Printer system capable of automatically feeding and taking paper and control method

Abstract

The invention provides an automatic paper feeding and taking printer system and a control method, and relates to the technical field of printing, the automatic paper feeding and taking printer system comprises a main control unit, a visual perception module, a mechanical arm execution unit and a material area management module; the material area management module comprises a blank paper storage area, a printer paper feeding mechanism, a printer paper discharging mechanism and a printed finished product temporary storage area; the visual perception module comprises an industrial camera. The mechanical arm execution unit comprises a multi-degree-of-freedom mechanical arm and a composite clamp installed at the tail end of the multi-degree-of-freedom mechanical arm. The composite clamp comprises a lateral clamping mechanism used for clamping from the side face of the paper pile and an up-down fixing mechanism used for pressing and fixing the upper surface and the lower surface of the paper pile. The main control unit is in communication connection with the visual perception module and the mechanical arm execution unit.

Description

Technical Field

[0001] This invention relates to the field of printing technology, and in particular to an automatic paper feeding and dispensing printer system and control method. Background Technology

[0002] In office or industrial printing environments, large-scale continuous printing jobs require printers to have a continuous supply of blank paper and to promptly transfer finished printed paper to a storage area.

[0003] Traditional printers typically rely on manual loading of blank paper and unloading of printed products. Operators need to stay by the equipment for long periods of time and frequently perform paper loading and unloading operations. This is not only labor-intensive, but also prone to human error, which can lead to paper shortages in the paper feed mechanism or paper overflow in the paper output mechanism, affecting the overall printing efficiency.

[0004] Especially when paper stacks in the storage area or output tray become tilted, offset, or cross irregularly due to repeated stacking, the difficulty of picking up and putting down paper increases further. Existing printing systems lack the ability to perceive and adaptively process the actual stacking state of paper stacks, making it difficult to reliably pick up and move paper stacks without relying on manual intervention, which restricts the development of large-volume continuous printing tasks towards full-process automation. Summary of the Invention

[0005] This invention provides an automatic paper feeding and paper picking printer system and control method to solve the technical problem that existing printing systems lack the ability to perceive and adaptively process the stacking state of paper piles, making it difficult to reliably pick up and transport irregularly stacked paper piles.

[0006] This invention provides an automatic paper feeding and paper picking printer system, comprising: a main control unit, a vision perception module, a robotic arm execution unit, and a material area management module; The material area management module includes: a blank paper storage area, a printer paper feeding mechanism, a printer paper output mechanism, and a printed product temporary storage area; The visual perception module includes an industrial camera, and is configured to: acquire images of the paper stack in the material area management module and perform visual analysis, and output the status information of the paper stack; the status information of the paper stack includes at least the three-dimensional spatial posture of the paper stack and the abnormal state of the paper stack; The robotic arm execution unit includes: a multi-degree-of-freedom robotic arm and a composite clamp mounted on the end of the multi-degree-of-freedom robotic arm; the composite clamp includes a lateral clamping mechanism for clamping from the side of the paper stack, and an upper and lower fixing mechanism for pressing and fixing the upper and lower surfaces of the paper stack. The main control unit is communicatively connected to both the visual perception module and the robotic arm execution unit, and the main control unit is configured as follows: When the paper quantity of the printer's paper feed mechanism is detected to be lower than the preset paper replenishment threshold, or the paper quantity of the printer's paper output mechanism reaches the preset paper dispensing threshold, the visual perception module is triggered to perform visual analysis on the paper pile in the current target operation area to obtain the status information of the paper pile. The target grasping pose is generated based on the three-dimensional spatial posture of the paper stack, and the corresponding execution command is generated based on the abnormal state of the paper stack. The robotic arm execution unit is driven to approach the paper stack according to the target grasping posture, and through the lateral clamping mechanism and the upper and lower fixing mechanism, a three-dimensional physical constraint on the paper stack is established in coordination. According to the execution command, the paper stack is transported and released from the target operation area to the target placement area.

[0007] According to the automatic paper feeding and paper picking printer system provided by the present invention, the lateral clamping mechanism is a pair of symmetrically arranged left and right pressing flat jaws, the left and right pressing flat jaws are configured to open and close synchronously, and are used to apply pre-tightening force from both sides of the paper stack for lateral limiting. The upper and lower fixing mechanisms include: a top fixing plate and a bottom fixing plate, which are used to apply vertical pressing force to the upper and lower surfaces of the paper stack, respectively. The mechanism for establishing the three-dimensional physical constraint is as follows: the main control unit is configured to first control the left and right pressing plate jaws to close synchronously to apply lateral preload, and then control the top fixed plate and the bottom fixed plate to apply the vertical pressing force, forming a three-dimensional physical constraint with the combined action of lateral limiting and upper and lower pressing.

[0008] According to the automatic paper feeding and paper picking printer system provided by the present invention, the industrial camera includes: a first industrial camera deployed directly above the blank paper storage area, a second industrial camera deployed directly above the printer paper output mechanism, and a third industrial camera deployed at the end of the multi-degree-of-freedom robotic arm. The physical layout of the blank paper storage area, the printer paper feeding mechanism, the printer paper output mechanism, and the printed product temporary storage area are all marked in a unified system coordinate system.

[0009] According to the automatic paper feeding and dispensing printer system provided by the present invention, the state information of the paper stack further includes: two-dimensional contour information of the paper stack and thickness information of the paper stack; the visual perception module is specifically used for: The collected images of the paper pile are input into a pre-trained semantic segmentation model to perform pixel-level semantic segmentation, obtain a binary mask of the paper pile, and extract the two-dimensional contour information of the paper pile based on the binary mask of the paper pile. The pixels within the binary mask of the paper stack are back-projected into a three-dimensional point cloud in the world coordinate system using a pre-calibrated camera intrinsic parameter matrix and installation height parameters. A plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. Based on the optimal plane equation, the pitch angle and yaw angle of the upper surface of the paper stack relative to the reference plane are calculated. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The pitch angle, yaw angle, and surface flatness of the paper stack are used as the three-dimensional spatial attitude of the paper stack. The net height of the paper stack is obtained by subtracting the height of the pre-calibrated reference surface in the paperless state from the average height value of the five feature points. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

[0010] In the automatic paper feeding and dispensing printer system provided by the present invention, the visual perception module is further used for: Convex hull analysis is performed on the binary mask of the paper stack to calculate the convex defect features between the two-dimensional contour information of the paper stack and the convex hull; local curvature estimation is performed on the three-dimensional point cloud to obtain the curvature feature values ​​of the paper stack surface; Based on the convex defect characteristics, the curvature characteristic value, and the surface flatness of the paper stack, the abnormal state of the paper stack is determined; the abnormal state includes at least one of the following: Batch staggered stacking state: The convex defect feature exists and the thickness of the corresponding part of the convex defect feature exceeds the specified thickness threshold, while the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold. Batch folding and warping state: The curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold. Single sheet staggered stacking state: The convex defect feature exists and the thickness of the portion corresponding to the convex defect feature does not exceed the specified thickness threshold; Single sheet folding and warping state: The proportion of the abnormal area on the surface to the total area of ​​the paper stack is less than a preset area proportion threshold, and the local height change value is greater than a preset change threshold.

[0011] According to the automatic paper feeding and dispensing printer system provided by the present invention, the main control unit is further configured to: When the abnormal state is the batch staggered stacking state, an attitude compensation grasping command is generated to drive the robotic arm execution unit to adjust the end of the composite fixture to be parallel to the upper surface of the paper stack before performing the grasping. When the abnormal state is the batch folding and warping state, a batch waste instruction is generated, driving the robotic arm execution unit to grab and transport the entire paper stack to the waste paper area. When the abnormal state is the single sheet staggered stacking state, an edge pushing command is generated, which drives the robotic arm execution unit to control the composite clamp to gently push the edge of the paper stack along the protruding direction to make the misaligned paper return to its position. After the pushing is completed, the visual perception module is triggered to re-execute the visual analysis to confirm the alignment effect. After the confirmation is passed, the gripping is performed. When the abnormal state is the single sheet folded and warped state, a local waste instruction is generated, which drives the robotic arm execution unit to grab and transport the top layer of the paper stack containing the abnormal paper to the waste paper area.

[0012] According to the automatic paper feeding and paper picking printer system provided by the present invention, the main control unit is further configured to run a torque anomaly monitoring mechanism during the handling process of the robotic arm execution unit; the torque anomaly monitoring mechanism is used to detect the torque value of each joint of the multi-degree-of-freedom robotic arm in real time, and when the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and triggers emergency retraction, controlling the multi-degree-of-freedom robotic arm to retract along the original trajectory to a safe position; The main control unit is also configured to perform segmented speed control on the robotic arm execution unit: performing deceleration motion when approaching the paper stack, performing uniform motion when transporting the paper stack, and performing slow motion when placing the paper stack.

[0013] The present invention also provides a control method applied to the system described in any of the above claims, comprising: The paper level of the printer's paper feed mechanism and paper output mechanism is monitored in real time. When the paper level of the printer's paper feed mechanism is lower than the preset paper replenishment threshold, or the paper level of the printer's paper output mechanism reaches the preset paper take-out threshold, the current target operation area is determined. The visual perception module is controlled to acquire images of the paper pile in the target operation area. The acquired images of the paper pile are semantically segmented to obtain a binary mask of the paper pile. The pixels in the binary mask of the paper pile are mapped to a three-dimensional point cloud. The three-dimensional spatial pose of the paper pile relative to the reference plane is calculated based on the three-dimensional point cloud. The thickness information of the paper pile is estimated based on the three-dimensional point cloud. Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, the abnormal state of the paper stack is determined, and the processing strategy corresponding to the abnormal state is determined. Based on the three-dimensional spatial posture of the paper stack, the normal vector and geometric center position of the upper surface of the paper stack are calculated. Combined with the thickness information of the paper stack, the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism are determined. The angles of each joint of the multi-degree-of-freedom robotic arm are calculated by inverse kinematics algorithm to obtain the target grasping pose, and the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position is planned. The multi-degree-of-freedom robotic arm is controlled to move along the obstacle avoidance trajectory to the target grasping position, and the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture are driven to work together to establish a three-dimensional physical constraint on the paper stack. The paper stack is then transported to the target placement area and released according to the processing strategy.

[0014] Optionally, calculating the three-dimensional spatial pose of the paper stack relative to the reference plane based on the three-dimensional point cloud includes: The RANSAC plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. The pitch and yaw angles of the upper surface of the paper stack relative to the reference plane are calculated based on the optimal plane equation. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The estimation of the thickness information of the paper stack based on the three-dimensional point cloud includes: Subtract the pre-calibrated reference surface height in the paperless state from the average height value of the five feature points to obtain the net height of the paper stack. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

[0015] Optionally, the abnormal state of the paper stack is determined, and a corresponding processing strategy is determined, including: When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features exceeds the specified thickness threshold, and the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold, the abnormal state is determined to be a batch staggered stacking state, and the processing strategy is to calculate the attitude compensation amount and directly perform gripping and handling on the paper stack. When the curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or when the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold, the abnormal state is determined to be a batch folding and warping state, and the processing strategy is to grab all the paper stacks and transport them to the waste paper area. When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features does not exceed the specified thickness threshold, the abnormal state is determined to be a single sheet staggered stacking state. The processing strategy is to control the composite clamp to gently push the edge of the paper stack along the protruding direction to put the misaligned paper back into place. After the pushing is completed, image acquisition and visual analysis are re-executed to confirm the alignment effect. After confirmation, the paper stack is gripped and transported. When the area of ​​the abnormal surface region is less than the proportion of the total area of ​​the paper pile to a preset area ratio threshold, and the local height change value is greater than the preset change threshold, the abnormal state is determined to be a single sheet folding and warping state. The processing strategy is to grab and transport the top preset number of papers containing the abnormal paper to the waste paper area.

