Robot drilling end effector and operating method thereof

Abstract

The present disclosure relates to an end effector attached to a robot and a method of operating the same. A method of operating an end effector attached to a robot according to one embodiment may include: a step of identifying a global path including curvature information for each section of a workpiece based on three-dimensional design information of a pre-set workpiece; a step of aligning the center of a drill bit of the end effector with respect to the first workpiece when it is identified that the end effector has entered a first workpiece along the global path; a step of adjusting the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first workpiece when the center of the drill bit is aligned; and a step of performing drilling on the first workpiece according to drill control information determined based on torque sensor data of the spindle motor of the end effector.

Description

Technology Field

[0001] The present disclosure relates to a robot drilling end effector and a method of operating the same. More specifically, it relates to an end effector that performs drilling on a workpiece by being detachably attached to a robot and a method of operating the same. Background Technology

[0002] The present invention relates to a robot system for performing a precision drilling process on a large workpiece having a curved surface, such as an aircraft fuselage, and in particular to an intelligent end-effector that fuses multi-sensor information to optimize the process in real time and autonomously avoids obstacles, and a method for controlling the same.

[0003] Large structures, such as fuselages and wings used in the aerospace industry, undergo an assembly process in which numerous parts are fastened together using rivets or fasteners. During this process, the drilling operation, which involves machining thousands to tens of thousands of holes with high precision, is a critical step that determines the quality and efficiency of the entire process. Conventionally, two main methods have been used for this drilling operation.

[0004] The first method involves a worker manually operating the drilling equipment, and the second involves using a large, fixed gantry-type CNC machine. Manual work had limitations, such as large variations in quality depending on the worker's skill level, low productivity, and safety issues like musculoskeletal disorders. Gantry-type CNC machines, introduced to overcome these limitations, provide high rigidity and precision, but they had the disadvantage that the equipment itself was very large and expensive, and it was difficult to flexibly respond to workpieces of various shapes because it was fixed to a specific work line.

[0005] As an alternative to overcome the limitations of such stationary equipment, automated drilling systems utilizing articulated industrial robots are being researched. Articulated robots have the advantage of being able to flexibly approach workpieces with complex curved surfaces based on their wide working radius and high degrees of freedom. However, existing industrial robot-based drilling systems have faced the following inherent technical limitations.

[0006] First, while industrial robots offer high repeatability, their low absolute positional precision makes it difficult to accurately reach target points on CAD data without error. Second, while actual workpieces may experience minute shape errors or accumulated tolerances during the manufacturing and assembly processes, existing robot systems lacked the ability to recognize and correct these discrepancies between CAD data and the actual shapes.

[0007] Third, particularly on curved surfaces such as aircraft fuselages, the drill bit must always remain perpendicular to the surface to ensure hole quality; however, it was very difficult to secure precise verticality due to robot position errors and the uncertainty of the actual curved surface. Fourth, there was a lack of intelligent process control functions that could detect in real time and actively respond to quality degradation factors, such as chip clogging occurring during the drilling process or burrs generated at the moment of penetrating the material.

[0008] Finally, the lack of autonomous path movement capabilities to detect obstacles such as temporary tack rivets or permanent fasteners in the work path in advance and avoid them without collision became a factor hindering process stability and full automation.

[0009] Therefore, there is an urgent need for innovative technologies capable of maximizing the flexibility of articulated robots while achieving the accuracy and process intelligence levels of high-precision CNC machines. Specifically, it is an urgent task to develop next-generation robotic drilling end effectors and their control technologies that can fuse multi-sensor information in real time to autonomously correct robot position errors and uncertainties in the actual working environment, intelligently optimize the entire drilling process, and autonomously avoid unexpected obstacles. Prior art literature

[0010] Korean Registered Patent No. 2508871 The problem to be solved

[0011] According to one embodiment, an end effector that is detachably attached to a robot and a method of operating the same may be provided.

[0012] According to one embodiment, an end effector and a method of operation thereof may be provided, which automatically aligns drilling points and moves to the next drilling point when drilling is completed at the first point. means of solving the problem

[0013] As a means for achieving the technical problem described above, a method of operation of an end effector attached to and detached from a robot may be provided, comprising: a step of identifying a global path including section-by-section curvature information of a workpiece based on three-dimensional design information of a pre-set workpiece; a step of aligning the center of a drill bit of the end effector with respect to a first work point when it is identified that the end effector has entered a first work point along the global path; a step of adjusting the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first work point when the center of the drill bit is aligned; and a step of performing drilling on the first work point according to drill control information determined based on torque sensor data of the spindle motor of the end effector.

[0014] According to one embodiment, the method may further include the step of moving to the second work point based on a movement path connecting the first work point and the second work point identified on the global path when work is completed at the first work point.

[0015] According to one embodiment, the step of performing the drilling may include: determining cutting energy based on the torque and angular velocity of the spindle motor; determining the time rate of change of the determined cutting energy; identifying the state in which the drill bit has penetrated the workpiece when the determined time rate of change decreases below a preset threshold; and stopping the feed of the drill bit when the penetration state is identified. According to another example, it is obvious that the state in which the drill bit has penetrated the workpiece may be identified when the time rate of change falls within a preset range or when the time rate of change falls outside a preset range.

[0016] According to one embodiment, the movement path may include a geodesic line, which is the shortest surface path, determined by projecting the three-dimensional coordinates of the first work point and the second work point onto a three-dimensional curved surface model, and a variable contour path determined by offsetting each point on the determined geodesic line by a constant safe separation distance in the direction of a normal vector perpendicular to the surface.

[0017] According to one embodiment, the segment type includes an interest segment type in which the curvature value exceeds a threshold, and a general segment type in which the curvature value identified for each segment does not exceed a threshold, and the global path may be provided with a different density of command points for generating control commands for controlling the operation of the end effector depending on the segment type.

[0018] According to one embodiment, the density of the command points for the interest section type can be set higher than the density of the command points for the general section type.

[0019] According to one embodiment, the step of moving to the second work point may include: acquiring camera images of the path ahead from the camera of the end effector while moving along the path; checking whether a suspected curvature section different from the curvature information in the 3D design information is identified based on the optical flow between the acquired camera images; if the suspected curvature section is identified, activating displacement sensors located on the attachment surface of the attachment frame of the end effector to check the actual curvature information of the work object surface; modifying the command point density pre-assigned by the section type based on the identified actual curvature information; and moving to the second work point along a global path including the modified command point density.

