Path planning method and full-automatic carrier tape braiding packaging equipment

Abstract

The invention discloses a path planning method and full-automatic carrier tape braiding packaging equipment, and relates to the technical field of automatic packaging. The method comprises the following steps: acquiring an initial position coordinate of a mechanical arm and identifying a target position of a to-be-grabbed semiconductor device; constructing a feasible position set, wherein the feasible position set comprises all candidate positions which can be safely reached and have no collision risk; traversing the set, calculating the distance from the current position to each candidate position, and selecting the nearest position as a next target; removing the data from the set and updating the set; and when the mechanical arm reaches the target position, grabbing is executed, the final position is recorded, then the position is set as the initial position of the next mechanical arm, and the steps are repeated till all tasks are completed or the target position is empty. When the position of the device in the vibration disc changes or a new device capable of being grabbed appears, the path can be automatically re-planned, the grabbing strategy can be automatically adjusted, invalid movement can be reduced, the work period is shortened, and the production efficiency of the full-automatic carrier tape braiding packaging equipment is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automatic packaging technology, in particular to a path planning method and a full-automatic tape carrier package (TCP) packaging equipment. BACKGROUND

[0002] In order to cope with the changes in the market demand for semiconductor device packaging, a full-automatic TCP packaging equipment has emerged. This kind of equipment integrates mechanical transmission, optical detection, precision control and other technologies to realize the full-process automation operation from device feeding, positioning and recognition, grabbing and placing to sealing and packaging. The performance of the mechanical arm as the execution mechanism of the packaging equipment directly determines the overall efficiency of the equipment. Although the early single mechanical arm system can complete the basic grabbing and placing task, the processing capacity of the single execution unit becomes a constraint factor when facing mass production demand. Therefore, the multi-robot collaborative working mode gradually becomes the industry development trend. By configuring multiple mechanical arms for parallel operation, the processing speed and production efficiency of the equipment can be significantly improved.

[0003] However, the introduction of the multi-robot system also brings new technical challenges. When multiple robots operate collaboratively in the same workspace, how to reasonably allocate tasks, plan motion paths and avoid mutual interference becomes a key technical problem. In particular, in the design scheme of the mechanical arms sharing a mobile platform, the motion of each mechanical arm will affect each other, and the traditional independent path planning method is no longer applicable. SUMMARY

[0004] The purpose of the present application is to provide a path planning method and a full-automatic TCP packaging equipment to solve the problems raised in the background art.

[0005] In a first aspect, an embodiment of the present application provides a path planning method applied to multiple mechanical arms of a full-automatic TCP packaging equipment. The multiple mechanical arms are fixed to the same base and sequentially grab semiconductor devices in a vibration tray. The method comprises: obtaining the starting position coordinates of the current mechanical arm and identifying the target position coordinate set of all semiconductor devices to be grabbed; constructing a feasible position set of the current mechanical arm, the feasible position set including all candidate positions that can be safely reached without collision risk; traversing the feasible position set, calculating the distance from the current position of the current mechanical arm to each candidate position, and selecting the nearest candidate position as the next moving target position of the current mechanical arm; removing the next moving target position from the feasible position set and updating the feasible position set; the current mechanical arm performs the grabbing task when reaching the target position and records the final position after completing the grabbing task; the final position is set as the starting position of the next mechanical arm to be actuated, and the above steps are repeated until all mechanical arms complete their grabbing tasks or the target position coordinate set is empty.

[0006] With reference to the first aspect, in some implementations of the first aspect, the traversing the set of feasible positions, calculating distances from the current position of the current robot arm to each candidate position, and selecting a candidate position closest to the current robot arm as a next movement target position of the current robot arm comprises: obtaining a current position coordinate of the base as a reference point of the current robot arm position; calculating a base position offset after a previous robot arm action based on a predicted movement trajectory of the previous robot arm and an expected position after completion of the task, to obtain a base displacement compensation factor, which represents a position difference vector between an actual position of the base after the previous robot arm completes the grasping task and an initial position of the base before the current robot arm starts to move; superimposing the base displacement compensation factor on the current robot arm position to obtain a compensated starting position of the robot arm; calculating straight-line distances from the compensated starting position of the robot arm to each candidate position using the Euclidean distance formula; and selecting a candidate position with the shortest straight-line distance from the compensated starting position of the robot arm as the next movement target position of the current robot arm.

[0007] With reference to the first aspect, in some implementations of the first aspect, the traversing the set of feasible positions, calculating distances from the current position of the current robot arm to each candidate position, and selecting a candidate position closest to the current robot arm as a next movement target position of the current robot arm comprises: performing multi-dimensional evaluation of each candidate position in the set of feasible positions, which includes movement distance evaluation, obstacle avoidance complexity evaluation, and base displacement impact evaluation; calculating straight-line movement distances from the current position of the current robot arm to each candidate position using the Euclidean distance formula, and converting the distance values into movement distance evaluation scores after normalization; analyzing the number of obstacles to be avoided and the length of the detour path in the movement path from the current position of the current robot arm to each candidate position, to calculate the obstacle avoidance complexity evaluation score; calculating the base displacement impact evaluation score based on the expected displacement of the base caused by the execution of the task of the previous robot arm; and calculating a comprehensive evaluation score of each candidate position based on the movement distance evaluation score, the obstacle avoidance complexity evaluation score, and the base displacement impact evaluation score using weighted summation, and selecting a candidate position with the highest comprehensive evaluation score as the next movement target position of the current robot arm.

[0008] With reference to the first aspect, in some implementations of the first aspect, the constructing a set of feasible positions of the current robot arm comprises: establishing a working area boundary constraint condition, defining an X-Y plane range of a movable base of the robot arm and a Z-axis working height of the robot arm; setting a set of obstacle avoidance regions, including a set of coordinates of obstacle regions inside the device that need to be avoided, to ensure the safety of the movement path of the robot arm; calculating a feasible detour path when a straight-line path from the current position of the current robot arm to the target position intersects with an obstacle; and including all candidate positions that can be safely reached and have no collision risk into the set of feasible positions.

[0009] With reference to the first aspect, in some implementations of the first aspect, the current robot arm performs a grabbing task when reaching the target position, and records the final position after the grabbing task is completed, comprising: controlling the current robot arm to move to the target position and perform a standard grabbing action sequence of lowering-grabbing-raising; recording the final coordinate data and timestamp information of the base after the grabbing task is completed, for subsequent setting of the starting position of the robot arm.

[0010] With reference to the first aspect, in some implementations of the first aspect, the method further comprises: monitoring the running state of the robot arm in real time, and when a mechanical failure or communication interruption is detected, dynamically reassigning the assignment task of the failed robot arm to other normal robot arms; when the displacement of the base exceeds the preset compensation range, triggering a re-calibration operation; when the position of the semiconductor device to be grabbed changes or the state is abnormal, updating the set of feasible positions and recalculating the optimal candidate position.

[0011] With reference to the first aspect, in some implementations of the first aspect, after setting the final position as the starting position of the next robot arm to be operated, the method further comprises: tracking the work progress of each robot arm through the task assignment matrix and the device grabbing state; when the robot arms all complete grabbing and reach the release position, triggering a preparation signal for a unified rotation release action; and performing the unified rotation release action to simultaneously release the grabbed semiconductor devices into the carrier tape.

[0012] In a second aspect, an embodiment of the present application provides a full-automatic tape carrier package (TCP) packaging device, which comprises: a rack fixedly arranged and supporting the overall device structure; a plurality of mechanical arms fixed to a movable base connected with the rack and movable in a plane, each mechanical arm being configured with an end effector for sequentially grabbing semiconductor devices; a device supply device comprising a vibrating disc for providing semiconductor devices to be grabbed and making the semiconductor devices in a distributed state; a detection device comprising a visual recognition module and a position sensor, the visual recognition module being used for identifying a target position coordinate set of all semiconductor devices to be grabbed in the vibrating disc, and the position sensor being used for acquiring a starting position coordinate of the mechanical arm in real time; and a control system connected with the plurality of mechanical arms and the detection device, comprising: a position coordinate processing module for receiving and processing the target position coordinate set and the starting position coordinate acquired by the detection device; a feasible position set construction module for constructing, based on a working area boundary and an obstacle avoidance requirement, a feasible position set comprising all candidate positions that can be safely reached and have no collision risk for the current mechanical arm; a distance calculation and optimization selection module for traversing the feasible position set, calculating distances from a current position of the current mechanical arm to each candidate position, and selecting, based on a greedy algorithm, a candidate position closest to the current position as a next moving target position; a dynamic set updating module for removing the next moving target position from the feasible position set and updating available candidate positions of the feasible position set in real time; a task execution control module for controlling the current mechanical arm to move to the target position and execute a grabbing task, and recording a final position of the movable base after the grabbing task is completed; a mechanical arm coordination switching module for setting the final position as a starting position of a next mechanical arm to be actuated, so as to realize sequential coordination control among the plurality of mechanical arms; and a loop control module for repeatedly executing the above position planning and task control processes until all the mechanical arms complete respective grabbing tasks or the target position coordinate set is empty.

