed along the time sorting direction so that the change in the same direction forms an accumulated result at adjacent time positions, and the direction accumulation calculation sequence is obtained. S2-2. The cumulative results of each time position in the direction cumulative calculation sequence are mapped to the corresponding coordinate direction in the spatial coordinate system in the order of time. Starting from the spatial position in the initial visual data, the cumulative results of each time position are calculated point by point according to the time order, so that each time position forms a continuous spatial position node. The spatial position nodes are connected along the time direction to form a continuous change path, thereby constructing an attitude-driven change field covering a continuous time interval. S2-3. Extract the spatial change component corresponding to the current time position in the attitude-driven change field, combine the change components in each coordinate direction to obtain the attitude offset at the current time, and perform position superposition processing on the attitude offset at the current time and the attitude information in the visual data to determine the state change.

5. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 4, characterized in that, S2-3 specifically refers to: In the attitude-driven change field, the spatial position node corresponding to the current time position is read according to the time order. The position difference between the spatial position node corresponding to the current time position and the spatial position node in the initial visual data is calculated. The position change amount in each coordinate direction is extracted. At the same time, the directional change information in the attitude-driven change field is separated according to the corresponding coordinate direction and combined to form the spatial change component corresponding to the current time position. The position change and direction change information in the spatial change components corresponding to the current time position are fused and calculated to generate the current attitude offset. The current attitude offset is then superimposed on the attitude information in the visual data to form the updated attitude information. The updated attitude information is compared with the attitude information in the original visual data item by item. When the attitude angle change or spatial position change exceeds the preset judgment range in the difference comparison, it is determined that the abnormally damaged smoke box has changed its state.

6. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 1, characterized in that, S3 includes the following sub-steps: S3-1. Based on the spatial position nodes and change direction information in the attitude-driven change field, obtain the initial spatial coordinates of the damaged smoke box in the visual coordinates, and perform direction-by-direction migration calculation on the initial spatial coordinates according to the change amount of each coordinate direction in the change direction information. Arrange the migrated spatial coordinates according to the time position to obtain the migrated visual coordinates corresponding to different time positions. S3-2. Associate the migrated visual coordinates corresponding to different time positions with the attitude direction information in the attitude-driven change field, combine the spatial coordinate information in the migrated visual coordinates with the attitude direction information at the corresponding time position, and sort them according to time position to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information. S3-3. Determine the gripping area on the surface of the damaged cigarette box based on the spatial coordinate information in the attitude compensation coordinate set, and calculate the contact direction between the end effector and the damaged cigarette box based on the attitude direction information in the attitude compensation coordinate set, and redetermine the gripping contact point and gripping direction. S3-4. Transform the newly determined gripping contact point and gripping direction into the execution coordinate system of the palletizing robot, and combine the posture information of the damaged cigarette box after the state change to perform spatial position adjustment and orientation posture matching, and generate the gripping pose after the corresponding state change, including gripping position and gripping direction.

7. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 6, characterized in that, S3-2 specifically refers to: The migrated visual coordinates corresponding to different time positions are indexed according to time position, and the attitude direction information corresponding to the time position in the attitude-driven change field is read. The spatial coordinate position in the migrated visual coordinates is matched with the direction change parameter in the attitude direction information to form a combination of spatial coordinate information and attitude direction information containing time identifier. The combined data of spatial coordinate information and attitude direction information are arranged in chronological order. The spatial coordinate information and attitude direction information corresponding to the same time position are bound together, and the combined data corresponding to consecutive time positions are connected in sequence to form an attitude compensation coordinate set containing spatial coordinate information and attitude direction information.

8. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 1, characterized in that, S4 specifically refers to: The damaged smoke box is visually sampled again. The spatial position data and attitude parameter data corresponding to the resampling time are obtained according to the sampling time sequence. The spatial position data and attitude parameter data obtained by continuous sampling are associated with time markers. The spatial position data and attitude parameter data corresponding to adjacent sampling time positions are combined according to the sampling order. The combined sampling data are arranged in sequence to form an instantaneous attitude sampling sequence. The spatial position data and attitude parameter data in the instant attitude sampling sequence are mapped to the corresponding time positions in the attitude compensation coordinate set according to the time position. The spatial coordinate information and attitude direction information at the same time position are calculated to calculate the difference. The difference data corresponding to each time position are arranged in chronological order to form the difference mapping data between the instant attitude sampling sequence and the attitude compensation coordinate set. The convergence calculation is performed on the continuous time position difference changes in the difference mapping data. The spatial position change difference and the attitude direction change difference corresponding to adjacent time positions are continuously compared. The trend of the difference change gradually decreasing within the continuous time position is extracted. The convergence amount of spatial position change and the convergence amount of attitude direction change are combined to generate the pose offset convergence amount. The spatial position and orientation of the grasping pose are adjusted based on the pose offset convergence amount, and the adjusted grasping position and orientation are replaced in the current grasping pose to complete the pose correction.

9. The adaptive decision-making method for gripping posture of a palletizing robot facing a damaged cigarette box according to claim 1, characterized in that, S5 specifically refers to: During the palletizing robot's grasping and execution process, the current spatial position data and attitude parameter data of the damaged cigarette box are continuously acquired according to the grasping and execution time sequence. The spatial position data and attitude parameter data corresponding to the newly added time position are added to the attitude evolution trajectory sequence. The attitude evolution trajectory sequence is updated according to the spatial change nodes and attitude change direction information corresponding to the continuous time positions in the attitude evolution trajectory sequence. The attitude driving change field is regenerated based on the updated attitude evolution trajectory sequence. Based on the updated attitude-driven change field, the spatial change information corresponding to the current time position is read. The spatial change information is matched with the motion position and direction in the current grasping trajectory of the palletizing robot. The motion position and direction in the grasping trajectory are adjusted according to the spatial change information, and the palletizing robot is controlled to perform the grasping action according to the adjusted grasping trajectory.

Images