Automated battery packaging method and system

Abstract

This disclosure provides an automated battery packaging method and system. The method includes: setting a delivery area on a preset motion path of a robotic arm, and dividing the delivery area into multiple sub-regions; acquiring the actual pose of the packaging box in each sub-region relative to the region coordinate system, and updating the delivery pose parameters of the sub-region based on the actual pose and preset delivery pose parameters; responding to the detection of a sub-region in place, using the robotic arm to pick up a battery from a picking station, and moving the target sub-region along a preset motion path; after the robotic arm reaches a preset safety height of the target sub-region, performing a placement action according to the compensated and updated target delivery pose parameters to place the battery into the packaging box in the target sub-region; responding to the packaging box in the current target sub-region reaching a preset loading condition, switching to the next sub-region. Using the above technical solution can improve the continuity, accuracy, and overall production efficiency of automated battery packaging.

Description

Technical Field

[0001] This disclosure relates to the field of battery packaging technology, and in particular to an automated battery packaging method and system. Background Technology

[0002] With the development of the new energy industry and automated manufacturing technology, battery products are increasingly using automated equipment such as robotic arms to complete handling and assembly operations in the production, testing, and packaging stages. In the automated battery packaging process, batteries are typically picked up from a material handling station and placed into a packaging box to complete the boxing, positioning, and subsequent sealing processes. Due to the high dimensional accuracy requirements of batteries and the relatively complex structure of packaging boxes, the battery boxing process places high demands on the stability of placement, orientation, and contact.

[0003] In existing automated battery packaging methods, robotic arms mostly perform placement actions based on pre-calibrated fixed coordinate systems and placement posture parameters. However, in actual production processes, packaging boxes are prone to positional shifts, posture changes, or slight deformations after transportation, loading, or long-term use, which can affect the accuracy of battery placement and even cause problems such as collisions and jamming.

[0004] Therefore, there is an urgent need to provide an automated battery packaging method to improve the stable, smooth and efficient placement control during the battery loading process, thereby enhancing the stability and reliability of the automated battery packaging process. Summary of the Invention

[0005] In view of this, the present disclosure provides an automated packaging method and system for batteries, which can improve the continuity, accuracy and overall production efficiency of automated battery packaging.

[0006] In a first aspect, embodiments of this disclosure provide an automated packaging method for batteries, comprising: A delivery area is set on the preset motion path of the robotic arm, and the delivery area is divided into multiple sub-areas, with a packaging box placed in each sub-area; wherein, each sub-area has a corresponding area coordinate system, and each sub-area has a corresponding preset delivery pose parameter; Obtain the actual pose of the packaging box in each sub-region relative to the region coordinate system, and update the delivery pose parameters of the corresponding sub-region based on the actual pose and the preset delivery pose parameters; In response to the detection of a sub-region in the in-place state, the robotic arm picks up the battery from the picking station and moves the target sub-region along the preset motion path. After the robotic arm reaches the preset safe height of the target sub-region, it performs a placement action according to the compensated and updated target placement pose parameters to put the battery into the packaging box in the target sub-region. In response to the packaging boxes in the current target sub-region reaching the preset loading conditions, the delivery target is switched to the next sub-region adjacent to the target sub-region and located on the preset movement path.

[0007] Optionally, obtaining the actual pose of the packaging box relative to the coordinate system of each sub-region, and updating the delivery pose parameters of the corresponding sub-region based on the actual pose and preset delivery pose parameters, includes: Obtain the spatial point set or edge features of the opening area of ​​the packaging box; The box opening plane is obtained by fitting the spatial point set or edge features, and the normal vector of the box opening plane is calculated. Using the normal vector of the box opening plane as an attitude constraint, and combining it with at least one edge direction or at least two feature point directions within the box opening plane, the three-dimensional spatial pose of the packaging box relative to the regional coordinate system of the corresponding sub-region is obtained. Based on the three-dimensional spatial pose, the pose deviation between the target delivery pose parameters and the actual pose is determined, pose compensation parameters are generated, and the delivery pose parameters corresponding to the target sub-region are updated based on the pose compensation parameters.

[0008] Optionally, the placement action includes: During the lowering phase near the opening of the packaging box, position control is used to constrain the lateral displacement and attitude of the end effector of the robotic arm, so that the attitude of the end effector is consistent with the normal vector constraint of the box opening plane. After the end effector enters the preset approach distance along the lowering direction, it switches to a force / position hybrid control mode, which maintains the lateral pose and attitude unchanged or restricted in change, while performing closed-loop control on the contact force in the lowering direction. Release conditions are determined based on the contact force signal, displacement signal, or force / position change rate detected in the force / position hybrid control mode. When the release conditions are met, the end effector is controlled to release the battery to complete the placement of the battery. The release conditions include at least one of the following: the contact force in the downward direction reaches a preset threshold and is maintained for a preset time; the displacement of the end effector reaches the target placement displacement and the contact force change rate is less than a preset threshold; the end effector pose change is less than a preset pose drift threshold after the contact force reaches the threshold.

[0009] Optionally, automated packaging methods include: During multiple consecutive battery packaging cycles, at least one process parameter related to the robotic arm's battery placement process is collected; Based on the process parameters, a parameter time series is constructed, and a process state model is established based on the parameter time series to characterize the evolution of the packaging interface contact state during battery placement. The process state model is used to describe the cumulative impact of the process parameters on the contact stability between the battery and the packaging box during continuous placement. Based on the process state model, the parameter interaction relationship between the process parameters and the battery placement quality results through the process state is established to distinguish between control parameters that have a direct impact on battery placement quality and auxiliary parameters that have a constraint effect on the control parameters. Based on the process state model and the relationship between the parameters, the trend of process state change in at least one future packaging cycle is predicted, and when the prediction result indicates that the process state has the risk of entering a preset abnormal state, a corresponding packaging quality risk assessment result is generated. Under the premise of meeting the safety constraints of robotic arm movement and packaging process constraints, based on the packaging quality risk assessment results, the adjustment priority and adjustment direction of multiple packaging control parameters in the control parameter combination are jointly constrained, and a control parameter combination for the current packaging cycle is generated. According to the combination of control parameters, the corresponding automatic battery packaging operation is performed.

[0010] Optionally, constructing the corresponding parameter time series includes: The packaging cycle of the robotic arm entering the packaging box is divided into the battery removal segment, the posture adjustment segment, the box positioning segment, the box pushing segment, the release segment, and the reset segment. The contact signal detected by the end effector of the robotic arm when the battery first comes into contact with the support surface of the packaging box is taken as the zero point of the packaging cycle. Around the zero point of time, the pushing force curve, displacement curve and pushing speed curve of the pushing segment into the packaging box are collected; The peak pushing force, force rise slope, and hysteresis area of ​​the force-displacement curve are extracted from the pushing force curve; the infeed stroke deviation and end rebound amount are extracted from the displacement curve; and the speed fluctuation amplitude is extracted from the pushing speed curve. The extracted results are concatenated into a periodic feature vector for the packaging cycle, and the periodic feature vectors of multiple consecutive packaging cycles are constructed into a parametric time series in chronological order.

[0011] Optionally, the construction and updating of the process state model includes the following process: The process states are defined as the cumulative deformation state of the packaging box structure, the alignment drift state between the battery and the box, and the attenuation state of the gripping stability at the end of the robotic arm, and constitute a single-cycle implicit process state vector. Based on the implicit process state estimate of the previous packaging cycle and the robotic arm control parameters of the current packaging cycle, a predicted process state is generated. A predicted cycle feature vector is generated based on the predicted process state, and the feature residual is calculated with the actual obtained cycle feature vector. The predicted process state is corrected based on the feature residual, and the estimated process state value of the current packaging cycle is output. The set values ​​of the robotic arm's box-pushing speed and clamping force are collected, and the deviation between them and the actual executed values ​​is used as an execution deviation vector to introduce the implicit process state prediction process.

