Multi-station cooperative soft package lithium battery independent grabbing intelligent control method and system

Abstract

The application provides a multi-station cooperative soft package lithium battery independent grabbing intelligent control method and system, relates to the technical field of automatic control, detects abnormal information generated when a transfer mechanical arm grabs a soft package lithium battery at a preset handover point, responds to the abnormal information, selects and activates a local abnormality processing scheme from a pre-stored local abnormality processing scheme library, controls the transfer mechanical arm to perform a grabbing adjustment action or a safety retreat action according to the scheme, and controls the grabbing mechanical arm to switch its preset motion path to an avoidance path or a temporary hovering path. After the transfer mechanical arm completes the abnormality processing, the grabbing mechanical arm and the transfer mechanical arm resume normal production work, effectively solving the problems of grabbing failure, workpiece damage and mechanical arm collision caused by slight displacement or rotation of the soft package lithium battery in the high-speed transfer process in the prior art.

Description

Technical Field

[0001] This application relates to the field of automation control technology, and more specifically, to a multi-station collaborative intelligent control method and system for independently grasping soft-pack lithium batteries. Background Technology

[0002] In modern industrial production, automation technology has been widely applied to high-precision, high-efficiency manufacturing processes, especially in the production of pouch lithium batteries. To meet market demands, production lines often employ multiple robotic arms working collaboratively to achieve rapid battery gripping, transfer, and assembly. This collaborative operation places extremely high demands on the positioning accuracy and motion coordination of the robotic arms; any minute deviation can affect production efficiency and product quality. However, in actual operation, due to long-term equipment wear, subtle changes in workpiece characteristics, and the inertial effect of high-speed motion, the actual position and posture of the workpiece may deviate from the preset target during gripping and transfer. These deviations are not obvious initially but accumulate as production progresses, eventually leading to gripping failures, workpiece damage, or even accidental collisions between robotic arms, causing production line downtime and economic losses.

[0003] Specifically, on an automated production line for pouch lithium batteries, a gripping robotic arm uses vacuum suction cups to pick up the pouch lithium batteries and transports them at high speed to a pre-set transfer point, where a transfer robotic arm takes over the transfer. However, the vacuum suction cups wear down over time, and microscopic differences exist on the surface of the pouch lithium batteries, leading to a decrease in the stability of the suction cup's adsorption force. During high-speed transfer, the pouch lithium batteries may experience minute and unpredictable relative displacement or rotation within the suction cup, causing their actual position and orientation at the transfer point to deviate from the pre-set target. When this deviation exceeds the tolerance limit of the transfer robotic arm's grippers, the transfer robotic arm will fail to grasp the batteries, potentially causing the pouch lithium batteries to fall or be damaged. More seriously, when the transfer robotic arm fails to grasp the batteries and stops urgently, due to system response delays, the gripping robotic arm or other downstream robotic arms may not be able to avoid the collision in time, resulting in accidental collisions, causing equipment hardware damage and prolonged production line downtime. Summary of the Invention

[0004] This application provides a multi-station collaborative intelligent control method and system for independently gripping soft-pack lithium batteries, aiming to solve the problem in the prior art that soft-pack lithium batteries are prone to slight displacement or rotation during high-speed transfer, leading to gripping failure, workpiece damage, or even accidental collisions between robotic arms, causing production line downtime and economic losses.

[0005] On the one hand, this application provides a multi-station collaborative intelligent control method for independently grasping soft-pack lithium batteries, including:

[0006] Detect abnormal information generated when the transfer robotic arm picks up a soft-pack lithium battery at a preset handover point;

[0007] In response to the abnormal information, a local abnormality handling scheme is selected and activated from a pre-stored local abnormality handling scheme library. The local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm.

[0008] According to the local anomaly handling scheme, control the transfer robotic arm to perform a grasping adjustment action or a safe retreat action, and control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path.

[0009] After the transfer robotic arm completes the abnormality handling, the gripping robotic arm and the transfer robotic arm are coordinated to resume normal production operations.

[0010] Optionally, the step of responding to the exception information by selecting and activating a local exception handling scheme from a pre-stored local exception handling scheme library includes:

[0011] In response to the abnormal information, when the transfer robotic arm attempts to grasp the soft-pack lithium battery, a multi-axis force signal is collected during the contact process between the gripper of the transfer robotic arm and the soft-pack lithium battery; and, within a preset time before and after the transfer robotic arm attempts to grasp the soft-pack lithium battery, an image sequence of the handover area is collected.

[0012] The multi-axis force signal is analyzed in the time and frequency domains to identify the fluctuation pattern or periodic change of the multi-axis force signal and obtain the force signal analysis results; and the image sequence is tracked for feature points and reconstructed for motion trajectory to estimate the instantaneous motion vector of the soft-pack lithium battery and obtain the micro-motion analysis results of the image sequence.

[0013] Based on the force signal analysis results and the image sequence micro-motion analysis results, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state.

[0014] When it is determined that the static offset state is in the state, the static offset re-grabbing scheme is selected and activated, and the transfer robot arm is controlled to adjust its position and attitude and then grab the soft-pack lithium battery again. When it is determined that the dynamic unstable state is in the state, the flexible contact dynamic stabilization scheme or the tactile guidance adaptive grasping scheme is selected and activated, and the gripper of the transfer robot arm is controlled to contact the surface of the soft-pack lithium battery with controlled pressure to perform stable operation or adjust the grasping parameters according to real-time tactile feedback.

[0015] Based on the judgment result, the grasping robotic arm is controlled to switch to an avoidance path or a temporary hovering path until the transfer robotic arm completes the abnormality handling.

[0016] Optionally, the step of determining whether the pouch lithium battery is in a static offset state or a dynamic unstable state based on the force signal analysis results and the image sequence micro-motion analysis results includes:

[0017] The confidence level of the force signal analysis results is evaluated based on the signal-to-noise ratio, clarity of the fluctuation pattern, and stability of the periodic changes of the multi-axis force signal.

[0018] The confidence level of the micro-motion analysis results of the image sequence is evaluated based on the stability of feature point tracking, the smoothness of motion trajectory reconstruction, and the amplitude of instantaneous motion vectors; wherein, the image occlusion interference caused by the movement of the grasping robotic arm is excluded in the stability evaluation of feature point tracking.

[0019] The weights of the force signal analysis results and the image sequence micro-motion analysis results are adjusted based on the confidence levels of the force signal analysis results and the image sequence micro-motion analysis results.

[0020] Based on the adjusted weights and the force signal analysis results, as well as the micro-motion analysis results of the image sequence, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state.

[0021] Optionally, after the step of determining whether the pouch lithium battery is in a static offset state or a dynamically unstable state based on the adjusted weights, the force signal analysis results, and the image sequence micro-motion analysis results, the following is included:

[0022] When the confidence levels of the force signal analysis results and the micro-motion analysis results of the image sequence are both lower than the preset confidence threshold, the transfer robotic arm is controlled to approach the soft-pack lithium battery again at a preset low speed and with a gentle contact force. During this process, the multi-axis force signal and the image sequence are collected again for secondary discrimination.

[0023] Simultaneously, the gripping robotic arm is controlled to maintain a temporary hovering path until the secondary judgment is completed.

[0024] Optionally, the step of controlling the transfer robotic arm to adjust its position and attitude before grabbing the pouch lithium battery again includes:

[0025] The multi-axis force signal is compared with the preset ideal contact force distribution to identify the force distribution deviation caused by the gripper contact surface of the transfer robot or the elastic characteristics of the soft-pack lithium battery.

[0026] Based on the force distribution deviation, the approach trajectory and gripping posture of the gripper of the transfer robotic arm are adjusted to gradually increase the gripping force and complete the grasping of the soft-pack lithium battery, so as to achieve smooth contact and uniform force with the surface of the soft-pack lithium battery.

[0027] Optionally, the step of selecting and activating a flexible contact dynamic stabilization scheme or a tactile-guided adaptive gripping scheme when the dynamic unstable state is determined, and controlling the gripper of the transfer robotic arm to contact the surface of the soft-pack lithium battery with controlled pressure for stable operation or adjusting the gripping parameters according to real-time tactile feedback includes:

[0028] Based on the force signal analysis results and the image sequence micro-motion analysis results, the composite dynamic instability mode of the soft-pack lithium battery is identified;

[0029] Based on the composite dynamic instability mode, a matching scheme is selected from the preset flexible contact dynamic stabilization scheme library or the haptic-guided adaptive grasping scheme library;

[0030] According to the matching scheme, the contact point, contact area, contact pressure, gripping speed and gripping posture of the gripper of the transfer robot arm are adjusted, and the gripper of the transfer robot arm is controlled to perform stable operation or adaptive gripping with the adjusted parameters.

[0031] Simultaneously, collaborative instructions for the gripping robotic arm are generated, including controlling the gripping robotic arm to reduce the safe distance threshold of the avoidance path, and / or extending the duration of the temporary hovering path.