[0016] Optionally, the step of calculating the normal vector and geometric center position of the upper surface of the paper stack based on the three-dimensional spatial orientation of the paper stack further includes: The two-dimensional outline of the paper stack is shrunk, and a safe gripping area is generated inside the two-dimensional outline of the paper stack. The shrunk process reserves a preset edge margin. The clamping direction of the lateral clamping mechanism is determined based on the direction of the minimum circumscribed rectangle of the two-dimensional contour of the paper stack. The attitude compensation of the end of the composite fixture is performed based on the pitch and yaw angles in the three-dimensional spatial attitude of the paper stack, so that the end of the composite fixture is parallel to the upper surface of the paper stack. Based on the geometric center position, the safe gripping area, the gripping direction, and the posture of the gripper end after posture compensation, the angles of each joint of the multi-degree-of-freedom robotic arm are calculated using the inverse kinematics algorithm.

[0017] Optionally, the planned obstacle avoidance trajectory from the current position to the grab position and then to the release position includes: Smooth and continuous joint space trajectories are generated using polynomial interpolation or trapezoidal velocity curve planning. A fast exploration random tree algorithm is used to perform local dynamic obstacle avoidance by combining a pre-stored environment map; The obstacle avoidance trajectory is subject to segmented speed control: deceleration is performed when approaching the paper stack, uniform speed is performed when transporting the paper stack, and slow speed is performed when placing the paper stack.

[0018] Optionally, the lateral clamping mechanism and the upper and lower fixing mechanism that drive the composite fixture work together to establish a three-dimensional physical constraint on the paper stack, including: First, control the left and right pressing plate jaws of the lateral clamping mechanism to close synchronously to apply lateral pre-tightening force, and then control the top fixing plate and bottom fixing plate of the upper and lower fixing mechanism to apply vertical pressing force. After establishing the three-dimensional physical constraints, the success of pressing the upper and lower fixing mechanisms is determined by pressure threshold detection. During the handling process, the torque value of each joint of the multi-degree-of-freedom robotic arm is detected in real time. When the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and the multi-degree-of-freedom robotic arm is controlled to retreat along the original trajectory to a safe position. The release includes: after reaching the target placement area, first releasing the top fixed plate and the bottom fixed plate, and then releasing the left and right pressing plate clamps.

[0019] Optionally, when the target operating area is the blank paper storage area, the target placement area is the printer paper feeding mechanism; when the target operating area is the printer paper output mechanism, the target placement area is the printed product temporary storage area. When the target operating area is the printer's paper output mechanism, after the paper stack is transported and released, the process further includes: updating the batch count of completed printing tasks and recording the paper retrieval time and paper quantity.

[0020] Optionally, the method further includes: When the binary mask of the paper stack indicates that the paper stack is in a severely scattered state, the multi-degree-of-freedom robotic arm is controlled to drive the composite clamp to lightly press the top of the paper stack and apply a small-amplitude high-frequency vibration so that the paper naturally aligns under the action of gravity and friction. After alignment is complete, image acquisition is performed again to obtain an updated image of the stack of paper.

[0021] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the control method described above.

[0022] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the control method as described above.

[0023] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the control method as described above.

[0024] The automatic paper feeding and picking printer system and control method provided by this invention acquires and analyzes images of the paper stack through a visual perception module to obtain the three-dimensional spatial posture and abnormal state of the paper stack. The main control unit generates a target grasping posture and execution command that is adapted to the current paper stack state, and drives the composite gripper at the end of the robotic arm execution unit to establish three-dimensional physical constraints on the paper stack through the coordinated action of the lateral gripping mechanism and the upper and lower fixing mechanism to complete the handling and release. This allows the system to adaptively adjust the grasping strategy based on visual feedback, and reliably complete the gripping and handling of the paper stack even when the paper stack is tilted, offset, or in other irregular stacking states. This improves the reliability and automation of automatic paper feeding and picking operations in high-volume continuous printing scenarios. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of the working structure of the automatic paper feeding and paper picking printer system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the robotic arm execution unit provided in an embodiment of the present invention; Figure 3 This is the working state of the clamp for batch paper gripping provided in the embodiments of the present invention; Figure 4 A schematic diagram of the robotic arm's posture during the paper feeding process provided by the present invention; Figure 5 This is a schematic diagram of the robotic arm posture during the paper-picking process provided in an embodiment of the present invention; Figure 6 A schematic diagram of a control method provided by the present invention; Figure 7 The system workflow diagram provided for this invention; Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0028] The automatic paper feeding and dispensing printer system and control method provided in this invention can be applied to scenarios involving large-volume continuous printing tasks, such as batch document output, invoice printing, and label printing. The control logic of this system is executed by a main control unit, which can be implemented using an embedded SOC, FPGA, industrial computer, or other processing devices with computing and communication capabilities.

[0029] The following describes the specific implementation method using an embedded SOC as the main control unit, illustrating the detailed implementation process.

[0030] Figure 1 This is a schematic diagram of the working structure of the automatic paper feeding and dispensing printer system provided in an embodiment of the present invention. Figure 1 As shown, the automatic paper feeding and paper picking printer system provided in this embodiment includes four core modules: a main control unit, a vision perception module, a robotic arm execution unit 11, and a material area management module; the vision perception module may include multiple cameras 12; the four modules work together to achieve fully automatic paper handling.

[0031] The material area management module defines the storage and circulation areas for various types of paper within the system. For example... Figure 1 As shown, the material area management module includes four functional areas: a blank paper storage area, i.e. Figure 1 The paper feed area 13 is used to centrally store unused blank paper; users can simply insert a sufficient amount of paper at once before activating the system; this is the printer's paper feed mechanism. Figure 1 The paper tray 14 in the printer refers to the paper box or tray inside the printer used to receive the paper to be printed; the printer's paper output mechanism, i.e. Figure 1 The end paper tray 16 is the stacking position after the printer outputs the printed paper; the finished product storage area, i.e. Figure 1 The paper storage area 15 is used to temporarily store printed products retrieved from the paper output mechanism by the robotic arm, for manual collection later. The physical locations of the above four functional areas are fixed, and their coordinates were determined during the system installation and debugging phase.

[0032] The visual perception module is responsible for providing the entire system with environmental awareness capabilities. For example... Figure 1 As shown, the visual perception module includes an industrial camera, namely... Figure 1The cameras are deployed in key locations within the system to cover the areas that need to be observed.

[0033] The core function of the visual perception module is to acquire images of paper stacks in the material area management module, perform visual analysis, and finally output the status information of the paper stacks.

[0034] Here, "paper stack" refers to a complete stack of papers placed in any functional area, and its status information includes at least two aspects: The three-dimensional spatial orientation of the paper stack, namely the tilt angle and orientation information of the upper surface of the paper stack relative to the horizontal reference plane in three-dimensional space; the abnormal state of the paper stack, namely whether the paper stack has irregular conditions such as interlacing, folding, and warping.

[0035] The specific algorithm flow for visual analysis can be implemented in different ways according to actual needs. For example, edge detection methods based on traditional image processing or semantic segmentation methods based on deep learning can be used, as long as the above state information can be reliably output.

[0036] Figure 2 This is a schematic diagram of the structure of the robotic arm execution unit provided in an embodiment of the present invention. Figure 3 This describes the working state of the clamp for batch paper gripping provided in the embodiments of the present invention. For example... Figure 2 and Figure 3 As shown, the robotic arm execution unit includes a multi-degree-of-freedom robotic arm and a composite gripper mounted at the end of the robotic arm. The multi-degree-of-freedom robotic arm is mounted on the side of the printer or on an external independent bracket, and its working range covers all four functional areas mentioned above.

[0037] The number of degrees of freedom of the robotic arm can be selected according to actual layout requirements, such as a 4-DOF, 5-DOF, or 6-DOF collaborative robotic arm, as long as it can deliver the composite gripper to each functional area in a suitable pose. The composite gripper is a special gripping device installed on the flange at the end of the robotic arm. The core idea of ​​its structural design is to achieve reliable clamping of the paper stack through multi-directional constraints.

[0038] The composite clamp consists of two parts: one part is a lateral clamping mechanism, which is used to clamp the paper stack horizontally from the left and right sides to prevent the paper from slipping laterally during handling; the other part is an upper and lower fixing mechanism, which is used to apply vertical pressure from the top and bottom of the paper stack to prevent the paper from loosening or falling off.

[0039] The lateral clamping mechanism can be implemented as a flat plate gripper, an arc-shaped gripper, or other forms of lateral constraint structure, while the upper and lower fixing mechanism can be implemented as a fixed flat plate, a vacuum adsorption surface, or other forms of vertical constraint structure.

[0040] The two mechanisms work together to apply constraints to the paper stack from both horizontal and vertical dimensions, thereby establishing a three-dimensional physical constraint on the paper stack—that is, simultaneously restricting the paper stack's degrees of freedom of movement in the lateral and vertical directions in three-dimensional space, ensuring the stability of the paper during handling.

[0041] The main control unit is the control core of the entire system, and it communicates with the vision perception module and the robotic arm execution unit.

[0042] Communication can be done via wired or wireless connection, as long as the data transmission is real-time and reliable.

[0043] The main control unit is configured to perform the following three-stage coordinated control: The first stage is triggering visual analysis. The main control unit detects the paper level status of the printer's paper feed and output mechanisms in real time or periodically. The paper level can be obtained by receiving status signals such as paper shortage or full paper output through communication protocols with the printer host (such as serial port protocols or network protocols), or by directly detecting it through independently deployed sensors (such as weight sensors or photoelectric sensors).

[0044] When the paper quantity of the paper feeding mechanism is detected to be lower than the preset paper replenishment threshold, the main control unit determines the current target operating area as the blank paper storage area; when the paper quantity of the paper output mechanism reaches the preset paper dispensing threshold, the target operating area is determined as the paper output mechanism.

[0045] Subsequently, the main control unit sends a data acquisition and analysis command to the visual perception module, triggering the visual perception module to perform visual analysis on the paper stack in the target operation area and obtain the status information of the paper stack. The paper replenishment threshold and paper retrieval threshold can be configured by the user according to actual operation needs. For example, the paper replenishment threshold can be set to 20% of the maximum capacity of the paper feeding mechanism, and the paper retrieval threshold can be set to 80% of the maximum capacity of the paper output mechanism.

[0046] The second stage is generating control commands. After receiving the status information of the paper stack, the main control unit generates the target grasping pose based on the three-dimensional spatial attitude of the paper stack—that is, the spatial coordinates (x, y, z) and orientation of the composite gripper end that should reach when approaching the paper stack. At the same time, it generates corresponding execution commands based on abnormal states of the paper stack—that is, the operation plan determined for the current paper stack condition, including but not limited to direct grasping, sorting before grasping, and disposal.