[0013] In conjunction with the second aspect, in some implementations of the second aspect, the movable base is driven by a drive device to move in the XY plane, the drive device including an X-axis stepper motor and a Y-axis stepper motor; each robotic arm also includes: a telescopic device that extends and retracts the robotic arm in the Z-axis direction to achieve a grasping action; a rotating device installed at the end of the robotic arm for adjusting the orientation of the semiconductor device; a position sensor connected to the control system and providing real-time feedback of the robotic arm's position coordinates; a device supply device including a vibratory feeder that evenly distributes the semiconductor devices on the feeder surface; a detection device including a vision detection device that identifies the target position coordinate set of the semiconductor devices and transmits it to the control system; the control system also includes: a base position reference acquisition module for acquiring the current position coordinates of the movable base in real time through the position sensor and setting these coordinates as the reference point for calculating the current robotic arm position; and a trajectory prediction and displacement compensation calculation module for calculating the position of the preceding robotic arm. The system calculates the expected position offset of the base when the preceding robotic arm completes its movement, based on the predicted movement trajectory and the expected dwell position after the task is completed. This results in a base displacement compensation factor, which is represented as the position difference vector between the actual position of the base after the preceding robotic arm completes its grasping task and the initial position of the base before the current robotic arm begins its movement. A compensation distance calculation module adds the base displacement compensation factor to the current robotic arm position to obtain the compensated starting position. The Euclidean distance formula is used to calculate the straight-line distance from the compensated starting position of the robotic arm to each candidate position. A compensation optimization selection module selects the candidate position with the shortest straight-line distance from the compensated starting position of the robotic arm to each candidate position as the next target position for the current robotic arm. A dynamic compensation update module updates the base position reference point in real time after each robotic arm completes its grasping task and recalculates the base displacement compensation factor for the next robotic arm to move.

[0014] In conjunction with the second aspect, in some implementations of the second aspect, the control system further includes: a comprehensive evaluation construction module, which establishes a multi-dimensional evaluation framework for candidate positions, including a movement distance evaluation dimension, an obstacle avoidance complexity evaluation dimension, and a base displacement impact evaluation dimension; a movement distance evaluation calculation unit, which uses the Euclidean distance formula to calculate the straight-line movement distance from the current position of the robotic arm to each candidate position, and converts the distance values ​​into movement distance evaluation scores after normalization; and an obstacle avoidance complexity evaluation calculation unit, which analyzes the obstacles that need to be avoided in the movement path from the current position of the robotic arm to each candidate position. The system calculates the obstacle avoidance complexity evaluation score based on the number of obstacles and the length of the detour path; the base displacement impact evaluation calculation unit calculates the base displacement impact evaluation score based on the expected displacement of the base caused by the previous robotic arm task execution; the weighted comprehensive evaluation processor calculates the comprehensive evaluation score of each candidate position by weighted summation based on the scores of the movement distance evaluation, the obstacle avoidance complexity evaluation, and the base displacement impact evaluation; and the optimal position decision selector compares the comprehensive evaluation scores of each candidate position in the feasible position set and selects the candidate position with the highest comprehensive evaluation score as the next movement target position of the current robotic arm.

[0015] Compared with the prior art, the beneficial effects of the present invention are:

[0016] 1. This application employs an improved greedy algorithm for dynamic path planning. For the special working mode of multiple robotic arms sharing a fixed base, it achieves intelligent optimization of the grasping path. Based on the current position of the robotic arm and the distribution of the target device, it dynamically calculates the optimal grasping sequence, minimizing the total time consumed by multiple grasping actions. Compared to traditional sequential grasping methods, this method reduces the ineffective movement distance of the robotic arm, shortens the work cycle time, and thus significantly improves the overall production efficiency of the fully automated carrier tape and packaging equipment.

[0017] 2. This application establishes a real-time monitoring and dynamic adjustment mechanism. A CCD vision inspection device continuously monitors the state changes of semiconductor devices in the vibratory feeder, and a position sensor acquires the robotic arm's position information in real time, ensuring timely response to dynamic changes in the working environment. When the position of a device in the vibratory feeder changes or a new graspable device appears, the path can be automatically replanned and the grasping strategy adjusted, ensuring the equipment maintains a highly efficient and stable working state in complex and ever-changing production environments. Attached Figure Description

[0018] Figure 1 A schematic diagram of a fully automatic carrier tape taping and packaging equipment provided in an embodiment of this application;

[0019] Figure 2 A schematic flowchart illustrating a path planning method provided in an embodiment of this application;

[0020] Figure 3 A flowchart illustrating a path planning method provided in another embodiment of this application;

[0021] Figure 4 A flowchart illustrating a path planning method provided in yet another embodiment of this application;

[0022] Figure 5 A flowchart illustrating a path planning method provided in yet another embodiment of this application;

[0023] Figure 6 A flowchart illustrating a path planning method provided in yet another embodiment of this application;

[0024] Figure 7 A flowchart illustrating a path planning method provided in yet another embodiment of this application;

[0025] Figure 8 This is a flowchart illustrating a path planning method provided in yet another embodiment of this application. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] Exemplary system

[0028] One embodiment of this application provides a fully automatic carrier tape packaging equipment. The equipment adopts a modular design and achieves efficient automated packaging of semiconductor devices through the coordinated operation of mechanical systems, detection systems, and control systems.

[0029] The equipment includes a frame that is fixedly mounted and supports the overall structure. The frame is made of high-strength aluminum alloy and includes a support system consisting of a body frame, anchor bolts, and casters, ensuring stability and mobility during operation. Multiple robotic arms are fixed to a single movable base, which is connected to the frame and can move within a plane. Each robotic arm is equipped with an end effector for sequentially grasping semiconductor devices. For example, five robotic arms (numbered 1-5) can be configured, coordinating their work by sharing a base. Each robotic arm is equipped with a vacuum suction cup as an end effector. The device supply device includes a vibratory feeder for providing the semiconductor devices to be grasped and distributing them. The device supply device also includes a hopper, a vibratory feeder, a guide trough, and a cleaning device to ensure a continuous and stable supply of semiconductor devices. The detection device includes a vision recognition module and a position sensor. The vision recognition module identifies the target position coordinates of all semiconductor devices to be grasped in the vibratory feeder, and the position sensor acquires the starting position coordinates of the robotic arms in real time. The vision recognition module uses an industrial-grade CCD sensor, and the position sensor uses photoelectric encoder technology.

[0030] Figure 1 This is a schematic diagram of a fully automatic carrier tape taping and packaging equipment according to an embodiment of this application. Figure 1 As shown, the device also includes a control system 800, which is connected to multiple robotic arms and detection devices. The control system 800 includes: a position coordinate processing module 801, used to receive and process the target position coordinate set and starting position coordinates obtained by the detection devices; a feasible position set construction module 802, which, based on the work area boundary and obstacle avoidance requirements, constructs a feasible position set for the current robotic arm, including all candidate positions that are safely reachable and without collision risk; a distance calculation and optimization selection module 803, which traverses the feasible position set, calculates the distance from the current position of the robotic arm to each candidate position, and selects the closest candidate position as the next movement target position based on a greedy algorithm; and a dynamic set update module 80... 4. Remove the next target position from the set of feasible positions and update the available candidate positions in the set of feasible positions in real time; Task execution control module 805 controls the current robotic arm to move to the target position and execute the grasping task, and records the final position of the movable base after the grasping task is completed; Robotic arm coordination switching module 806 sets the final position as the starting position of the next robotic arm to be moved, realizing sequential coordination control between multiple robotic arms; Loop control module 807 repeats the above position planning and task control process until all robotic arms complete their respective grasping tasks or the target position coordinate set is empty.

[0031] In some implementations, the movable base is driven to move in the XY plane by a drive device, which includes an X-axis stepper motor and a Y-axis stepper motor. The X-axis and Y-axis stepper motors can be model 57HS22-A, with a step angle of 1.8 degrees and a movement range of 0-800 mm in the X-axis direction and 0-600 mm in the Y-axis direction.

[0032] Each robotic arm includes: a telescopic device that extends and retracts along the Z-axis to perform grasping actions. The telescopic device is cylinder-driven with a stroke of 50 mm and a working height range of 5 mm to 100 mm from the vibratory feeder surface. A rotating device, mounted at the end of the robotic arm, adjusts the orientation of the semiconductor devices. The rotating device uses a 20BYJ46 micro stepper motor, enabling 360-degree continuous rotation with an angular resolution of 0.18 degrees. A position sensor, connected to the control system 800, provides real-time feedback of the robotic arm's position coordinates. The position sensor uses a photoelectric encoder with a resolution of 1024 pulses per revolution, transmitting data via the RS485 communication protocol. The device supply device includes a vibratory feeder that evenly distributes the semiconductor devices on its surface. The vibratory feeder is equipped with a vibratory motor, a base, device recesses, and a protective cover to ensure stable device supply and operational safety. The detection device includes a vision detection unit that identifies the target position coordinates of the semiconductor devices and transmits them to the control system 800. The visual inspection device includes a 1280×1024 pixel CCD camera, an LED ring light source, an ARM Cortex-A9 data processing unit, and an Ethernet communication module.