[0032] Optionally, the step of identifying the combined dynamic instability mode of the pouch lithium battery based on the force signal analysis results and the image sequence micro-motion analysis results includes:

[0033] By comparing the timestamps of the force signal analysis results and the image sequence micro-motion analysis results, the time deviation between the force signal analysis results and the image sequence micro-motion analysis results is identified;

[0034] Based on the time deviation, the analysis results with earlier timestamps are subjected to time compensation processing based on historical data or predicted trends to ensure the synchronization of the force signal analysis results and the image sequence micro-motion analysis results in the time dimension.

[0035] Based on the force signal analysis results after time compensation processing and the micro-motion analysis results of the image sequence, when there is a continuous time delay or data loss in the force signal analysis results or the micro-motion analysis results of the image sequence, predictive identification of the pattern is performed according to the current state and historical trend of the other analysis result, and the predicted composite dynamic instability pattern is output.

[0036] Optionally, the step of predictively identifying patterns and outputting predicted composite dynamic instability patterns based on the current state and historical trends of the other analysis result when there is a continuous time delay or data gap in the force signal analysis result or the image sequence micro-motion analysis result further includes:

[0037] Once the data with time delay or missing data is recovered, the predicted composite dynamic instability pattern is compared with the recovered actual composite dynamic instability pattern, and the deviation between the two is calculated.

[0038] Based on the deviation value, adjust the historical trend weights or current state influence factors used in the predictive identification process;

[0039] Based on the adjusted historical trend weights or current state influence factors, predictive pattern identification is performed, and the predicted composite dynamic unstable pattern is output.

[0040] Optionally, the composite dynamic instability mode includes high-frequency vibration, low-frequency swaying, and minute rotation.

[0041] On the other hand, this application provides a multi-station collaborative independent gripping intelligent control system for soft-pack lithium batteries, the system comprising:

[0042] The abnormal information acquisition module is used to detect abnormal information generated when the transfer robotic arm grabs a soft-pack lithium battery at a preset handover point;

[0043] The scheme activation module is used to respond to the abnormal information, select and activate a local abnormality handling scheme from a pre-stored local abnormality handling scheme library, and the local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm.

[0044] The robotic arm control module is used to control the transfer robotic arm to perform grasping and adjustment actions or safe retreat actions according to the local anomaly handling scheme, and to control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path.

[0045] The recovery coordination module is used to coordinate the gripping robot arm and the transfer robot arm to resume normal production operations after the transfer robot arm completes the abnormal handling.

[0046] This application relates to a multi-station collaborative intelligent control method and system for independently gripping pouch lithium batteries, which effectively solves problems in the prior art such as gripping failure, workpiece damage, and robotic arm collisions caused by minute displacements or rotations during high-speed transport of pouch lithium batteries. Through intelligent anomaly detection and collaborative processing mechanisms, this application can significantly improve the gripping success rate and production efficiency of automated pouch lithium battery production lines, reduce equipment damage risks and downtime, thereby bringing significant economic benefits and improved production stability. Attached Figure Description

[0047] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0048] Figure 1 The diagram above illustrates a flowchart of an intelligent control method for independently grasping multi-station collaborative soft-pack lithium batteries.

[0049] Figure 2 The diagram above illustrates a schematic of a multi-station collaborative intelligent control system for independently grasping soft-pack lithium batteries.

[0050] Figure reference numerals: 100, Multi-station collaborative independent gripping intelligent control system for soft-pack lithium batteries; 10, Abnormal information acquisition module; 20, Scheme activation module; 30, Robotic arm control module; 40, Recovery and coordination module. Detailed Implementation

[0051] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0052] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0053] In modern industrial production, automation technology has been widely applied to high-precision, high-efficiency manufacturing processes, especially in the production of pouch lithium batteries. To meet market demands, production lines often employ multiple robotic arms working collaboratively to achieve rapid battery gripping, transfer, and assembly. This collaborative operation places extremely high demands on the positioning accuracy and motion coordination of the robotic arms; any minute deviation can affect production efficiency and product quality. However, in actual operation, due to long-term equipment wear, subtle changes in workpiece characteristics, and the inertial effect of high-speed motion, the actual position and posture of the workpiece may deviate from the preset target during gripping and transfer. These deviations are not obvious initially but accumulate as production progresses, eventually leading to gripping failures, workpiece damage, or even accidental collisions between robotic arms, causing production line downtime and economic losses.

[0054] like Figure 1 The diagram illustrates an exemplary flow chart of a multi-station collaborative intelligent control method for independently grasping pouch lithium batteries. This application proposes a multi-station collaborative intelligent control method for independently grasping pouch lithium batteries, comprising:

[0055] S10, detect abnormal information generated when the transfer robotic arm grabs a soft-pack lithium battery at a preset handover point;

[0056] Among them, pouch lithium batteries refer to lithium-ion batteries packaged in flexible packaging materials. They are characterized by high shape flexibility and high energy density, but this also places higher demands on the mechanical control during the gripping and transfer process to avoid damage to the battery itself. Transfer robotic arms and gripping robotic arms are two types of industrial robotic arms with different functions. The gripping robotic arm is typically responsible for picking up pouch lithium batteries from upstream workstations and transferring them at high speed to a pre-designated handover point, while the transfer robotic arm receives the pouch lithium batteries at the handover point and transfers them to downstream workstations. The coordinated operation of these two types of robotic arms is key to achieving efficient production. The pre-designated handover point refers to the specific spatial location on the automated production line where the gripping robotic arm and the transfer robotic arm exchange pouch lithium batteries.

[0057] S20, in response to the abnormal information, select and activate a local abnormality handling scheme from a pre-stored local abnormality handling scheme library, wherein the local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm.

[0058] S30, according to the local anomaly handling scheme, control the transfer robotic arm to perform a grasping adjustment action or a safe retreat action, and control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path.

[0059] S40, after the transfer robotic arm completes the abnormality handling, coordinate the gripping robotic arm and the transfer robotic arm to resume normal production operations.

[0060] Specifically, this method first involves "detecting abnormal information generated when the transfer robot arm grasps the pouch lithium battery at a preset handover point." Detecting abnormal information is the foundation of the entire intelligent control method. For example, force sensors can be installed on the grippers of the transfer robot arm to monitor the contact force of the grippers when grasping the pouch lithium battery in real time. When the contact force exceeds a preset threshold, or the force distribution is uneven, it can be identified as abnormal information. Another approach is to deploy vision sensors, such as high-speed cameras, near the preset handover point to continuously capture images or video sequences of the pouch lithium battery at the handover point. Through image processing algorithms, it is possible to analyze whether the position and orientation of the pouch lithium battery deviate from the preset target, or whether there is any unstable state such as shaking or rotation. When a significant deviation or unstable state is detected, abnormal information is generated. Furthermore, by monitoring the kinematic parameters of the transfer robot arm, such as joint angles, speed, and acceleration, unexpected fluctuations in these parameters during the grasping process can also be identified as abnormal information.

[0061] Upon detecting an anomaly, the system "responds to the anomaly by selecting and activating a local anomaly handling scheme from a pre-stored local anomaly handling scheme library." This library contains various strategies for handling different anomalies. For example, a slight displacement of the pouch lithium battery might trigger a "static displacement re-grasp scheme"; a swaying motion of the pouch lithium battery during high-speed movement might trigger a "flexible contact dynamic stabilization scheme." Scheme selection and activation can be achieved through preset decision logic, such as matching the type and severity of the anomaly with the real-time production status. Each local anomaly handling scheme includes coordinated action instructions for the transfer robot and the gripping robot, ensuring that the two robots work collaboratively during anomaly handling to avoid mutual interference. For example, when the transfer robot needs to adjust its gripping position, the gripping robot may need to temporarily avoid or hover.

[0062] Subsequently, "according to the local anomaly handling scheme, the transfer robotic arm will be controlled to perform a gripping adjustment action or a safe retreat action, and the gripping robotic arm will be controlled to switch its preset motion path to an avoidance path or a temporary hovering path." Once a specific local anomaly handling scheme is selected and activated, the transfer robotic arm and the gripping robotic arm will be precisely controlled according to the preset instructions in that scheme. For example, if the anomaly information indicates a slight positional or orientation deviation of the pouch lithium battery, the transfer robotic arm may perform a "gripping adjustment action," such as fine-tuning the approach trajectory, gripping angle, or gripping force of the gripper to re-attempt the gripping. If the anomaly is more severe, such as the pouch lithium battery about to fall or a significant positional deviation, the transfer robotic arm may perform a "safe retreat action," that is, quickly withdrawing the gripper to avoid damaging the pouch lithium battery or colliding with the gripping robotic arm. Simultaneously, the gripping robotic arm will also switch its preset motion path according to the coordinated action instructions. For example, when the transfer robotic arm performs a gripping adjustment, the gripping robotic arm may switch to an "avoidance path," that is, temporarily moving away from the handover area to provide sufficient working space for the transfer robotic arm. When the transfer robotic arm performs a safe retreat maneuver, the grasping robotic arm may switch to a "temporary hovering path," which means pausing its movement at a safe distance and waiting for the transfer robotic arm to complete the abnormality handling.