[0012] Optionally, the identification of the causal transmission path includes the following process: While keeping the battery model and packaging box specifications unchanged, apply a micro-perturbation of a preset amplitude to a single control parameter of the robotic arm's box-pushing speed or clamping force, while keeping the other control parameters unchanged. Within the micro-perturbation period, the implicit process state changes and corresponding packaging result data are obtained; Compare the implicit changes in process state and packaging results between the micro-disturbance period and the non-micro-disturbance period; When the comparison result meets the preset discrimination condition, the corresponding control parameter is determined to be a key control parameter that has a direct causal impact on packaging quality, and its direction of action is recorded. Based on the results of multiple micro-perturbations, the control parameters are sorted according to the magnitude of the changes in process state they cause, and the control parameters with a change magnitude higher than a preset threshold are selected to form a control parameter set. Optionally, the packaging quality risk assessment includes the following process: The packaging result data is limited to include at least one of the following: battery misalignment, inability of the lid to close properly, or indentation on the box. Establish the correspondence between the packaging result data and the estimated process status; Using at least one of the following, namely, battery misalignment in the box, inability of the box lid to close properly, or equivalent packaging abnormalities, as one of the failure conditions, the state threshold of battery alignment drift or box structure deformation accumulation is obtained by reverse calculation. Based on the state threshold, a failure state region containing implicit process states is constructed; The current implicit process state estimate is sampled multiple times, and the implicit process state of future packaging cycles is predicted. The proportion of predicted states falling into the failure state region is statistically analyzed and used as the packaging quality risk assessment result.

[0013] Optionally, the generation of the control parameter combination includes the following process: Based on the set of control parameters and their direction of action, the permissible adjustment directions of packaging control parameters, including at least the robotic arm's box-pushing speed and clamping force, are defined. Set upper and lower limits for the pushing speed and clamping force, as well as the maximum variation between adjacent packaging cycles; Generate the robot arm control parameter combination for the current packaging cycle under the conditions of allowing adjustment direction, upper and lower limits, and maximum change; Based on the results of the packaging quality risk assessment, determine the adjustment priority of each key control parameter, and prioritize adjusting the control parameters that contribute the most to the risk.

[0014] Secondly, embodiments of this disclosure provide an automated battery packaging system, comprising: A robotic arm used to pick up batteries and put them into a packaging box; A controller, connected to the robotic arm, is used to control the robotic arm to sequentially dispense the batteries into the packaging boxes in each sub-area of ​​the delivery area using the automated battery packaging method described in any of the foregoing embodiments.

[0015] Compared with the prior art, the technical solution of the present disclosure has the following advantages: In the automated battery packaging method provided in this disclosure, a delivery area is set on a preset motion path of a robotic arm and further subdivided into multiple sub-areas. An independent regional coordinate system and delivery posture parameters are established for each sub-area, enabling the packaging boxes to be identified and managed in an orderly and accurate manner along the same path. By acquiring the actual posture of the packaging boxes in each sub-area in real time and dynamically updating the delivery posture parameters, the position and posture deviations of the packaging boxes during placement, transportation, or cumulative use are effectively compensated. When a sub-area is detected to be in place, the robotic arm performs grasping and delivery operations along a predetermined safe path, and completes precise placement based on the compensated and updated delivery posture after reaching a safe height, reducing the risk of collision and misplacement. After the packaging box reaches the preset loading conditions, the delivery target is automatically switched to an adjacent sub-area, realizing a continuous and uninterrupted automated packaging process, significantly improving the continuity, accuracy, and overall production efficiency of automated battery packaging. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of an automated battery packaging method according to an embodiment of the present disclosure; Figure 2 This is a flowchart illustrating the updating of projection pose parameters in an embodiment of the present disclosure; Figure 3 This is a flowchart of a battery placement control method according to an embodiment of the present disclosure; Figure 4 This is a flowchart of a battery packaging adaptive update method according to an embodiment of the present disclosure; Figure 5 This is a schematic diagram of the structure of an automated battery packaging system according to an embodiment of the present disclosure. Detailed Implementation

[0018] As can be seen from the background technology, existing automated battery packaging methods have problems that affect the accuracy of battery insertion into the box, and may even cause collisions or jamming, resulting in low packaging efficiency and requiring manual intervention.

[0019] To address at least one of the aforementioned technical problems, this disclosure provides an automated battery packaging method. By setting a delivery area on a preset motion path of a robotic arm and further subdividing it into multiple sub-areas, and establishing an independent coordinate system and delivery posture parameters for each sub-area, the packaging boxes can be identified and managed systematically and accurately along the same path. By acquiring the actual posture of the packaging boxes within each sub-area in real time and dynamically updating the delivery posture parameters, the method effectively compensates for positional and orientation deviations that occur during placement, transport, or cumulative use. When a sub-area is detected to be in place, the robotic arm performs grasping and delivery operations along a predetermined safe path, and completes precise placement based on the compensated and updated delivery posture after reaching a safe height, reducing the risk of collisions and misplacement. After the packaging box reaches the preset loading conditions, the delivery target is automatically switched to an adjacent sub-area, achieving a continuous and uninterrupted automated packaging process, significantly improving the continuity, accuracy, and overall production efficiency of automated battery packaging.

[0020] In other words, the embodiments of this disclosure, through a collaborative mechanism of regional management, pose adaptive compensation, and sequential switching delivery, upgrade the traditional packaging method that relies on manual calibration or a single fixed coordinate to an intelligent delivery control scheme for multiple workstations and multiple packaging boxes, which significantly improves the continuity, accuracy, and overall production efficiency of automatic battery packaging.

[0021] To enable those skilled in the art to better understand and implement this disclosure, the following detailed description of the specific solutions, principles, advantages, and effects of this disclosure is provided with reference to the accompanying drawings and specific embodiments.

[0022] See Figure 1 , Figure 1 The flowchart below shows an automated battery packaging method according to an embodiment of this disclosure, which may include the following steps: S101, a delivery area is set on the preset motion path of the robotic arm, and the delivery area is divided into multiple sub-areas, with a packaging box placed in each sub-area; wherein, each sub-area has a corresponding area coordinate system, and each sub-area has a corresponding preset delivery pose parameter.

[0023] Specifically, in automated production lines, robotic arms typically perform pick-and-place operations along a pre-planned motion path. This pre-planned motion path can be set according to the on-site spatial layout, the robotic arm's degrees of freedom, and the positional relationship between the pick-up station and the placement station.

[0024] For example, the preset motion path can be a spatial trajectory extending along a straight line, a broken line, or a curve, and the embodiments disclosed herein do not impose specific limitations on this.

[0025] A delivery area is set on one or both sides of a preset motion path to hold multiple battery boxes to be loaded. Furthermore, to achieve orderly delivery of multiple boxes, the delivery area is divided into multiple sub-areas, each corresponding to an individual box. In this way, multiple boxes can be arranged sequentially along the preset motion path, facilitating the robotic arm to perform delivery operations in order.

[0026] In this embodiment, each sub-region has a corresponding regional coordinate system. The regional coordinate system can be a local coordinate system established with a fixed reference point within the sub-region (such as the center point of the bottom of the packaging box, the lower left corner of the sub-region, etc.) as the origin, used to describe the spatial pose information of the packaging box within the sub-region.

[0027] In addition, each sub-region is pre-set with corresponding preset placement pose parameters. These preset placement pose parameters can include target position and orientation information of the robotic arm's end effector during placement, such as displacement parameters (X, Y, Z) and orientation parameters (rotation angles around each axis) in three-dimensional space. By setting preset placement pose parameters for each sub-region, it is possible to accurately place the battery into the corresponding packaging box within the appropriate sub-region under ideal conditions.

[0028] It should be noted that the number and arrangement of the above-mentioned placement areas and sub-areas are only illustrative examples. In actual applications, they can be flexibly adjusted according to the packaging box specifications, battery size, and production line cycle time. This disclosed embodiment does not impose any limitations on this.

[0029] S102, obtain the actual pose of the packaging box in each sub-region relative to the region coordinate system, and update the delivery pose parameters of the corresponding sub-region based on the actual pose and the preset delivery pose parameters.

[0030] Specifically, in actual production, due to factors such as placement errors of the packaging box, vibration, human intervention, or equipment tolerances, the actual position and posture of the packaging box within a sub-area often deviate from the theoretical settings. If the preset placement posture parameters are still used directly for placement, it is easy to cause inaccurate battery placement, collision with the edge of the packaging box, or even placement failure.

[0031] Therefore, in this embodiment, the preset placement pose parameters are dynamically corrected by acquiring the actual pose of the packaging box relative to the corresponding region's coordinate system within each sub-region. The methods for acquiring the actual pose may include, but are not limited to, using visual recognition systems (such as industrial cameras or 3D cameras), laser sensors, displacement sensors, etc., to detect the spatial position and orientation information of the packaging box.