[0033] Furthermore, the control system 800 also includes: a base position reference acquisition module 808, used to acquire the current position coordinates of the movable base in real time through a position sensor, and set the coordinates as the reference point for calculating the current position of the robotic arm; and a trajectory prediction and displacement compensation calculation module 809, used to calculate the expected position offset of the base when the preceding robotic arm completes its action based on the predicted movement trajectory of the preceding robotic arm and the expected dwell position after the task is completed, and obtain the base displacement compensation factor. The base displacement compensation factor is represented as the position difference vector between the actual position of the base after the preceding robotic arm completes the grasping task and the initial position of the base before the current robotic arm starts moving. The compensation distance calculation module is used to superimpose the base displacement compensation factor onto the current robotic arm position to obtain the compensated starting position of the robotic arm, and calculate the straight-line distance from the compensated starting position of the robotic arm to each candidate position using the Euclidean distance formula. The compensation optimization selection module 810 is used to select the candidate position with the shortest straight-line distance from the compensated starting position of the robotic arm to each candidate position as the next moving target position of the current robotic arm. The dynamic compensation update module 811 updates the base position reference point in real time after each robotic arm completes the grasping task, and recalculates the base displacement compensation factor for the next robotic arm to be moved.

[0034] Furthermore, the control system 800 also includes: a comprehensive evaluation construction module 812, which establishes a multi-dimensional evaluation framework for candidate positions, including a movement distance evaluation dimension, an obstacle avoidance complexity evaluation dimension, and a base displacement impact evaluation dimension; a movement distance evaluation calculation unit 813, which uses the Euclidean distance formula to calculate the straight-line movement distance from the current position of the robotic arm to each candidate position, and converts the distance values ​​into movement distance evaluation scores after normalization; and an obstacle avoidance complexity evaluation calculation unit 814, which analyzes the number of obstacles to be avoided and the detour paths in the movement path from the current position of the robotic arm to each candidate position. The system calculates the obstacle avoidance complexity evaluation score based on the length of the obstacle avoidance distance; the base displacement impact evaluation calculation unit 815 calculates the base displacement impact evaluation score based on the expected displacement of the base caused by the previous robotic arm task execution; the weighted comprehensive evaluation processor 816 calculates the comprehensive evaluation score of each candidate position by weighted summation based on the scores of the movement distance evaluation, the obstacle avoidance complexity evaluation, and the base displacement impact evaluation; and the optimal position decision selector 817 compares the comprehensive evaluation scores of each candidate position in the feasible position set and selects the candidate position with the highest comprehensive evaluation score as the next movement target position of the current robotic arm.

[0035] Based on the above-mentioned structural configuration of the fully automatic carrier tape taping and packaging equipment, the following is combined with... Figures 2-8 The path planning method provided in this application is described in detail.

[0036] Figure 2 This is a flowchart illustrating a path planning method provided in one embodiment of this application. Figure 2 As shown, the path planning method provided in this application includes the following steps:

[0037] Step 100: Obtain the starting position coordinates of the current robotic arm and identify the target position coordinate set of all semiconductor devices to be grasped.

[0038] Specifically, position sensors acquire the real-time position coordinates P_base_current of the five robotic arms sharing a fixed base. These coordinates represent the precise position of the robotic arms in the XY working plane. Simultaneously, a CCD vision inspection device performs real-time image acquisition and processing of the vibratory feeder area. Using edge detection and shape recognition algorithms, it identifies all front-facing semiconductor devices, establishing a target position coordinate set P_target_set = {P_target_1, P_target_2, ..., P_target_n}. Each target position contains the XY coordinate information and rotation angle parameters of the device. Then, the physical coordinates are mapped to digital coordinates that the control system can process using a coordinate transformation matrix.

[0039] Step 101: Construct a set of feasible positions for the current robotic arm. The set of feasible positions includes all candidate positions that are safely accessible and without collision risk.

[0040] Specifically, a feasible location set P_available is established based on the work area boundary constraints and obstacle avoidance requirements. The work area boundary is set to 0-800 mm in the X-axis direction and 0-600 mm in the Y-axis direction; locations outside this range are excluded. Obstacle zones within the equipment are identified, including the edges of the vibratory feeder, conveyor devices, and other fixed structures, and an obstacle avoidance zone set Obstacle_zones is established. For cases where a straight path intersects with an obstacle, a simplified version of the A* algorithm is used to calculate the detour path. The work area is divided into a 10×10 mm grid for path search, ensuring that all candidate locations can be safely reached without collision risk.

[0041] Step 102: Traverse the set of feasible positions, calculate the distance from the current position of the current robotic arm to each candidate position, and select the candidate position with the closest distance as the next target position for the current robotic arm to move.

[0042] Specifically, the Euclidean distance formula is used for calculation:

[0043]

[0044] Where P_current is the current position of the robotic arm base, P_i is the i-th candidate position, x_i and y_i are the coordinates of the i-th candidate position, and x_current and y_current are the coordinates of the current robotic arm base. Based on distance calculation, a greedy strategy is applied to select the optimal position:

[0045] P_next=argmin{distance(P_base_current,P_i)|P_i∈P_available};

[0046] This involves selecting the location closest to the current base position from the set of feasible locations as the next target location. Considering path feasibility scoring, and taking into account both distance and obstacle avoidance complexity, the selected path is ensured to be both short and safe. The scoring formula is:

[0047] Score=0.7×distance+0.3×obstacle_penalty.

[0048] Step 103: Remove the next target location from the set of feasible locations and update the set of feasible locations.

[0049] Specifically, the dynamic set update module removes the selected next move target position from the feasible position set and updates the available candidate positions in the feasible position set in real time. It also updates the current path sequence record of the robotic arm, and the dynamic compensation update module prepares for state changes.

[0050] Step 104: When the current robotic arm reaches the target position, it performs the grasping task and records the final position after the grasping task is completed.

[0051] Specifically, the task execution control module controls the robotic arm to move to the target position, completing the full sequence of actions for the grasping task through the telescopic device, rotating device, and vacuum suction cup. The task execution control module records the final position P_base_final of the movable base after the grasping task is completed, providing a positional reference for the subsequent coordinated work of the robotic arm.

[0052] Step 105: Set the final position as the starting position of the next robotic arm to be moved. Repeat the above steps until all robotic arms have completed their respective grasping tasks or the target position coordinate set is empty.

[0053] Specifically, the robotic arm coordination and switching module sets the final position as the starting position of the next robotic arm to be operated, realizing sequential coordinated control among multiple robotic arms. The dynamic compensation update module recalculates the base displacement compensation factor to ensure that the path planning of subsequent robotic arms is based on accurate position information. The loop control module repeatedly executes the above position planning and task control process until all robotic arms complete their respective grasping tasks or the target position coordinate set is empty.

[0054] The beneficial effects of this application's embodiments are as follows: Real-time position acquisition and target recognition, distance calculation, and greedy strategy selection ensure that each robotic arm can select the optimal grasping path in its current state, effectively reducing ineffective movement time and waiting time. The dynamic set update and position transfer mechanism solves the coordination problem of multiple robotic arms sharing a base, avoiding the efficiency bottleneck of traditional sequential work modes, enabling five robotic arms to complete the grasping task in the optimal sequence, significantly shortening the overall operation cycle. A multi-layered safety protection system is established. Through work area boundary restrictions and obstacle recognition, it ensures that the robotic arms always operate within a safe range, effectively preventing equipment collisions and personnel injury risks. The A* algorithm is used for detour path planning, combined with a path feasibility scoring mechanism, ensuring the safety of path selection while pursuing efficiency. Task execution control and real-time status monitoring ensure the standardization and high success rate of each grasping operation. Simultaneously, it can respond in real-time to changes in the working environment, automatically adjusting the path planning strategy to adapt to changes in device position and the appearance of new devices. Dynamic compensation mechanisms and cyclic control logic ensure stable operation in complex production environments, realizing fully automated management from task initiation to completion.

[0055] like Figure 3 As shown, in the path planning method provided in this application embodiment, the feasible location set is traversed, the distance from the current position of the current robotic arm to each candidate position is calculated, and the candidate position with the closest distance is selected as the next movement target position of the current robotic arm, including:

[0056] Step 200: Obtain the current position coordinates of the base as the reference point for the current position of the robotic arm.

[0057] Specifically, position sensors installed at the end of the robotic arms are used to acquire the current position coordinates of the five robotic arms sharing a fixed base in real time via RS485 communication protocol at a baud rate of 9600. A base position reference point is then established.

[0058] P_base_current=(X_base_current,Y_base_current);

[0059] Where X_base_current is the coordinate value of the base in the X-axis direction, and Y_base_current is the coordinate value of the base in the Y-axis direction.

[0060] The physical coordinates acquired by the position sensors are converted into a standard working coordinate system that the control system can process using the `Transform_matrix`. A working coordinate system is established for the device, with an X-axis range of 0-800 mm, a Y-axis range of 0-600 mm, and a Z-axis working height set at 5 mm above the vibratory feeder surface. To ensure time synchronization for coordinated operation of multiple robotic arms, a millisecond-level timestamp (`Timestamp_base_current`) is added to each reference point acquisition operation. This timestamp information is used for timing control in subsequent position compensation calculations and path planning, ensuring the real-time nature and validity of the position data.