[0020] According to one embodiment, the step of identifying the chip jamming pattern may include: generating a frequency spectrum by performing a fast Fourier transform on the torque sensor data; and identifying the chip jamming pattern when, from the generated frequency spectrum, a sideband component adjacent to the rotational frequency of the spindle motor and the harmonic component of the rotational frequency is identified, or when a rise in the noise floor level of the generated frequency spectrum is identified.

[0021] According to one embodiment, the step of identifying the workpiece in a penetrating state may include: identifying a penetration prediction section before the end of the drill bit penetrates the workpiece based on thickness information of the workpiece and shape information of the drill bit; activating a penetration detection algorithm that shortens the size of the time window of a moving average that calculates the time rate of change of the cutting energy to a value shorter than the normal cutting section when the feed distance of the drill bit enters the penetration prediction section; and identifying the drill bit in a penetrating state when the time rate of change of the cutting energy calculated through the shortened time window decreases to below the preset negative threshold.

[0022] According to one embodiment, the step of moving to the second work point may include: a step of identifying an obstacle according to the classification information of the obstacle by applying a pre-trained deep learning-based object detection model to the acquired camera images; a step of generating a local avoidance path when a collision between the identified obstacle and the end effector is predicted; and a step of moving to the second work point according to the generated local avoidance path.

[0023] According to one embodiment, the deep learning-based object detection model may be a neural network-based model that outputs classification information indicating the type of obstacle by being pre-trained based on an image dataset of temporary fixing rivets and permanent fixing fasteners.

[0024] According to one embodiment, the step of generating the local avoidance path may include: generating a vertical avoidance path by modifying the movement path upward through spline interpolation so that a vertical separation distance from the temporary fixing rivet is secured when the obstacle is identified as a temporary fixing rivet according to the obstacle classification information; and generating a horizontal avoidance path by gently bypassing the movement path to the side so that a preset horizontal separation distance from the fastener is secured when the obstacle is identified as a permanent fixing fastener according to the obstacle classification information.

[0025] According to one embodiment, the step of performing the drilling may include: gradually applying a preset pressure to the surface of the first work point through the drill bit; identifying the stiffness of the surface of the workpiece based on the displacement values ​​of the displacement sensors of the end effector according to the pressure; and correcting the feed rate included in the drill control information based on the identified stiffness. However, according to another example, the step of performing the drilling may not perform the step of identifying the stiffness of the surface of the workpiece, but may simply correct the feed rate included in the drill control information based only on the displacement values ​​of the displacement sensors of the end effector.

[0026] According to one embodiment, the step of identifying the normal vector may include: obtaining displacement values ​​from three displacement sensors located at three preset points on the attachment surface of the attachment frame of the end effector to the surface of the workpiece; determining three point coordinates in three-dimensional space corresponding to the three points based on the three displacement values; and identifying the normal vector of the surface of a plane that includes all three point coordinates. However, according to another example, four or more preset displacement sensors may be attached to the attachment surface of the attachment frame of the end effector, or a smaller number of displacement sensors may be attached. The attachment surface of the attachment frame of the end effector according to the present disclosure may include at least one displacement sensor.

[0027] According to one embodiment, the step of identifying the normal vector may include: acquiring displacement values ​​from each of four or more displacement sensors located at each of four or more points pre-set on the attachment surface of the attachment frame of the end effector to the surface of the workpiece; determining four or more point coordinates in three-dimensional space corresponding to the four or more points through the acquired four or more displacement values; applying a Least Squares Method fitting algorithm to the four or more point coordinates to calculate a best-fit plane that minimizes the error; and identifying the normal vector of the calculated best-fit plane.

[0028] According to one embodiment, the step of adjusting the verticality of the central axis may include, prior to the step of detecting the initial contact of the drill bit, a step of projecting a structured light pattern onto the surface of the first work point through the projection unit of the end effector, analyzing the distorted pattern acquired by the camera of the end effector to identify a first normal vector, and then adjusting the first verticality of the central axis according to the first normal vector; and a step of adjusting the verticality of the central axis according to a second normal vector of the surface of the work object identified based on the displacement values ​​of the plurality of displacement sensors acquired after the adjustment of the first verticality is completed.

[0029] According to another embodiment as a technical means for achieving the technical problem described above, an end effector detachably attached to a robot comprises: a frame having a mounting portion detachably attached to the robot and an attachment surface that contacts the surface of a workpiece; a head supported by the frame and including a drill bit and a spindle for driving the drill bit; a camera attached to the frame to acquire an image by photographing the workpiece; and at least one displacement sensor positioned on the attachment surface to acquire displacement values ​​for the surface of the workpiece. An end effector may be provided, comprising: a control unit that identifies a global path including segment-by-segment curvature information of the workpiece based on three-dimensional design information pre-set for the workpiece, identifies that the end effector has entered a first work point along the global path, aligns the center of the drill bit of the end effector with respect to the first work point, and when the center of the drill bit is aligned, adjusts the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first work point, generates drill control information based on torque sensor data of the spindle motor of the end effector, and controls the operation of the end effector based on the drill control information.

[0030] According to another embodiment as a technical means for achieving the technical problem described above, a computer-readable recording medium storing a program for performing a method of operating an end effector attached to and detached from a robot may be provided, the method comprising: a step of identifying a global path including curvature information for each section of a workpiece based on three-dimensional design information of a pre-set workpiece; a step of aligning the center of a drill bit of the end effector with respect to a first work point when it is identified that the end effector has entered a first work point along the global path; a step of adjusting the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first work point when the center of the drill bit is aligned; and a step of performing drilling on the first work point according to drill control information determined based on torque sensor data of the spindle motor of the end effector. Effects of the invention

[0031] According to one embodiment, a drilling target point can be accurately identified on the surface of a workpiece including an aircraft part having a fine curved surface, such as an aircraft fuselage, or various aircraft parts having a flat surface.

[0032] According to one embodiment, the verticality of the surface of the workpiece can be precisely controlled.