[0032] After obtaining the actual pose of the packaging box, the actual pose is compared and analyzed with the preset delivery pose parameters of the corresponding sub-region. The pose deviation between the two is calculated, and the original delivery pose parameters are compensated and updated based on the deviation, so as to obtain the updated delivery pose parameters.

[0033] By using the above method, the robotic arm can be controlled based on the real-time corrected placement posture parameters when performing the placement action, thereby effectively improving the accuracy and stability of battery placement.

[0034] S103, in response to detecting a sub-region in the in-place state, the robotic arm picks up the battery from the picking station and moves the target sub-region along a preset motion path. After the robotic arm reaches the preset safe height of the target sub-region, it performs a placement action according to the compensated and updated target placement pose parameters to put the battery into the packaging box in the target sub-region.

[0035] Specifically, after updating the placement pose parameters for each sub-region, the status information of each sub-region is monitored in real time. When a sub-region is detected to be in place, it indicates that the packaging box in that sub-region is in place and meets the conditions for performing the placement operation. At this time, the sub-region is determined as the current target sub-region.

[0036] Subsequently, the robotic arm is controlled to pick up the batteries to be delivered from the picking station. The picking station can be a battery buffer area, the discharge port of upstream equipment, or a designated location on a conveyor belt; this embodiment of the disclosure is not limited to this.

[0037] After grasping the battery, the robotic arm moves towards the target sub-area along a preset motion path. When the robotic arm moves to a preset safe height above the target sub-area, the preset safe height is used to ensure a safe distance between the end effector of the robotic arm and the packaging box, avoiding collisions during posture adjustment or path correction.

[0038] After reaching the preset safe height, the robotic arm performs precise positioning control based on the compensated and updated target placement posture parameters, and executes the placement action to smoothly place the battery into the packaging box in the target sub-area.

[0039] By introducing a preset safety height and compensated updated deployment posture parameters, the risk of mechanical interference during deployment can be effectively reduced, and the reliability of operation can be improved.

[0040] S104, in response to the packaging box in the current target sub-region reaching the preset loading condition, the delivery target is switched to the next sub-region adjacent to the target sub-region and located on the preset movement path.

[0041] Specifically, after completing one or more battery deployment operations, the system determines in real time whether the packaging boxes in the current target sub-area have met the preset loading conditions. The preset loading conditions may include, but are not limited to: the number of batteries in the packaging box reaching a set threshold, the weight of the packaging box reaching a preset value, or the packaging box being determined to be fully loaded.

[0042] When the system detects that the packaging box in the current target sub-region has reached the preset loading conditions, it means that the packaging box is no longer suitable for continuing to place batteries. At this time, it is necessary to switch the placement target. The system will automatically switch the placement target to the next sub-region adjacent to the current target sub-region and located on the preset movement path, thereby realizing the sequential placement of multiple sub-regions.

[0043] By using the above target switching method, the robotic arm does not need to frequently adjust its overall motion strategy. It only needs to complete the delivery tasks of each sub-area sequentially along a predetermined path, thereby improving the overall delivery efficiency.

[0044] In some embodiments, see Figure 2 The flowchart shown in this embodiment of the present disclosure illustrates an update process for projection pose parameters, as follows: Figure 2 As shown, the following update steps can be performed: S201, Obtain the spatial point set or edge features of the opening area of ​​the packaging box.

[0045] Specifically, when performing automatic delivery, gripping, placement, or loading alignment of packaging boxes, it is necessary to accurately determine the position and orientation information of the box opening in three-dimensional space. To this end, the opening area of ​​the packaging box is first sensor-sampled to obtain data that can characterize the geometric shape of the box opening.

[0046] In this embodiment, the spatial point set can be a set of three-dimensional point clouds located near the box opening, which are collected by sensors such as depth cameras, structured light, TOF, binocular vision or three-dimensional lasers. Each point in the point cloud contains three-dimensional coordinate information (e.g. (X, Y, Z) in the camera coordinate system).

[0047] Edge features can be: box edge segments, edge point sequences, corner points, boundary curve sets, etc., obtained through image edge detection, line segment extraction, contour fitting, or point cloud boundary extraction.

[0048] As a non-limiting example, when the opening of the packaging box is approximately rectangular, the four edge segments or four corner points of the opening can be extracted as edge features; when the opening is approximately circular or irregular in shape, the set of boundary points of the opening can be obtained and used to fit the opening plane.

[0049] S202, the box opening plane is obtained by fitting based on the spatial point set or edge features, and the normal vector of the box opening plane is calculated.

[0050] Specifically, the box opening area can be considered as an approximate planar structure in local geometry. Therefore, the plane parameters of the box opening can be obtained through fitting calculations, thereby determining the spatial orientation of the plane.

[0051] In this embodiment, when the input is a set of spatial points, the box opening plane equation can be obtained by using methods such as least squares plane fitting, RANSAC plane fitting, or weighted least squares fitting.

[0052] When the input is edge features, the edge point set can be fitted to a plane first; or several sampling points on the edge line segment can be used as fitting samples to obtain the plane parameters.

[0053] Furthermore, when the edge feature contains multiple edge lines, the spatial distribution of the edge lines can be obtained first, and then the best fitting plane can be calculated to obtain stronger noise resistance.

[0054] By using the above method, the problem of unstable attitude solution caused by uncertainty in the positive and negative normal vectors of the fitting results on the same plane can be avoided.

[0055] S203, using the normal vector of the box opening plane as the attitude constraint, and combining it with at least one edge direction or at least two feature point directions within the box opening plane, the three-dimensional spatial pose of the packaging box relative to the regional coordinate system of the corresponding sub-region is obtained.

[0056] Specifically, the plane normal vector alone can only determine the tilt direction of the box opening plane, but cannot determine the rotation angle of the box opening within that plane (i.e., the rotational degree of freedom about the direction of the normal vector). Therefore, this step, in addition to using the box opening plane normal vector as an attitude constraint, also needs to introduce in-plane orientation information to fully determine the three-dimensional attitude of the packaging box.

[0057] In this embodiment, the regional coordinate system can be: the local coordinate system corresponding to the target sub-region (such as a certain area of ​​the delivery station, a sub-grid divided by the conveyor belt, a partition in the material frame, a palletizing cell, etc.), whose origin and coordinate axis direction can be obtained by calibration plate, mechanical base coordinate system conversion or tooling reference definition.

[0058] In this design, the unit normal vector of the box opening plane is used as an axial constraint on the box's attitude. As a non-restrictive example, it can be aligned with the Z-axis direction of the box's attitude coordinate system or kept in a fixed relationship with it, thereby determining the pitch and roll angles.

[0059] When there is at least one edge direction, such as extracting a long side direction vector to the box opening, the long side direction vector can be used as the X-axis (or Y-axis) direction reference of the packaging box posture coordinate system in the box opening plane, and then normalized and orthogonalized.

[0060] When there are at least two feature point directions, such as extracting two corner points of the box opening, a direction vector can be constructed and projected onto the box opening plane to obtain a stable direction in the plane.

[0061] After determining the long side direction vector and the direction vector, a third mutually orthogonal axial vector can be constructed to form the rotation matrix of the packaging box relative to the regional coordinate system.

[0062] Simultaneously, displacement information within the box opening plane can be obtained based on the spatial point set or edge features, such as the box opening center point (which can be obtained by fitting the center of edge points, the mean of corner points, or the center of the bounding box), thus yielding the translation vector. Ultimately, the three-dimensional spatial pose of the packaging box relative to the corresponding sub-region is obtained.

[0063] It should be noted that in some embodiments, when the symmetry of the box opening edge leads to directional ambiguity (e.g., square box opening, circular box opening), this embodiment can also introduce additional discrimination information to eliminate ambiguity.

[0064] S204. Based on the three-dimensional spatial pose, determine the pose deviation between the target delivery pose parameters and the actual pose, generate pose compensation parameters, and update the delivery pose parameters corresponding to the target sub-region based on the pose compensation parameters.

[0065] Specifically, in automated delivery tasks, the system typically pre-configures target delivery pose parameters (such as delivery point coordinates, end effector attitude, descent height, and entry angle) for each sub-region. However, due to factors such as sensor errors, equipment setup deviations, tooling deformation, conveyor belt vibration, and box size variations, the actual detected packaging box pose often deviates from the preset target. To improve delivery accuracy, this step calculates and compensates for this deviation.