[0061] Step 201: Based on the estimated movement trajectory of the preceding robotic arm and the expected dwell position after the task is completed, calculate the expected position offset of the base when the preceding robotic arm completes its action, and obtain the base displacement compensation factor. The base displacement compensation factor is represented as the position difference vector between the actual position of the base after the preceding robotic arm completes the grasping task and the initial position of the base before the current robotic arm starts its action.

[0062] Specifically, for the preceding robotic arm numbered il, a complete trajectory prediction model is established. The trajectory prediction comprises three stages: the first stage is the movement stage, from the current position P_base_current to the target device position P_target_previous; the second stage is the grasping stage, executing a standard descent-grab-ascend sequence at the target position; and the third stage is the return stage, moving from the target position to the carrier conveyor position P_carrier_position. The movement distance in each stage is calculated using the Euclidean distance formula:

[0063]

[0064] The expected dwell position P_expected_final after the preceding robotic arm completes its grasping task is defined as the device release position above the carrier tape conveyor. The coordinates of this position are determined based on the carrier tape hole distribution and the device placement accuracy requirements; the specific coordinates are as follows:

[0065] P_expected_final=(X_carrier+Offset_x,Y_carrier+Offset_y);

[0066] Offset_x and Offset_y are the position offsets calculated based on the device size and carrier tape specifications.

[0067] The base displacement compensation factor `Compensation_vector` represents the position difference vector between the actual position of the base after the preceding robotic arm completes its grasping task and the initial position of the base before the current robotic arm begins its movement. The calculation formula is:

[0068] Compensation_vector=P_expected_final-P_base_initial;

[0069] Where: P_base_initial is the initial position of the base recorded before the current robotic arm starts moving.

[0070] Compensation_vector=(ΔX_compensation,ΔY_compensation);

[0071] ΔX_compensation=X_expected_final-X_base_initial;

[0072] ΔY_compensation=Y_expected_final-Y_base_initial;

[0073] Considering the inertia and positioning error of the mechanical system, a dynamic adjustment mechanism is introduced. The actual application value of the compensation factor is corrected by the adjustment coefficient K_adjustment:

[0074] Compensation_vector_adjusted=K_adjustment×Compensation_vector;

[0075] K_adjustment is determined based on historical operating data and the characteristics of the mechanical system, with a typical value of 0.95, and is used to compensate for the accumulation of errors in the mechanical transmission process.

[0076] Step 202: Add the base displacement compensation factor to the current position of the robotic arm to obtain the compensated starting position of the robotic arm.

[0077] Specifically, the base displacement compensation factor Compensation_vector_adjusted calculated in step 201 is superimposed on the current reference position of the robotic arm to obtain the compensated starting position P_compensated_start of the robotic arm. The calculation formula is:

[0078] P_compensated_start=P_base_current+Compensation_vector_adjusted;

[0079] Expanded to:

[0080] X_compensated_start=X_base_current+ΔX_compensation×K_adjustment;

[0081] Y_compensation_start=Y_base_current+ΔY_compensation×K_adjustment;

[0082] Perform a working area boundary check on the compensated starting position to ensure that the compensated position is within the allowable working range of the equipment. If the compensated position exceeds the boundary, automatically adjust it to the nearest valid boundary point and record the boundary adjustment event for optimization.

[0083] Boundary check conditions are:

[0084] 0≤X_compensated_start≤800mm;

[0085] 0≤Y_compensated_start≤600mm;

[0086] The accuracy of the compensation calculation is ensured by comparing and verifying the results with real-time feedback from the position sensor. The verification formula is as follows:

[0087] Position_error=|P_compensated_start-P_sensor_feedback|;

[0088] When Position_error > 0.1 mm, a recalibration procedure is triggered to ensure that the accuracy of position compensation meets the requirements of the grasping task.

[0089] Step 203: Calculate the straight-line distance from the compensated starting position of the robotic arm to each candidate position using the Euclidean distance formula.

[0090] Specifically, candidate positions P_candidate_i are extracted one by one from the feasible position set P_available, where i = 1, 2, ..., n, and n is the total number of feasible candidate positions. Each candidate position P_candidate_i = (X_candidate_i, Y_candidate_i) represents a position of a graspable semiconductor device in the vibratory feeder.

[0091] The straight-line distance from the compensated starting position of the robotic arm to each candidate position is calculated using the standard Euclidean distance formula. All calculated distance values ​​are then used to construct a distance matrix `Distance_matrix` with dimensions 1×n, where `Distance_matrix[i]` represents the distance from the compensated starting position to the i-th candidate position. Simultaneously, the candidate position index corresponding to each distance is recorded, and an index mapping relationship `Index_mapping` is established.

[0092] To improve the accuracy and efficiency of distance calculation, a fast distance estimation algorithm is used for pre-screening. First, Manhattan distance is used for coarse sorting, and then Euclidean distance is calculated for the k closest candidate locations, with the value of k dynamically adjusted according to the device distribution density.

[0093] Step 204: Select the candidate position with the shortest straight-line distance from the compensated starting position of the robotic arm to each candidate position as the next target position for the robotic arm's movement.

[0094] Specifically, the minimum distance value is found from the distance matrix Distance_matrix, the candidate position index Index_optimal corresponding to the minimum distance is recorded, and the optimal candidate position is determined, which is the target position for the next movement of the current robotic arm.

[0095] Before selecting the optimal candidate position, multiple constraint verifications are performed to ensure the rationality of the selection: check whether there are obstacles blocking the path from the compensation starting position to the optimal candidate position, confirm that the candidate device is still in a graspable state through a CCD vision inspection device, confirm that the XY axis stepper motor of the base can complete the required displacement action, and verify that moving to this position will not cause time conflicts with the work of other robotic arms.

[0096] To address situations where the optimal choice is infeasible, a suboptimal alternative mechanism is established. The distance matrix is ​​sorted in ascending order, creating a priority queue for candidate positions. When the optimal choice fails to be verified, the next feasible candidate position in the queue is automatically selected, ensuring the continuity and reliability of path planning. After determining the final next target position P_next_target, the following recording and transmission operations are performed: the current target position record of the robotic arm is updated, the selected position is removed from the set of feasible positions, a movement command is sent to the robotic arm control module, and the timestamp and relevant parameters of the selection decision are recorded.

[0097] The beneficial effects of the embodiments of this application are that, through position compensation, intelligent path planning and fault handling mechanisms, it can maintain a stable and efficient operating state in a complex multi-arm collaborative environment, effectively reduce the operation error rate, improve the automation level of the production line and the consistency of product quality, and significantly improve the overall efficiency and quality of semiconductor device gripping operations.

[0098] Figure 4 The diagram shown is a flowchart illustrating a path planning method provided in another embodiment of this application. Figure 4 As shown, in the path planning method provided in this application embodiment, the feasible location set is traversed, the distance from the current position of the current robotic arm to each candidate position is calculated, and the candidate position with the closest distance is selected as the next movement target position of the current robotic arm, including:

[0099] Step 300: Perform multi-dimensional evaluation on each candidate location in the feasible location set. The multi-dimensional evaluation includes movement distance evaluation, obstacle avoidance complexity evaluation, and base displacement impact evaluation.

[0100] Specifically, a comprehensive evaluation framework is established, expanding the single distance factor into a three-dimensional evaluation system. The evaluation framework adopts a hierarchical structure design. The first layer is the basic data collection layer, responsible for collecting the original evaluation data of each candidate location; the second layer is the single-dimensional calculation layer, which calculates the movement distance evaluation, obstacle avoidance complexity evaluation, and base displacement impact evaluation respectively; the third layer is the comprehensive decision-making layer, which obtains the final evaluation score through a weighted summation mechanism.

[0101] A weighting mechanism for evaluation dimensions is established, with weight coefficients determined based on the actual production environment and equipment characteristics. The weight w1 for movement distance is set at 0.5, reflecting the dominant role of distance optimization in improving grasping efficiency. The weight w2 for obstacle avoidance complexity is set at 0.3, reflecting the importance of safety considerations. The weight w3 for base displacement impact is set at 0.2, considering the influencing factors of coordinated operation of multiple robotic arms.

[0102] An evaluation data structure is established for each candidate location P_candidate_i in the feasible location set P_available. The data structure includes the candidate location coordinates (X_candidate_i, Y_candidate_i), the device orientation angle θ_device_i, and the grabbable status flag Status_grabbable_i. Three evaluation score storage arrays are also initialized:

[0103] Score_distance_array[n], Score_obstacle_array[n], Score_base_impact_array[n];

[0104] Where n is the total number of candidate positions.

[0105] The Euclidean distance formula is used to calculate the straight-line distance from the current position of the robotic arm to each candidate position. The distance values ​​are then normalized and converted into scores for the movement distance evaluation. Based on the compensated starting position P_compensated_start obtained in step 202, the Euclidean distance formula is used to calculate the straight-line distance to each candidate position. To eliminate the influence of distance dimensions on the evaluation results, the original distance values ​​are normalized. First, the maximum and minimum values ​​of the distance range are determined. Then, a linear normalization formula is applied to convert the normalized distance values ​​into movement distance evaluation scores. A reverse scoring method is used, with evaluation scores ranging from 0 to 100 points; the closer the distance, the higher the score. A distance grading evaluation standard is established: Excellent (90-100 points) corresponds to a distance less than 50 mm, Good (70-89 points) corresponds to a distance of 50-150 mm, Average (50-69 points) corresponds to a distance of 150-300 mm, and Poor (0-49 points) corresponds to a distance greater than 300 mm.