[0033] According to one embodiment, obstacles such as fasteners and temporary rivets can be identified and avoided on the work path of a work object. Brief explanation of the drawing

[0034] FIG. 1 is a diagram schematically illustrating the operation of an end effector according to one embodiment and a robot drilling system to which the end effector is applied. FIG. 2 is a drawing showing an example of an end effector according to one embodiment. FIG. 3 is a flowchart of the operation method of an end effector according to one embodiment. FIG. 4 is a flowchart of the operation method of an end effector according to another embodiment. FIG. 5 is a flowchart of a specific method for aligning the center of a drill bit according to one embodiment. FIG. 6 is a flowchart of a specific method for adjusting the verticality of a drill bit according to one embodiment. FIG. 7 is a flowchart of a specific method for performing drilling according to one embodiment. FIG. 8 is a flowchart of a specific method for moving to a second work point after completing work at a first work point according to one embodiment. FIG. 9 is a block diagram of an end effector according to one embodiment. FIG. 10 is a block diagram of an end effector according to another embodiment. Specific details for implementing the invention

[0035] The terms used in this specification will be briefly explained, and the present disclosure will be described in detail.

[0036] The terms used in this disclosure have been selected to be as widely used and general as possible, taking into account their functions within this disclosure; however, these terms may vary depending on the intent of those skilled in the art, case law, the emergence of new technologies, etc. Additionally, in specific cases, terms have been selected at the applicant's discretion, and in such cases, their meanings will be described in detail in the relevant description of the invention. Therefore, terms used in this disclosure should be defined not merely by their names, but based on their meanings and the overall content of this disclosure.

[0037] When a part of a specification is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Furthermore, terms such as "...part" or "module" as used in the specification refer to a unit that processes at least one function or operation, and this may be implemented in hardware or software, or as a combination of hardware and software.

[0038] Embodiments of the present disclosure are described below with reference to the attached drawings so that those skilled in the art can easily implement them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present disclosure in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0040] FIG. 1 is a diagram schematically illustrating the operation of an end effector according to one embodiment and a robot drilling system to which the end effector is applied.

[0041] The aircraft component manufacturing management system (10) may include a robot drilling system (100) and a server (3000). The robot drilling system (100) may be connected to the server (3000) via a network. The robot drilling system (100) according to the present disclosure may identify work points for various aircraft components having surfaces such as flat or curved surfaces, and may autonomously perform drilling operations at the identified work points.

[0042] The server (3000) can store master data that forms the basis of the entire drilling process and perform preliminary calculations. For example, the server (3000) can store 3D design information in advance, such as 3D CAD data of the workpiece and a list of coordinates of all target drilling points. Furthermore, the server (3000) can analyze the 3D design information in an offline stage before the actual drilling operation begins, calculate the change in curvature of the workpiece surface, and generate an initial global path that connects all target points by the shortest distance based on this. In this process, the server (3000) can generate optimized global path information by differentially assigning high-density command points to sections with steep curvature and low-density command points to sections with gentle curvature, and transmit this to the robot drilling system (100). However, according to another example, the end effector can perform drilling operations independently with the robot arm without being connected to the server (3000).

[0043] According to one embodiment, the robot drilling system (100) may include an end effector (1000) and a robot (2000). The robot (2000) may include a robot arm that operates based on control information. The end effector (1000) may be a device that is detachably attached to the robot (2000) to perform a desired operation on an aircraft component.

[0044] More specifically, the robot drilling system (100) can perform actual drilling operations based on a global path received from the server (3000). Specifically, the robot (2000) may include a multi-joint robot arm, etc., and may be responsible for macroscopic movement to transport the end effector (1000) along the global path to near the target drilling point.

[0045] According to one embodiment, an end effector (1000) attached to a robot can perform drilling operations on a workpiece based on control information. For example, the end effector (1000) can perform drilling operations on multiple target work points by contacting aircraft parts having flat or curved surfaces. The end effector (1000) according to the present disclosure can accurately identify target drilling points on the surface of a workpiece, such as a fine curved surface, a general curved surface, or a flat surface, such as an aircraft fuselage, can precisely control the perpendicularity to the surface of the workpiece, and can identify and avoid obstacles such as fasteners and temporary rivets on the work path of the workpiece.

[0046] More specifically, the end effector (1000) can be detachably coupled to the end of the robot (2000) and is equipped with its own control unit and multiple sensors, so that it can autonomously perform microscopic and intelligent tasks such as precise alignment after reaching a target point, securing verticality, executing intelligent drilling, and real-time path optimization when moving between points. That is, the end effector (1000) can perform the drilling process by actively responding to the uncertainty of the actual work environment based on sensor data acquired in real time from built-in sensors and control information received from the server (3000). Through this, the end effector (1000) according to the present disclosure and the robot drilling system (100) including it can have the effect of maximizing the precision and productivity of the entire process by highly accurately identifying and aligning the drilling target point on the surface of the workpiece, precisely controlling the verticality with respect to the surface, and identifying in real time and autonomously avoiding obstacles such as pre-fastened rivets or fasteners present on the movement path.

[0048] FIG. 2 is a drawing showing an example of an end effector according to one embodiment.

[0049] Referring to the figure of FIG. 2, the end effector (1000) may include a frame that protects and supports various internal components. The frame may include a robot attachment surface (212) on one side that is coupled to the robot arm of the robot (2000), and on the other side, a workpiece attachment surface (214) that directly contacts the surface of the workpiece or maintains a certain distance during a drilling operation.

[0050] According to one embodiment, the end effector (1000) may include multiple sensor units for performing a precise drilling process. For example, at least one vision sensor (not shown) may be provided on the front of the end effector (1000) to capture the shape and position of a pilot hole, which is a target point to be drilled, and to acquire image data. The vision sensor (not shown) may include a camera, and the acquired image data may be used in a vision servoing process to correct the position error of the robot (2000) or the position error of the end effector in real time and to precisely align the center of the drill bit with the center of the pilot hole.

[0051] Additionally, a plurality of displacement sensors (222, 224, 226) for measuring the distance to the surface of the workpiece may be placed on the workpiece attachment surface (214) or in an adjacent area. In one embodiment, three or more displacement sensors (222, 224, 226) may be spaced apart and, by using the deviation of the displacement values ​​measured from each displacement sensor, the end effector (1000) can accurately calculate the local inclination of the workpiece surface at the current position, i.e., the normal vector.