[0066] The target deployment pose parameters can be: a pre-set reference pose facing the target sub-region, which can be obtained from offline calibration, teaching, process database or simulation planning.

[0067] Based on the actual pose obtained in step S203, the pose deviation between the two is calculated, and the pose deviation is mapped to pose compensation parameters.

[0068] In summary, by employing the above method and combining box opening plane fitting, normal vector posture constraints, and in-plane direction supplementary constraints, the complete solution of the three-dimensional pose of the packaging box is achieved. Furthermore, compensation parameters are generated based on the deviation between the target pose and the actual pose to update the placement pose parameters of the target sub-region. This improves placement alignment accuracy and robustness, reduces placement failure rate and equipment collision risk caused by pose errors, and ultimately enhances the overall automated operation efficiency and yield of the production line.

[0069] Correspondingly, a flowchart of a battery placement control method in an embodiment of this disclosure, as shown in Figure 3, can also be further executed, such as... Figure 3 As shown: S301, during the lowering stage near the opening area of ​​the packaging box, position control is used to constrain the lateral displacement and attitude of the end effector of the robotic arm, so that the attitude of the end effector is consistent with the normal vector constraint of the box opening plane.

[0070] Specifically, during the process of lowering the battery from above into the opening area of ​​the packaging box, the opening of the packaging box typically corresponds to a box opening plane (e.g., a plane obtained by geometric fitting of the box opening edge), which has a unique normal vector direction. To avoid the end effector scraping, colliding, or getting stuck with the box opening edge due to lateral offset or tilting when approaching the box opening, this embodiment prioritizes the use of position control mode during the lowering stage to constrain and control the lateral displacement and attitude of the end effector.

[0071] In this embodiment, lateral displacement can refer to two degrees of freedom displacement perpendicular to the lowering direction (e.g., X / Y direction displacement defined in the world coordinate system or the box-mouth coordinate system), while the lowering direction can refer to the direction opposite to gravity or the normal to the box-mouth (e.g., the Z direction). In position control mode, the target pose of the end effector in the lateral direction can be set as a pre-aligned pose, that is: the lateral position of the end effector is constrained within a preset trajectory range, and the attitude of the end effector is constrained to be consistent with the normal vector of the box-mouth plane or consistent within the allowable error range.

[0072] As a non-limiting example, the end effector attitude constraint can be expressed as follows: the angle between the Z-axis of the end effector's tool coordinate system and the normal vector of the box opening plane does not exceed a preset angle threshold (e.g., ≤1° or ≤2°), thereby maintaining an attitude facing the box opening when approaching it; at the same time, the end effector's lateral displacement constraint can be expressed as follows: the projection point of the end effector onto the box opening plane is always located within a preset safe area (e.g., the central area of ​​the box opening or the buffer area of ​​the avoidance edge).

[0073] It should be noted that the role of position control in this step is to establish deterministic geometric constraints to ensure that the lateral degree of freedom and attitude of the end effector do not drift uncontrollably in the high-risk stage of approaching the box opening, thereby providing stable and repeatable entry conditions for the force control switching in the subsequent contact stage.

[0074] S302, after the end effector enters the preset approach distance along the lowering direction, switches to force / position hybrid control mode, while maintaining the lateral pose and attitude unchanged or restricted from changing, and performs closed-loop control of the contact force in the lowering direction.

[0075] Specifically, when the end effector continues to move in the downward direction and gradually approaches the opening area of ​​the packaging box and the internal support surface (such as the bottom of the box, the inner lining, the positioning step or the limiting structure), if pure position control is still used, small positioning errors in the downward direction, box deformation, clamp compliance errors or battery size tolerances may cause impact force at the moment of contact, which may result in pressure damage to the battery surface, damage to the packaging lining or rebound of the end effector.

[0076] Therefore, in this embodiment, after detecting that the end effector has entered a preset approach distance along the downward direction (i.e., the distance between the end effector and the target placement plane or the reference height inside the box is less than or equal to a preset distance threshold), the execution control mode is switched: from position control mode to force / position hybrid control mode.

[0077] In this embodiment, the preset approach distance can be calculated based on displacement sensors, visual ranging, box opening geometry models, or robotic arm encoders.

[0078] As a non-limiting example, the preset approach distance can be a value in the range of 1mm to 10mm, or can be dynamically set according to the end effector, gripper compliance and battery weight.

[0079] In the force / position hybrid control mode, different control strategies can be applied to different degrees of freedom, as shown in the following example: In the lateral and attitude directions, position control or strong constraint control is continued to be used to keep the lateral position and attitude unchanged, or only allow fine adjustment within a preset limit (e.g., allow ±0.2mm lateral fine adjustment, allow ±0.5° attitude fine adjustment) to avoid the end effector from lateral slippage due to contact, which would cause scratches on the edge of the box.

[0080] In the downward direction, a force control closed loop is used to adjust the contact force of the end effector in real time, so that the force in the downward direction tracks the target force (such as a light touch force or a slow force increase curve), thereby achieving compliant contact and stable fit.

[0081] It should be noted that the force / position hybrid control in this step can be implemented using impedance control, admittance control, or an explicit force controller; its core is to switch the lowering direction from position-dominated to force-dominated in order to reduce contact impact and improve the robustness of the placement process, while maintaining lateral and attitude constraints to ensure that geometric alignment is not compromised.

[0082] S303, based on the contact force signal, displacement signal or force / position change rate detected in the force / position hybrid control mode, the release judgment condition is determined. When the release judgment condition is met, the end effector is controlled to release the battery to complete the battery placement. The release judgment condition includes at least one of the following: the contact force in the downward direction reaches a preset threshold and is maintained for a preset time; the displacement of the end effector reaches the target placement displacement and the contact force change rate is less than the preset threshold; after the contact force reaches the threshold, the end effector pose change is less than the preset pose drift threshold.

[0083] Specifically, in the force / position hybrid control mode, the contact force signal in the lowering direction (such as the component of a six-dimensional force sensor, the signal from a wrist torque sensor, or the force signal estimated by the motor current) and the end displacement signal (such as encoder displacement, visual feedback displacement, or displacement relative to the reference surface inside the box) can be acquired in real time. By filtering, denoising, and trend analysis of the above signals, a judgment logic for stable positioning can be constructed to determine when to perform the battery release action.

[0084] In this embodiment, the release determination condition is used to characterize that the battery has been reliably contacted and stably supported, and that after release, it will not obviously bounce, tip over, slide sideways, or get stuck between the clamp and the box opening.

[0085] The release determination condition must include at least one of the following (these can be used individually or in combination to form a more reliable determination): (1) The contact force in the downward direction reaches the preset threshold and is maintained for a preset time.

[0086] Specifically, when the contact force in the downward direction reaches a preset force threshold F_th (e.g., the minimum force indicating that the battery has compacted into contact with the inner liner / support surface), and this contact force remains within the allowable fluctuation range (e.g., ±ΔF) within the duration T_hold, it can be determined that the battery has formed a stable contact. At this time, the end effector is controlled to perform a release action.

[0087] As a non-limiting example, F_th can be set according to the battery weight, the stiffness of the inner liner material, and the allowable compression; T_hold can be set from 50ms to 500ms to avoid misjudging transient spikes.

[0088] (2) The end effector displacement reaches the target placement displacement and the rate of change of contact force is less than the preset threshold. Specifically, when the end effector's displacement along the downward direction reaches the target placement displacement Z_target (e.g., reaching the designed placement height inside the box or reaching the safe height at the end of the stroke), and at the same time, the contact force change rate |dF / dt| is detected to be less than the preset threshold K_F (indicating that the contact has entered a steady state and there is no obvious trend of continued compression or rebound), it can be determined that the battery has been placed and the force is stable, thereby executing the release action.

[0089] It should be noted that by introducing the contact force change rate condition, we can avoid erroneous releases that occur when the displacement is in place but the object is still in the compression rising phase or the rebound recovery phase, thus improving placement consistency.

[0090] (3) After the contact force reaches the threshold, the change in end pose is less than the preset pose drift threshold.

[0091] Specifically, when the contact force reaches the threshold, if the change in the lateral pose and attitude of the end effector within a certain decision window is less than the preset drift threshold (e.g., lateral drift ≤ D_th, attitude change ≤ A_th), it indicates that no obvious lateral slip, rotation or edge jamming occurred after contact, the battery is in a stable and restricted state, and the release action can be performed.