[0047] The third stage is the driven execution of the paper transport. The main control unit drives the robotic arm execution unit to approach the paper stack according to the target grasping posture, and controls the lateral gripping mechanism and the upper and lower fixing mechanism to coordinate their actions according to preset timing and force parameters to establish three-dimensional physical constraints on the paper stack. After the constraints are established, the paper stack is transported from the target operation area to the target placement area and released according to the execution command. The target placement area is determined according to the task type: in a paper replenishment task, it is the printer paper feeding mechanism; in a paper retrieval task, it is the temporary storage area for printed products.

[0048] In this application, the visual perception module provides the posture and abnormal state information of the paper stack in three-dimensional space. Based on this, the main control unit adaptively generates the grasping pose and execution instructions. The composite fixture establishes three-dimensional physical constraints through the synergistic effect of lateral clamping and upper and lower fixation, forming a complete closed loop. This significantly reduces the frequency of manual intervention and improves the work efficiency in high-volume continuous printing scenarios.

[0049] Optionally, the lateral clamping mechanism is a pair of symmetrically arranged left and right pressing plate grippers, which are configured to open and close synchronously to apply pre-tightening force from both sides of the paper stack for lateral positioning. The upper and lower fixing mechanisms include: a top fixing plate and a bottom fixing plate, which are used to apply vertical pressing force to the upper and lower surfaces of the paper stack, respectively. The mechanism for establishing the three-dimensional physical constraint is as follows: the main control unit is configured to first control the left and right pressing plate jaws to close synchronously to apply lateral preload, and then control the top fixed plate and the bottom fixed plate to apply the vertical pressing force, forming a three-dimensional physical constraint with the combined action of lateral limiting and upper and lower pressing.

[0050] like Figure 2 and Figure 3 As shown, the lateral clamping mechanism is specifically implemented as symmetrically arranged left and right pressing flat grippers. These two grippers are located on the left and right sides of the composite fixture, respectively, and have flat contact surfaces. The material of the contact surfaces can be non-slip rubber or engineering plastics with a high coefficient of friction to increase the friction between the gripper and the edge of the paper.

[0051] The left and right clamping plates open and close synchronously under the control of the main control unit. That is, the left and right clamping plates close inward or open outward at the same speed and stroke, which is used to apply pre-tightening force from both sides of the paper stack for lateral limiting.

[0052] Pre-tightening force refers to the horizontal clamping force applied to both sides of the paper stack before the handling begins, so as to generate sufficient friction between the layers of paper to resist inertia and vibration during the handling process.

[0053] The upper and lower fixing mechanisms include a top fixing plate and a bottom fixing plate. For example... Figure 3 As shown, the top fixing plate is located above the composite clamp, and the bottom fixing plate is located below the composite clamp. Both can move independently toward the upper and lower surfaces of the paper stack and apply vertical pressing force. When gripping, the top fixing plate presses down onto the upper surface of the paper stack, while the bottom fixing plate inserts under the paper stack to apply an upward supporting force to the bottom surface. Together, they form a vertical upper-pressing and lower-supporting structure.

[0054] like Figure 3 As shown, the establishment of three-dimensional physical constraints follows a specific timing sequence: the main control unit first controls the left and right pressing plate jaws to close synchronously to apply lateral preload, and then controls the top fixed plate and bottom fixed plate to apply vertical pressing force, ultimately forming a three-dimensional physical constraint with the combined action of lateral limiting and upper and lower pressing.

[0055] The reason for adopting the sequence of first applying lateral pressure and then vertical pressure is that the lateral pre-tightening force is used to initially constrain and align the layers of paper in the paper stack, making the edges of the paper more even. Then, the vertical pressure is applied, which can more evenly compact the paper stack and avoid misaligned paper being pressed off or broken due to the vertical force being applied first.

[0056] The release operation also follows a specific timing sequence, which is the opposite of the establishment timing sequence: after reaching the target placement area, the main control unit first controls the top fixed plate and the bottom fixed plate to release, and then controls the left and right pressing plate grippers to release.

[0057] The order of releasing the vertical pressure first and then the lateral pressure is because releasing the vertical pressure first allows the paper stack to naturally unfold and fall smoothly into the target tray or compartment under its own weight. Releasing the lateral constraints then effectively avoids the paper from spreading out or shifting due to elastic recovery caused by releasing all constraints at the same time.

[0058] In this application, through the above-mentioned structural design and timing control, the left and right pressing plate grippers cooperate with the top fixed plate and the bottom fixed plate in terms of structure and act sequentially in terms of timing control, ensuring that the paper does not scatter, slip or be damaged during the handling process, and that the paper can fall smoothly and undisturbed into the target position during the release process.

[0059] Optionally, the industrial camera includes: a first industrial camera deployed directly above the blank paper storage area, a second industrial camera deployed directly above the printer paper output mechanism, and a third industrial camera deployed at the end of the multi-degree-of-freedom robotic arm; The physical layout of the blank paper storage area, the printer paper feeding mechanism, the printer paper output mechanism, and the printed product temporary storage area are all marked in a unified system coordinate system.

[0060] In this application, three industrial cameras are deployed in the visual perception module. The first industrial camera is fixedly installed directly above the blank paper storage area, i.e. Figure 1 The camera marked above the paper dispensing area has a field of view covering the entire upper surface of the paper pile in the blank paper storage area, and is used to capture images of the paper pile in this area during the automatic paper replenishment process.

[0061] Because the first industrial camera is fixedly installed, its relative position to the blank paper storage area is known and constant, which is beneficial for improving the accuracy and processing speed of visual analysis. The second industrial camera is fixedly installed directly above the printer's paper output mechanism, i.e. Figure 1 The camera, marked near the middle end of the paper tray, has a field of view covering the entire upper surface of the paper stack in the paper output mechanism. It is used to capture images of the printed paper stack during the automatic paper feeding process.

[0062] The third industrial camera is mounted at the end of a multi-degree-of-freedom robotic arm, i.e. Figure 1 The camera on the head of the robotic arm moves with the arm. The third industrial camera can capture high-precision images of the paper stack from a closer distance when the robotic arm moves above each work area. It is particularly suitable for secondary confirmation of local details of the paper stack, such as edge alignment and surface wrinkles, before the composite gripper is about to perform a gripping operation, thereby further improving the gripping accuracy.

[0063] In this application, the physical layout of the blank paper storage area, the printer paper feeding mechanism, the printer paper output mechanism, and the printed product temporary storage area are all marked in a unified system coordinate system.

[0064] The system coordinate system is established as a three-dimensional rectangular coordinate system based on a preset spatial origin, and the position coordinates of each region are expressed in this coordinate system. The coordinate calibration process is completed through joint calibration between the third industrial camera (installed at the end of the robotic arm) and the first and second industrial cameras. Specifically, the transformation matrix between the coordinate systems of each camera and the system coordinate system can be determined using a hand-eye calibration method.

[0065] In a unified system coordinate system, the reference plane is the plane with z=0, and the three-dimensional spatial orientation of the paper stack is the spatial angular relationship with respect to this reference plane.

[0066] Three industrial cameras cover two fixed fields of view (paper infeed and paper outfeed) and one servo field of view, enabling multi-view, full-coverage perception of each work area. A unified system coordinate system ensures that the paper stack position information output by the vision perception module and the motion space of the multi-degree-of-freedom robotic arm are under the same coordinate reference, eliminating coordinate transformation errors and improving gripping and positioning accuracy.

[0067] Optionally, the state information of the paper stack further includes: two-dimensional contour information of the paper stack and thickness information of the paper stack; the visual perception module is specifically used for: The collected images of the paper pile are input into a pre-trained semantic segmentation model to perform pixel-level semantic segmentation, obtain a binary mask of the paper pile, and extract the two-dimensional contour information of the paper pile based on the binary mask of the paper pile. The pixels within the binary mask of the paper stack are back-projected into a three-dimensional point cloud in the world coordinate system using a pre-calibrated camera intrinsic parameter matrix and installation height parameters. A plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. Based on the optimal plane equation, the pitch angle and yaw angle of the upper surface of the paper stack relative to the reference plane are calculated. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The pitch angle, yaw angle, and surface flatness of the paper stack are used as the three-dimensional spatial attitude of the paper stack. The net height of the paper stack is obtained by subtracting the height of the pre-calibrated reference surface in the paperless state from the average height value of the five feature points. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

[0068] In this application, the state information of the paper stack also includes the two-dimensional contour information and the thickness information of the paper stack. The two-dimensional contour information refers to the boundary shape description of the paper stack on the image plane, which can be expressed as an ordered sequence of pixel coordinate points. The thickness information refers to the actual physical thickness value of the paper stack from bottom to top, which will directly affect the setting of the opening distance and pressing stroke of the subsequent composite fixture.

[0069] The complete process of visual analysis performed by the visual perception module is as follows: The visual perception module first inputs the acquired image of the paper stack into the pre-trained semantic segmentation model, performs pixel-level semantic segmentation, and obtains the binary mask of the paper stack.

[0070] The semantic segmentation model can be implemented based on the improved U-Net architecture or the YOLOv8-seg architecture. The model input is a high-resolution RGB image that has been preprocessed with illumination equalization and noise suppression, and the output is a single-channel binary mask, where regions with a pixel value of 1 identify regions belonging to the paper stack, and regions with a pixel value of 0 identify the background.

[0071] The model was trained on a real-world dataset containing various stacking states, such as regular stacking, tilted stacking, misaligned stacking, warped stacking, and cross stacking, to ensure robust extraction of the outer contour of the paper stack even in complex backgrounds. Based on a binary mask, an edge tracking algorithm is used to extract the two-dimensional contour information of the paper stack.

[0072] Subsequently, the visual perception module backprojects the pixels within the binary mask into a 3D point cloud in the world coordinate system using a pre-calibrated camera intrinsic matrix and installation height parameters.

[0073] Specifically, for each pixel with a value of 1 in the binary mask, its two-dimensional image coordinates are mapped to three-dimensional spatial coordinates (x, y, z) in the world coordinate system based on camera intrinsic parameters and installation height. The z-coordinate reflects the height of the paper stack surface corresponding to that pixel. After all valid pixels are mapped, the three-dimensional point cloud of the upper surface of the paper stack is obtained.

[0074] Next, a plane fitting algorithm (such as RANSAC algorithm or least squares method) is applied to the 3D point cloud to solve the optimal plane equation ax+by+cz+d=0 for the upper surface of the paper stack, where the normal vector (a,b,c) represents the orientation of the upper surface of the paper stack in 3D space.

[0075] Based on the optimal plane equation, calculate the pitch angle (rotation angle about the y-axis) and yaw angle (rotation angle about the z-axis) of the upper surface of the paper stack relative to the reference plane.

[0076] Based on the attitude calculation, the visual perception module extracts the height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack.

[0077] The four corner points are the three-dimensional point cloud height values ​​corresponding to the four vertices of the minimum bounding rectangle of the two-dimensional outline of the paper stack, and the center point is the height value corresponding to the geometric center of the minimum bounding rectangle.

[0078] The mean height values ​​of five feature points are calculated to reflect the average height level of the paper stack's upper surface, the standard deviation (which reflects the dispersion of height at each location; a smaller standard deviation indicates a smoother surface), and the maximum value (which reflects the location of the highest point and is used to detect local protrusions). These statistics are defined as the surface smoothness of the paper stack. Finally, the pitch angle, yaw angle, and surface smoothness are combined to determine the three-dimensional spatial attitude of the paper stack.

[0079] The thickness information is estimated by subtracting the pre-calibrated reference surface height in the paperless state from the average height value of the five feature points, and obtaining the reference height value by calibrating the surface of the empty pallet without placing any paper, to obtain the net height of the paper stack.