[0106] Step 302: Analyze the number of obstacles to be avoided and the length of the detour path in the movement path from the current position of the robotic arm to each candidate position, and calculate the score of obstacle avoidance complexity evaluation.

[0107] Specifically, obstacle detection is performed on the movement path based on a preset obstacle avoidance zone set, Obstacle_zones. The obstacle avoidance zones include fixed obstacles such as the vibratory feeder edge protection cover, the mechanical structure of the carrier conveyor, and the workspace of other robotic arms. A ray-mapping algorithm is used to determine whether the straight path from P_compensated_start to P_candidate_i intersects with the obstacle zones.

[0108] When a straight path intersects with an obstacle, a simplified version of the A* algorithm is used to calculate the detour path. The working area is divided into a 10×10 mm grid, with obstacle areas marked as impassable. The formula for calculating the detour path length is:

[0109]

[0110] Where m is the total number of nodes in the detour path, and X_node_k and Y_node_k are the coordinates of the k-th path node.

[0111] The obstacle avoidance complexity evaluation comprehensively considers the number of obstacles, Obstacle_count[i], and the detour increase distance, Detour_additional[i]. The calculation formula is as follows:

[0112]

[0113] Where Detour_additional[i] = Detour_length[i] - Distance_raw[i], it represents the additional distance of the detour relative to the straight path. The evaluation score is based on a 100-point scale, with a perfect straight path receiving full marks, each obstacle deducting 10 points, and 1 point deducted for every 10 millimeters of additional detour distance. Distance_raw[i] represents the original distance value to the i-th candidate position.

[0114] Step 303: Calculate the score for the impact evaluation of base displacement based on the expected displacement of the base caused by the execution of the preceding robotic arm task.

[0115] Specifically, the impact of the preceding robotic arm (numbered il) completing its grasping task on the efficiency of subsequent robotic arms is analyzed. Influencing factors include the base position change (Base_displacement), the replanning complexity of subsequent task replanning, and the time delay impact (Time_delay_impact). The base position change is calculated by estimating the final position.

[0116] Base_displacement=|P_expected_final-P_base_current|;

[0117] The impact of selecting the current candidate position on the working efficiency of subsequent robotic arms (numbered i+1 to 5) is quantitatively analyzed. The impact assessment is achieved by calculating the average distance change from subsequent robotic arms to the remaining components.

[0118]

[0119] Where Distance_after[j][k] represents the distance from the robotic arm j to the remaining target k after selecting the current candidate position, and Distance_before[j][k] represents the corresponding distance before selection.

[0120] The evaluation score for the impact of base displacement is calculated based on the influence coefficient and the base displacement. The evaluation score is based on a 100-point scale. 0.2 points are deducted for every 1 mm increase in base displacement, and 5 points are deducted for every 0.1 increase in the negative impact coefficient. This ensures that the selected candidate positions meet both the efficiency requirements of the current robotic arm and the overall coordination.

[0121] Step 304: Based on the scores of the movement distance evaluation, the obstacle avoidance complexity evaluation, and the base displacement impact evaluation, a weighted summation is used to calculate the comprehensive evaluation score of each candidate position, and the candidate position with the highest comprehensive evaluation score is selected as the next target position for the current robotic arm movement.

[0122] Specifically, a weighted summation method is used to calculate the comprehensive evaluation score of each candidate position, integrating the evaluation results of the three dimensions. The comprehensive evaluation score ranges from 0 to 100 points, with higher scores indicating better candidate positions.

[0123] The candidate position with the highest comprehensive evaluation score is selected as the optimal target. The selected optimal position undergoes final verification to ensure the feasibility of the decision. Verification includes reconfirmation of the device's graspability status, path reachability check, and time window conflict detection. Alternative solutions are established, with the top three candidate positions based on comprehensive evaluation scores as backup options. If the optimal selection fails verification, the suboptimal solution is automatically activated. The selected optimal target position P_next_optimal is then passed to the robotic arm control system to update relevant state parameters.

[0124] The beneficial effects of this application's embodiments include establishing a comprehensive evaluation framework that expands the traditional single-distance consideration into a three-dimensional evaluation system, including movement distance evaluation, obstacle avoidance complexity evaluation, and base displacement impact evaluation. Through a hierarchical structure design and a weighted summation mechanism, comprehensive path selection decisions are achieved, effectively avoiding the local optima problem caused by a single indicator. By introducing base displacement impact evaluation, the impact of the current robotic arm movement on the subsequent robotic arm's work efficiency can be predicted, and the impact of base position changes on the overall system coordination can be quantitatively analyzed.

[0125] Figure 5 The diagram shown is a flowchart illustrating a path planning method provided in another embodiment of this application. Figure 5 As shown, in the path planning method provided in this application embodiment, constructing the set of feasible positions for the current robotic arm includes:

[0126] Step 400: Establish boundary constraints for the working area, and define the movable XY plane range of the robot arm's fixed base and the working height of the robot arm's Z-axis.

[0127] Specifically, the movement boundary constraints of the fixed base are established based on the equipment's mechanical structure and safety requirements. The X-axis movement range is set to 0-800 mm, determined by the maximum stroke capacity of the 57HS22-A X-axis stepper motor and the physical limitations of the guide rail system. The Y-axis movement range is set to 0-600 mm, also based on the technical specifications and mechanical structure constraints of the Y-axis stepper motor. The working height of the robotic arm in the vertical direction is determined by the cylinder stroke of the telescopic device. The standard gripping working height Z_working_standard is set to 5 mm from the surface of the vibratory feeder, ensuring that the vacuum suction cup can reliably adsorb semiconductor devices without collision. The maximum working height Z_working_max is set to 100 mm, corresponding to the maximum extension stroke of the cylinder (50 mm) plus the base height of the robotic arm. The Z-axis constraint conditions are expressed as follows:

[0128] Z_constraint={Z_height|5mm≤Z_height≤100mm};

[0129] To ensure the safe operation of the robotic arm, a safety buffer zone is set inside the physical boundary. The safety buffer distance Buffer_X in the X-axis direction is set to 20 mm, and the safety buffer distance Buffer_Y in the Y-axis direction is set to 15 mm. The actual usable working area is calculated as follows:

[0130] Safe_workspace_X={X_position|Buffer_X≤X_position≤800mm-Buffer_X};

[0131] Safe_Workspace_Y={Y_position|Buffer_Y≤Y_position≤600mm-Buffer_Y};

[0132] Establish a real-time boundary constraint verification procedure to check the boundary compliance of the robotic arm's target position. The boundary verification function returns a Boolean value; if the target position exceeds the safe working area, it returns and triggers a boundary warning, adjusting the out-of-bounds target to the nearest valid boundary point.

[0133] Step 401: Set the obstacle avoidance area set, including the coordinate set of obstacle areas inside the device that need to be avoided, to ensure the safety of the robotic arm's movement path.

[0134] Specifically, all fixed obstacles inside the equipment are identified and their geometric models are established. The main obstacles include the vibratory feeder protective cover area, the mechanical structure of the carrier conveyor, the support frame of the cover conveyor, and the central processing system installation area. Each obstacle area is represented by a rectangular envelope. Dynamic obstacle updates are established to account for temporary obstacle areas formed by other robotic arms during operation. When robotic arm i is in a moving state, its workspace forms a temporary obstacle avoidance zone, Temp_obstacle_zone_i. The geometry of the temporary obstacle zone is determined based on the robotic arm's movement trajectory and safety distance requirements, with the safety distance set at 50 mm.

[0135] To ensure the absolute safety of the robotic arm's movement, all obstacle areas are safely extended. The extension distance, Safety_margin, is set to 30 mm. The extended obstacle avoidance zone set, Obstacle_zones_safe, is used for subsequent path collision detection and detour calculations.

[0136] Establish a standardized data structure for obstacle information. Each obstacle record includes the area boundary coordinates, obstacle type, priority level, and safety extension parameters.

[0137] Step 402: When the straight path from the current robotic arm position to the target position intersects with an obstacle, calculate a feasible detour path.

[0138] Specifically, a ray-rectangle intersection algorithm is used to detect whether the straight path from the current robotic arm position P_compensated_start to the target position P_candidate_i intersects with an obstacle region. The collision detection algorithm traverses each obstacle region in the obstacle avoidance zone set Obstacle_zones_safe, calculating the intersection point of the ray and the rectangle boundary. When the function returns, it indicates that the straight path has a collision risk and detour path planning is required.