[0063] Finally, after the transfer robotic arm completes the anomaly handling, it will "coordinate the gripping robotic arm and the transfer robotic arm to resume normal production operations." Once the transfer robotic arm successfully handles the anomaly, such as successfully gripping a pouch lithium battery or safely retracting, it will issue a command to coordinate the two robotic arms to return to the normal production process. This may include the gripping robotic arm returning from the avoidance path or temporary hovering path to its preset normal movement path and continuing to transfer the next pouch lithium battery to the handover point; simultaneously, the transfer robotic arm will also resume its normal gripping and transfer operations. This recovery mechanism ensures the continuity and efficiency of the production line, minimizing the impact of anomalies on the overall production schedule.

[0064] The multi-station collaborative independent gripping intelligent control method for soft-pack lithium batteries proposed in this application effectively solves the problems of gripping failure, equipment collision, and production line downtime caused by workpiece deviation in traditional multi-station collaborative operations through the aforementioned series of collaborative control steps. Traditional methods often rely on preset fixed paths and simple collision detection. Once there is a slight deviation in the position or posture of the workpiece, it is easy to cause gripping failure or even collision between robotic arms. For example, in the prior art, when the transfer robotic arm fails to grip and stops urgently, due to the response delay, the gripping robotic arm or other downstream robotic arms may not be able to avoid the collision in time, resulting in accidental collisions, causing equipment hardware damage and long-term production line downtime.

[0065] In contrast, the core innovation of this application lies in the introduction of a real-time detection and intelligent response mechanism for abnormal information, as well as dynamic collaborative control between the transfer robotic arm and the gripping robotic arm. By detecting abnormal information generated when the transfer robotic arm grips pouch lithium batteries at a preset handover point in real time, potential problems can be identified promptly. More importantly, this application can intelligently select and activate the most suitable handling scheme from a pre-stored library of local abnormality handling schemes based on the type and severity of the abnormal information. These schemes not only include gripping adjustment or safe retreat actions of the transfer robotic arm, but also synchronously coordinate the gripping robotic arm to switch to an avoidance path or a temporary hovering path, thereby avoiding potential collisions between robotic arms and ensuring equipment safety. Furthermore, after the abnormality is handled, the two robotic arms can be coordinated to quickly resume normal production operations, minimizing production interruption time. This intelligent abnormality handling and collaborative recovery mechanism significantly improves the stability and efficiency of the pouch lithium battery production line, reduces economic losses caused by abnormal situations, and demonstrates the significant progress of this application in the field of automation control.

[0066] In some embodiments, the step of selecting and activating a local exception handling scheme from a pre-stored local exception handling scheme library in response to the exception information includes:

[0067] In response to the abnormal information, when the transfer robotic arm attempts to grasp the soft-pack lithium battery, a multi-axis force signal is collected during the contact process between the gripper of the transfer robotic arm and the soft-pack lithium battery; and, within a preset time before and after the transfer robotic arm attempts to grasp the soft-pack lithium battery, an image sequence of the handover area is collected.

[0068] The multi-axis force signal is analyzed in the time and frequency domains to identify the fluctuation pattern or periodic change of the multi-axis force signal and obtain the force signal analysis results; and the image sequence is tracked for feature points and reconstructed for motion trajectory to estimate the instantaneous motion vector of the soft-pack lithium battery and obtain the micro-motion analysis results of the image sequence.

[0069] Based on the force signal analysis results and the image sequence micro-motion analysis results, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state.

[0070] When it is determined that the static offset state is in the state, the static offset re-grabbing scheme is selected and activated, and the transfer robot arm is controlled to adjust its position and attitude and then grab the soft-pack lithium battery again. When it is determined that the dynamic unstable state is in the state, the flexible contact dynamic stabilization scheme or the tactile guidance adaptive grasping scheme is selected and activated, and the gripper of the transfer robot arm is controlled to contact the surface of the soft-pack lithium battery with controlled pressure to perform stable operation or adjust the grasping parameters according to real-time tactile feedback.

[0071] Based on the judgment result, the grasping robotic arm is controlled to switch to an avoidance path or a temporary hovering path until the transfer robotic arm completes the abnormality handling.

[0072] Specifically, when the robotic arm attempts to grasp a pouch-type lithium battery, multi-axis force sensors mounted on the grippers of the robotic arm collect multi-axis force signals in real time during the contact process between the grippers and the pouch-type lithium battery. These multi-axis force signals reflect the magnitude, direction, and distribution of the contact force. Simultaneously, within a preset time period before and after the robotic arm attempts to grasp the pouch-type lithium battery, vision sensors positioned in the junction area collect image sequences of that area. These image sequences are used to capture minute movements or positional changes of the pouch-type lithium battery near the junction point.

[0073] The analysis of multi-axis force signals, including time-domain and frequency-domain analyses, aims to deeply explore the intrinsic characteristics of these signals. Time-domain analysis can identify instantaneous changes, peak values, and durations of the force signal, while frequency-domain analysis can reveal fluctuation patterns, periodic variations, or abnormal vibration frequencies, thus yielding the force signal analysis results. For example, continuous low-frequency fluctuations may indicate shaking of the pouch lithium battery, while high-frequency vibrations may indicate friction or impact between the gripper and the pouch lithium battery.

[0074] Furthermore, using computer vision algorithms, key feature points on the surface of the pouch lithium battery are identified and tracked in consecutive image frames. Based on the movement of these feature points, the motion trajectory of the pouch lithium battery in three-dimensional space is reconstructed, enabling feature point tracking and motion trajectory reconstruction of image sequences. By analyzing the motion trajectory, the instantaneous motion vector of the pouch lithium battery, i.e., its velocity and direction at a certain moment, can be accurately estimated, thus obtaining the micro-motion analysis results of the image sequence.

[0075] Furthermore, based on the force signal analysis results and the micro-motion analysis results of the image sequence, it is possible to determine whether the pouch lithium battery is currently in a static offset state or a dynamic unstable state. A static offset state typically refers to a slight but relatively stable deviation in the position or orientation of the pouch lithium battery at the junction point, such as a position deviating from the preset gripping point or an angle tilt. A dynamic unstable state refers to the pouch lithium battery being in continuous motion, shaking, or vibration, for example, due to airflow, mechanical vibration, or residual momentum from previous processes preventing it from remaining stable.

[0076] When a static offset state is detected, the static offset re-grip scheme is selected and activated. This scheme aims to precisely adjust the position and attitude of the transfer robot arm so that its grippers can accurately align and re-grip the pouch lithium battery. When a dynamic instability state is detected, either the flexible contact dynamic stabilization scheme or the tactile-guided adaptive gripping scheme is selected and activated. The flexible contact dynamic stabilization scheme typically involves the transfer robot arm's grippers contacting the pouch lithium battery surface with controlled, gentle pressure, gradually stabilizing it from a dynamically unstable state through buffering and stabilization. The tactile-guided adaptive gripping scheme goes a step further, dynamically adjusting gripping parameters, such as clamping force and contact point, based on real-time tactile feedback to adapt to the dynamic changes of the pouch lithium battery and achieve stable gripping.

[0077] During this process, based on the judgment result, the gripping robotic arm will be controlled to switch to an avoidance path or a temporary hovering path. An avoidance path means the gripping robotic arm temporarily deviates from its preset normal movement path to avoid collisions with the transfer robotic arm or the pouch lithium battery. A temporary hovering path means the gripping robotic arm pauses its movement at a safe position, waiting for the transfer robotic arm to complete the abnormality handling.

[0078] This application's solution significantly improves the accuracy and response capability to abnormal states of pouch lithium batteries by introducing the acquisition and analysis of multi-axis force signals and image sequences. Specifically, when the transfer robotic arm attempts to grasp the pouch lithium battery, the multi-axis force signals provide detailed information on the contact force between the gripper and the pouch lithium battery, such as whether there is uneven force, impact, or vibration. Simultaneously, the analysis of image sequences can visually capture the actual position, orientation, and minute movements of the pouch lithium battery, overcoming the limitations of single force or visual information. By performing time-domain and frequency-domain analysis, feature point tracking, and motion trajectory reconstruction on these two heterogeneous data sets, it is possible to comprehensively determine whether the pouch lithium battery is in a static offset state (relatively stationary but with positional deviation) or a dynamic unstable state (continuous movement or vibration). This refined state discrimination is key to overcoming the aforementioned limitations because it allows for the selection and activation of more targeted processing schemes from a pre-set library of local anomaly handling schemes. For example, for static offset, a re-grasping scheme using position and orientation adjustment is more efficient; while for dynamic instability, a flexible contact or tactile-guided adaptive grasping scheme is needed to ensure stability. In addition, the synchronous avoidance or hovering of the gripping robotic arm ensures that secondary collisions or interference are avoided during the handling of anomalies by the transfer robotic arm, thereby ensuring the smooth progress of anomaly handling.