[0092] After the release determination conditions are met, this embodiment controls the end effector to perform a release action. The release action may include: releasing the gripper, releasing the vacuum adsorption, opening the electromagnetic clamp, or other fixing methods. After the release is completed, the end effector can be withdrawn in the reverse direction to a safe height and return to the initial position for the next cycle to enter the next battery handling and placement process.

[0093] It should be noted that the release action can be set to release in stages, such as first reducing the clamping force to a preset value and then completely releasing the clamp to reduce the disturbance at the moment of release; at the same time, force / position hybrid control can be maintained for a short time after release to confirm that the battery has not been lifted or has not rebounded significantly.

[0094] By employing the above methods, positional constraints are used to ensure geometric alignment during the approach stage to the box opening, force / position hybrid control is used to achieve compliant placement during the contact stage, and release judgment conditions are constructed through multi-source signals to ensure stable placement. This effectively reduces placement impact, minimizes the risk of scratches and jamming, improves the consistency and yield of battery placement in the packaging box, and thus enhances the reliability and production efficiency of the entire automated packaging line.

[0095] In the above embodiments, the present invention achieves automatic, sequential, and precise placement of batteries in multiple packaging boxes by dividing the placement area into sub-regions, compensating and updating the placement posture parameters based on the actual posture of the packaging box, and combining the preset motion path and placement control strategy of the robotic arm.

[0096] However, in the continuous automated packaging process, factors such as battery batch differences, packaging box size tolerances, end effector wear, and environmental disturbances may affect the stability and consistency of packaging quality. Although the pose compensation and placement control within a single packaging cycle can ensure the current placement accuracy, they may still have a cumulative effect over multiple packaging cycles.

[0097] Based on this, the present invention further proposes a packaging control parameter optimization scheme based on process state modeling. By collecting process parameters related to the battery packaging process in multiple packaging cycles, an implicit process state model is constructed to characterize the internal process evolution of the battery packaging process. Based on this model, the process state in future packaging cycles is predicted, thereby assessing the packaging quality risk and generating a combination of packaging control parameters for the current packaging cycle. This further improves the stability of the automatic battery packaging process and the consistency of packaging quality without changing the basic automatic packaging process.

[0098] See Figure 4 The flowchart shown in this disclosure illustrates an adaptive update method for battery packaging, as follows: Figure 4 As shown, the following steps can be performed: S401, during multiple consecutive battery packaging cycles, collect at least one process parameter related to the robotic arm's battery placement process.

[0099] Specifically, automated battery packaging typically uses a packaging cycle as the basic repeating unit. Each packaging cycle can include actions such as: taking out the battery, aligning it, pressing / placing it, releasing it, removing it, and moving on to the next cycle.

[0100] Because the robotic arm repeatedly performs placement actions in consecutive cycles, the contact and interaction between the end effector, gripping mechanism, packaging box, and battery exhibit continuity and cumulativeity over time, leading to drift or abrupt changes in placement quality over multiple cycles. Therefore, collecting process parameters over several consecutive packaging cycles helps to capture this cumulative effect and provide a basis for subsequent modeling.

[0101] In this embodiment, the process parameters may include at least any one of the following categories or combinations thereof (for example only, not as a limitation): robotic arm motion and dynamics parameters: end-effector pose (position / attitude), placement path deviation, speed, acceleration, joint torque, servo following error, stopping stabilization time, etc.; end-effector and clamping parameters: clamping force / clamping current, release sequence, clamp opening and closing displacement, adsorption pressure (such as vacuum adsorption), end-effector compliant mechanism compression, etc.; contact and packaging interface parameters: downward contact force, contact force change rate, contact duration, friction estimate, collision event flag, contact surface vibration amplitude, etc.; battery and packaging box characteristic parameters: battery dimensional deviation, weight, surface friction characteristics, packaging box positioning deviation, inner lining elastic rebound characteristics, etc.; environmental and operating condition parameters: temperature, humidity, production line cycle time, equipment vibration, lubrication status, etc.

[0102] It should be noted that, firstly, process parameters can be obtained from the robotic arm controller, force / torque sensor, vision system, displacement sensor, and clamp current detection module; secondly, process parameters can be obtained by sampling at multiple moments within the same cycle, sampling triggered by key events (such as the moment of contact or release), or high-frequency sampling throughout the process; and thirdly, to facilitate cross-cycle alignment, a unified time reference or event reference is usually established for each packaging cycle (for example, the moment when the contact force is detected to exceed the threshold for the first time is taken as the zero point within the cycle).

[0103] In this embodiment, step S401 may include: A1) The packaging cycle of the robotic arm entering the packaging box is divided into the battery removal segment, posture adjustment segment, box positioning segment, box pushing segment, release segment, and reset segment.

[0104] A2) The contact signal detected by the end effector of the robotic arm when the battery first makes contact with the support surface of the packaging box is taken as the zero point of the packaging cycle.

[0105] A3) Around time zero, collect the pushing force curve, displacement curve, and pushing speed curve of the pushing segment into the packaging box. A4) Extract the peak pushing force, force rise slope and force-displacement hysteresis area from the pushing force curve; extract the infeed stroke deviation and end rebound amount from the displacement curve; and extract the speed fluctuation amplitude from the pushing speed curve.

[0106] A5) The extracted results are concatenated into a periodic feature vector for the packaging cycle, and the periodic feature vectors of multiple consecutive packaging cycles are constructed into a parametric time series in chronological order.

[0107] By subdividing a single packaging cycle into multiple functional stages and using the first contact signal as a unified time zero point, the timing alignment of process parameters in different cycles is achieved. Force, displacement, and velocity curves are collected in the key box-in pushing section, and multi-dimensional feature parameters that can reflect contact characteristics and dynamic response are extracted, so that each packaging cycle can be abstracted into a structured feature vector. The feature vectors of continuous cycles are constructed into a parameter time series, providing a high-information-density data foundation for subsequent process state modeling.

[0108] S402, construct a parameter time series based on process parameters, and establish a process state model based on the parameter time series to characterize the evolution of the packaging interface contact state during battery placement. The process state model is used to describe the cumulative impact of process parameters on the contact stability between the battery and the packaging box during continuous placement.

[0109] Specifically, by organizing the process parameters collected in step S401 according to the packaging cycle sequence, a parameter time series can be obtained. The parameter time series not only reflects the instantaneous state of a single cycle, but also reflects the trend changes, fluctuation amplitudes, and cumulative offsets between multiple cycles, thus providing a calculable representation of the contact state evolution.

[0110] In this embodiment, the parameter time series can be represented as: for the k-th packaging cycle, extracting several feature vectors. ,in It can be composed of key statistics or events within the period, such as: maximum contact force, contact force rise slope, peak end attitude deviation, release delay, average clamping force, vibration energy, etc. Arranging the characteristics of continuous periods in sequence yields the corresponding time series.

[0111] Furthermore, a process state model is established based on the parameter time series to characterize the evolution of the contact state at the packaging interface. The process state can be understood as the comprehensive state of the battery during placement, including contact stability, fit, alignment consistency, and implicit changes caused by equipment wear, misalignment, temperature drift, etc. This state is usually difficult to observe directly from a single parameter, but can be inferred from the correlation changes of multiple parameters over time.

[0112] In this embodiment, the process state model is used to describe the cumulative impact of process parameters on the contact stability between the battery and the packaging box during continuous placement. For example, slight drift in the clamping and release timing causes the battery placement posture to gradually deviate after multiple cycles; small fluctuations in the pressing speed cause repeated impacts of the contact force, resulting in the accumulation of local compression deformation of the inner liner, which in turn reduces the contact stability in subsequent cycles; visual alignment errors can be absorbed by the compliant structure in the short term, but gradually approach the limit under continuous cycle time, eventually triggering placement abnormalities.

[0113] The form of the process state model is not limited; it can be a state-space model, a hidden Markov model, a recursive filter model, a machine learning-based temporal network model, etc. As a non-restrictive example, the process state can be denoted as... Its evolution is recursively determined by the period: Status Update: It is used to characterize the cumulative effect.

[0114] Observation mapping: It is used to characterize the contact performance or quality signs that can be observed in the current cycle.

[0115] In this example, the process state model construction and update in step S402 includes the following process: B1) Define the process state as the cumulative deformation state of the packaging box structure, the alignment drift state between the battery and the box, and the attenuation state of the gripping stability at the end of the robotic arm, and form a single-cycle implicit process state vector.