[0139] To support detour path calculation, the entire working area is divided into a regular grid. The grid cell size (Grid_size) is set to 10×10 mm to ensure that the path planning accuracy meets the grasping requirements. Each grid cell is marked as passable or impassable, and grid cells within obstacle areas are marked as impassable. A simplified version of the A* algorithm is used to search for the optimal detour path in the gridded working area. The algorithm maintains open and closed lists and evaluates the priority of each grid node through a heuristic function. Detour path smoothing is performed. The grid paths generated by the A* algorithm often contain multiple right-angle turns, which are unsuitable for the smooth movement of the robotic arm. A path smoothing algorithm is used to optimize the detour path, converting the polyline path into a smooth curve through B-spline curve fitting. The smoothed path point coordinate sequence (Detour_path_smooth) ensures that the robotic arm can complete the detour with a continuous motion trajectory, avoiding the impact of sudden stops and starts on the mechanical system.

[0140] Step 403: Include all candidate locations that are safely accessible and pose no risk of collision into the set of feasible locations.

[0141] Specifically, the safety of all semiconductor device locations identified by the CCD vision inspection device is verified. The verification procedure includes location coordinate validity checking, device status confirmation, and grasping accessibility analysis. Location coordinate validity checking confirms that the device location is within the working area boundary, and device status confirmation verifies through image analysis that the device is facing upwards and undamaged. Grasping accessibility analysis calculates whether there is a feasible path for the robotic arm from its current position to the device location. A detailed collision risk assessment is performed on each candidate device location. The assessment process includes straight-line collision detection, detour path calculation, and robotic arm workspace conflict analysis. The collision risk assessment function comprehensively considers path safety, time feasibility, and mechanical accessibility. The risk assessment results are divided into four levels: safe (no collision risk), low risk (requires simple detour), medium risk (requires complex detour), and high risk (not recommended). Only candidate locations at the safe and low-risk levels are included in the feasible location set. The candidate locations that pass the safety verification are sorted according to their comprehensive evaluation scores to construct an ordered feasible location set P_available_sorted. The set construction algorithm first filters candidate locations that meet the safety conditions, and then sorts them based on the multi-dimensional evaluation results in steps 300-304. The data structure of the feasible location set includes information such as location coordinates, evaluation score, risk level, and path type.

[0142] A dynamic maintenance system for the feasible location set is established to respond in real time to changes in the working environment. When the CCD vision inspection device detects a new graspable device, it automatically performs safety verification and evaluation calculations, adding the new device location that meets the criteria to the feasible location set. When a device is grasped or its status changes, the corresponding location is immediately removed from the set. The dynamic update frequency is set to perform a set status check every 100 milliseconds to ensure that the feasible location set always reflects the current real working environment.

[0143] The beneficial effects of this application's embodiments are as follows: By establishing boundary constraints for the working area, the robotic arm is ensured to operate within a predefined safe range. By setting limits on the XY plane movement range and Z-axis working height, combined with a safety buffer mechanism, equipment damage or safety accidents caused by the robotic arm exceeding physical boundaries are effectively prevented. Real-time boundary verification further enhances safety assurance capabilities, enabling timely detection and correction of out-of-bounds operations. Regarding obstacle avoidance, a complete obstacle avoidance area collection management system is established, covering both fixed obstacles and dynamic temporary obstacle areas. By employing a ray-and-rectangle intersection algorithm for collision detection, combined with the A* algorithm to calculate detour paths, safe and efficient path planning can be achieved in complex environments. Path smoothing ensures the continuity of the robotic arm's movement, avoiding the impact and wear caused by sudden stops and starts on the mechanical system.

[0144] Figure 6 The diagram shown is a flowchart illustrating a path planning method provided in another embodiment of this application. Figure 6 As shown, in the path planning method provided in this application embodiment, when the current robotic arm reaches the target position, it performs a grasping task and records the final position after the grasping task is completed, including:

[0145] Step 500: Control the current robotic arm to move to the target position and execute the standard grasping action sequence of descending-grabbing-ascending.

[0146] Specifically, based on the optimal target position P_next_optimal determined in step 304, the robotic arm base is controlled to move from the current compensated position P_compensated_start to the target position. Movement control is achieved through the coordinated drive of X-axis and Y-axis stepper motors, using the 57HS22-A model with a step angle of 1.8 degrees. The movement trajectory planning employs a trapezoidal velocity curve, comprising three stages: acceleration, constant speed, and deceleration, ensuring smooth movement and positioning accuracy.

[0147] After the robotic arm reaches the target position, it confirms its arrival and verifies its positioning accuracy using position sensors. The arrival determination condition is that the deviation between the actual position and the target position is within the allowable range. When the positioning error exceeds the tolerance range, fine-tuning control is executed, and precise positioning is achieved through small-amplitude step movements.

[0148] After the robotic arm reaches the target position and completes positioning verification, it executes a standardized grasping sequence. The sequence comprises four sub-stages, each with clearly defined control parameters and execution time. Stage 1: Descending action – the cylinder of the telescopic device drives the robotic arm to descend from a safe height to the grasping working height. Stage 2: Device orientation adjustment – ​​the rotary motor calculates the required rotation angle based on the device's current orientation (θ_device_current) identified by the CCD vision inspection device and the required orientation (θ_required) of the carrier tape aperture. Stage 3: Vacuum adsorption execution – the vacuum generator activates when the robotic arm reaches 5 mm above the device, generating a negative pressure of -80 kPa to complete device adsorption. Adsorption confirmation is achieved by monitoring vacuum changes through a pressure sensor; successful adsorption is determined when the pressure reaches or exceeds -75 kPa and remains stable for at least 0.2 seconds. Stage 4: Ascending action – after confirming successful device adsorption, the robotic arm ascends back to the safe moving height, with the ascent speed also set to 20 mm / s.

[0149] Real-time status monitoring is performed during the grasping process, and multi-sensor fusion is used to ensure a high success rate. Monitoring parameters include vacuum pressure, position sensor feedback, and rotation angle confirmation. An anomaly handling system is established so that when adsorption failure, excessive positional deviation, or abnormal rotation is detected, a retry procedure is automatically executed or the system switches to an alternative target location.

[0150] Step 501: Record the final coordinate data and timestamp information of the base after the grasping task is completed, which will be used for setting the starting position of the robotic arm in the future.

[0151] Specifically, after the robotic arm completes the device gripping, the system controls the base to move to the device release position P_carrier_position above the carrier conveyor. The coordinates of the release position are determined based on the carrier tape hole distribution and the device placement accuracy requirements, and are calculated as follows:

[0152] P_carrier_position=(X_carrier_base+Hole_offset_X,Y_carrier_base+Hole_offset_Y);

[0153] Where X_carrier_base and Y_carrier_base are the reference position coordinates of the carrier tape conveyor, and Hole_offset_X and Hole_offset_Y are the hole offsets calculated based on the carrier tape specifications and device dimensions. The robotic arm performs the device placement action at the release position and shuts down the vacuum generator to complete the device release.

[0154] The final coordinates of the base, P_base_final_current, are precisely measured and recorded after the grasping task is completed using position sensors. The final position record employs high-precision measurement, and the recorded data structure contains complete position information and related metadata.

[0155] Millisecond-level timestamps are used to record key time nodes of the grasping task. Timestamp information includes task start time, target arrival time, grasp completion time, and release completion time. Time synchronization is ensured through the system clock to guarantee timing coordination between the robotic arms. The recorded final base position P_base_final_current is set as the starting position reference for the next robotic arm (numbered i+l). The position transfer mechanism is implemented through the robotic arm coordination and switching module of the control system, ensuring accurate transmission and real-time updates of position information.

[0156] The beneficial effects of the embodiments of this application are that, through multi-sensor fusion for real-time status monitoring, key parameters such as vacuum pressure, position feedback and rotation angle can be monitored. The established anomaly handling mechanism can automatically execute the retry procedure when adsorption failure or position deviation is detected, which greatly improves reliability and grasping success rate.

[0157] Figure 7 The diagram shown is a flowchart illustrating a path planning method provided in another embodiment of this application. Figure 7 As shown in the embodiments of this application, the path planning method further includes:

[0158] Step 600: Monitor the operating status of the robotic arm in real time. When a mechanical failure or communication interruption is detected, dynamically reassign the assigned tasks of the faulty robotic arm to other normal robotic arms.

[0159] Specifically, a real-time monitoring system for the robotic arm's operating status based on multi-sensor fusion is established. The monitoring system includes three levels of status detection mechanisms: mechanical system status monitoring monitors the deviation between the actual position and the expected position of the robotic arm through position sensors, and triggers a position abnormality alarm when the deviation exceeds ±0.1 mm; electrical system status monitoring monitors the operating current of the X-axis and Y-axis stepper motors through current sensors, with a normal operating current range of 1.5-2.5 amps, and abnormal current changes indicate mechanical resistance or motor failure; vacuum system status monitoring monitors the operating status of the vacuum generator through pressure sensors, with a normal vacuum pressure range of -75 to -85 kPa.

[0160] A heartbeat mechanism is used to monitor the communication status between each robotic arm and the control system. The heartbeat signal transmission frequency is set to once every 100 milliseconds, and the timeout judgment period is 300 milliseconds. The communication fault detection algorithm maintains a communication status counter Comm_failure_counter[i]. When robotic arm i fails to respond for 3 consecutive heartbeat cycles, it is determined that the communication is interrupted. A fault type identification system is established to classify and handle detected abnormalities. Fault types include: Level 1 faults (position deviation, slight communication delay) can be resolved through automatic correction; Level 2 faults (motor overcurrent, abnormal vacuum pressure) require pausing the current task for inspection; Level 3 faults (complete communication interruption, mechanical jamming) require immediate shutdown of the robotic arm.