[0080] The actual physical thickness of the paper stack is then obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient. The thickness coefficient is obtained by calibrating a standard paper stack of known thickness and is used to convert the visually measured height value into the actual physical thickness.

[0081] For example, assuming the average height of the five feature points is 25.3 mm and the reference plane height is 2.1 mm, the net height of the paper stack is 23.2 mm; assuming the thickness coefficient is 0.95 and the zero-point offset is 0.2 mm, the actual physical thickness is 23.2 × 0.95 + 0.2 ≈ 22.24 mm. Based on the estimate that the thickness of a single standard A4 sheet of paper is about 0.1 mm, the current paper stack contains about 222 sheets of paper.

[0082] This embodiment starts with a two-dimensional image, obtains an accurate binary mask of the paper stack through semantic segmentation, obtains the spatial pose through three-dimensional back projection and plane fitting, quantifies the surface flatness through five-point height statistics, and obtains the physical thickness through net height and thickness coefficient, forming a complete link from two-dimensional perception to three-dimensional state representation, providing sufficient data support for subsequent anomaly detection and grasping pose planning.

[0083] Optionally, the visual perception module is further configured to: Convex hull analysis is performed on the binary mask of the paper stack to calculate the convex defect features between the two-dimensional contour information of the paper stack and the convex hull; local curvature estimation is performed on the three-dimensional point cloud to obtain the curvature feature values ​​of the paper stack surface; Based on the convex defect characteristics, the curvature characteristic value, and the surface flatness of the paper stack, the abnormal state of the paper stack is determined; the abnormal state includes at least one of the following: Batch staggered stacking state: The convex defect feature exists and the thickness of the corresponding part of the convex defect feature exceeds the specified thickness threshold, while the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold. Batch folding and warping state: The curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold. Single sheet staggered stacking state: The convex defect feature exists and the thickness of the portion corresponding to the convex defect feature does not exceed the specified thickness threshold; Single sheet folding and warping state: The proportion of the abnormal area on the surface to the total area of ​​the paper stack is less than a preset area proportion threshold, and the local height change value is greater than a preset change threshold.

[0084] In this application, after the visual perception module completes the calculation of the above-mentioned three-dimensional spatial pose and thickness information, it further performs the judgment of abnormal state.

[0085] First, convex hull analysis is performed on the binary mask of the paper stack to calculate the convex defect features between the two-dimensional contour of the paper stack and the convex hull.

[0086] A convex hull is the smallest convex polygon that can completely enclose the two-dimensional outline of a paper stack. A convex defect is a concave area that exists between the actual outline of the paper stack and the convex hull, including parameters such as the location, depth, and area of ​​the concave defect.

[0087] The presence of convex defects usually means that paper protrudes or recedes from the normal rectangular outline. Simultaneously, local curvature estimation is performed on the 3D point cloud to obtain the curvature characteristic values ​​of the paper stack surface.

[0088] Local curvature estimation can be achieved by performing principal component analysis in the neighborhood of each point and calculating the minimum eigenvalue λ_min. The larger λ_min is, the more pronounced the curvature of the region.

[0089] After acquiring the convex defect features, curvature feature values, and previously calculated surface flatness, the visual perception module determines the abnormal state of the paper stack based on these three-dimensional features.

[0090] Abnormal states include at least one of the following: batch staggered stacking state, batch folding and warping state, single staggered stacking state, and single folding and warping state.

[0091] Batch staggered stacking refers to a situation where multiple sheets of paper in a stack have undergone lateral shift as a whole. The criteria for determination are: the presence of a convex defect feature and the thickness of the corresponding portion exceeding a specified thickness threshold, for example, 0.5mm, corresponding to approximately the thickness of 5 sheets of paper; and the standard deviation of surface flatness not exceeding a preset flatness threshold. Since it is a whole-staggered shift, the surface of the paper stack remains relatively flat, which is a normal and usable condition; only posture compensation is needed for normal grasping.

[0092] Batch folding and warping refers to a situation where a large area of ​​the paper stack has undergone structural bending deformation. The determination criteria are: the curvature characteristic value of a portion of the paper stack surface area exceeds a preset curvature threshold, or the aspect ratio of the binary mask deviates from the nominal value by more than a preset deviation threshold. If either of these conditions is met, the paper stack is considered to be in this state, and it cannot be restored; the entire batch should be discarded.

[0093] A single sheet staggered stacking state refers to a situation where only a single sheet or a very small number of sheets protrude from the edge of the stack. The determination condition is: the convex defect feature exists, but the thickness of the corresponding part of the convex defect does not exceed the specified thickness threshold mentioned above, indicating that it is not an overall offset but a local misalignment of a small number of sheets. In this case, it can be restored to the normal state by edge pushing operation.

[0094] Single-sheet folding and warping refers to a situation where only a small area on the surface of the paper stack exhibits height abnormalities, such as a single sheet of paper being wrinkled or warped. The criteria for determination are: the proportion of the abnormal area to the total area of ​​the paper stack is less than a preset area proportion threshold, but the local height abrupt change value is greater than a preset abrupt change threshold. Since the composite fixture cannot reliably separate the abnormal single sheet of paper, the top layer of papers, including the abnormal sheet, must be discarded along with it.

[0095] When none of the above four abnormal states are triggered, the paper stack is considered to be in a normal state, and the main control unit directly generates the execution instruction for regular grasping.

[0096] Through the above-mentioned multi-dimensional anomaly discrimination mechanism, the system can accurately distinguish four typical paper stack anomaly states, providing a reliable basis for subsequent differentiated processing strategies for different anomaly types.

[0097] Optionally, the main control unit is further configured to: When the abnormal state is the batch staggered stacking state, an attitude compensation grasping command is generated to drive the robotic arm execution unit to adjust the end of the composite fixture to be parallel to the upper surface of the paper stack before performing the grasping. When the abnormal state is the batch folding and warping state, a batch waste instruction is generated, driving the robotic arm execution unit to grab and transport the entire paper stack to the waste paper area. When the abnormal state is the single sheet staggered stacking state, an edge pushing command is generated, which drives the robotic arm execution unit to control the composite clamp to gently push the edge of the paper stack along the protruding direction to make the misaligned paper return to its position. After the pushing is completed, the visual perception module is triggered to re-execute the visual analysis to confirm the alignment effect. After the confirmation is passed, the gripping is performed. When the abnormal state is the single sheet folded and warped state, a local waste instruction is generated, which drives the robotic arm execution unit to grab and transport the top layer of the paper stack containing the abnormal paper to the waste paper area.

[0098] In this application, when the abnormal state is a batch interleaved stacking state, the main control unit generates an attitude compensation grasping command.

[0099] Since the essence of batch staggered stacking is that the entire paper stack tilts or shifts laterally while the upper surface remains flat, it is only necessary to perform corresponding rotational compensation on the end attitude of the composite fixture based on the pitch and yaw angles obtained from the aforementioned visual analysis. After the bottom surface of the fixture is parallel and in contact with the upper surface of the paper stack, the normal gripping process can be directly executed. This strategy does not require any paper stack arrangement operations and has the highest execution efficiency.

[0100] When the abnormal state is a batch of folded and warped paper, the main control unit generates a batch discard command. This command drives the robotic arm to grab the entire stack of paper and transport it to the waste paper area. The waste paper area is a dedicated collection area located near the material area management module, used to receive defective paper. Since the entire stack of paper has undergone structural bending and deformation that cannot be repaired, continued use will lead to paper jams or quality degradation, therefore the entire batch must be discarded. The system simultaneously records the abnormal event and reports it to the main control log, facilitating subsequent investigation by maintenance personnel to determine the cause.

[0101] When the abnormal state is a single sheet staggered stack, the main control unit generates an edge pushing command. The main control unit determines the direction and displacement of the protruding paper based on the characteristics of the convex defect, controls the multi-degree-of-freedom robotic arm to move the side of the composite fixture to the edge position of the protruding paper, and then applies a small pushing force in the opposite direction to the protruding direction, so that the protruding paper returns to the position aligned with the overall paper stack.

[0102] After the alignment is completed, the main control unit triggers the visual perception module to re-execute the visual analysis to confirm the alignment effect. If it is confirmed that the two-dimensional outline of the paper stack has been restored to the standard rectangular range, normal grasping continues; if there are still misalignments, the alignment can be repeated or manual intervention can be initiated.

[0103] When the abnormal state is a single sheet folded or warped, the main control unit generates a localized waste disposal command. This command drives the robotic arm execution unit to grab a preset number of sheets, such as 5 to 10 sheets, from the top layer of the paper stack containing the abnormal paper and transport them to the waste paper area.

[0104] The reason for adopting the joint discard strategy is that the composite clamp is a batch clamping structure and cannot accurately separate a single abnormal paper. Therefore, it is necessary to remove and discard the abnormal paper and a small number of normal papers nearby together to ensure the quality reliability of the remaining paper pile.

[0105] In this application, through the aforementioned differentiated processing strategy, the system directly captures batches of interlacing that can be resolved by posture compensation, discards batches of warping due to structural defects, performs push-and-restore on repairable single-sheet misalignment, and discards small batches of irreparable single-sheet warping, thus balancing processing efficiency and paper quality control.

[0106] Optionally, the main control unit is further configured to run a torque anomaly monitoring mechanism during the handling process performed by the robotic arm execution unit; the torque anomaly monitoring mechanism is used to detect the torque value of each joint of the multi-degree-of-freedom robotic arm in real time, and when the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and an emergency retraction is triggered, controlling the multi-degree-of-freedom robotic arm to retract along the original trajectory to a safe position; The main control unit is also configured to perform segmented speed control on the robotic arm execution unit: performing deceleration motion when approaching the paper stack, performing uniform motion when transporting the paper stack, and performing slow motion when placing the paper stack.

[0107] In this application, Figure 4 This is a schematic diagram of the robotic arm posture during the paper feeding process provided by the present invention. Figure 5 This is a schematic diagram of the robotic arm posture during the paper-picking process provided in an embodiment of the present invention. The main control unit has an abnormal torque monitoring mechanism during the handling process of the robotic arm execution unit, which is used to detect the torque values ​​of each joint of the multi-degree-of-freedom robotic arm in real time.

[0108] The torque value can be directly measured by torque sensors installed in each joint, or indirectly calculated by the current value of the motor in each joint.

[0109] The main control unit compares the torque values ​​of each joint with the preset normal range. The preset normal range refers to the reasonable fluctuation range of the torque values ​​of each joint under normal paper stack handling conditions. This range is determined by statistical data from multiple no-load and load operations.

[0110] When the torque value exceeds the preset normal range, the main control unit determines that there is unexpected resistance, such as the robotic arm colliding with the surrounding environment, paper getting stuck on the printer's structural components, or unexpected obstacles appearing on the movement path. It immediately triggers an emergency retreat, controlling the multi-degree-of-freedom robotic arm to retreat along the original trajectory to a safe position.

[0111] The safe position is a known safe coordinate point on the current trajectory of the robotic arm, at a preset distance from the starting point. Instead of executing a new path, the robotic arm retracts along the original trajectory, which can avoid secondary collisions caused by path planning uncertainties in emergency situations.