[0052] This can provide key information for correcting the posture so that the central axis of the drill bit always remains perpendicular to the surface of the workpiece. Meanwhile, the end effector (1000) may include a head unit that performs the actual drilling operation. The head unit may include a spindle (232) that mounts a drill bit (not shown) and rotates it at high speed.

[0053] The spindle (232) may be driven by an internal spindle motor (not shown), and a torque sensor provided in the spindle motor may be used to monitor the cutting load generated during the drilling process in real time. Additionally, the spindle (232) may perform a feed operation of the drill bit by moving forward or backward (e.g., in the Z-axis direction) along a spindle axis (not shown) defined by a drive system such as a ball screw and a servo motor. Through this, the end effector (1000) can precisely control the drilling depth and intelligently control the drilling process by detecting specific process events such as chip jamming or penetration.

[0055] FIG. 3 is a flowchart of the operation method of an end effector according to one embodiment.

[0056] FIG. 3 is a diagram showing the flow of the overall operation method of an end effector (1000) according to one embodiment of the present disclosure. Each step illustrated in FIG. 3 can be performed by a control unit embedded in the end effector (1000), an external server (3000) communicating with it, or by the embedded control unit and the external server interacting with each other.

[0057] In S310, the end effector (1000) can identify a global path including segment-by-segment curvature information of the workpiece based on pre-set three-dimensional design information of the workpiece (e.g., 3D CAD data, but not limited thereto). The global path may include an optimal basic movement path connecting all target drilling points, and in particular may include segment-by-segment curvature information of the workpiece surface (e.g., segments connecting multiple work points, for example, the segment connecting the first work point and the second work point may be defined as the first segment, and the segment connecting the second work point and the third work point may be defined as the second segment). The control unit (950) can identify a global path including curvature information for the first segment and the second segment. For example, the control unit (950) can analyze three-dimensional design information to identify a ‘preliminary interest section’ where the curvature of the surface exceeds a preset threshold and other ‘general sections,’ and generate or verify a global path by assigning different command point densities according to the characteristics of each section.

[0058] More specifically, step S310 can be performed through the following steps. For example, the end effector (1000) can determine a pre-determined curvature value for each section of the surface of the workpiece based on the three-dimensional design information. The end effector (1000) can obtain mesh data or point cloud data representing the surface shape of the workpiece from three-dimensional CAD data. Subsequently, the control unit (950) can calculate curvature indices, such as principal curvatures or Gaussian curvature, for each point or pre-defined section of the surface data. Through this, the end effector (1000) can determine how much and in which direction the surface is curved as quantitative numerical data.

[0059] Subsequently, the end effector (1000) can determine the section type of the workpiece surface, which is predetermined based on whether the curvature value confirmed for each section exceeds a preset threshold. For example, a section where the calculated curvature value exceeds the threshold and the shape changes rapidly can be classified as a 'Region of Interest' type, which requires precise attitude change control of the robot, and a section where the curvature value shows a gradual change below the threshold can be classified as a 'General Section' type, which allows for high-speed movement. This selective classification of section types can be utilized as key basic information for dynamically adjusting the robot's movement speed or control precision according to the characteristics of each section in the future.

[0060] Additionally, the end effector (1000) can verify the global path containing information regarding the curvature values ​​for each section and the identified section type. Specifically, the end effector (1000) can generate a basic path connecting all target drilling points in order of shortest distance, and the generated global path may have a data structure that includes not only the 3D coordinate information of each path point, but also the calculated curvature value of the point and an identifier indicating whether it is a 'section of interest' or a 'general section' as attribute information. A global path containing such diverse geometric information can enable intelligent robot control in subsequent steps, such as an optimal control strategy suited to the characteristics of each section, for example, differentially allocating the density of command points.

[0061] According to one embodiment, the segment type includes a segment of interest type, which is a segment where the curvature value exceeds a threshold, and a general segment type, which is a segment where the curvature value identified for each segment does not exceed a threshold. Depending on the segment type, the global path may have a different density of command points where control commands for controlling the operation of the end effector are generated. For example, the density of command points for the segment of interest type may be set higher than the density of command points for the general segment type.

[0062] In S320, when the end effector (1000) is identified as having entered a first work point along the global path, the end effector can align the center of the end effector's drill bit with respect to the first work point. For example, the end effector can perform the step of precisely aligning the center of the drill bit with the center of the first work point (e.g., the center of the pilot hole). Step S320 may be performed through a rough position arrival determination based on encoder data of the robot (2000) and subsequent precise vision servoing using the vision sensor (222) of the end effector (1000).

[0063] In S330, the end effector (1000) can adjust the perpendicularity of the center axis of the drill bit to the surface of the workpiece containing the first work point when the center of the drill bit is aligned. For example, this can be achieved by the end effector (1000) precisely detecting the moment when the drill bit first contacts the surface of the workpiece through a change in the torque of the spindle motor, immediately calculating the local surface normal vector of the corresponding point using a plurality of displacement sensors (224, 226), and then fine-tuning the posture of the robot (2000) so that the center axis of the drill bit is parallel to the calculated normal vector.

[0064] In S340, the end effector (1000) can perform drilling on the first work point according to drill control information determined based on torque sensor data of the spindle motor of the end effector. For example, the end effector (1000) performs actual drilling on the first work point when both position and verticality are secured.

[0065] Step S340 can be performed according to intelligent drill control information determined by continuously analyzing data collected in real time from the torque sensor of the spindle motor, going beyond simply rotating and moving the drill bit. For example, the control unit (950) can perform various intelligent controls, such as stopping drilling, retracting, or performing peck drilling by detecting chip jamming through frequency analysis of torque data, or predicting the moment of material penetration and minimizing burr generation by detecting a rapid decrease in cutting energy. Additionally, according to one embodiment, if chip jamming is detected through frequency analysis of torque data, the control unit (950) can perform intelligent operations for chip evacuation by repeatedly retracting and advancing the drill bit to a certain depth or a certain distance.

[0067] FIG. 4 is a flowchart of the operation method of an end effector according to another embodiment.

[0068] S410 to S440 may correspond to S310 to S340 disclosed in FIG. 3, so a detailed description is omitted. Through these steps, the end effector (1000) can complete precise alignment, verticality securing, and intelligent drilling for the first work point.