[0161] When a robotic arm malfunction is detected, a dynamic task reassignment procedure is initiated. First, the unfinished tasks of the malfunctioning robotic arm are identified. The target device corresponding to the malfunctioning robotic arm is extracted from the task assignment matrix Task_assignment_matrix. Then, the task is reassigned to the normally operating robotic arm based on the principle of proximity. The reassignment decision is based on the dual considerations of shortest distance and workload balance.

[0162] Step 601: When the base displacement is detected to exceed the preset compensation range, a recalibration operation is triggered.

[0163] Specifically, real-time monitoring of the cumulative displacement error of the base is established, and the cumulative error is identified by comparing the theoretically calculated position with the actual measured position. When the error exceeds the preset compensation range, a calibration procedure is triggered, and the compensation range threshold is determined based on the transmission accuracy and positioning requirements of the mechanical system. When the recalibration condition is triggered, the system executes a standardized calibration procedure. The calibration process includes four stages: reference position regression, where the robot arm base moves to the preset calibration reference position P_calibration_reference, which is usually selected as the center point of the working area (400 mm, 300 mm); sensor zero-point calibration, where the actual coordinates of the reference position are remeasured using position sensors, and the deviation from the theoretical coordinates is calculated; compensation parameter update, where the displacement compensation coefficient and coordinate transformation matrix are updated based on the calibration results; and calibration accuracy verification, where standard test movement is performed to verify the calibration effect. After the calibration procedure is completed, all relevant position parameters and status indicators are updated. The current base position P_base_current is reset to the accurate position after calibration, all robot arms are reset to the "READY" state, the cumulative error counter is cleared, and the task assignment matrix Task_assignment_matrix is ​​recalculated based on the calibrated position information to obtain the optimal assignment scheme.

[0164] Step 602: When the position of the semiconductor device to be captured changes or its state becomes abnormal, update the set of feasible positions and recalculate the optimal candidate position.

[0165] Specifically, the CCD vision inspection device continuously monitors the vibratory feeder area at a frequency of 30 frames per second, detecting changes in device position using an image difference algorithm. Position change detection is based on comparative analysis of consecutive frame images; a change in device center position exceeding 2 millimeters is considered a positional change. Image analysis algorithms identify device anomalies, including device flipping (from face up to back up), device overlap, device damage, and device detachment from the vibratory feeder. Anomaly detection employs a combination of template matching and contour analysis, determining device status by comparing the similarity between the current image features of the device and a standard template.

[0166] When a change in device location or an abnormal status is detected, the feasible location set P_available is immediately updated. The update process includes three operations: removing abnormal device locations, removing device locations with abnormal status from the feasible location set; adding newly discovered locations, adding newly detected normal device locations to the feasible set after security verification; and reordering the set, recalculating the comprehensive evaluation score of each candidate location based on the updated device distribution.

[0167] Based on the updated set of feasible locations, the multi-dimensional evaluation and optimal selection procedure in steps 300-304 is re-executed. The recalculation process considers the current real-time position of the robotic arm, P_compensated_start, and the updated target device distribution, ensuring that the selection decision is based on the latest working environment state. If the previously selected target location, P_next_optimal, becomes invalid due to abnormal device status, a new optimal target is automatically selected from the reordered candidate locations. This decision update mechanism ensures the continuity and reliability of path planning, avoiding task interruptions caused by environmental changes.

[0168] The beneficial effects of this application's embodiments include establishing a three-layer monitoring system. Through multi-dimensional real-time monitoring using position sensors, current sensors, and pressure sensors, it can identify abnormal states of the mechanical, electrical, and vacuum systems. When a robotic arm malfunction is detected, the dynamic task reassignment function can immediately identify incomplete tasks and reassign them to normal equipment based on proximity principles and load balancing strategies. This avoids production interruptions caused by single-point failures, ensures continuous production line operation, and maximizes equipment utilization. When the cumulative error of the base position displacement exceeds the compensation range, a standardized calibration procedure is automatically executed, including reference position regression, sensor zero-point calibration, compensation parameter updates, and accuracy verification, ensuring that positioning accuracy is maintained during long-term operation.

[0169] Figure 8 The diagram shown is a flowchart illustrating a path planning method provided in another embodiment of this application. Figure 8As shown, in the path planning method provided in this application embodiment, after setting the final position as the starting position of the next robotic arm to be operated, it further includes:

[0170] Step 700: Track the work progress of each robotic arm through the task allocation matrix and device grasping status.

[0171] Specifically, a 5×n-dimensional task assignment matrix, Task_assignment_matrix, is maintained, where rows represent robotic arm numbers (1-5) and columns represent target device numbers (1 to n, where n is the total number of devices that can be grasped in the vibratory feeder). Matrix elements use Boolean values ​​to represent the assignment relationship: true indicates that the device has been assigned to the corresponding robotic arm, and false indicates that it has not been assigned.

[0172] A device grabbing status tracking array, Device_grabbed_status[n], is established to record the current processing status of each device. Status types include PENDING, ASSIGNED, GRABBING, GRABBED, and RELEASED. A quantitative assessment of work progress is established, using multiple indicators to comprehensively evaluate the working status of each robotic arm. Work progress indicators include task completion percentage, current stage execution time, and estimated remaining time.

[0173] Establish state synchronization among robotic arms to ensure the timing consistency of coordinated operation among multiple robotic arms. State synchronization is achieved through the loop control module of the control system, with a synchronization frequency set to perform a state check and update every 50 milliseconds. The synchronization control maintains a global state vector, Global_status_vector, which contains the current state, position information, and task progress of all robotic arms.

[0174] Step 701: When the robotic arms have completed their gripping and reached the release position, a preparation signal for a unified rotational release action is triggered.

[0175] Specifically, the robot status array Robot_status[5] is traversed to detect whether all robots have completed the grasping task and reached the carrier release position. The completion status detection conditions include grasping status confirmation, position arrival verification, and device adsorption status check.

[0176] The positioning accuracy of each robotic arm at the carrier release position is finally verified. Higher accuracy is required for the release position, with the tolerance narrowed to ±0.05 mm to ensure accurate placement of the device into the carrier hole. Position verification is achieved by comparing the current position P_base_current[i] with the target release position P_carrier_position[i]. Before performing a unified release, the orientation of all grasped devices is finally verified. The current orientation θ_device_final[i] of each device is confirmed to be consistent with the required orientation of the carrier hole using the angle sensor of the rotary motor. When all robotic arms meet the release conditions, a preparation signal for the unified rotational release action is triggered. After the preparation signal is triggered, the unified release mode is entered, and all robotic arms enter a standby state awaiting the synchronous release command.

[0177] Step 702: Perform a unified rotation release action to simultaneously release the gripped semiconductor devices into the carrier tape.

[0178] Specifically, a timing control system is established to ensure that all five robotic arms can simultaneously perform device release actions. Synchronous control employs a master-slave mode, with the control system acting as the master controller sending unified release commands, and each robotic arm controller acting as a slave device receiving and executing the commands. The rotary motors of all robotic arms are controlled to synchronously perform final orientation adjustments. The rotation action ensures that the orientation of each device perfectly matches the corresponding carrier tape hole position; rotation control is based on a pre-calculated final rotation angle θ_final_rotation[i]. After unified rotation is completed, the vacuum generators of all robotic arms are synchronously shut down, achieving simultaneous device release. Vacuum shutdown employs a delay control mechanism, ensuring that the vacuum is only shut off after the device is in the correct position. Vacuum shutdown confirmation is achieved by monitoring changes in pressure sensor values; successful device release is confirmed when the pressure rises from -80 kPa to above -10 kPa. After device release, the complete execution data of this grasping cycle is recorded, and relevant status parameters are updated.

[0179] The beneficial effects of this application's embodiments are that, by establishing a task allocation matrix and a device grasping status tracking array, the task allocation status, execution progress, and device processing status of each robotic arm can be monitored in real time, ensuring timing coordination and information consistency among multiple robotic arms and effectively avoiding conflicts and duplicate operations between devices. Through rigorous completion status detection, it is ensured that all robotic arms enter a unified release mode after meeting the release conditions.

[0180] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A path planning method, characterized in that, Multiple robotic arms applied to a fully automated carrier tape taping and packaging equipment, the multiple robotic arms being fixed to the same base, sequentially grasp semiconductor devices to be grasped from a vibratory feeder, the method comprising: Obtain the starting position coordinates of the current robotic arm and identify the target position coordinate set of all semiconductor devices to be grasped; Construct a set of feasible locations for the current robotic arm, which includes all candidate locations that are safely accessible and without collision risk; Traverse the set of feasible positions, calculate the distance from the current position of the current robotic arm to each candidate position, and select the candidate position with the closest distance as the next target position for the current robotic arm to move; Remove the next target location from the set of feasible locations and update the set of feasible locations; When the current robotic arm reaches the target position, it performs a grasping task and records the final position after the grasping task is completed; Set the final position as the starting position of the next robotic arm to be operated, and repeat the above steps until all robotic arms have completed their respective grasping tasks or the target position coordinate set is empty.