[0116] B2) Based on the implicit process state estimate of the previous packaging cycle and the robotic arm control parameters of the current packaging cycle, generate the predicted process state.

[0117] B3) Generate a predicted periodic feature vector based on the predicted process state, and calculate the feature residual with the actual obtained periodic feature vector.

[0118] Specifically, the predicted process state can be mapped to the corresponding packaging process characteristics, such as displacement characteristics, force change characteristics, and posture stability characteristics during the pushing process, thereby generating a predicted cycle feature vector.

[0119] Meanwhile, the actual cycle feature vector is obtained by collecting data from the actual packaging process through sensors. The predicted cycle feature vector is compared with the actual cycle feature vector, and the feature residual between the two is calculated to characterize the deviation between the predicted state and the actual process behavior.

[0120] B4) Correct the predicted process state based on the feature residuals and output the estimated process state value for the current packaging cycle.

[0121] B5) Collect the set values ​​of the robotic arm's box-pushing speed and clamping force, and the deviation between them and the actual executed values, and introduce the deviation as an execution deviation vector into the implicit process state prediction process.

[0122] By abstracting the process state into implicit states such as box structure deformation, battery alignment drift, and clamping stability decay, the system can trace from macroscopic quality results to microscopic cumulative effects. The system uses the process state of the previous cycle and the current control parameters to predict the state of the next cycle and corrects it through characteristic residuals to achieve dynamic estimation of the process state. An execution deviation vector is introduced to incorporate the difference between the set value and the actual execution into the state prediction, thereby improving the model's adaptability to real operating conditions.

[0123] S403, based on the process state model, establishes the parameter interaction relationship between process parameters and battery placement quality results through process state, in order to distinguish between control parameters that have a direct impact on battery placement quality and auxiliary parameters that have a constraint effect on control parameters.

[0124] Specifically, in actual packaging, not all process parameters affect placement quality in the same way: some parameters directly alter the contact behavior between the battery and the packaging box (e.g., pressing speed, end posture, clamping and releasing sequence), thus directly impacting placement quality; others do not directly determine the placement result but create constraints, boundaries, or coupling relationships with the directly influencing parameters (e.g., ambient temperature causing clamping force calibration drift, equipment vibration limiting the upper limit of usable acceleration, and liner rebound affecting the effective range of pressing force). Therefore, it is necessary to establish a parameter-process state-quality outcome interaction chain to achieve parameter hierarchical and interpretable control.

[0125] In this embodiment, based on the process state model, parameter interaction relationships are established for description: 1) How process parameters affect the process state (i.e., the driving effect of parameters on the evolution of contact stability).

[0126] 2) How the process condition affects the quality result (i.e. how the condition deterioration or improvement is reflected in the placement quality fluctuation).

[0127] 3) The different roles of different parameters in this action chain.

[0128] The battery placement quality results may include, but are not limited to: placement position deviation, posture deviation, gap between the battery and the packaging box, whether jamming / collision occurred, battery stability after packaging (amount of shaking / loosening), and first-pass yield.

[0129] Based on the above parameter interactions, we can distinguish: Control parameters: Parameters that have a direct impact on placement quality and can be adjusted by the control system, such as placement speed curve, pressure depth / force control threshold, release time, end attitude compensation, compliance stiffness setting, etc.

[0130] Auxiliary parameters: Parameters that constrain or indirectly affect control parameters and are usually not suitable for direct or frequent adjustment, such as ambient temperature and humidity, equipment wear indicators, fixture hysteresis, and statistical quantities of incoming material deviation for packaging boxes.

[0131] It should be noted that the distinction between control parameters and auxiliary parameters is not fixed: under different equipment configurations or different production line strategies, the same parameter may be used as a control parameter in some embodiments and as an auxiliary parameter in other embodiments. This disclosure does not limit this.

[0132] In some embodiments, the identification of causal transmission paths includes the following process: C1) While keeping the battery model and packaging box specifications unchanged, apply a micro-perturbation of a preset amplitude to a single control parameter of the robotic arm's box-entry pushing speed or clamping force, while keeping the other control parameters unchanged.

[0133] C2) Obtain the implicit process state changes and corresponding packaging result data within the micro-perturbation period.

[0134] C3) Compare the implicit process state changes and packaging results data between the micro-disturbance period and the non-micro-disturbance period.

[0135] C4) When the comparison results meet the preset discrimination conditions, the corresponding control parameter is determined to be a key control parameter that has a direct causal impact on packaging quality, and its direction of action is recorded.

[0136] C5) Based on the results of multiple micro-perturbations, each control parameter is sorted according to the magnitude of the change in process state it causes, and the control parameters with a change magnitude higher than the preset threshold are selected to form a control parameter set.

[0137] S404, based on the process state model and parameter interaction relationship, predicts the trend of process state change within at least one packaging cycle in the future, and generates the corresponding packaging quality risk assessment result when the prediction result indicates that the process state has the risk of entering a preset abnormal state.

[0138] Specifically, since the process status evolves over time, historical parameter time series and current process status estimates can be used to predict the status change trend for at least one packaging cycle (or multiple cycles). The purpose of the prediction is to identify the deteriorating trend of the contact status in advance before quality results actually occur, and to provide a risk assessment output that can be used for control decisions.

[0139] In this embodiment, abnormal states can be preset, such as: the contact stability index is lower than the threshold; the predicted posture deviation will exceed the allowable tolerance; the predicted peak contact force will exceed the safety limit and may cause a collision; the predicted risk of loosening after placement is increased (e.g., the fitting gap shows a significant growth trend).

[0140] When the prediction result indicates that the process status poses a risk of entering a preset abnormal state, a corresponding packaging quality risk assessment result is generated. The risk assessment result may include, but is not limited to: risk level (high / medium / low), risk type (displacement risk, collision risk, loosening risk, etc.), expected triggering cycle range, main contributing parameters and their contribution ranking, and suggested adjustment direction, etc.

[0141] It should be noted that the risk assessment results are not required to be equivalent to the final quality judgment, but are used to provide early warnings and guide parameter adjustments in order to reduce the probability of anomalies or mitigate their consequences.

[0142] Correspondingly, packaging quality risk assessment includes the following process: D1) The packaging result data shall be limited to include at least one of the following: battery misalignment, inability of the lid to close properly, or indentation on the box.

[0143] D2) Establish the correspondence between packaging result data and process status estimates.

[0144] Specifically, in order to infer the inherent state of the process from the observable packaging results, which is difficult to observe directly, this embodiment establishes a correspondence between packaging result data and process state estimates.

[0145] In some embodiments, the process state estimate can be used to characterize the state variables implicit in the packaging process that have a decisive impact on the packaging quality, such as: drift states related to battery and box alignment (e.g., assembly reference offset, fixture positioning drift, feeding posture deviation, etc.); states related to the cumulative deformation of the box structure (e.g., clamping fatigue, micro-deformation caused by force accumulation, material springback changes, etc.).

[0146] The above-mentioned process status estimates can be obtained through multi-source data fusion.

[0147] As a non-restricted example, equipment sensor data (force, displacement, vibration), fixture positioning feedback, visual inspection features, environmental variables, historical anomaly records, etc., can be input into the state estimation model to obtain the implicit process state estimate at the current moment.

[0148] Furthermore, the "correspondence" in this step can be implemented in various forms, such as: establishing a mapping based on statistical learning or regression / classification models so that a given process state estimate can predict the probability of abnormal packaging results; establishing a relationship based on causal or mechanistic constraints, such as an increase in alignment drift leading to an increase in the probability of battery insertion bias and closure failure; and establishing a correspondence based on rule thresholds and combinational logic, such as triggering the insertion bias tag when the alignment drift exceeds a certain range.

[0149] It should be noted that the correspondence is not limited to a one-to-one correspondence; the same packaging abnormality may be caused by multiple process states, and the same process state may also cause multiple packaging abnormalities. This embodiment allows the establishment of many-to-many mapping relationships.

[0150] D3) Using at least one of the following as failure conditions, namely battery misalignment in the box or failure of the box cover to close properly, or equivalent packaging abnormalities, the state threshold of battery alignment drift or box structure deformation accumulation is obtained by reverse calculation.

[0151] Specifically, after establishing the correspondence, this embodiment introduces packaging abnormalities as a failure condition to achieve the reverse determination of the implicit process state threshold.