[0069] In S450, when the end effector (1000) completes the work for the first work point, it may move to the second work point based on a movement path connecting the first work point and the second work point identified on the global path. Step S450 may not simply connect the two points, but may include an intelligent path generation and optimization process to maximize safety and efficiency.

[0070] According to one embodiment, the movement path may include a geodesic, which is the shortest surface path, determined by projecting the three-dimensional coordinates of the first work point and the second work point onto a three-dimensional curved surface model, and a variable contour path determined by offsetting each point on the determined geodesic by a certain safe distance in the direction of a normal vector perpendicular to the surface.

[0071] More specifically, the end effector (1000) can first project the 3D coordinates of the first work point and the 3D coordinates of the second work point onto a 3D curved surface model of the work object obtained from 3D design information. Then, the end effector (1000) can calculate the shortest path, i.e., a geodesic, when moving between these two points on the curved surface. While a straight path in 3D space carries a risk of collision with the curved surface, a geodesic path can provide the most efficient surface movement path that reduces non-processing time by minimizing unnecessary movement.

[0072] Next, the end effector (1000) can perform an offset process to ensure safety based on the calculated geodesic path. Specifically, the end effector (1000) can calculate a normal vector perpendicular to the surface of each point at each path point constituting the geodesic line. Subsequently, points in three-dimensional space located at a predetermined 'safety separation distance' (e.g., 50 mm) in the direction of the normal vector calculated at each path point can be set as new path points. By applying this process over the entire geodesic line, the final movement path can be generated as a 'variable contour path' that follows the curved shape of the workpiece while always maintaining a constant height.

[0073] This path generation method allows the end effector (1000) to move to the next target point in the shortest possible time while maintaining a safe distance from the surface, regardless of the curvature of the curved surface, thereby maximizing the productivity and stability of the entire drilling process.

[0074] Furthermore, while the end effector (1000) moves along this variable contour path, the control unit of the end effector can perform dynamic path replanning to avoid unexpected obstacles in real time. According to one embodiment, the step of moving to the second work point may include: identifying an obstacle according to the classification information of the obstacle by applying a deep learning-based object detection model that has been pre-trained on the acquired camera images; generating a local avoidance path when a collision between the identified obstacle and the end effector is predicted; and moving to the second work point according to the generated local avoidance path.

[0075] More specifically, the control unit (950) may apply a deep learning-based object detection model, such as YOLO (You Only Look Once) or SSD (Single Shot MultiBox Detector), to a camera image taken of the path ahead. Alternatively, the control unit according to the present disclosure may perform a rule-based object detection process without using the aforementioned conventional deep learning-based object detection model. According to one embodiment, the deep learning-based object detection model may be a neural network-based model pre-trained based on an image dataset containing a large number of images of temporary fixing rivets (e.g., Cleco) and permanent fixing fasteners (e.g., Hi-Lok). This model may output not only bounding box information indicating the location and size of an obstacle within the camera image, but also classification information indicating the type of obstacle that distinguishes whether the obstacle is a 'temporary fixing rivet' or a 'permanent fixing fastener'. The control unit (950) can immediately generate a local avoidance path if a collision possibility is predicted based on the location of the identified obstacle and the current path of the end effector (1000).

[0076] According to one embodiment, the step of the control unit generating the local avoidance path may apply an optimal avoidance strategy differentiated according to the 'classification information' of the previously identified obstacle. For example, if the obstacle is identified as a 'temporary fixing rivet' according to the obstacle classification information, the control unit (950) may perform the step of generating a vertical avoidance path. Temporary fixing rivets generally have a low protrusion height. Therefore, the control unit (950) can generate a vertical avoidance path that ensures a minimum vertical separation distance sufficient to pass over the top of the temporary fixing rivet by smoothly modifying the existing movement path upward using a spline interpolation technique, etc., without needing to deviate significantly from the path to the left or right. This enables efficient avoidance by minimizing the increase in movement distance.

[0077] On the other hand, if an obstacle is identified as the 'permanent fixing fastener' according to the obstacle classification information, the control unit (950) may perform the step of generating a horizontal avoidance path. Depending on the type, the permanent fixing fastener may have a high protrusion height or be a critical component that has already been finally fastened, requiring a more conservative approach. Therefore, the control unit (950) may generate a horizontal avoidance path that gently bypasses the existing movement path to the side to secure a sufficient horizontal separation distance that is pre-set for the fastener. In this way, the present disclosure goes beyond simply detecting obstacles to intelligently classify their types and dynamically generates and applies an optimal avoidance path (vertical avoidance or horizontal avoidance) suitable for the characteristics of each type. Through this, the end effector (1000) can achieve excellent effects that ensure collision safety for all types of obstacles while minimizing unnecessary bypass maneuvers, thereby shortening non-processing time and maximizing the productivity of the entire process.

[0079] FIG. 5 is a flowchart of a specific method for aligning the center of a drill bit according to one embodiment.

[0080] FIG. 5 is a flowchart for further explaining the step (S320) of aligning the center of a drill bit according to one embodiment of the present disclosure. This alignment process can be performed in two main steps: a rough approach step (S510) to the vicinity of a target point and a precision correction step (S520) using a vision sensor.

[0081] In S510, the end effector (1000) can identify that it has entered the first work point based on coordinate information for the first work point appearing in the global path and encoder coordinate information according to the encoding data of the end effector's servo motor. More specifically, the end effector (1000) can collect encoder data in real time from servo motors attached to each joint of the robot (2000) and apply forward kinematics operations to the encoder values ​​to calculate the current 3D spatial coordinates corresponding to the tool center point (TCP) of the drill bit. It is not limited to the above 3D spatial coordinate calculation process, and other coordinate calculation algorithms for identifying the position of the first work point based on encoder coordinate information can be used.

[0082] Additionally, the end effector (1000) can retrieve target 3D coordinate information of the first work point from a previously verified global path. The end effector (1000) continuously calculates the Euclidean distance between the calculated current coordinates and the target coordinates, and when this distance enters within a preset 'precision search start radius', it can stop high-speed movement and switch to the next step, the precision alignment sequence. According to another example, control information for performing step S510 can be received from a server or a robot, and the end effector may perform step S510 based on the control information determined by the robot or the server.

[0083] In S520, when the end effector (1000) is identified as having entered the first work point, it can identify a plurality of candidate groups representing the edges of the pilot hole centered on the center coordinates of the pilot hole from the camera image of the first work point, and align the center of the drill bit such that the error between the center of one of the candidate groups and the center of the drill bit is less than or equal to a threshold.