2. The path planning method according to claim 1, characterized in that, The step of traversing the set of feasible positions, calculating the distance from the current position of the robotic arm to each candidate position, and selecting the closest candidate position as the next target position for the robotic arm includes: The current position coordinates of the base are obtained as the reference point for the current position of the robotic arm; Based on the estimated movement trajectory of the preceding robotic arm and the expected dwell position after the task is completed, the expected position offset of the base when the preceding robotic arm completes its action is calculated, and the base displacement compensation factor is obtained. The base displacement compensation factor is expressed as the position difference vector between the actual position of the base after the preceding robotic arm completes the grasping task and the initial position of the base before the current robotic arm starts its action. The base displacement compensation factor is superimposed on the current robotic arm position to obtain the compensated robotic arm starting position; The straight-line distance from the compensated starting position of the robotic arm to each candidate position is calculated using the Euclidean distance formula; The candidate position with the shortest straight-line distance from the compensated starting position of the robotic arm to each candidate position is selected as the next target position for the current robotic arm movement.

3. The path planning method according to claim 1, characterized in that, The step of traversing the set of feasible positions, calculating the distance from the current position of the robotic arm to each candidate position, and selecting the closest candidate position as the next target position for the robotic arm includes: Each candidate location in the set of feasible locations is evaluated in multiple dimensions, including movement distance evaluation, obstacle avoidance complexity evaluation, and base station displacement impact evaluation. The straight-line movement distance from the current position of the robotic arm to each candidate position is calculated using the Euclidean distance formula. The distance values ​​are then normalized and converted into scores for the movement distance evaluation. Analyze the number of obstacles to be avoided and the length of the detour path in the movement path from the current position of the current robotic arm to each candidate position, and calculate the score of the obstacle avoidance complexity evaluation. The score for the impact evaluation of the base displacement is calculated based on the expected displacement of the base caused by the execution of the preceding robotic arm task. Based on the scores of the movement distance evaluation, the obstacle avoidance complexity evaluation, and the base displacement impact evaluation, a weighted summation is used to calculate the comprehensive evaluation score of each candidate position, and the candidate position with the highest comprehensive evaluation score is selected as the next target position for the current robotic arm movement.

4. The path planning method according to claim 1, characterized in that, The construction of the current set of feasible positions for the robotic arm includes: Establish boundary constraints for the working area, and define the movable XY plane range of the fixed base of the robotic arm and the working height of the robotic arm along the Z axis; Set up an obstacle avoidance zone set, including the coordinate set of obstacle areas inside the device that need to be avoided, to ensure the safety of the robotic arm's movement path; When the straight path from the current robotic arm position to the target position intersects with an obstacle, a feasible detour path is calculated; All candidate locations that are safely accessible and pose no risk of collision are included in the set of feasible locations.

5. A path planning method according to any one of claims 1 to 3, characterized in that, When the current robotic arm reaches the target position, it performs a grasping task and records the final position after the grasping task is completed, including: Control the current robotic arm to move to the target position and execute a standard grasping motion sequence of descent-grasp-ascend; Record the final coordinate data and timestamp information of the base after the grasping task is completed, which will be used for setting the starting position of the robotic arm in the future.

6. A path planning method according to any one of claims 1 to 3, characterized in that, Also includes: The robot arm's operating status is monitored in real time. When a mechanical failure or communication interruption is detected, the assigned tasks of the faulty robot arm are dynamically reassigned to other normal robot arms. When a base displacement is detected to exceed the preset compensation range, a recalibration operation is triggered; When the position of the semiconductor device to be grasped changes or its state becomes abnormal, the set of feasible positions is updated and the optimal candidate position is recalculated.

7. The path planning method according to claim 1, characterized in that, After setting the final position as the starting position of the next robotic arm to be operated, the method further includes: The working progress of each robotic arm is tracked through a task allocation matrix and device grasping status; When all the robotic arms have completed their grasping and reached the release position, a preparation signal is triggered to initiate a unified rotation and release action; Perform a unified rotational release action to simultaneously release the grasped semiconductor devices into the carrier tape.

8. A fully automatic carrier tape taping and packaging equipment, characterized in that, include: The frame is a fixed fixture that supports the overall equipment structure. Multiple robotic arms are fixed to the same movable base, which is connected to the frame and can move in a plane. Each robotic arm is equipped with an end effector for sequentially grasping semiconductor devices. A device supply device includes a vibratory feeder for providing semiconductor devices to be grasped and distributing the semiconductor devices; The detection device includes a vision recognition module and a position sensor. The vision recognition module is used to identify the target position coordinate set of all semiconductor devices to be grasped in the vibratory feeder, and the position sensor is used to obtain the starting position coordinates of the robotic arm in real time. A control system, connected to the plurality of robotic arms and the detection device, includes: The position coordinate processing module is used to receive and process the target position coordinate set and the starting position coordinates obtained by the detection device; The feasible location set construction module, based on the work area boundary and obstacle avoidance requirements, constructs a feasible location set for the current robotic arm, including all candidate locations that are safely accessible and without collision risk. The distance calculation and optimization selection module traverses the set of feasible positions, calculates the distance from the current position of the current robotic arm to each candidate position, and selects the closest candidate position as the next movement target position based on a greedy algorithm. The dynamic set update module removes the next move target location from the feasible location set and updates the available candidate locations in the feasible location set in real time. The task execution control module controls the current robotic arm to move to the target position and perform the grasping task, and records the final position of the movable base after the grasping task is completed; The robotic arm coordination and switching module sets the final position as the starting position of the next robotic arm to be operated, thereby realizing sequential coordination control among multiple robotic arms. The loop control module repeatedly executes the above position planning and task control process until all robotic arms complete their respective grasping tasks or the target position coordinate set is empty.

9. The fully automatic carrier tape taping and packaging equipment according to claim 8, characterized in that, The movable base is driven to move in the XY plane by a driving device, which includes an X-axis stepper motor and a Y-axis stepper motor. Each of the robotic arms further includes: a telescopic device that allows the robotic arm to extend and retract in the Z-axis direction to achieve a grasping action; A rotating device, mounted at the end of the robotic arm, is used to adjust the orientation of the semiconductor device; A position sensor is connected to the control system and provides real-time feedback of the robotic arm's position coordinates. The device supply device includes a vibratory feeder, which causes the semiconductor devices to be evenly distributed on the surface of the feeder. The detection device includes a vision detection device, which identifies the target position coordinate set of the semiconductor device and transmits it to the control system. The control system further includes: a base position reference acquisition module, used to acquire the current position coordinates of the movable base in real time through the position sensor, and set the coordinates as the reference point for calculating the current position of the robotic arm; The trajectory prediction and displacement compensation calculation module is used to calculate the expected position offset of the base when the preceding robotic arm completes its action, based on the predicted movement trajectory of the preceding robotic arm and the expected dwell position after the task is completed, and to obtain the base displacement compensation factor. The base displacement compensation factor is expressed as the position difference vector between the actual position of the base after the preceding robotic arm completes the grasping task and the initial position of the base before the current robotic arm starts its action. The compensation distance calculation module is used to superimpose the base displacement compensation factor onto the current robot arm position to obtain the compensated robot arm starting position; and to calculate the straight-line distance from the compensated robot arm starting position to each candidate position using the Euclidean distance formula. The compensation optimization selection module is used to select the candidate position with the shortest straight-line distance from the compensated starting position of the robotic arm to each candidate position as the next target position for the current robotic arm movement. The dynamic compensation update module updates the base position reference point in real time after each robotic arm completes the grasping task, and recalculates the base displacement compensation factor for the next robotic arm to be operated.

10. The fully automatic carrier tape taping and packaging equipment according to claim 8, characterized in that, The control system further includes: a comprehensive evaluation construction module, which establishes a multi-dimensional evaluation framework for candidate positions, the evaluation framework including a movement distance evaluation dimension, an obstacle avoidance complexity evaluation dimension, and a base position displacement impact evaluation dimension; The movement distance evaluation calculation unit uses the Euclidean distance formula to calculate the straight-line movement distance from the current position of the robotic arm to each candidate position, and converts the distance value into a score for the movement distance evaluation after normalization. The obstacle avoidance complexity evaluation calculation unit analyzes the number of obstacles to be avoided and the length of the detour path in the movement path from the current position of the current robotic arm to each candidate position, and calculates the score of the obstacle avoidance complexity evaluation. The base displacement impact assessment calculation unit calculates the score of the base displacement impact assessment based on the expected displacement of the base caused by the execution of the preceding robotic arm task. The weighted comprehensive evaluation processor calculates the comprehensive evaluation score of each candidate position by weighted summation based on the scores of the movement distance evaluation, the obstacle avoidance complexity evaluation, and the base position displacement impact evaluation. The optimal position decision selector compares the comprehensive evaluation scores of each candidate position in the feasible position set and selects the candidate position with the highest comprehensive evaluation score as the next target position for the current robotic arm.

Images