[0112] like Figure 4 and Figure 5 As shown, the main control unit also performs segmented speed control on the robotic arm's execution unit, which includes three stages: During the approach phase of the robotic arm moving from the standby position to the paper stack, a deceleration motion is performed to reduce airflow disturbance and vibration when the composite gripper approaches the paper stack, and to prevent the paper from being blown or shaken apart before being gripped. During the handling phase, when the robotic arm has established three-dimensional physical constraints and is moving towards the target placement area, uniform motion is performed to reduce the impact of inertial forces generated by acceleration and deceleration on the stability of the paper stack. During the placement phase, when the robotic arm reaches the vicinity of the target placement area and is ready to release, a slow motion is performed to ensure that the composite gripper is precisely aligned with the target placement position at an extremely low speed, reducing the impact force at the moment of release.

[0113] In this application, the torque anomaly monitoring mechanism provides real-time collision detection and emergency protection capabilities, while segmented speed control reduces disturbance to the paper stack during handling from a kinematic perspective. The two work together to effectively reduce the risk of paper scattering, equipment collisions, and paper breakage.

[0114] Figure 6 A schematic diagram of a control method provided by the present invention is shown below. Figure 6 As shown, it includes: Step 610: Monitor the paper level status of the printer's paper feed mechanism and paper output mechanism in real time. When the paper level of the printer's paper feed mechanism is lower than the preset paper replenishment threshold, or the paper level of the printer's paper output mechanism reaches the preset paper take-out threshold, determine the current target operation area. Step 620: Control the visual perception module to acquire images of the paper pile in the target operation area. Perform semantic segmentation on the acquired images of the paper pile to obtain a binary mask of the paper pile. Map the pixels in the binary mask of the paper pile to a three-dimensional point cloud. Calculate the three-dimensional spatial pose of the paper pile relative to the reference plane based on the three-dimensional point cloud. Estimate the thickness information of the paper pile based on the three-dimensional point cloud. Step 630: Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, determine the abnormal state of the paper stack and determine the processing strategy corresponding to the abnormal state. Step 640: Calculate the normal vector and geometric center position of the upper surface of the paper stack based on the three-dimensional spatial posture of the paper stack. Combine the thickness information of the paper stack to determine the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism. Solve the joint angles of the multi-degree-of-freedom robotic arm through inverse kinematics algorithm to obtain the target grasping pose, and plan the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position. Step 650: Control the multi-degree-of-freedom robotic arm to move along the obstacle avoidance trajectory to the target grasping position, drive the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture to work together to establish a three-dimensional physical constraint on the paper stack, and transport the paper stack to the target placement area and release it according to the processing strategy.

[0115] In this application, the main control unit periodically queries or receives the current paper quantity information of the paper feeding mechanism and the paper output mechanism through the communication interface between the main control unit and the printer host. When the paper quantity of the paper feeding mechanism is lower than the preset replenishment threshold, the main control unit determines the current target operating area as the blank paper storage area and starts the automatic replenishment process; when the paper quantity of the paper output mechanism reaches the preset paper picking threshold, the target operating area is determined as the paper output mechanism and the automatic paper picking process is started.

[0116] After determining the target operating area, the main control unit controls the visual perception module to acquire images of the paper stacks in that area.

[0117] The visual perception module performs semantic segmentation on the acquired paper stack image to obtain a binary mask of the paper stack, maps the pixels in the binary mask to a three-dimensional point cloud, calculates the three-dimensional spatial pose of the paper stack relative to the reference plane based on the three-dimensional point cloud, and estimates the thickness information of the paper stack based on the three-dimensional point cloud.

[0118] The main control unit sends an acquisition command to the visual perception module. After the visual perception module completes the acquisition, it transmits the paper stack image to the main control unit, or the built-in processor of the visual perception module executes the above calculation process and outputs the binary mask, three-dimensional spatial pose and thickness information.

[0119] Based on the visual analysis results, the main control unit determines the abnormal state of the paper stack based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature values ​​of the paper stack surface, and determines the corresponding processing strategy. The types of processing strategies include direct grasping after posture compensation, batch rejection, grasping after edge pushing, and rejection along with local areas.

[0120] Subsequently, the main control unit calculates the normal vector and geometric center position of the upper surface of the paper stack based on the three-dimensional spatial posture of the paper stack. Combined with the thickness information, it determines the opening distance of the lateral clamping mechanism and the pressing stroke of the upper and lower fixing mechanisms of the composite fixture. The inverse kinematics algorithm is used to solve the angles of each joint of the multi-degree-of-freedom robotic arm to obtain the target grasping pose, and plans the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position.

[0121] Among them, the normal vector determines the orientation of the end of the composite gripper, the position of the geometric center determines the gripping center in the horizontal direction, the thickness information determines the opening distance of the gripper and the pressing stroke, and the inverse kinematics algorithm converts the above end pose parameters into the angle values ​​of each joint.

[0122] Finally, the main control unit controls the multi-degree-of-freedom robotic arm to move along the obstacle avoidance trajectory to the target grasping position, drives the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture to work together to establish a three-dimensional physical constraint on the paper stack, and transports the paper stack to the target placement area and releases it according to the processing strategy.

[0123] The above control method realizes a complete closed loop from paper quantity monitoring, visual perception, anomaly judgment, pose planning to physical execution. It can adaptively cope with various posture deviations and abnormal states that may occur in the paper stack under actual working conditions, and significantly improve the reliability of automatic paper feeding and paper picking operations.

[0124] Optionally, calculating the three-dimensional spatial pose of the paper stack relative to the reference plane based on the three-dimensional point cloud includes: The RANSAC plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. The pitch and yaw angles of the upper surface of the paper stack relative to the reference plane are calculated based on the optimal plane equation. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The estimation of the thickness information of the paper stack based on the three-dimensional point cloud includes: Subtract the pre-calibrated reference surface height in the paperless state from the average height value of the five feature points to obtain the net height of the paper stack. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

[0125] In this application, when calculating the three-dimensional spatial pose of a stack of paper based on three-dimensional point clouds, the RANSAC plane fitting algorithm is first applied to the three-dimensional point clouds.

[0126] The specific execution process of the RANSAC algorithm is as follows: three non-collinear points are randomly selected from the 3D point cloud to determine a candidate plane equation. The distance from all points to the candidate plane is calculated. Points with a distance less than a preset inlier threshold are marked as inliers. The random selection and inlier counting process is repeated for a preset number of iterations, such as 1000 times. The candidate plane with the most inliers is selected as the optimal plane equation for the upper surface of the paper stack.

[0127] Compared to the ordinary least squares method, the RANSAC algorithm is more robust to noise points and outliers, and can still accurately fit the principal plane of the upper surface of the paper stack even when there is local warping or foreign objects on the paper stack surface.

[0128] Based on the optimal plane equation, with the normal vector n=(n_x,n_y,n_z) and the reference plane normal vector n0=(0,0,1), the pitch angle is pitch=arctan(n_x / n_z) and the yaw angle is yaw=arctan(n_y / n_z).

[0129] More specifically, the height values ​​of five feature points—the four corner points and the center point at the top of the paper stack—are extracted, and the mean, standard deviation, and maximum value of the five height values ​​are calculated to quantify the surface flatness of the paper stack.

[0130] The process of estimating thickness information is as follows: subtract the height of the pre-calibrated reference surface in the paperless state from the average height value of the five feature points to obtain the net height of the paper stack, and then obtain the actual physical thickness of the paper stack based on the net height of the paper stack and the pre-calibrated thickness coefficient.

[0131] To illustrate with a specific numerical example: Assuming the average height of the five feature points is 25.3 mm and the height of the reference plane is 2.1 mm, the net height of the paper stack is 23.2 mm; assuming the thickness coefficient is 0.95 and the zero-point offset is 0.2 mm, the actual physical thickness is 23.2 × 0.95 + 0.2 ≈ 22.24 mm.

[0132] Based on the thickness of a standard A4 sheet of paper being approximately 0.1 mm, the current paper stack can be estimated to contain approximately 222 sheets. This value can be used to determine whether to trigger a paper replenishment or paper retrieval process, while also providing precise parameters for the opening distance and pressing stroke of the composite clamp.

[0133] In this application, principal plane parameters are robustly extracted from noisy point clouds through RANSAC plane fitting, and surface flatness is quantified by five-point height statistics and the visual height is converted into physical thickness through thickness coefficient, providing accurate quantitative data for subsequent anomaly detection and fixture parameter setting.

[0134] Optionally, the abnormal state of the paper stack is determined, and a corresponding processing strategy is determined, including: When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features exceeds the specified thickness threshold, and the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold, the abnormal state is determined to be a batch staggered stacking state, and the processing strategy is to calculate the attitude compensation amount and directly perform gripping and handling on the paper stack. When the curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or when the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold, the abnormal state is determined to be a batch folding and warping state, and the processing strategy is to grab all the paper stacks and transport them to the waste paper area. When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features does not exceed the specified thickness threshold, the abnormal state is determined to be a single sheet staggered stacking state. The processing strategy is to control the composite clamp to gently push the edge of the paper stack along the protruding direction to put the misaligned paper back into place. After the pushing is completed, image acquisition and visual analysis are re-executed to confirm the alignment effect. After confirmation, the paper stack is gripped and transported. When the area of ​​the abnormal surface region is less than the proportion of the total area of ​​the paper pile to a preset area ratio threshold, and the local height change value is greater than the preset change threshold, the abnormal state is determined to be a single sheet folding and warping state. The processing strategy is to grab and transport the top preset number of papers containing the abnormal paper to the waste paper area.

[0135] In this application, when the convex defect features of the two-dimensional contour of the paper stack exist and the thickness of the corresponding part of the convex defect exceeds a specified thickness threshold, such as 0.5 mm, and the standard deviation of the surface flatness is not greater than a preset flatness threshold, such as 2 mm, the abnormal state is determined to be a batch staggered stacking state. The processing strategy is to calculate the attitude compensation amount and directly perform gripping and handling on the paper stack. The attitude compensation amount is the pitch angle and yaw angle calculated above. The main control unit rotates the end of the composite fixture by the corresponding angle to make it parallel to the upper surface of the paper stack before performing normal gripping.

[0136] When the curvature characteristic value of more than a preset proportion, such as 30% of the paper stack surface area, exceeds the preset curvature threshold, or when the aspect ratio of the binary mask deviates from the nominal value by more than a preset deviation threshold, such as 15%, the abnormal state is determined to be a batch folding and warping state. The handling strategy is to grab the entire paper stack and transport it to the waste paper area. When the main control unit executes this strategy, it changes the target placement area from the normal target position to the coordinates of the waste paper area.

[0137] When a convex defect exists and the thickness of the corresponding part of the convex defect does not exceed the specified thickness threshold, the abnormal state is determined to be a single sheet staggered stacking state. The handling strategy is to control the composite clamp to gently push the edge of the paper stack along the protruding direction to put the misaligned paper back into place. After the push is completed, image acquisition and visual analysis are re-executed to confirm the alignment effect. After confirmation, the paper stack is gripped and transported.

[0138] If alignment cannot be confirmed after a preset number of push-and-align operations, such as 3, the main control unit generates a prompt for manual intervention.

[0139] When the area of ​​the abnormal surface region accounts for less than the proportion of the total area of ​​the paper stack, such as 10%, and the local height change value is greater than the preset change threshold, such as 3mm, the abnormal state is determined to be a single sheet folding and warping state. The processing strategy is to grab a preset number of sheets of paper, such as 5 to 10 sheets, from the top layer of the paper stack containing the abnormal paper and transport them to the waste paper area.

[0140] In this application, by binding the judgment conditions with the processing strategies one by one, the main control unit can immediately select the optimal processing path after determining the abnormal state, without the need for an additional decision-making process, thereby improving the system's response speed and processing efficiency.