[0084] According to one embodiment, the step S520 of aligning the center of the drill bit can be implemented through visual servoing technology that controls the robot's posture in real time based on image feedback. In particular, the following series of processes can be repeated at high speed as a single closed-loop control cycle until the pixel error between the center of the drill bit and the center of the pilot hole converges within a preset allowable range.

[0085] First, the end effector (1000) may perform the step of acquiring a camera image of a first work point. Specifically, a camera (930) equipped in the end effector (1000) may capture a high-resolution digital image in one frame by photographing the surface of the work object included in the current field of view. Next, the control unit (950) may perform the step of binarizing the acquired camera image. This is to analyze each pixel value within the image to clearly distinguish between a surface area that appears bright due to lighting and a pilot hole area that appears dark because light is not reflected.

[0086] For example, the end effector (1000) can convert the image into a binary image composed of black and white pixels by applying an adaptive thresholding technique such as Otsu's algorithm, thereby separating the pixel regions corresponding to the pilot hole from the background. Subsequently, the control unit (950) can perform the step of detecting the edges of the pixel regions corresponding to the pilot hole from the binarized camera image.

[0087] For example, a set of edge pixels corresponding to the boundary line, i.e., the border, of the pilot hole region in a binarized image can be precisely extracted using a Canny edge detection algorithm or a Sobel filter. Then, the control unit (950) can perform the step of determining one candidate among the plurality of candidate groups by applying a Hough Circle Transform algorithm to the detected edges. The Hough Circle Transform algorithm is a technique for finding the center point and radius that best satisfy the equation of a circle from a set of edge pixels.

[0088] The end effector (1000) can, through this algorithm, finally determine the most distinct and reliable circle, that is, the actual pilot hole, among a plurality of circle candidates that may consist of extracted edge pixels, and mathematically calculate the center pixel coordinates (u, v) and radius r of the candidate. Finally, the end effector (1000) can perform the step of aligning the center of the drill bit by repeatedly performing the step of acquiring the camera image and the step of determining the one candidate so that the error between the center of the determined candidate and the center of the drill bit is below a threshold.

[0089] Specifically, the control unit (950) can calculate a pixel error (Δu, Δv) between the center pixel (u, v) of the calculated candidate and the pixel coordinates (uc, vc) corresponding to the center of the camera image. This pixel error can be converted into a physical error (ΔX, ΔY) that must be moved in real space using the intrinsic parameter of the camera and the height information (Z-depth) of the current end effector. The control unit (950) can calculate the amount of rotation of each joint of the robot (2000) required to offset this physical error through an inverse Jacobian matrix or inverse kinematics operation, and drive the servo motor by the calculated amount to move the center of the drill bit finely. This entire loop, consisting of 'image acquisition - binarization - edge detection - candidate determination - error calculation - robot movement,' is performed repeatedly until the pixel error converges to below a preset threshold (e.g., ±1 pixel) and it is determined that no further correction is required. Through this, the center of the drill bit can be highly precisely aligned with the center of the pilot hole to a level below a certain threshold, despite the robot's kinematic errors or uncertainties in the external environment.

[0091] FIG. 6 is a flowchart of a specific method for adjusting the verticality of a drill bit according to one embodiment.

[0092] In S610, the end effector (1000) can detect the initial contact of the drill bit with the surface of the workpiece based on the rate of change of the torque sensor data of the spindle motor after the step of aligning the center of the drill bit. For example, the end effector (1000) can slowly lower the drill bit toward the surface of the workpiece while driving the spindle motor in an idling state beforehand.

[0093] The end effector (1000) can monitor torque data output from the torque sensor of the spindle motor in real time while performing step S610. When the end of the drill bit touches the surface, the torque increases rapidly due to cutting resistance, and the control unit (950) can detect the point when the rate of change of this torque value per hour (dTorque / dt) exceeds a preset threshold as 'initial contact' and immediately stop the feed of the drill bit.

[0094] In S620, when the end effector (1000) detects the initial contact of the drill bit with the surface of the workpiece, it can identify (or extract) the normal vector of the surface of the workpiece based on the displacement values ​​of a plurality of displacement sensors located on the attachment surface of the attachment frame of the end effector while the transfer of the drill bit is stopped. A specific method for identifying the normal vector may include various embodiments as follows, depending on the number of sensors and the algorithm applied.

[0095] According to one embodiment, the step of identifying the normal vector may include: obtaining displacement values ​​from three displacement sensors located at three preset points (e.g., triangle vertices) on the attachment surface of the attachment frame (1014) of the end effector (1000) to the surface of the workpiece; determining three point coordinates in three-dimensional space corresponding to the three points based on the three displacement values; and identifying a normal vector of a plane that includes all three point coordinates.

[0096] According to another example, as disclosed in this specification, when there are four displacement sensors, four displacement values ​​can be obtained, and a normal vector can be identified based on the four obtained displacement values.

[0097] More specifically, the control unit can calculate the coordinates of three points P1, P2, and P3 in three-dimensional space by combining the measured values ​​of each displacement sensor with information on the mounting positions of the known sensors within the end effector coordinate system. Since geometrically, three points that are not on the same straight line define a unique plane, the control unit (950) can generate two vectors (e.g., vector V1 = P2-P1, vector V2 = P3-P1) using these three points and calculate the cross product of these two vectors to precisely identify a normal vector perpendicular to the plane.

[0098] As another embodiment, to ensure higher accuracy and robustness, the step of identifying the normal vector may use four or more displacement sensors. In this case, the step of identifying the normal vector may include: a step of acquiring displacement values ​​from four or more displacement sensors located at four or more points each pre-set on the attachment surface of the attachment frame of the end effector to the surface of the workpiece; a step of determining four or more point coordinates in three-dimensional space corresponding to the four or more points through the acquired four or more displacement values; a step of calculating a best-fit plane that minimizes the error by applying a least squares method fitting algorithm to the four or more point coordinates; and a step of identifying the normal vector of the calculated best-fit plane.