[0079] Through the above technical solution, this application can achieve accurate identification and intelligent response to abnormal states during the handling of pouch lithium batteries. Compared with basic solutions that rely solely on general anomaly information, this application, by combining multi-axis force signals and image sequence depth analysis, can accurately distinguish between static offset and dynamic instability states of pouch lithium batteries, thereby avoiding blind or inefficient anomaly handling. This refined discrimination capability enables the activation of more suitable and efficient local anomaly handling solutions, such as a re-handling solution for static offset and a flexible contact or tactile-guided adaptive handling solution for dynamic instability. This not only significantly improves the success rate and efficiency of anomaly handling, reducing production interruptions and material losses caused by misjudgment or improper handling, but also ensures stable and efficient handling of pouch lithium batteries under complex working conditions through the synchronous and coordinated movements of the gripping robotic arm.

[0080] For example, suppose that during the process of transferring the pouch lithium battery from the upstream station to the gripping station, the transfer robotic arm detects an anomaly when it attempts to grip the pouch lithium battery at a preset handover point.

[0081] Specifically, when the gripper of the transfer robotic arm approaches and slightly contacts the pouch lithium battery, a multi-axis force sensor mounted on the gripper begins to collect multi-axis force signals. Simultaneously, a high-speed camera begins to acquire image sequences of the transfer area.

[0082] If the force signal analysis shows uneven contact force distribution, and the image sequence micro-motion analysis shows a small, continuous displacement of the pouch lithium battery in the horizontal direction without obvious shaking or vibration, then the pouch lithium battery is determined to be in a static offset state. In this case, the static offset re-grabbing scheme is activated. The transfer robotic arm is controlled to fine-tune the position and attitude of its grippers based on the force signal and image analysis results, for example, translating 2 mm to the left and rotating 1 degree, and then attempting to grab the pouch lithium battery again. During this period, the gripping robotic arm switches to a temporary hovering path, waiting in a safe area for the transfer robotic arm to complete the re-grabbing operation.

[0083] On the other hand, if force signal analysis shows periodic high-frequency vibrations, and image sequence micro-motion analysis shows continuous low-frequency shaking of the pouch lithium battery, then the pouch lithium battery is determined to be in a dynamically unstable state. In this case, a flexible contact dynamic stabilization scheme is activated. The gripper of the transfer robot arm contacts the surface of the pouch lithium battery with a preset, gentle pressure. The flexible material of the gripper and the controlled contact force absorb vibrations and shaking, gradually stabilizing the battery. Alternatively, a tactile-guided adaptive grasping scheme is activated. Based on real-time force signals, the gripping force is dynamically adjusted; for example, a smaller force is used for initial contact, and the gripping force is gradually increased after the pouch lithium battery stabilizes to complete the grasp. During this process, the grasping robot arm switches to an avoidance path to ensure a sufficient safe distance from the transfer robot arm until the transfer robot arm successfully stabilizes and grasps the pouch lithium battery.

[0084] In some embodiments, the step of determining whether the pouch lithium battery is in a static offset state or a dynamic unstable state based on the force signal analysis results and the image sequence micro-motion analysis results includes:

[0085] The confidence level of the force signal analysis results is evaluated based on the signal-to-noise ratio, clarity of the fluctuation pattern, and stability of the periodic changes of the multi-axis force signal.

[0086] The confidence level of the micro-motion analysis results of the image sequence is evaluated based on the stability of feature point tracking, the smoothness of motion trajectory reconstruction, and the amplitude of instantaneous motion vectors; wherein, the image occlusion interference caused by the movement of the grasping robotic arm is excluded in the stability evaluation of feature point tracking.

[0087] The weights of the force signal analysis results and the image sequence micro-motion analysis results are adjusted based on the confidence levels of the force signal analysis results and the image sequence micro-motion analysis results.

[0088] Based on the adjusted weights and the force signal analysis results, as well as the micro-motion analysis results of the image sequence, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state.

[0089] Specifically, the reliability of multi-axis force signals is determined by analyzing their quality characteristics, thus establishing the confidence level of the force signal analysis results. For example, a higher signal-to-noise ratio indicates a purer signal, less susceptibility to noise interference, and higher confidence; higher clarity of the fluctuation pattern and better stability of periodic changes also mean more reliable force signal analysis results. These indicators quantify the quality of force signal data, providing a basis for subsequent judgments.

[0090] The reliability of image processing is assessed by analyzing key indicators to evaluate the confidence level of micro-motion analysis results for image sequences. For example, higher stability in feature point tracking indicates a higher degree of accurate identification and tracking of image feature points in consecutive frames; higher smoothness in motion trajectory reconstruction means that the estimated motion trajectory better conforms to actual physical motion laws and is less affected by random errors; the amplitude of the instantaneous motion vector reflects the intensity of the motion, and excessively large or small amplitudes may indicate data anomalies. In particular, in the stability evaluation of feature point tracking, it is necessary to eliminate image occlusion interference caused by the movement of the gripping robotic arm to avoid misinterpreting the movement of the robotic arm itself as the movement of the pouch lithium battery, thereby improving the accuracy of the evaluation.

[0091] In practical applications, adjusting the weights of force signal analysis results and image sequence micro-motion analysis results dynamically allocates their influence in the final judgment based on their respective confidence levels. For example, when the confidence level of the force signal analysis results is high while the confidence level of the image sequence micro-motion analysis results is low, the weight of the force signal analysis results can be appropriately increased while the weight of the image sequence micro-motion analysis results is decreased, making the judgment result more inclined towards the high-confidence data source. The reverse is also true. This dynamic weight adjustment mechanism ensures that, under different data quality conditions, multi-source information can be more intelligently integrated to make more accurate judgments.

[0092] Through the above technical solution, this application effectively solves the problem of decreased discrimination accuracy caused by fluctuations in sensor data quality in traditional methods. By evaluating the confidence level of multi-axis force signals and image sequence micro-motion analysis results, and dynamically adjusting their weights based on the evaluation results, it enables more intelligent fusion of multi-source information when determining whether a pouch lithium battery is in a static offset state or a dynamically unstable state. This not only improves the accuracy and reliability of the discrimination results and reduces the false judgment rate, but also enhances the adaptability and robustness in complex and variable production environments. This provides a solid foundation for the accurate selection and efficient execution of subsequent anomaly handling solutions, further improving the overall performance of the multi-station collaborative intelligent control method for independent grasping of pouch lithium batteries.

[0093] For example, suppose that when the transfer robotic arm attempts to grasp a pouch lithium battery, it acquires multi-axial force signals and image sequences of the junction area.

[0094] First, the multi-axis force signal is analyzed to obtain the force signal analysis results. Simultaneously, the signal-to-noise ratio of the force signal is calculated to be 15 dB, the wave pattern clarity to be 0.8, and the periodicity stability to be 0.9. Based on these indicators, the confidence level of the force signal analysis results is assessed as 0.85.

[0095] Next, feature point tracking and motion trajectory reconstruction were performed on the image sequence to obtain the micro-motion analysis results. During this process, a brief occlusion of the gripping robotic arm in the image was detected. However, by using an algorithm to eliminate occlusion interference, the stability of feature point tracking was still evaluated as 0.7, the smoothness of motion trajectory reconstruction as 0.8, and the amplitude of the instantaneous motion vector as 0.05 m / s. Based on these indicators, the confidence level of the obtained micro-motion analysis results of the image sequence was evaluated as 0.75.

[0096] Because the confidence level of the force signal analysis result (0.85) is higher than that of the image sequence micro-motion analysis result (0.75), the weights of the two will be dynamically adjusted. For example, the weight of the force signal analysis result can be adjusted to 0.6, and the weight of the image sequence micro-motion analysis result can be adjusted to 0.4.

[0097] Finally, a comprehensive judgment is made based on the adjusted weights, the original force signal analysis results, and the micro-motion analysis results of the image sequence. For example, if the force signal analysis results strongly indicate static offset, while the micro-motion analysis results of the image sequence slightly indicate dynamic instability, the final judgment will lean more towards the static offset state because the force signal analysis results have a higher weight. This allows for a more accurate assessment of the actual state of the pouch lithium battery and the selection of appropriate anomaly handling schemes.

[0098] In some embodiments, after the step of determining whether the pouch lithium battery is in a static offset state or a dynamically unstable state based on the adjusted weights, the force signal analysis results, and the image sequence micro-motion analysis results, the following is included:

[0099] When the confidence levels of the force signal analysis results and the micro-motion analysis results of the image sequence are both lower than the preset confidence threshold, the transfer robotic arm is controlled to approach the soft-pack lithium battery again at a preset low speed and with a gentle contact force. During this process, the multi-axis force signal and the image sequence are collected again for secondary discrimination.