[0141] Optionally, the step of calculating the normal vector and geometric center position of the upper surface of the paper stack based on the three-dimensional spatial orientation of the paper stack further includes: The two-dimensional outline of the paper stack is shrunk, and a safe gripping area is generated inside the two-dimensional outline of the paper stack. The shrunk process reserves a preset edge margin. The clamping direction of the lateral clamping mechanism is determined based on the direction of the minimum circumscribed rectangle of the two-dimensional contour of the paper stack. The attitude compensation of the end of the composite fixture is performed based on the pitch and yaw angles in the three-dimensional spatial attitude of the paper stack, so that the end of the composite fixture is parallel to the upper surface of the paper stack. Based on the geometric center position, the safe gripping area, the gripping direction, and the posture of the gripper end after posture compensation, the angles of each joint of the multi-degree-of-freedom robotic arm are calculated using the inverse kinematics algorithm.

[0142] In this application, after calculating the surface normal vector and geometric center position based on the three-dimensional spatial attitude of the paper stack, the main control unit also needs to perform the following fine-grained planning process.

[0143] First, the two-dimensional outline of the paper stack is shrunk to create a safe gripping area inside the outline. The preset edge margin reserved for the shrunk process is set to, for example, 8 to 12 mm.

[0144] The inward shrinkage process involves shrinking the two-dimensional outline of the paper stack inward along the normal direction by a margin distance, generating an inward shrinkage outline that is slightly smaller than the original outline. The area enclosed by the inward shrinkage outline is the safe gripping area. This design ensures that the composite clamp will not cause the edge paper to curl or be damaged when gripping the paper stack.

[0145] Then, the clamping direction of the lateral clamping mechanism is determined based on the direction of the minimum circumscribed rectangle of the two-dimensional profile of the paper stack. The minimum circumscribed rectangle is the rectangle with the smallest area that can completely enclose the two-dimensional profile of the paper stack. The orientation of its long and short sides determines the direction from which the left and right pressing plate jaws should clamp. Usually, clamping along the short side is chosen to minimize the jaw travel and ensure the most stable clamping.

[0146] More specifically, based on the pitch and yaw angles in three-dimensional space, attitude compensation is performed on the end effector of the composite gripper to make the end effector parallel to the upper surface of the paper stack. The main control unit calculates the angles that the end effector needs to rotate around the x and y axes based on the pitch and yaw angles, and adds the rotation amount to the target attitude of the robotic arm end effector.

[0147] Finally, based on the geometric center position, safe gripping area, gripping direction, and compensated gripper end posture, the angles of each joint of the multi-degree-of-freedom robotic arm are calculated using an inverse kinematics algorithm.

[0148] The input to the inverse kinematics algorithm is the target position (x, y, z) and target pose (roll, pitch, yaw) of the end of the composite fixture in the world coordinate system, and the output is the angle value (θ1, θ2, θ3, θ4, θ5, θ6) that each joint (e.g., 6 rotary joints) should reach.

[0149] For a 6-DOF robotic arm, the inverse kinematics algorithm can be an analytical solution based on DH (Denavit-Hartenberg) parameters or an iterative numerical solution based on the Jacobian matrix.

[0150] In this embodiment, by designing a safety margin for contour shrinkage, selecting the optimal clamping angle in the direction of the minimum circumscribed rectangle, and automatically compensating based on the actual paper stack posture, the composite fixture is ensured to approach the paper stack in the optimal posture, thereby improving the success rate of gripping and reducing disturbance to the paper stack.

[0151] Optionally, the planned obstacle avoidance trajectory from the current position to the grab position and then to the release position includes: Smooth and continuous joint space trajectories are generated using polynomial interpolation or trapezoidal velocity curve planning. A fast exploration random tree algorithm is used to perform local dynamic obstacle avoidance by combining a pre-stored environment map; The obstacle avoidance trajectory is subject to segmented speed control: deceleration is performed when approaching the paper stack, uniform speed is performed when transporting the paper stack, and slow speed is performed when placing the paper stack.

[0152] In this application, when planning the obstacle avoidance trajectory from the current position to the grab position and then to the release position, it is first necessary to generate a smooth and continuous joint space trajectory.

[0153] Quintic polynomial interpolation can be used, with the position, velocity, and acceleration of the starting and ending points as boundary conditions to solve for the six coefficients of the quintic polynomial, thereby generating a joint motion trajectory with continuous position, velocity, and acceleration. Alternatively, trapezoidal velocity curve planning can be used, dividing the joint motion into three stages: acceleration, uniform velocity, and deceleration. Both methods can ensure the smoothness of the joint motion and avoid abrupt accelerations or impact forces.

[0154] Based on a smooth trajectory, a fast exploration random tree algorithm is used to perform local dynamic obstacle avoidance in conjunction with a pre-stored environment map.

[0155] The pre-stored environment map is a three-dimensional spatial information containing fixed obstacles such as the printer body, desktop, and stand, obtained through 3D modeling or laser scanning during the system installation and debugging phase.

[0156] The fast-exploration random tree algorithm starts from the current position of the robotic arm, randomly samples feasible motion nodes in the joint space or Cartesian space, and gradually builds a search tree extending towards the target position. At the same time, it checks whether the robotic arm configuration corresponding to each edge collides with obstacles in the environment map, and finally outputs a collision-free motion path.

[0157] In this application, segmented speed control of the obstacle avoidance trajectory specifically refers to: decelerating during the approach phase to reduce airflow disturbance; maintaining a constant speed during the paper stack handling phase to reduce inertial effects; and moving at a slow speed during the paper stack placement phase to ensure accurate placement. The specific speed parameters for each phase can be configured according to the actual operating environment and the state of the paper stack.

[0158] In this application, polynomial interpolation or trapezoidal velocity curves ensure smooth motion, fast exploration random tree algorithm ensures motion safety, and segmented velocity control ensures the matching of motion characteristics at each stage with operational requirements. The combination of these three elements forms a reliable motion trajectory planning scheme.

[0159] Optionally, the lateral clamping mechanism and the upper and lower fixing mechanism that drive the composite fixture work together to establish a three-dimensional physical constraint on the paper stack, including: First, control the left and right pressing plate jaws of the lateral clamping mechanism to close synchronously to apply lateral pre-tightening force, and then control the top fixing plate and bottom fixing plate of the upper and lower fixing mechanism to apply vertical pressing force. After establishing the three-dimensional physical constraints, the success of pressing the upper and lower fixing mechanisms is determined by pressure threshold detection. During the handling process, the torque value of each joint of the multi-degree-of-freedom robotic arm is detected in real time. When the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and the multi-degree-of-freedom robotic arm is controlled to retreat along the original trajectory to a safe position. The release includes: after reaching the target placement area, first releasing the top fixed plate and the bottom fixed plate, and then releasing the left and right pressing plate clamps.

[0160] In this application, when the lateral clamping mechanism and the upper and lower fixing mechanism of the drive composite fixture work together to establish three-dimensional physical constraints, they are executed in the order of lateral first and then upper and lower. First, the left and right pressing plate jaws are controlled to close synchronously to apply lateral preload, and then the top fixing plate and the bottom fixing plate are controlled to apply vertical pressing force.

[0161] After the constraints are established, the system determines whether the pressing of the upper and lower fixing mechanisms is successful by detecting pressure thresholds. Pressure sensors are installed on the top fixing plate and / or the bottom fixing plate. When the pressing force detected by the pressure sensor reaches the preset pressure threshold, the pressing is considered successful.

[0162] If the pressing force value does not reach the threshold, it may be due to the paper stack being too thin or the clamp position being misaligned. The main control unit will fine-tune the pressing stroke of the upper and lower fixing mechanisms and re-detect until the pressure requirement is met or the maximum number of retries is reached, at which point an alarm will be triggered.

[0163] During the handling process, the system monitors the torque values ​​of each joint of the multi-degree-of-freedom robotic arm in real time. When the torque value exceeds the preset normal range, it is determined that there is unexpected resistance, and the robotic arm is controlled to retract along the original trajectory to a safe position.

[0164] Upon reaching the target placement area, the release operation is performed in the order of releasing the top and bottom plates first, followed by releasing the side plates. First, release the top and bottom fixed plates, then release the left and right clamping plates to ensure that the paper stack falls smoothly into the target position under its own weight.

[0165] In this embodiment, a complete closed-loop control system is formed, encompassing constraint establishment, constraint confirmation, transport monitoring, and orderly release. The dual safeguards of pressure threshold detection and torque anomaly monitoring ensure the reliability and safety of each stage of grasping, transporting, and releasing.

[0166] Optionally, when the target operating area is the blank paper storage area, the target placement area is the printer paper feeding mechanism; when the target operating area is the printer paper output mechanism, the target placement area is the printed product temporary storage area. When the target operating area is the printer's paper output mechanism, after the paper stack is transported and released, the process further includes: updating the batch count of completed printing tasks and recording the paper retrieval time and paper quantity.

[0167] Figure 7 The system workflow diagram provided for this invention is as follows: Figure 7 As shown, after power-on, the system first enters the initialization phase. The main control unit performs a power-on self-test to check whether each communication link is normal; the joints of the multi-degree-of-freedom robotic arm move sequentially to the preset zero position, and the composite gripper retracts to its initial configuration. The vision perception module sequentially activates three industrial cameras to perform initial scans of the blank paper storage area and the printer paper output mechanism, recording the reference plane height values ​​of each area. During this phase, the user neatly places a sufficient amount of blank paper into the blank paper storage area. After initialization is complete, the system enters normal operation.

[0168] After the printer starts executing a printing task normally, the main control unit continuously monitors two key indicators through the communication protocol between the printer host and the printer host: whether the current paper amount of the printer's paper feed mechanism is lower than the preset paper replenishment threshold, and whether the paper amount of the printer's paper output mechanism has reached the preset paper pick-up threshold.

[0169] Meanwhile, the main control unit also monitors for equipment malfunction signals or user manual intervention commands. During periods when neither of these indicators is triggered, the robotic arm remains stationary in the standby position, and the system does not send any operation commands.

[0170] When the paper level in the paper feeding mechanism falls below the replenishment threshold, the system triggers an automatic paper replenishment process. The main control unit instructs the first industrial camera to acquire images of the blank paper storage area, and then sequentially performs semantic segmentation to extract the binary mask of the paper stack, 3D point cloud back projection, plane fitting to calculate the 3D spatial pose, five-point height extraction to quantify surface flatness, and estimation of the physical thickness of the paper stack based on the thickness coefficient.

[0171] After visual analysis, the system proceeds to the anomaly detection stage. If the paper stack is severely disordered, the main control unit first drives the composite gripper to lightly press the top of the paper stack and applies small-amplitude high-frequency vibration to perform pre-sorting, then re-scans and analyzes. If it is determined to be a batch of staggered stacks, the attitude compensation is calculated and the system directly enters the gripping stage. If it is a single sheet of staggered stacks, edge pushing is performed and alignment is reconfirmed before gripping. If it is a batch of folded and warped stacks, the entire paper stack is transferred to the waste paper area. If it is a single sheet of folded and warped stacks, the top layer of a preset number of sheets is discarded along with the paper and sent to the waste paper area. If the paper stack is normal, the system directly enters the gripping planning stage.

[0172] After the grasping plan is completed, the robotic arm moves at a reduced speed along the obstacle avoidance trajectory to approach the stack of paper. After reaching the target grasping pose, it establishes three-dimensional physical constraints according to the sequence of first side to back and up and down, and confirms success through pressure threshold detection.