[0152] In this embodiment, the failure condition includes at least one of the following: battery misalignment upon insertion into the box; failure of the box lid to close properly; or a packaging abnormality equivalent to the above-mentioned abnormality (such as excessive insertion misalignment, excessive closing gap, or failure to fasten, which have the same quality consequences or rework / scrap consequences as the target abnormality).

[0153] Based on the failure condition, this step inversely calculates at least one of the following state thresholds: Battery and housing alignment drift state threshold: used to distinguish whether the alignment drift is within an acceptable range or within a high-risk range.

[0154] State threshold for cumulative deformation of box structure: used to characterize the structural changes of box caused by long-term stress, clamping or closing action.

[0155] In some embodiments, the reverse process can be manifested as: finding a set of samples that meet the failure conditions in historical samples and determining the boundary point in the corresponding process state estimate distribution; or determining the state value in the model that makes the failure probability reach a preset level (e.g., reach a certain risk probability) as a threshold.

[0156] It should be noted that this step can yield a single threshold or multiple thresholds (e.g., alignment drift threshold and deformation accumulation threshold are determined separately), and the thresholds can be single-dimensional thresholds, two-dimensional joint thresholds, or a set of multi-dimensional thresholds. This embodiment of the invention does not impose any restrictions on this.

[0157] D4) Construct failure state regions with hidden process states based on state thresholds.

[0158] D5) Sample the current implicit process state estimate multiple times and predict the implicit process state of future packaging cycles. Calculate the proportion of predicted states that fall into the failure state area and use it as the result of packaging quality risk assessment.

[0159] Specifically, this embodiment uses the implicit process state estimate at the current moment, takes into account the uncertainty of the state estimate and the evolution trend of the future process state, and generates a set of possible future state trajectories by multiple sampling, and assesses the packaging quality risk accordingly.

[0160] In this step, the following procedure can be performed: Multiple sampling: Multiple random samplings are performed around the current implicit process state estimate to construct a set of possible true states. As an unrestricted example, sampling can reflect estimation errors, measurement noise, or model uncertainties, such that each sampling yields a possible initial implicit state.

[0161] Predicting the implicit process state of future packaging cycles: Input the sampled initial implicit state into the state evolution model or prediction model to obtain the predicted value of the implicit process state for one or more future packaging cycles (e.g., the next N cycles).

[0162] The proportion of predicted states falling into the failure state region is statistically analyzed: all sampled prediction results are statistically analyzed, and the percentage of times or trajectories of predicted implicit process states falling into the failure state region within the future prediction period is calculated. This percentage is used as the result of packaging quality risk assessment.

[0163] It should be noted that when the prediction period covers multiple packaging periods, the statistical method in this step can be "any period entering the failure area is counted as high risk" or "the cumulative percentage of periods entering the failure area". This embodiment does not limit this, as long as it can output risk assessment results related to packaging quality.

[0164] Using the above method, a correspondence is established between packaging result data (battery insertion bias, box lid failure to close properly, box indentation, etc.) and implicit process state estimates. Furthermore, thresholds are calculated and failure state regions are constructed. Probabilistic risk assessment is then achieved through multiple sampling predictions. This method can identify the process state deterioration trend before packaging anomalies occur in large numbers, thereby providing a basis for early maintenance, process parameter adjustment, or line stop calibration. This reduces the probability of packaging failure and improves production line stability and yield, thereby reducing manufacturing costs.

[0165] S405, under the premise of meeting the safety constraints of robotic arm movement and packaging process constraints, and based on the packaging quality risk assessment results, jointly constrains the adjustment priority and adjustment direction of multiple packaging control parameters in the control parameter combination, and generates a control parameter combination for the current packaging cycle.

[0166] Specifically, robotic arm control is not a problem of adjusting a single parameter independently: for example, increasing the downward pressing speed may shorten the cycle time, but may increase the impact force; early release may reduce clamping marks, but may cause attitude instability; increasing attitude compensation may improve alignment, but may approach joint limits or cause path interference. Therefore, under safety and process constraints, and in conjunction with risk assessment results, it is necessary to jointly constrain the adjustment priority and direction of multiple control parameters to form an executable combination of control parameters.

[0167] In this embodiment, the safety constraints for the robotic arm's motion may include, but are not limited to: upper limits for joint speed / acceleration / torque; minimum safe distance between the end effector and surrounding structures; collision detection threshold and emergency stop strategy; trajectory smoothness and jitter limitation, etc.

[0168] Packaging process constraints may include, but are not limited to: permissible placement / posture tolerances; maximum permissible contact force and indentation depth; clamping / release timing window; cycle time limit and production line synchronization constraints; and restrictions on the deformation of packaging boxes and inner lining materials.

[0169] Based on the packaging quality risk assessment results, this embodiment can determine that: Adjustment priority: Prioritize adjusting control parameters that contribute the most to the current risk, or prioritize adjusting parameters that have a direct inhibitory effect on safety risks.

[0170] Adjustment direction: For example, when the predicted collision risk increases, the downward speed can be reduced, the contact force threshold can be reduced, or a deceleration phase can be added; when the predicted deviation risk increases, the visual alignment weight can be increased, the attitude compensation gain can be improved, or the stabilization waiting time can be extended.

[0171] Joint constraints: Avoid adjusting one parameter causing another risk to increase. For example, when reducing speed leads to insufficient cycle time, optimize the path or reduce redundancy to compensate for the cycle time; when increasing compensation gain leads to increased vibration, suppress vibration by increasing damping or limiting strategies.

[0172] The final result is a combination of control parameters for the current packaging cycle, which reduces the risk of predicted anomalies while meeting safety and process constraints, and maintains capacity and stability as much as possible.

[0173] In one embodiment, the generation of the control parameter combination includes the following process: Based on the set of control parameters and their direction of action, the permissible adjustment directions of packaging control parameters, including at least the robotic arm's box-feeding speed and clamping force, are defined; upper and lower limits are set for the pushing speed and clamping force, as well as the maximum variation in adjacent packaging cycles; the combination of robotic arm control parameters for the current packaging cycle is generated under the condition that the permissible adjustment direction, upper and lower limits, and maximum variation are met; the adjustment priority of each key control parameter is determined based on the packaging quality risk assessment results, and the control parameters that contribute the most to the risk are adjusted first.

[0174] S406 executes the corresponding automatic battery packaging operation according to the combination of control parameters.

[0175] Specifically, the control parameters obtained in step S405 are sent to the robotic arm controller, the end effector control module, and related sensing / calibration modules, so that the robotic arm can perform pick-up, placement, alignment, pressing, release, and evacuation actions according to the parameters within the current packaging cycle, thereby completing the automatic battery packaging operation.

[0176] It should be noted that while performing the current cycle control, process parameters can still be continuously collected and the parameter time series can be updated so that the process status model and risk assessment results can be continuously updated in subsequent cycles, thereby forming a closed-loop prediction-adjustment-verification mechanism, so that the placement quality and stability of the continuous packaging process can be continuously optimized.

[0177] Using the above method, the evolution of contact state is characterized by parameter time series over multiple consecutive packaging cycles, and an interpretable interaction relationship between process parameters, process state, and quality results is established. Then, combined with trend prediction and risk assessment, a parameter combination with joint constraints is generated. Compared with control strategies that rely solely on fixed thresholds or single-cycle real-time feedback, this approach can identify quality degradation trends earlier, rationally allocate the adjustment priority of multiple parameters, and reduce the probability of placement abnormalities while meeting the safety and process constraints of the robotic arm. This improves the consistency and yield of automated packaging and reduces the overall manufacturing costs caused by rework and line stoppages.

[0178] The automated battery packaging method has been described in detail above through some embodiments. In order to enable those skilled in the art to better understand and implement it, the corresponding system is also described in detail below through some embodiments.

[0179] See Figure 5 The diagram shown below illustrates the structure of an automated battery packaging system according to an embodiment of this disclosure. Figure 5 As shown, the automated battery packaging system 500 may include: Robotic arm 510 is used to pick up batteries and put them into a packaging box; The controller 520, connected to the robotic arm 510, is used to control the robotic arm to sequentially dispense batteries into packaging boxes in each sub-area of ​​the dispensing area using any of the battery automated packaging methods described in the foregoing embodiments.

[0180] For more details regarding the robotic arm 510 and the controller 520, please refer to the aforementioned example.

[0181] 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. The computer program can be stored in a medium or transferred from one medium to another computer-readable storage medium. For example, a computer program can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means.