[0100] Simultaneously, the gripping robotic arm is controlled to maintain a temporary hovering path until the secondary judgment is completed.

[0101] Specifically, the preset confidence threshold refers to a pre-defined judgment standard used to measure the reliability of force signal analysis results and image sequence micro-motion analysis results. When the confidence of either analysis result is lower than this threshold, it indicates that the analysis result may have high uncertainty or noise interference. When both are lower than this threshold, it is considered that the current environment or data quality is insufficient to support reliable discrimination. The preset low speed and gentle contact force refer to a safe operating mode adopted by the transfer robotic arm when it approaches the pouch lithium battery again. Low speed aims to reduce potential collision risks and provide a more stable environment for sensor data acquisition; gentle contact force ensures that it will not damage the pouch lithium battery when contacting it, while also being able to more sensitively capture minute mechanical changes. In practical applications, secondary discrimination refers to re-acquiring data and re-analyzing and discriminating when the first discrimination fails to reach a reliable conclusion due to insufficient confidence, in order to obtain more accurate pouch lithium battery state information. This process aims to improve the accuracy of discrimination by increasing the amount of data and optimizing the acquisition conditions. In addition, controlling the gripping robotic arm to maintain a temporary hovering path until the secondary judgment is completed aims to ensure that during the secondary judgment of the transfer robotic arm, the gripping robotic arm will not perform actions that may interfere with the operation of the transfer robotic arm or cause safety hazards due to misjudgment or uncertainty, thereby ensuring the safety of the entire collaborative operation process.

[0102] Through the above technical solution, this application effectively addresses the challenge posed by low confidence levels in initial sensor data analysis under complex or uncertain environments. By introducing an intelligent secondary discrimination mechanism, when the reliability of the initial data is confirmed to be insufficient, a more cautious strategy can be proactively adopted, namely, re-approaching and collecting data with low speed and gentle contact force, thereby obtaining more accurate pouch lithium battery status information. This significantly improves the accuracy and robustness of abnormal state discrimination, avoiding misjudgments and improper operations caused by insufficient information. Furthermore, by synchronously controlling the gripping robotic arm to maintain a temporary hovering path, the safety of the entire collaborative gripping process is further enhanced, ensuring that other robotic arms do not cause interference or safety hazards when the transfer robotic arm is performing uncertain operations. Therefore, the solution of this application greatly improves the reliability and safety of the pouch lithium battery gripping process while ensuring production efficiency.

[0103] For example, suppose that during the transfer of pouch lithium batteries, the robotic arm detects an anomaly when it grasps the battery at a preset handover point. Subsequently, multi-axis force signals and image sequences of the handover area are acquired using the method described above, and preliminary analysis is performed. However, due to sudden changes in ambient light causing significant noise in the image sequences, or reflections on the pouch lithium battery surface leading to unstable feature point tracking, and the multi-axis force sensor experiencing minor external impacts at the moment of contact, the signal-to-noise ratio and fluctuation pattern clarity of the force signal analysis results are both low. After evaluation, the confidence level of the force signal analysis results is 0.4, and the confidence level of the image sequence micro-motion analysis results is 0.3, both lower than the preset confidence threshold of 0.5.

[0104] At this point, according to the scheme of this application, the judgment and scheme activation will not be immediately based on these low-confidence results, but a secondary judgment process will be triggered. Specifically, the transfer robotic arm will be instructed to slowly approach the pouch lithium battery again at a low speed of, for example, 5 mm / s, and the contact force of its gripper will be limited to a gentle contact force range of, for example, 0.5 N. During the process of the transfer robotic arm approaching and gently contacting the pouch lithium battery again, multi-axis force signals and image sequences will be acquired in real time again. For example, during the second acquisition, the camera exposure parameters may be adjusted to adapt to changes in light, or a more advanced filtering algorithm may be used to process the force signal. Through the re-acquired data, time-domain and frequency-domain analysis, feature point tracking, and motion trajectory reconstruction will be performed again to obtain new force signal analysis results and image sequence micro-motion analysis results. Assuming that the confidence levels of the new analysis results are 0.8 and 0.7, respectively, both higher than the preset threshold, it is possible to accurately determine whether the pouch lithium battery is in a static offset state or a dynamically unstable state based on these high-confidence results, and activate the corresponding local anomaly handling scheme. Throughout this secondary judgment process, the gripping robotic arm remains on a temporary hovering path, for example, stationary at a distance of 100mm above the handover point, until the transfer robotic arm completes the secondary judgment and begins executing the determined anomaly handling plan. In this way, even with poor initial data quality, an intelligent retry mechanism can ultimately achieve accurate judgment of the status of the pouch lithium battery and safe and efficient anomaly handling.

[0105] In some embodiments, the step of controlling the transfer robotic arm to adjust its position and orientation before grabbing the pouch lithium battery again includes:

[0106] The multi-axis force signal is compared with the preset ideal contact force distribution to identify the force distribution deviation caused by the gripper contact surface of the transfer robot or the elastic characteristics of the soft-pack lithium battery.

[0107] Based on the force distribution deviation, the approach trajectory and gripping posture of the gripper of the transfer robotic arm are adjusted to gradually increase the gripping force and complete the grasping of the soft-pack lithium battery, so as to achieve smooth contact and uniform force with the surface of the soft-pack lithium battery.

[0108] Specifically, the multi-axis force signal refers to the multi-dimensional mechanical data collected by force sensors mounted on the gripper during the contact process between the gripper of the transfer robotic arm and the pouch lithium battery. This includes forces and torques in the X, Y, and Z axes. The preset ideal contact force distribution refers to the pre-set mechanical distribution pattern between the gripper and the battery surface during the gripping process, based on the material characteristics, structural dimensions, and safe gripping requirements of the pouch lithium battery. This ensures uniform gripping force within a safe range. The force distribution deviation refers to the difference between the force distribution reflected by the actual collected multi-axis force signal and the ideal contact force distribution. This deviation can indicate problems such as unevenness of the gripper contact surface, local deformation of the pouch lithium battery, or uneven elasticity. Adjusting the approach trajectory and gripping posture of the gripper of the transfer robotic arm according to the force distribution deviation means dynamically correcting the movement path of the transfer robotic arm and the posture parameters such as the tilt angle and rotation angle of the gripper based on the identified force distribution deviation, so that the gripper can approach and contact the pouch lithium battery in a manner more consistent with the ideal contact force distribution. Gradually increasing the clamping force means that during the gripping process, the maximum clamping force is not applied all at once, but rather increased slowly in stages until the minimum force required for stable gripping is reached. The purpose is to avoid damage to the pouch lithium battery from sudden impacts and to allow for fine-tuning during the force increase. Through these adjustments, a smooth contact and uniform force distribution with the pouch lithium battery surface are ultimately achieved, ensuring that there is no violent impact when the grippers contact the battery surface, and that the clamping force is evenly distributed across the contact surface, avoiding localized stress concentration.

[0109] Through the above technical solution, this application can significantly improve the success rate and safety of the transfer robotic arm in handling statically offset pouch lithium batteries. By finely analyzing multi-axis force signals and identifying force distribution deviations, precise perception and control of the gripping process can be achieved, effectively avoiding problems such as surface damage and internal structural destruction of pouch lithium batteries caused by excessive or uneven gripping force. Furthermore, by adjusting the approach trajectory and gripping posture of the grippers and gradually increasing the gripping force, the compliance and mechanical uniformity of the gripping process are ensured. This maximizes the protection of the integrity and performance of the pouch lithium batteries while maintaining high-efficiency production, reducing the scrap rate and improving the overall efficiency of the production line.

[0110] For example, suppose a robotic arm attempts to grasp a pouch lithium battery in a static offset state, with its grippers contacting the battery surface. At this time, multi-axis force sensors mounted on the grippers collect force signals in real time during the contact process. These real-time force signals are compared with a pre-stored ideal contact force distribution model set for that type of pouch lithium battery. For example, if the ideal model requires uniform force on both contact surfaces of the gripper, but the actual detection shows that the pressure on one contact surface is significantly higher than the other, a force distribution deviation will be identified. Specifically, when a force distribution deviation is detected, the adjustment amount required by the robotic arm gripper is calculated based on the magnitude and direction of the deviation. For example, if excessive force is found on the left side of the gripper, the robotic arm may be instructed to fine-tune its approach trajectory, shifting it slightly to the right, while simultaneously adjusting the gripper's posture to make it more parallel to the contact surface of the pouch lithium battery. During the second grasping attempt, the gripping force will no longer be applied instantaneously, but will increase slowly at a preset gradient, for example, starting from 0.5N and increasing by 0.1N every 0.1 seconds until a stable gripping force of 2N is reached. During this process, multi-axis force signals are continuously monitored to ensure that the force distribution remains within the ideal range, and fine adjustments are made based on real-time feedback to ultimately achieve a smooth and uniform gripping of the soft-pack lithium battery.