[0173] The robotic arm then moves at a constant speed to transport the paper stack to the printer's paper feed mechanism. Upon arrival, it moves slowly to precisely align with the tray position and releases the paper stack in a sequence of first releasing from the top and bottom, then releasing laterally. After release, the robotic arm returns to the standby position, and the paper replenishment process ends.

[0174] When the paper level in the paper output mechanism reaches the paper-fetching threshold, the system triggers an automatic paper-fetching process. The main control unit instructs the second industrial camera to acquire images of the paper output mechanism, performing the same visual analysis and anomaly detection process as the paper replenishment process. Because the printed paper has been passed through the printer's internal rollers multiple times, the paper pile in the output area is more prone to misalignment or uneven edges. The main control unit selects the corresponding processing strategy based on the anomaly type and then proceeds with the grasping plan.

[0175] The robotic arm moves along an obstacle avoidance trajectory to the paper output mechanism to complete the grabbing, then transports the stack of printed paper to the designated cell in the printed paper storage area and releases it smoothly. After release, the main control unit updates the batch count and records the paper retrieval time and quantity in the operation log. The robotic arm returns to the standby position, and the paper retrieval process ends.

[0176] If the main control unit detects a device malfunction or receives a manual pause command, the system immediately stops the current operation. The robotic arm retracts along its original trajectory to a safe position and enters standby mode, awaiting troubleshooting or a reset command from the user. When all printing tasks are completed and no new tasks are scheduled, the system switches to a low-power standby mode, and any remaining blank paper and printed products are collected and disposed of by staff.

[0177] In this embodiment, the system realizes a fully automated closed loop from status monitoring, visual perception, anomaly handling, grasping planning to physical handling, enabling unattended operation in large-volume continuous printing scenarios, significantly reducing the frequency of manual intervention and improving printing efficiency.

[0178] Optionally, the method further includes: When the binary mask of the paper stack indicates that the paper stack is in a severely scattered state, the multi-degree-of-freedom robotic arm is controlled to drive the composite clamp to lightly press the top of the paper stack and apply a small-amplitude high-frequency vibration so that the paper naturally aligns under the action of gravity and friction. After alignment is complete, image acquisition is performed again to obtain an updated image of the stack of paper.

[0179] In this application, in actual operating environments, paper stacks may be severely scattered due to improper placement by users, transportation vibrations, or other external forces. In such cases, directly determining the abnormal state and planning the handling process often fails to yield effective results. To address this issue, the control method adds a pre-sorting step after visual analysis and before abnormal determination.

[0180] When the binary mask of the paper stack indicates that the paper stack is in a severely disordered state, such as the outline shape presented by the binary mask deviating significantly from the standard rectangle, having multiple disconnected areas, having jagged edges on the outline, or having an outline area much smaller than the expected value, the main control unit controls the multi-degree-of-freedom robotic arm to drive the composite clamp to lightly press the top of the paper stack and apply small-amplitude high-frequency vibration.

[0181] Light pressure refers to controlling the top fixed plate to press down on the upper surface of the paper stack with a small force, such as between 0.5N and 2N, which is enough to restrain the paper but will not cause damage.

[0182] Small-amplitude high-frequency vibration refers to controlling the end effector of a robotic arm to perform small-amplitude, high-frequency, reciprocating motion along the horizontal or vertical plane while maintaining a light pressure state. Under this vibration, scattered papers gradually return to their correct positions under the combined action of gravity and friction between adjacent papers.

[0183] After alignment, the main control unit re-triggers the visual perception module to acquire images of the paper stack, obtaining an updated image of the paper stack. Subsequently, semantic segmentation, 3D pose calculation, and anomaly detection are re-performed based on the updated image. These processes ensure that subsequent analysis and retrieval operations are based on the more organized state information of the paper stack after pre-sorting.

[0184] In this application, the ability to handle extreme paper stack conditions is expanded by introducing a pre-sorting stage, further reducing the frequency of manual intervention required due to poor initial paper stack conditions, and improving the robustness of the entire system under complex actual working conditions.

[0185] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 8 As shown, the electronic device may include a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a control method, which includes: real-time monitoring of the paper level status of the printer's paper feed mechanism and the printer's paper output mechanism; when the paper level of the printer's paper feed mechanism is lower than a preset paper replenishment threshold, or the paper level of the printer's paper output mechanism reaches a preset paper dispensing threshold, determining the current target operating area; The visual perception module is controlled to acquire images of the paper pile in the target operation area. The acquired images of the paper pile are semantically segmented to obtain a binary mask of the paper pile. The pixels in the binary mask of the paper pile are mapped to a three-dimensional point cloud. The three-dimensional spatial pose of the paper pile relative to the reference plane is calculated based on the three-dimensional point cloud. The thickness information of the paper pile is estimated based on the three-dimensional point cloud. Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, the abnormal state of the paper stack is determined, and the processing strategy corresponding to the abnormal state is determined. Based on the three-dimensional spatial posture of the paper stack, the normal vector and geometric center position of the upper surface of the paper stack are calculated. Combined with the thickness information of the paper stack, the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism are determined. The angles of each joint of the multi-degree-of-freedom robotic arm are calculated by inverse kinematics algorithm to obtain the target grasping pose, and the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position is planned. The multi-degree-of-freedom robotic arm is controlled to move along the obstacle avoidance trajectory to the target grasping position, and the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture are driven to work together to establish a three-dimensional physical constraint on the paper stack. The paper stack is then transported to the target placement area and released according to the processing strategy.

[0186] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0187] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being stored on a non-transitory computer-readable storage medium, the computer program being executed by a processor, the computer being able to execute the control methods provided by the above methods, the method including: real-time monitoring of the paper quantity status of the printer paper feed mechanism and the printer paper output mechanism, and determining the current target operation area when the paper quantity of the printer paper feed mechanism is lower than a preset paper replenishment threshold, or the paper quantity of the printer paper output mechanism reaches a preset paper take-out threshold; The visual perception module is controlled to acquire images of the paper pile in the target operation area. The acquired images of the paper pile are semantically segmented to obtain a binary mask of the paper pile. The pixels in the binary mask of the paper pile are mapped to a three-dimensional point cloud. The three-dimensional spatial pose of the paper pile relative to the reference plane is calculated based on the three-dimensional point cloud. The thickness information of the paper pile is estimated based on the three-dimensional point cloud. Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, the abnormal state of the paper stack is determined, and the processing strategy corresponding to the abnormal state is determined. Based on the three-dimensional spatial posture of the paper stack, the normal vector and geometric center position of the upper surface of the paper stack are calculated. Combined with the thickness information of the paper stack, the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism are determined. The angles of each joint of the multi-degree-of-freedom robotic arm are calculated by inverse kinematics algorithm to obtain the target grasping pose, and the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position is planned. The multi-degree-of-freedom robotic arm is controlled to move along the obstacle avoidance trajectory to the target grasping position, and the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture are driven to work together to establish a three-dimensional physical constraint on the paper stack. The paper stack is then transported to the target placement area and released according to the processing strategy.

[0188] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the control method provided by the above methods, the method comprising: real-time monitoring of the paper quantity status of the printer's paper feed mechanism and the printer's paper output mechanism, and determining the current target operating area when the paper quantity of the printer's paper feed mechanism is lower than a preset paper replenishment threshold, or when the paper quantity of the printer's paper output mechanism reaches a preset paper take-up threshold; The visual perception module is controlled to acquire images of the paper pile in the target operation area. The acquired images of the paper pile are semantically segmented to obtain a binary mask of the paper pile. The pixels in the binary mask of the paper pile are mapped to a three-dimensional point cloud. The three-dimensional spatial pose of the paper pile relative to the reference plane is calculated based on the three-dimensional point cloud. The thickness information of the paper pile is estimated based on the three-dimensional point cloud. Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, the abnormal state of the paper stack is determined, and the processing strategy corresponding to the abnormal state is determined. Based on the three-dimensional spatial posture of the paper stack, the normal vector and geometric center position of the upper surface of the paper stack are calculated. Combined with the thickness information of the paper stack, the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism are determined. The angles of each joint of the multi-degree-of-freedom robotic arm are calculated by inverse kinematics algorithm to obtain the target grasping pose, and the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position is planned. The multi-degree-of-freedom robotic arm is controlled to move along the obstacle avoidance trajectory to the target grasping position, and the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture are driven to work together to establish a three-dimensional physical constraint on the paper stack. The paper stack is then transported to the target placement area and released according to the processing strategy.

[0189] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0190] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic paper feeding and dispensing printer system, characterized in that, include: The system consists of a main control unit, a vision perception module, a robotic arm execution unit, and a material area management module. The material area management module includes: a blank paper storage area, a printer paper feeding mechanism, a printer paper output mechanism, and a printed product temporary storage area; The visual perception module includes an industrial camera, and is configured to: acquire images of the paper stack in the material area management module and perform visual analysis, and output the status information of the paper stack; the status information of the paper stack includes at least the three-dimensional spatial posture of the paper stack and the abnormal state of the paper stack; The robotic arm execution unit includes: a multi-degree-of-freedom robotic arm and a composite clamp mounted on the end of the multi-degree-of-freedom robotic arm; the composite clamp includes a lateral clamping mechanism for clamping from the side of the paper stack, and an upper and lower fixing mechanism for pressing and fixing the upper and lower surfaces of the paper stack. The main control unit is communicatively connected to both the visual perception module and the robotic arm execution unit, and the main control unit is configured as follows: When the paper quantity of the printer's paper feed mechanism is detected to be lower than the preset paper replenishment threshold, or the paper quantity of the printer's paper output mechanism reaches the preset paper dispensing threshold, the visual perception module is triggered to perform visual analysis on the paper pile in the current target operation area to obtain the status information of the paper pile. The target grasping pose is generated based on the three-dimensional spatial posture of the paper stack, and the corresponding execution command is generated based on the abnormal state of the paper stack. The robotic arm execution unit is driven to approach the paper stack according to the target grasping posture, and through the lateral clamping mechanism and the upper and lower fixing mechanism, a three-dimensional physical constraint on the paper stack is established in coordination. According to the execution command, the paper stack is transported and released from the target operation area to the target placement area.

2. The automatic paper feeding and dispensing printer system according to claim 1, characterized in that, The lateral clamping mechanism consists of a pair of symmetrically arranged left and right pressing plate grippers, which are configured to open and close synchronously to apply pre-tightening force from both sides of the paper stack for lateral positioning. The upper and lower fixing mechanisms include: a top fixing plate and a bottom fixing plate, which are used to apply vertical pressing force to the upper and lower surfaces of the paper stack, respectively. The mechanism for establishing the three-dimensional physical constraint is as follows: the main control unit is configured to first control the left and right pressing plate jaws to close synchronously to apply lateral preload, and then control the top fixed plate and the bottom fixed plate to apply the vertical pressing force, forming a three-dimensional physical constraint with the combined action of lateral limiting and upper and lower pressing.

3. The automatic paper feeding and dispensing printer system according to claim 1, characterized in that, The industrial cameras include: a first industrial camera deployed directly above the blank paper storage area, a second industrial camera deployed directly above the printer paper output mechanism, and a third industrial camera deployed at the end of the multi-degree-of-freedom robotic arm. The physical layout of the blank paper storage area, the printer paper feeding mechanism, the printer paper output mechanism, and the printed product temporary storage area are all marked in a unified system coordinate system.