[0099] Using four or more sensors provides redundancy to the measurement data, which has the effect of minimizing the impact of minute measurement errors of a specific sensor or local irregularities on the surface of the workpiece on the overall calculation result. For the acquired four or more point coordinate data, the control unit can mathematically calculate the optimal approximation plane through the least squares method, such that the sum of the squared distances from each point to the plane is minimized. Then, the control unit (950) can finally identify the normal vector of the optimal approximation plane calculated in this way as the most reliable surface normal vector of the corresponding point.

[0100] As another embodiment, the step of adjusting the verticality of the central axis may be performed in a two-step (Coarse-to-Fine) manner combining non-contact pre-adjustment and contact precision adjustment to maximize efficiency and precision. In this case, the step of adjusting the verticality of the central axis may include the following. First, prior to the step of detecting the initial contact of the drill bit, the control unit may project a structured light pattern onto the surface of the first work point through the projection unit (not shown) of the end effector (1000), analyze the distorted pattern acquired by the camera (930) of the end effector to identify a first normal vector, and then perform the step of adjusting the first verticality of the central axis according to the first normal vector. This is a process of determining the approximate inclination of the surface in advance and roughly aligning the orientation of the drill bit before the end effector physically contacts the workpiece, that is, during the process of approaching the target point. This prevents the drill bit from contacting the surface in an excessively tilted state upon final contact, thereby increasing stability and reducing the time required for subsequent precision adjustments.

[0101] After the first verticality adjustment is completed, the control unit (950) can detect initial contact by slowly approaching the drill bit to the surface as described in the preceding embodiments, and identify a more precise second normal vector based on the displacement values ​​of the plurality of displacement sensors (940). Then, the control unit (950) performs a step of finally fine-tuning the verticality of the central axis according to this second normal vector. This two-stage verticality adjustment method organically combines the advantages of different methods, namely the speed of non-contact measurement and the precision of contact measurement. Through this, the present disclosure can provide an excellent effect of improving productivity by shortening the time required to secure overall verticality, while simultaneously maximizing final alignment accuracy to ensure the highest level of drilling quality.

[0102] In S630, the end effector (1000) can adjust the verticality of the center axis of the drill bit according to the direction of the identified normal vector. For example, the end effector (1000) can compare the direction of the identified normal vector with the direction of the current center axis of the drill bit to calculate the difference, and finely drive specific joints, such as the wrist joint of the robot (2000), to compensate for this difference. By doing so, the orientation of the center axis of the drill bit is adjusted so that it is perfectly parallel to the identified normal vector, thereby completing the securing of final verticality.

[0104] FIG. 7 is a flowchart of a specific method for performing drilling according to one embodiment.

[0105] FIG. 7 is a flowchart specifically illustrating internal sub-steps of an intelligent drilling execution step (S340) according to one embodiment of the present disclosure. This step is a process of processing the actual material after precise positioning and verticality have been secured, and may include a number of intelligent control techniques to maximize process quality and stability.

[0106] According to one embodiment, the step of performing the drilling may include a step of performing soft touch control, in which, when the adjustment of the verticality of the center axis is completed while contact of the drill bit is detected, the feed rate of the drill bit is gradually increased to start the initial cutting. Specifically, the end effector (1000) may start cutting by setting the feed rate of the drill bit to a preset very low initial value immediately after the verticality adjustment is completed. This is intended to minimize the impact at the moment the drill bit enters the surface of the workpiece, thereby preventing micro-chipping of the tool and ensuring the quality of the hole entrance.

[0107] Next, the end effector (1000) can monitor the torque sensor data of the spindle motor in real time and smoothly increase (ramp-up) the feed rate linearly or exponentially until it reaches a predefined 'settling torque' value, which indicates that the torque value has entered a stable cutting state. Through such soft-touch control, the initial interaction between the drill bit and the workpiece can be stabilized, and the reliability of the entire drilling process can be improved.

[0108] According to another embodiment, before full-scale cutting begins, the step of performing the drilling may further include a step of optimizing drilling parameters by identifying the local mechanical properties of the workpiece. Specifically, the step of performing the drilling may include a step of gradually applying a preset pressure to the surface of the first work point through the drill bit; a step of identifying the stiffness of the surface of the workpiece based on the displacement values ​​of the displacement sensors of the end effector according to the pressure; and a step of correcting the feed rate included in the drill control information based on the identified stiffness.

[0109] More specifically, the control unit of the end effector can gradually apply pressure to the surface of the workpiece through the end of the drill bit while temporarily stopping the rotation of the spindle or maintaining it at a very low speed after stable contact is confirmed by soft touch control. At this time, the control unit can control the feed until the pressure reaches a preset value (e.g., 50 N) through the force / torque sensor (1024). In a more preferred embodiment, if the end effector (1000) is equipped with a pressure foot pad, more accurate measurements may be possible by using the pressure foot pad instead of the drill bit to stably apply pressure to the surface. While this pressure is being applied, the control unit can measure the amount of change in minute displacement values ​​of the surface in real time from a plurality of displacement sensors located on the attachment surface of the end effector. The control unit can quantitatively identify the local stiffness of the corresponding point from the relationship between the measured displacement value and the applied pressure value. For example, stiffness can be calculated as the ratio of the change in surface displacement to the change in unit pressure.

[0110] Finally, the control unit of the end effector may perform a step of correcting preset drill control information, specifically the feed rate, in real time based on the identified stiffness. If the identified stiffness is higher than a reference value, it indicates that the point has internal stiffeners or the material is harder than expected; therefore, the control unit may lower the base feed rate to prevent tool breakage and ensure machining quality. Conversely, if the stiffness is lower than the reference value, the control unit may raise the feed rate within a safe range to improve productivity. Through such real-time stiffness identification and feed rate correction processes, the present disclosure can actively respond to local material property changes of the actual workpiece that cannot be determined solely from CAD data, and derive optimal machining conditions for each hole in real time, thereby simultaneously maximizing the quality, stability, and efficiency of the drilling process.

[0111] The stiffness measurement of the end effector control unit described above and the control process utilizing it are merely examples of one embodiment. Furthermore, the end effector according to the present disclosure utilizes a circular attachment surface instead of a pressure foot, thereby enabling the improvement of the quality, stability, and efficiency of the drilling process by utilizing a plurality of displacement sensors provided on the attachment surface and a camera positioned close to the attachment surface.