[0111] In some embodiments, the step of selecting and activating a flexible contact dynamic stabilization scheme or a tactile-guided adaptive gripping scheme when the dynamic unstable state is determined, and controlling the gripper of the transfer robotic arm to contact the surface of the soft-pack lithium battery with controlled pressure for stable operation or adjusting the gripping parameters according to real-time tactile feedback includes:

[0112] Based on the force signal analysis results and the image sequence micro-motion analysis results, the composite dynamic instability mode of the soft-pack lithium battery is identified;

[0113] Based on the composite dynamic instability mode, a matching scheme is selected from the preset flexible contact dynamic stabilization scheme library or the haptic-guided adaptive grasping scheme library;

[0114] According to the matching scheme, the contact point, contact area, contact pressure, gripping speed and gripping posture of the gripper of the transfer robot arm are adjusted, and the gripper of the transfer robot arm is controlled to perform stable operation or adaptive gripping with the adjusted parameters.

[0115] Simultaneously, collaborative instructions for the gripping robotic arm are generated, including controlling the gripping robotic arm to reduce the safe distance threshold of the avoidance path, and / or extending the duration of the temporary hovering path.

[0116] Specifically, identifying the composite dynamic instability modes of pouch lithium batteries refers to accurately determining the specific type and degree of the current dynamic instability state of the pouch lithium battery by comprehensively analyzing the time and frequency domain characteristics of multi-axis force signals, as well as the results of feature point tracking and motion trajectory reconstruction of image sequences. For example, composite dynamic instability modes may include high-frequency vibration, low-frequency swaying, and minute rotation. Among them, high-frequency vibration may manifest as significant fluctuations in the force signal within a high frequency range, while the image sequence shows small but rapid displacements; low-frequency swaying may manifest as slow periodic changes in the force signal and large, slow displacements in the image sequence; minute rotations may be identified by the relative positional changes of feature points in the image sequence.

[0117] Furthermore, based on the identified complex dynamic instability mode, the system selects the most suitable scheme from a pre-set library of flexible contact dynamic stabilization schemes or a library of haptic-guided adaptive grasping schemes. For example, for a high-frequency vibration mode, a flexible contact scheme with high response speed and damping characteristics might be selected; for a low-frequency swaying mode, a scheme that counteracts swaying by slowly adjusting the gripping posture and contact point might be selected; and for minute rotations, a scheme that corrects rotation by adjusting the relative angle of the grippers might be selected.

[0118] Based on this, and according to the selected matching scheme, the gripping parameters of the transfer robotic arm's grippers will be finely adjusted. These parameters include the contact point, contact area, contact pressure, gripping speed, and gripping posture between the grippers and the pouch lithium battery. For example, when dealing with high-frequency vibrations, the contact area may be increased to disperse vibration energy, and a lower contact pressure may be used to avoid damage; when dealing with low-frequency shaking, the contact point may be adjusted to provide better support, and the gripping speed may be slower; when dealing with minute rotations, the gripping posture may be adjusted to apply corrective torque. Through these fine adjustments, the transfer robotic arm's grippers can operate stably or adaptively with optimized parameters, effectively coping with various complex dynamic instabilities.

[0119] Simultaneously, to ensure the safety and efficiency of the entire multi-station collaborative operation, collaborative instructions for the gripping robotic arm will be generated concurrently. These collaborative instructions may include controlling the gripping robotic arm to reduce the safe distance threshold of the avoidance path, so as to resume normal operation more quickly after the transfer robotic arm completes the stabilization operation; and / or extending the duration of the temporary hovering path, so as to give the transfer robotic arm more time to complete complex dynamic stabilization processing and avoid the gripping robotic arm intervening too early and causing secondary interference.

[0120] Through the above technical solution, this application overcomes the problems of insufficient versatility and inaccurate response that may exist in existing technologies when dealing with complex dynamic instability situations. Specifically, by identifying complex dynamic instability modes and selecting matching schemes, customized processing of different types of dynamic instability states can be achieved, significantly improving the success rate and stability of the transfer robot arm in dynamic environments. Simultaneously, refined parameter adjustments enable the gripper to make smooth contact and uniform force with the pouch lithium battery in an optimized manner, effectively reducing the risk of damage to the pouch lithium battery during gripping. Furthermore, the dynamic adjustment of the collaborative commands of the gripping robot arm further optimizes the efficiency and safety of multi-robot collaborative operations, ensuring smooth continuity of the production process during abnormal handling and avoiding unnecessary stoppages or conflicts.

[0121] For example, suppose that when a robotic arm attempts to grasp a pouch lithium battery, force signal analysis shows high-frequency, small-amplitude periodic fluctuations, while image sequence micro-motion analysis shows slight, rapid shaking of the pouch lithium battery. Based on these analysis results, the pouch lithium battery is identified as being in a composite dynamic instability mode of "high-frequency vibration." In this case, a scheme specifically designed for high-frequency vibration is selected from a flexible contact dynamic stabilization scheme library. This scheme instructs the robotic arm's gripper to approach the pouch lithium battery with a larger contact area and lower contact pressure, operating at a slightly lower grasping speed than normal, while adjusting the gripping posture to provide additional damping. During this process, the grasping robotic arm receives a cooperative instruction to slightly reduce the safety distance threshold of its avoidance path, allowing it to enter its preset path more quickly after completing a stable grasp, thereby reducing the overall operation time.

[0122] In some embodiments, the step of identifying the combined dynamic instability mode of the pouch lithium battery based on the force signal analysis results and the image sequence micro-motion analysis results includes:

[0123] By comparing the timestamps of the force signal analysis results and the image sequence micro-motion analysis results, the time deviation between the force signal analysis results and the image sequence micro-motion analysis results is identified;

[0124] Based on the time deviation, the analysis results with earlier timestamps are subjected to time compensation processing based on historical data or predicted trends to ensure the synchronization of the force signal analysis results and the image sequence micro-motion analysis results in the time dimension.

[0125] Based on the force signal analysis results after time compensation processing and the micro-motion analysis results of the image sequence, when there is a continuous time delay or data loss in the force signal analysis results or the micro-motion analysis results of the image sequence, predictive identification of the pattern is performed according to the current state and historical trend of the other analysis result, and the predicted composite dynamic instability pattern is output.

[0126] Specifically, the "comparing the timestamps of the force signal analysis results and the image sequence micro-motion analysis results" involves checking the timestamps generated or acquired by each of the two analysis results upon receipt. For example, a temporal inconsistency can be determined by comparing the absolute values ​​of the two timestamps or calculating the difference between them. The "identifying the time deviation between the force signal analysis results and the image sequence micro-motion analysis results" involves quantifying the degree of this inconsistency, for example, determining how many milliseconds earlier or later one analysis result is than the other.

[0127] The phrase "performing time compensation processing on analysis results with earlier timestamps based on historical data or predicted trends according to the time deviation" can be understood as follows: if the timestamp of one data stream (e.g., a force signal) is found to be significantly earlier than that of another data stream (e.g., an image sequence), the historical data patterns of that data stream or a preset dynamic model will be used to predict its state at the later timestamp, thereby aligning it with the data stream with the later timestamp. The purpose is to eliminate information misalignment caused by asynchronous data acquisition or processing, ensuring the accuracy of subsequent pattern recognition. For example, time compensation can be achieved through linear interpolation, Kalman filtering, or time series prediction models based on machine learning.

[0128] In practical applications, the integrity and real-time performance of two data streams are continuously monitored to detect persistent time delays or data gaps in force signal analysis results or image sequence micro-motion analysis results. For example, if a data stream fails to update within a preset time window, or multiple consecutive data packets are lost, it is determined that there is a persistent time delay or data gap. Based on the current state and historical trends of the other analysis result, predictive pattern recognition is performed. When a problem occurs in one data stream, recognition does not simply stop, but rather utilizes the current information and historical trends of the other normal data stream, combined with an existing knowledge base of complex dynamic instability patterns, to infer the most likely complex dynamic instability pattern. The purpose is to provide a reasonable pattern recognition result even when data is incomplete, ensuring continuous operation and timely response. For example, if image sequence data is missing, but the force signal shows high-frequency vibrations, the correlation between high-frequency vibrations and specific complex dynamic instability patterns in historical data can be used to predict that the current state may be in that pattern.

[0129] Through the above technical solution, this application can effectively solve the problem of inconsistency in the time dimension of multi-source data, ensuring the synchronization of force signal analysis results and image sequence micro-motion analysis results, thereby significantly improving the accuracy of composite dynamic instability pattern recognition for pouch lithium batteries. Furthermore, even in extreme cases where sensor data experiences continuous delays or is missing, the predictive recognition mechanism of this application can still perform pattern inference based on existing reliable data. This enables more accurate and timely identification and handling of pouch lithium battery grasping anomalies in complex industrial environments, thereby reducing production interruptions and improving production efficiency and safety.