4. The automatic paper feeding and dispensing printer system according to claim 1, characterized in that, The state information of the paper stack also includes: the two-dimensional outline information of the paper stack and the thickness information of the paper stack; the visual perception module is specifically used for: The collected images of the paper pile are input into a pre-trained semantic segmentation model to perform pixel-level semantic segmentation, obtain a binary mask of the paper pile, and extract the two-dimensional contour information of the paper pile based on the binary mask of the paper pile. The pixels within the binary mask of the paper stack are back-projected into a three-dimensional point cloud in the world coordinate system using a pre-calibrated camera intrinsic parameter matrix and installation height parameters. A plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. Based on the optimal plane equation, the pitch angle and yaw angle of the upper surface of the paper stack relative to the reference plane are calculated. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The pitch angle, yaw angle, and surface flatness of the paper stack are used as the three-dimensional spatial attitude of the paper stack. The net height of the paper stack is obtained by subtracting the height of the pre-calibrated reference surface in the paperless state from the average height value of the five feature points. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

5. The automatic paper feeding and dispensing printer system according to claim 4, characterized in that, The visual perception module is also used for: Convex hull analysis is performed on the binary mask of the paper stack to calculate the convex defect features between the two-dimensional contour information of the paper stack and the convex hull. Local curvature estimation is performed on the three-dimensional point cloud to obtain the curvature characteristic value of the paper stack surface; Based on the convex defect characteristics, the curvature characteristic value, and the surface flatness of the paper stack, the abnormal state of the paper stack is determined; the abnormal state includes at least one of the following: Batch staggered stacking state: The convex defect feature exists and the thickness of the corresponding part of the convex defect feature exceeds the specified thickness threshold, while the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold. Batch folding and warping state: The curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold. Single sheet staggered stacking state: The convex defect feature exists and the thickness of the portion corresponding to the convex defect feature does not exceed the specified thickness threshold; Single sheet folding and warping state: The proportion of the abnormal area on the surface to the total area of ​​the paper stack is less than a preset area proportion threshold, and the local height change value is greater than a preset change threshold.

6. The automatic paper feeding and dispensing printer system according to claim 5, characterized in that, The main control unit is also used for: When the abnormal state is the batch staggered stacking state, an attitude compensation grasping command is generated to drive the robotic arm execution unit to adjust the end of the composite fixture to be parallel to the upper surface of the paper stack before performing the grasping. When the abnormal state is the batch folding and warping state, a batch waste instruction is generated, driving the robotic arm execution unit to grab and transport the entire paper stack to the waste paper area. When the abnormal state is the single sheet staggered stacking state, an edge pushing command is generated, which drives the robotic arm execution unit to control the composite clamp to gently push the edge of the paper stack along the protruding direction to make the misaligned paper return to its position. After the pushing is completed, the visual perception module is triggered to re-execute the visual analysis to confirm the alignment effect. After the confirmation is passed, the gripping is performed. When the abnormal state is the single sheet folded and warped state, a local waste instruction is generated, which drives the robotic arm execution unit to grab and transport the top layer of the paper stack containing the abnormal paper to the waste paper area.

7. The automatic paper feeding and dispensing printer system according to claim 1, characterized in that, The main control unit is also configured to run a torque anomaly monitoring mechanism during the handling process performed by the robotic arm execution unit; the torque anomaly monitoring mechanism is used to detect the torque value of each joint of the multi-degree-of-freedom robotic arm in real time. When the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and an emergency retraction is triggered, controlling the multi-degree-of-freedom robotic arm to retract along the original trajectory to a safe position. The main control unit is also configured to perform segmented speed control on the robotic arm execution unit: performing deceleration motion when approaching the paper stack, performing uniform motion when transporting the paper stack, and performing slow motion when placing the paper stack.

8. A control method applied to the automatic paper feeding and dispensing printer system according to any one of claims 1 to 7, characterized in that, include: The paper level of the printer's paper feed mechanism and paper output mechanism is monitored in real time. When the paper level of the printer's paper feed mechanism is lower than the preset paper replenishment threshold, or the paper level of the printer's paper output mechanism reaches the preset paper take-out threshold, the current target operation area is determined. The visual perception module is controlled to acquire images of the paper pile in the target operation area. The acquired images of the paper pile are semantically segmented to obtain a binary mask of the paper pile. The pixels in the binary mask of the paper pile are mapped to a three-dimensional point cloud. The three-dimensional spatial pose of the paper pile relative to the reference plane is calculated based on the three-dimensional point cloud. The thickness information of the paper pile is estimated based on the three-dimensional point cloud. Based on the convex defect features of the two-dimensional contour of the paper stack and the curvature feature value of the paper stack surface, the abnormal state of the paper stack is determined, and the processing strategy corresponding to the abnormal state is determined. Based on the three-dimensional spatial posture of the paper stack, the normal vector and geometric center position of the upper surface of the paper stack are calculated. Combined with the thickness information of the paper stack, the opening distance of the lateral clamping mechanism of the composite fixture and the pressing stroke of the upper and lower fixing mechanism are determined. The angles of each joint of the multi-degree-of-freedom robotic arm are calculated by inverse kinematics algorithm to obtain the target grasping pose, and the obstacle avoidance motion trajectory from the current position to the grasping position and then to the release position is planned. The multi-degree-of-freedom robotic arm is controlled to move along the obstacle avoidance trajectory to the target grasping position, and the lateral gripping mechanism and the upper and lower fixing mechanism of the composite fixture are driven to work together to establish a three-dimensional physical constraint on the paper stack. The paper stack is then transported to the target placement area and released according to the processing strategy.

9. The control method according to claim 8, characterized in that, The step of calculating the three-dimensional spatial pose of the paper stack relative to a reference plane based on the three-dimensional point cloud, and estimating the thickness information of the paper stack based on the three-dimensional point cloud, includes: The RANSAC plane fitting algorithm is applied to the three-dimensional point cloud to solve the optimal plane equation of the upper surface of the paper stack. The pitch and yaw angles of the upper surface of the paper stack relative to the reference plane are calculated based on the optimal plane equation. The height values ​​of five feature points, including the four corner points and the center point at the top of the paper stack, are extracted. The mean, standard deviation, and maximum value of the height values ​​of the five feature points are calculated to quantify the surface flatness of the paper stack. The estimation of the thickness information of the paper stack based on the three-dimensional point cloud includes: Subtract the pre-calibrated reference surface height in the paperless state from the average height value of the five feature points to obtain the net height of the paper stack. The thickness information of the paper stack is obtained based on the net height of the paper stack and the pre-calibrated thickness coefficient.

10. The control method according to claim 9, characterized in that, Determine the abnormal state of the paper stack and determine the corresponding processing strategy, including: When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features exceeds the specified thickness threshold, and the standard deviation of the surface flatness of the paper stack is not greater than the preset flatness threshold, the abnormal state is determined to be a batch staggered stacking state, and the processing strategy is to calculate the attitude compensation amount and directly perform gripping and handling on the paper stack. When the curvature characteristic value of the surface area of ​​the paper stack exceeds a preset curvature threshold, or when the aspect ratio of the outline of the binary mask of the paper stack deviates from the nominal value by more than a preset deviation threshold, the abnormal state is determined to be a batch folding and warping state, and the processing strategy is to grab all the paper stacks and transport them to the waste paper area. When the two-dimensional contour of the paper stack has convex defect features, and the thickness of the part corresponding to the convex defect features does not exceed the specified thickness threshold, the abnormal state is determined to be a single sheet staggered stacking state. The processing strategy is to control the composite clamp to gently push the edge of the paper stack along the protruding direction to put the misaligned paper back into place. After the pushing is completed, image acquisition and visual analysis are re-executed to confirm the alignment effect. After confirmation, the paper stack is gripped and transported. When the area of ​​the abnormal surface region is less than the proportion of the total area of ​​the paper pile to a preset area ratio threshold, and the local height change value is greater than the preset change threshold, the abnormal state is determined to be a single sheet folding and warping state. The processing strategy is to grab and transport the top preset number of papers containing the abnormal paper to the waste paper area.

11. The control method according to claim 8, characterized in that, The calculation of the normal vector and geometric center position of the upper surface of the paper stack based on the three-dimensional spatial orientation of the paper stack also includes: The two-dimensional outline of the paper stack is shrunk, and a safe gripping area is generated inside the two-dimensional outline of the paper stack. The shrunk process reserves a preset edge margin. The clamping direction of the lateral clamping mechanism is determined based on the direction of the minimum circumscribed rectangle of the two-dimensional contour of the paper stack. The attitude compensation of the end of the composite fixture is performed based on the pitch and yaw angles in the three-dimensional spatial attitude of the paper stack, so that the end of the composite fixture is parallel to the upper surface of the paper stack. Based on the geometric center position, the safe gripping area, the gripping direction, and the posture of the gripper end after posture compensation, the angles of each joint of the multi-degree-of-freedom robotic arm are calculated using the inverse kinematics algorithm.

12. The control method according to claim 8, characterized in that, The planned obstacle avoidance trajectory from the current position to the grab position and then to the release position includes: Smooth and continuous joint space trajectories are generated using polynomial interpolation or trapezoidal velocity curve planning. A fast exploration random tree algorithm is used to perform local dynamic obstacle avoidance by combining a pre-stored environment map; The obstacle avoidance trajectory is subject to segmented speed control: deceleration is performed when approaching the paper stack, uniform speed is performed when transporting the paper stack, and slow speed is performed when placing the paper stack.

13. The control method according to claim 8, characterized in that, The lateral clamping mechanism and the upper and lower fixing mechanism that drive the composite clamp work together to establish a three-dimensional physical constraint on the paper stack, including: First, control the left and right pressing plate jaws of the lateral clamping mechanism to close synchronously to apply lateral pre-tightening force, and then control the top fixing plate and bottom fixing plate of the upper and lower fixing mechanism to apply vertical pressing force. After establishing the three-dimensional physical constraints, the success of pressing the upper and lower fixing mechanisms is determined by pressure threshold detection. During the handling process, the torque value of each joint of the multi-degree-of-freedom robotic arm is detected in real time. When the torque value exceeds the preset normal range, it is determined that there is unexpected resistance and the multi-degree-of-freedom robotic arm is controlled to retreat along the original trajectory to a safe position. The release includes: after reaching the target placement area, first releasing the top fixed plate and the bottom fixed plate, and then releasing the left and right pressing plate clamps.

14. The control method according to claim 8, characterized in that, When the target operating area is the blank paper storage area, the target placement area is the printer paper feeding mechanism; when the target operating area is the printer paper output mechanism, the target placement area is the printed product temporary storage area. When the target operating area is the printer's paper output mechanism, after the paper stack is transported and released, the process further includes: updating the batch count of completed printing tasks and recording the paper retrieval time and paper quantity.

15. The control method according to claim 8, characterized in that, The method further includes: When the binary mask of the paper stack indicates that the paper stack is in a severely scattered state, the multi-degree-of-freedom robotic arm is controlled to drive the composite clamp to lightly press the top of the paper stack and apply a small-amplitude high-frequency vibration so that the paper naturally aligns under the action of gravity and friction. After alignment is complete, image acquisition is performed again to obtain an updated image of the stack of paper.

16. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the control method as described in any one of claims 8 to 15.

17. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the control method as described in any one of claims 8 to 15.

18. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the control method as described in any one of claims 8 to 15.

Images