[0182] While the embodiments disclosed in this application are as described above, the present invention is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. An automated packaging method for batteries, characterized in that, include: A delivery area is set on the preset motion path of the robotic arm, and the delivery area is divided into multiple sub-areas, with a packaging box placed in each sub-area; wherein, each sub-area has a corresponding area coordinate system, and each sub-area has a corresponding preset delivery pose parameter; Obtain the actual pose of the packaging box in each sub-region relative to the region coordinate system, and update the delivery pose parameters of the corresponding sub-region based on the actual pose and the preset delivery pose parameters; In response to the detection of a sub-region in the in-place state, the robotic arm picks up the battery from the picking station and moves the target sub-region along the preset motion path. After the robotic arm reaches the preset safe height of the target sub-region, it performs a placement action according to the compensated and updated target placement pose parameters to put the battery into the packaging box in the target sub-region. In response to the packaging boxes in the current target sub-region reaching the preset loading conditions, the delivery target is switched to the next sub-region adjacent to the target sub-region and located on the preset movement path.

2. The automated packaging method according to claim 1, characterized in that, The step of obtaining the actual pose of the packaging box relative to the coordinate system of each sub-region, and updating the delivery pose parameters of the corresponding sub-region based on the actual pose and preset delivery pose parameters, includes: Obtain the spatial point set or edge features of the opening area of ​​the packaging box; The box opening plane is obtained by fitting the spatial point set or edge features, and the normal vector of the box opening plane is calculated. Using the normal vector of the box opening plane as an attitude constraint, and combining it with at least one edge direction or at least two feature point directions within the box opening plane, the three-dimensional spatial pose of the packaging box relative to the regional coordinate system of the corresponding sub-region is obtained. Based on the three-dimensional spatial pose, the pose deviation between the target delivery pose parameters and the actual pose is determined, pose compensation parameters are generated, and the delivery pose parameters corresponding to the target sub-region are updated based on the pose compensation parameters.

3. The automated packaging method according to claim 1 or 2, characterized in that, The placement action includes: During the lowering phase near the opening of the packaging box, position control is used to constrain the lateral displacement and attitude of the end effector of the robotic arm, so that the attitude of the end effector is consistent with the normal vector constraint of the box opening plane. After the end effector enters the preset approach distance along the lowering direction, it switches to a force / position hybrid control mode, which maintains the lateral pose and attitude unchanged or restricted in change, while performing closed-loop control on the contact force in the lowering direction. Release conditions are determined based on the contact force signal, displacement signal, or force / position change rate detected in the force / position hybrid control mode. When the release conditions are met, the end effector is controlled to release the battery to complete the placement of the battery. The release conditions include at least one of the following: the contact force in the downward direction reaches a preset threshold and is maintained for a preset time; the displacement of the end effector reaches the target placement displacement and the contact force change rate is less than a preset threshold; the end effector pose change is less than a preset pose drift threshold after the contact force reaches the threshold.

4. The automated packaging method according to claim 1, characterized in that: During multiple consecutive battery packaging cycles, at least one process parameter related to the robotic arm's battery placement process is collected; Based on the process parameters, a parameter time series is constructed, and a process state model is established based on the parameter time series to characterize the evolution of the packaging interface contact state during battery placement. The process state model is used to describe the cumulative impact of the process parameters on the contact stability between the battery and the packaging box during continuous placement. Based on the process state model, the parameter interaction relationship between the process parameters and the battery placement quality results through the process state is established to distinguish between control parameters that have a direct impact on battery placement quality and auxiliary parameters that have a constraint effect on the control parameters. Based on the process state model and the relationship between the parameters, the trend of process state change in at least one future packaging cycle is predicted, and when the prediction result indicates that the process state has the risk of entering a preset abnormal state, a corresponding packaging quality risk assessment result is generated. Under the premise of meeting the safety constraints of robotic arm movement and packaging process constraints, based on the packaging quality risk assessment results, the adjustment priority and adjustment direction of multiple packaging control parameters in the control parameter combination are jointly constrained, and a control parameter combination for the current packaging cycle is generated. According to the combination of control parameters, the corresponding automatic battery packaging operation is performed.

5. The automated packaging method according to claim 4, characterized in that, The construction of the corresponding parameter time series includes: The packaging cycle of the robotic arm entering the packaging box is divided into the battery removal segment, the posture adjustment segment, the box positioning segment, the box pushing segment, the release segment, and the reset segment. The contact signal detected by the end effector of the robotic arm when the battery first comes into contact with the support surface of the packaging box is taken as the zero point of the packaging cycle. Around the zero point of time, the pushing force curve, displacement curve and pushing speed curve of the pushing segment into the packaging box are collected; The peak pushing force, force rise slope, and hysteresis area of ​​the force-displacement curve are extracted from the pushing force curve; the infeed stroke deviation and end rebound amount are extracted from the displacement curve; and the speed fluctuation amplitude is extracted from the pushing speed curve. The extracted results are concatenated into a periodic feature vector for the packaging cycle, and the periodic feature vectors of multiple consecutive packaging cycles are constructed into a parametric time series in chronological order.

6. The automated packaging method according to claim 4, characterized in that, The construction and updating of the process state model includes the following process: The process states are defined as the cumulative deformation state of the packaging box structure, the alignment drift state between the battery and the box, and the attenuation state of the gripping stability at the end of the robotic arm, and constitute a single-cycle implicit process state vector. Based on the implicit process state estimate of the previous packaging cycle and the robotic arm control parameters of the current packaging cycle, a predicted process state is generated. A predicted cycle feature vector is generated based on the predicted process state, and the feature residual is calculated with the actual obtained cycle feature vector. The predicted process state is corrected based on the feature residual, and the estimated process state value of the current packaging cycle is output. The set values ​​of the robotic arm's box-pushing speed and clamping force are collected, and the deviation between them and the actual executed values ​​is used as an execution deviation vector to introduce the implicit process state prediction process.

7. The automated packaging method according to claim 4, characterized in that, The identification of the causal transmission path includes the following process: While keeping the battery model and packaging box specifications unchanged, apply a micro-perturbation of a preset amplitude to a single control parameter of the robotic arm's box-pushing speed or clamping force, while keeping the other control parameters unchanged. Within the micro-perturbation period, the implicit process state changes and corresponding packaging result data are obtained; Compare the implicit changes in process state and packaging results between the micro-disturbance period and the non-micro-disturbance period; When the comparison result meets the preset discrimination condition, the corresponding control parameter is determined to be a key control parameter that has a direct causal impact on packaging quality, and its direction of action is recorded. Based on the results of multiple micro-perturbations, each control parameter is sorted according to the magnitude of the process state change it causes, and control parameters with a change magnitude higher than a preset threshold are selected to form a control parameter set.

8. The automated packaging method according to claim 6 or 7, characterized in that, The packaging quality risk assessment includes the following process: The packaging result data is limited to include at least one of the following: battery misalignment, inability of the lid to close properly, or indentation on the box. Establish the correspondence between the packaging result data and the estimated process status; Using at least one of the following, namely, battery misalignment in the box, inability of the box lid to close properly, or equivalent packaging abnormalities, as one of the failure conditions, the state threshold of battery alignment drift or box structure deformation accumulation is obtained by reverse calculation. Based on the state threshold, a failure state region containing implicit process states is constructed; The current implicit process state estimate is sampled multiple times, and the implicit process state of future packaging cycles is predicted. The proportion of predicted states falling into the failure state region is statistically analyzed and used as the packaging quality risk assessment result.

9. The automated packaging method according to claim 4, characterized in that, The generation of the control parameter combination includes the following process: Based on the set of control parameters and their direction of action, the permissible adjustment directions of packaging control parameters, including at least the robotic arm's box-pushing speed and clamping force, are defined. Set upper and lower limits for the pushing speed and clamping force, as well as the maximum variation between adjacent packaging cycles; Generate the robot arm control parameter combination for the current packaging cycle under the conditions of allowing adjustment direction, upper and lower limits, and maximum change; Based on the results of the packaging quality risk assessment, determine the adjustment priority of each key control parameter, and prioritize adjusting the control parameters that contribute the most to the risk.

10. An automated packaging system for batteries, characterized in that, include: A robotic arm used to pick up batteries and put them into a packaging box; A controller, connected to the robotic arm, is used to control the robotic arm to sequentially dispense the batteries into the packaging boxes in each sub-area of ​​the delivery area using the automated packaging method for batteries as described in any one of claims 1 to 9.

Images