[0130] For example, suppose that during a robotic arm's attempt to grasp a pouch lithium battery, multi-axis force signals and an image sequence of the junction area are simultaneously acquired. After analyzing this data, the timestamp of the force signal analysis result is T1, while the timestamp of the image sequence micro-motion analysis result is T1+50ms, resulting in a 50ms time discrepancy. This time discrepancy will be identified, and time compensation processing will be performed on the force signal analysis result with timestamp T1. For instance, the force signal state at time T1+50ms can be predicted based on historical force signal data prior to time T1, thus aligning the force signal analysis result with the image sequence micro-motion analysis result in time.

[0131] Furthermore, suppose that at some point, due to a temporary malfunction of the image sensor, image sequence data becomes continuously missing. In this case, pattern recognition does not stop. Instead, it utilizes the current state of the continuously received force signal analysis results (e.g., detecting continuous high-frequency vibration) and its historical trends (e.g., high-frequency vibration has often been accompanied by slight rotation of the pouch lithium battery in the past), combined with a pre-set pattern knowledge base, to predict that the current pouch lithium battery may be in a composite dynamic instability mode of "high-frequency vibration accompanied by slight rotation." This predicted mode is then output, and the corresponding flexible contact dynamic stabilization scheme is activated. When the image sensor malfunction is resolved, the previously predicted mode is compared with the actual recognized mode after the recovery, and the historical trend weights or current state influence factors in the prediction model are adjusted based on the comparison results to optimize future prediction accuracy. In this way, even with incomplete data, effective abnormal pattern recognition can be continuously provided, ensuring the continuity and safety of the production process.

[0132] In some embodiments, the step of predictively identifying patterns and outputting predicted composite dynamic instability patterns based on the current state and historical trends of the other analysis result when there is a continuous time delay or data gap in the force signal analysis result or the image sequence micro-motion analysis result further includes:

[0133] Once the data with time delay or missing data is recovered, the predicted composite dynamic instability pattern is compared with the recovered actual composite dynamic instability pattern, and the deviation between the two is calculated.

[0134] Based on the deviation value, adjust the historical trend weights or current state influence factors used in the predictive identification process;

[0135] Based on the adjusted historical trend weights or current state influence factors, predictive pattern identification is performed, and the predicted composite dynamic unstable pattern is output.

[0136] Specifically, once the time-delayed or missing data is recovered, the previously predicted composite dynamic instability pattern is compared with the recovered actual composite dynamic instability pattern. This comparison quantifies the difference between the prediction and the actual situation, calculating the deviation value. This deviation value can be understood as the degree of inaccuracy of the prediction model in a specific context, providing a quantitative basis for subsequent adjustments to the prediction model. Further, based on the calculated deviation value, the historical trend weights or current state influence factors used in the predictive identification process are adjusted. Historical trend weights refer to the degree of dependence on past data patterns during prediction, while the current state influence factor represents the degree of emphasis on the latest real-time data or state. By adjusting these parameters, the prediction model can better adapt to changes in the actual situation. For example, when the deviation value is large, the historical trend weights can be appropriately reduced and the current state influence factor increased to respond more quickly to the current situation. Therefore, based on the adjusted historical trend weights or current state influence factors, predictive identification of the pattern is performed again, and an updated predicted composite dynamic instability pattern is output. This iterative adjustment mechanism ensures that the predictive model can continuously learn and optimize, thereby improving its predictive accuracy and robustness even with incomplete data.

[0137] Through the above technical solution, this application can significantly improve the accuracy and robustness of predicting composite dynamic instability modes of soft-pack lithium batteries when there is a duration delay or data gap in the force signal analysis results or image sequence micro-motion analysis results. Specifically, by comparing the predicted mode with the recovered actual mode and calculating the deviation value, a quantitative evaluation of the prediction performance is obtained. On this basis, by dynamically adjusting the historical trend weights and current state influence factors in the prediction model, the prediction model can adaptively learn and optimize based on actual feedback. This mechanism effectively compensates for the limitations that may exist in prediction based solely on a single analysis result, ensuring that even under conditions of incomplete data or large fluctuations, it can continuously output high-precision predictions of composite dynamic instability modes. This provides a more reliable decision-making basis for the coordinated anomaly handling of the transfer robot arm and the gripping robot arm, further improving the intelligence level and production efficiency of the entire gripping control method.

[0138] In some embodiments, the composite dynamic instability mode includes high-frequency vibration, low-frequency swaying, and minute rotation.

[0139] High-frequency vibration refers to the rapid, small-amplitude reciprocating motion of the pouch lithium battery within a short period of time, usually caused by minor external disturbances or the micro-vibrations of the transfer robotic arm itself. Low-frequency swaying refers to the large-amplitude, low-frequency oscillation or shaking of the pouch lithium battery over a longer period of time, which may be caused by airflow, platform instability, or the slow movement of the transfer robotic arm. Micro-rotation refers to the slight angular deflection of the pouch lithium battery around its own axis or a specific point during the grasping process, which may lead to inaccurate grasping posture.

[0140] The above technical solutions enable a more comprehensive and detailed identification of the dynamic instability state of pouch lithium batteries, avoiding the confusion of different types of dynamic instability and thus improving the accuracy and effectiveness of anomaly handling scheme selection. This helps improve the success rate and stability of the transfer robotic arm in grasping pouch lithium batteries, reducing the risk of grasping failure or battery damage caused by dynamic instability.

[0141] This application also proposes a multi-station collaborative intelligent control system for independently grasping soft-pack lithium batteries, such as... Figure 2 As shown, a multi-station collaborative independent gripping intelligent control system 100 for soft-pack lithium batteries is disclosed. The system includes:

[0142] The abnormal information acquisition module 10 is used to detect abnormal information generated when the transfer robotic arm grabs a soft-pack lithium battery at a preset handover point;

[0143] The scheme activation module 20 is used to respond to the abnormal information, select and activate a local abnormality handling scheme from a pre-stored local abnormality handling scheme library, wherein the local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm.

[0144] The robotic arm control module 30 is used to control the transfer robotic arm to perform grasping and adjustment actions or safe retreat actions according to the local anomaly handling scheme, and to control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path.

[0145] The recovery coordination module 40 is used to coordinate the gripping robot arm and the transfer robot arm to resume normal production operations after the transfer robot arm completes the abnormal handling.

[0146] The multi-station collaborative independent gripping intelligent control system for pouch lithium batteries proposed in this application provides significant improvements to traditional automated production lines, addressing issues such as sluggish response, insufficient coordination, and susceptibility to equipment collisions and downtime when handling workpiece gripping anomalies. Traditional systems often rely on single sensor feedback or preset fixed logic, lacking the ability to finely identify and dynamically respond to complex anomalies. For example, when a pouch lithium battery experiences a slight shift or shaking at the junction point, existing systems may fail to accurately and promptly determine the type of anomaly, leading to gripping failure or inappropriate countermeasures.

[0147] This application constructs a highly integrated and intelligent control system by introducing an anomaly information acquisition module, a solution activation module, a robotic arm control module, and a recovery coordination module. The anomaly information acquisition module can perceive anomalies during the grasping process in real time and from multiple dimensions, providing data support for subsequent intelligent decision-making. The solution activation module can intelligently match and activate the optimal local processing solution based on the anomaly type, avoiding blind attempts or overreactions. The robotic arm control module realizes dynamic collaborative control between the transfer robotic arm and the grasping robotic arm, ensuring that the two robotic arms can coordinate their movements during anomaly handling, effectively avoiding collisions and ensuring equipment safety. The recovery coordination module ensures that production operations can be quickly and smoothly restored after anomaly handling. This systematic design makes the entire grasping process more flexible, intelligent, and efficient, significantly improving the stability and reliability of the production line and reducing operating costs and maintenance risks.

[0148] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A multi-station collaborative intelligent control method for independently grasping soft-pack lithium batteries, characterized in that, include: Detect abnormal information generated when the transfer robotic arm picks up a soft-pack lithium battery at a preset handover point; In response to the abnormal information, a local abnormality handling scheme is selected and activated from a pre-stored local abnormality handling scheme library. The local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm. According to the local anomaly handling scheme, control the transfer robotic arm to perform a grasping adjustment action or a safe retreat action, and control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path. After the transfer robotic arm completes the abnormality handling, the gripping robotic arm and the transfer robotic arm are coordinated to resume normal production operations. The step of responding to the abnormal information by selecting and activating a local abnormality handling scheme from a pre-stored local abnormality handling scheme library includes: In response to the abnormal information, when the transfer robotic arm attempts to grasp the soft-pack lithium battery, a multi-axis force signal is collected during the contact process between the gripper of the transfer robotic arm and the soft-pack lithium battery; and, within a preset time before and after the transfer robotic arm attempts to grasp the soft-pack lithium battery, an image sequence of the handover area is collected. The multi-axis force signal is analyzed in the time and frequency domains to identify the fluctuation pattern or periodic change of the multi-axis force signal and obtain the force signal analysis results; and the image sequence is tracked for feature points and reconstructed for motion trajectory to estimate the instantaneous motion vector of the soft-pack lithium battery and obtain the micro-motion analysis results of the image sequence. Based on the force signal analysis results and the image sequence micro-motion analysis results, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state. When it is determined that the static offset state is in the state, the static offset re-grabbing scheme is selected and activated, and the transfer robot arm is controlled to adjust its position and attitude and then grab the soft-pack lithium battery again. When it is determined that the dynamic unstable state is in the state, the flexible contact dynamic stabilization scheme or the tactile guidance adaptive grasping scheme is selected and activated, and the gripper of the transfer robot arm is controlled to contact the surface of the soft-pack lithium battery with controlled pressure to perform stable operation or adjust the grasping parameters according to real-time tactile feedback. Based on the judgment result, the grasping robotic arm is controlled to switch to an avoidance path or a temporary hovering path until the transfer robotic arm completes the abnormality handling.

2. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 1, characterized in that, The step of determining whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state based on the force signal analysis results and the image sequence micro-motion analysis results includes: The confidence level of the force signal analysis results is evaluated based on the signal-to-noise ratio, clarity of the fluctuation pattern, and stability of the periodic changes of the multi-axis force signal. The confidence level of the micro-motion analysis results of the image sequence is evaluated based on the stability of feature point tracking, the smoothness of motion trajectory reconstruction, and the amplitude of instantaneous motion vectors; wherein, the image occlusion interference caused by the movement of the grasping robotic arm is excluded in the stability evaluation of feature point tracking. The weights of the force signal analysis results and the image sequence micro-motion analysis results are adjusted based on the confidence levels of the force signal analysis results and the image sequence micro-motion analysis results. Based on the adjusted weights and the force signal analysis results, as well as the micro-motion analysis results of the image sequence, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state.

3. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 2, characterized in that, After the step of determining whether the pouch lithium battery is in a static offset state or a dynamically unstable state based on the adjusted weights, the force signal analysis results, and the image sequence micro-motion analysis results, the following is included: When the confidence levels of the force signal analysis results and the micro-motion analysis results of the image sequence are both lower than the preset confidence threshold, the transfer robotic arm is controlled to approach the soft-pack lithium battery again at a preset low speed and with a gentle contact force. During this process, the multi-axis force signal and the image sequence are collected again for secondary discrimination. Simultaneously, the gripping robotic arm is controlled to maintain a temporary hovering path until the secondary judgment is completed.

4. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 1, characterized in that, The step of controlling the transfer robotic arm to adjust its position and attitude before grabbing the pouch lithium battery again includes: The multi-axis force signal is compared with the preset ideal contact force distribution to identify the force distribution deviation caused by the gripper contact surface of the transfer robot or the elastic characteristics of the soft-pack lithium battery. Based on the force distribution deviation, the approach trajectory and gripping posture of the gripper of the transfer robotic arm are adjusted to gradually increase the gripping force and complete the grasping of the soft-pack lithium battery, so as to achieve smooth contact and uniform force with the surface of the soft-pack lithium battery.

5. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 1, characterized in that, The step of selecting and activating a flexible contact dynamic stabilization scheme or a tactile-guided adaptive gripping scheme when the dynamic unstable state is determined, and controlling the gripper of the transfer robotic arm to contact the surface of the soft-pack lithium battery with controlled pressure for stable operation or adjusting the gripping parameters according to real-time tactile feedback includes: Based on the force signal analysis results and the image sequence micro-motion analysis results, the composite dynamic instability mode of the soft-pack lithium battery is identified; Based on the composite dynamic instability mode, a matching scheme is selected from the preset flexible contact dynamic stabilization scheme library or the haptic-guided adaptive grasping scheme library. According to the matching scheme, the contact point, contact area, contact pressure, gripping speed and gripping posture of the gripper of the transfer robot arm are adjusted, and the gripper of the transfer robot arm is controlled to perform stable operation or adaptive gripping with the adjusted parameters. Simultaneously, collaborative instructions for the gripping robotic arm are generated, including controlling the gripping robotic arm to reduce the safe distance threshold of the avoidance path, and / or extending the duration of the temporary hovering path.

6. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 5, characterized in that, The step of identifying the combined dynamic instability mode of the pouch lithium battery based on the force signal analysis results and the image sequence micro-motion analysis results includes: By comparing the timestamps of the force signal analysis results and the image sequence micro-motion analysis results, the time deviation between the force signal analysis results and the image sequence micro-motion analysis results is identified; Based on the time deviation, the analysis results with earlier timestamps are subjected to time compensation processing based on historical data or predicted trends to ensure the synchronization of the force signal analysis results and the image sequence micro-motion analysis results in the time dimension. Based on the force signal analysis results after time compensation processing and the micro-motion analysis results of the image sequence, when there is a continuous time delay or data loss in the force signal analysis results or the micro-motion analysis results of the image sequence, predictive identification of the pattern is performed according to the current state and historical trend of the other analysis result, and the predicted composite dynamic instability pattern is output.

7. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 6, characterized in that, When there is a continuous time delay or data gap in the force signal analysis results or image sequence micro-motion analysis results, the step of predictively identifying the pattern based on the current state and historical trend of the other analysis result and outputting the predicted composite dynamic instability pattern includes: Once the data with time delay or missing data is recovered, the predicted composite dynamic instability pattern is compared with the recovered actual composite dynamic instability pattern, and the deviation between the two is calculated. Based on the deviation value, adjust the historical trend weights or current state influence factors used in the predictive identification process; Based on the adjusted historical trend weights or current state influence factors, predictive pattern identification is performed, and the predicted composite dynamic unstable pattern is output.

8. The intelligent control method for independent grasping of multi-station collaborative soft-pack lithium batteries according to claim 5, characterized in that, The composite dynamic instability mode includes high-frequency vibration, low-frequency shaking, and slight rotation; wherein, the slight rotation refers to the slight angular deflection of the soft-pack lithium battery around its own axis or a specific point during the grasping process.

9. A multi-station collaborative intelligent control system for independently grasping soft-pack lithium batteries, characterized in that, The system includes: The abnormal information acquisition module is used to detect abnormal information generated when the transfer robotic arm grabs a soft-pack lithium battery at a preset handover point; The scheme activation module is used to respond to the abnormal information, select and activate a local abnormality handling scheme from a pre-stored local abnormality handling scheme library, and the local abnormality handling scheme includes coordinated action instructions for the transfer robot arm and the gripping robot arm. The step of responding to the abnormal information by selecting and activating a local abnormality handling scheme from a pre-stored local abnormality handling scheme library includes: In response to the abnormal information, when the transfer robotic arm attempts to grasp the soft-pack lithium battery, a multi-axis force signal is collected during the contact process between the gripper of the transfer robotic arm and the soft-pack lithium battery; and, within a preset time before and after the transfer robotic arm attempts to grasp the soft-pack lithium battery, an image sequence of the handover area is collected. The multi-axis force signal is analyzed in the time and frequency domains to identify the fluctuation pattern or periodic change of the multi-axis force signal and obtain the force signal analysis results; and the image sequence is tracked for feature points and reconstructed for motion trajectory to estimate the instantaneous motion vector of the soft-pack lithium battery and obtain the micro-motion analysis results of the image sequence. Based on the force signal analysis results and the image sequence micro-motion analysis results, it is determined whether the soft-pack lithium battery is in a static offset state or a dynamic unstable state. When it is determined that the static offset state is in the state, the static offset re-grabbing scheme is selected and activated, and the transfer robot arm is controlled to adjust its position and attitude and then grab the soft-pack lithium battery again. When it is determined that the dynamic unstable state is in the state, the flexible contact dynamic stabilization scheme or the tactile guidance adaptive grasping scheme is selected and activated, and the gripper of the transfer robot arm is controlled to contact the surface of the soft-pack lithium battery with controlled pressure to perform stable operation or adjust the grasping parameters according to real-time tactile feedback. Based on the judgment result, the grasping robot arm is controlled to switch to an avoidance path or a temporary hovering path until the transfer robot arm completes the abnormality handling. The robotic arm control module is used to control the transfer robotic arm to perform grasping and adjustment actions or safe retreat actions according to the local anomaly handling scheme, and to control the grasping robotic arm to switch its preset motion path to an avoidance path or a temporary hovering path. The recovery coordination module is used to coordinate the gripping robot arm and the transfer robot arm to resume normal production operations after the transfer robot arm completes the abnormal handling.

Images