[0112] After the initial cutting has started stably, the end effector (1000) can detect in real time any abnormal process conditions that may occur during drilling and perform control in response. According to one embodiment, the step of performing the drilling may include: converting torque sensor data of the spindle motor into the frequency domain; identifying a preset chip jamming pattern in the converted frequency domain; and, when the chip jamming pattern is identified, performing peck drilling by stopping the drill bit or retracting and advancing it in a rotating state.

[0113] According to one embodiment, the step of identifying a chip jamming pattern may include generating a frequency spectrum by performing a fast Fourier transform on the torque sensor data, and identifying a sideband component adjacent to the rotational frequency of the spindle motor and the harmonic component of the rotational frequency from the generated frequency spectrum, or identifying a rise in the noise floor level of the generated frequency spectrum as the chip jamming pattern.

[0114] More specifically, the end effector (1000) can sample a torque signal from a torque sensor of a spindle motor at a constant period and apply a Fast Fourier Transform (FFT) to the acquired time-series torque data to convert it into a frequency spectrum. The frequency spectrum may include magnitude information of various frequency components constituting the torque signal.

[0115] Depending on the frequency spectrum, the end effector (1000) can identify a sideband component adjacent to the rotational frequency of the spindle motor and the harmonic component of the rotational frequency, or identify a rise in the noise floor level of the generated frequency spectrum, as a 'chip jamming pattern'.

[0116] More specifically, under normal cutting conditions, the rotational frequency of the spindle motor and its harmonic components appear as the main peaks; however, if a 'chip jamming' phenomenon occurs where the cut chips are not smoothly ejected and get stuck in the flute of the drill bit, periodic fluctuations are added to the torque signal, and a characteristic pattern may appear on the frequency spectrum.

[0117] Specifically, when a chip gets stuck in the flute of a drill bit during drilling, it acts as a disturbance that causes periodic load fluctuations in the spindle rotation system. These periodic load fluctuations have the effect of modulating the rotational frequency (carrier frequency) of the spindle motor, and this modulation phenomenon can generate new peaks, or sidebands, at regular intervals on either side of the original rotational frequency peak in the frequency spectrum. The appearance of these sidebands, which do not appear in normal conditions, can serve as a highly reliable indicator of chip jamming.

[0118] Additionally, 'noise floor level rise' is a phenomenon that occurs because the cutting process becomes irregular and friction increases due to chip jamming, causing an increase in non-periodic and broadband noise components in the torque signal. An end effector (1000) according to one embodiment of the present disclosure can determine the possibility of chip jamming by detecting that the overall baseline level of the frequency spectrum rises above a preset threshold.

[0119] If such a chip jamming pattern is identified, the end effector (1000) can perform immediate countermeasures. For example, the end effector (1000) can immediately stop the feed of the drill bit, or automatically perform a peck drilling operation in which, after stopping, the drill bit is retracted a certain distance from the drilled hole while maintaining spindle rotation, and then advanced again. This operation can have the effect of crushing the jammed chips or securing a discharge passage to facilitate chip discharge, and prevent tool breakage or machining defects. The end effector (1000) can automatically resume the drilling process after confirming that the chip jamming pattern has disappeared and the frequency spectrum has returned to normal.

[0120] In the final stage of the drilling process, precise penetration detection may be performed to minimize the generation of burrs on the opposite side of the workpiece. The process described above will be explained in detail below with reference to the drawings.

[0121] In S710, the end effector (1000) can determine cutting energy based on the torque and angular velocity of the spindle motor. In S720, the end effector (1000) can determine the time rate of change of the determined cutting energy. In particular, the workpiece penetration state identification step according to the present disclosure may include the following multi-stage precision detection algorithm to minimize burr generation.

[0122] First, the end effector (1000) may perform the step of identifying a specific section before the end of the drill bit penetrates the workpiece as a 'penetration prediction section' based on the thickness information of the workpiece and the shape information of the drill bit. For example, the control unit (950) may calculate the starting point of the penetration prediction section from the value obtained by subtracting the tip length of the drill bit and a small safety margin from the nominal thickness of the workpiece entered in advance.

[0123] Next, the end effector (1000) may perform the step of activating a high-sensitivity penetration detection algorithm that, when the feed distance of the drill bit enters the identified 'penetration prediction section,' shortens the time window size of the moving average calculating the time rate of change of the cutting energy to a preset shorter value compared to the normal cutting section. This may have the following technical significance. In the normal cutting section, slight noise may occur in the cutting energy due to minor material non-uniformity or vibration.

[0124] To prevent false detection caused by such noise, the control unit (950) can stably filter the signal by calculating a moving average using a relatively 'long time window'. However, since the moment of penetration is a phenomenon in which cutting energy drops rapidly over a very short period of time, if the 'long time window' is continuously used, this rapid change is slowed down by the average value, and the detection time may be delayed.

[0125] Accordingly, the end effector (1000) can intentionally shorten the size of the 'time window' at the moment it enters the 'penetration prediction section' where penetration is imminent, thereby maximizing the system's response speed and sensitivity to changes in cutting energy. In this high-sensitivity detection state, step S730 can be performed. In S730, the end effector (1000) can identify that the drill bit has penetrated the workpiece if the time rate of change of the cutting energy calculated through the shortened time window decreases below a preset negative threshold. That is, at the moment the energy rate of change calculated in a state of maximized sensitivity exceeds a negative threshold signifying a rapid decline, the control unit (950) can determine this as the moment of 'precision penetration' without delay. According to another example, the end effector (1000) may identify that the workpiece has penetrated if the time rate of change of the cutting energy decreases below a preset value or exceeds a preset range.

[0126] In S740, the end effector (1000) can stop the feed of the drill bit when the penetration state is identified. The end effector (1000) according to the present disclosure can effectively minimize the size of the exit-side burr that may occur due to unnecessary additional advance and complete a high-quality hole.

[0127] According to another example, the end effector (1000) may detect chip ejection by detecting motor torque or simply chip ejection sound with a microphone or sensor, rather than the multi-stage precision detection algorithm shown in FIG. 7. According to another example, the end effector (1000) according to the present disclosure may not perform a chip ejection detection operation and may only perform a drill control operation.

[0129] FIG. 8 is a flowchart of a specific method for moving to a second work point after completing work at a first work point according to one embodiment.

[0130] The method illustrated in FIG. 8 is characterized by including a dynamic path correction mechanism that goes beyond simply following a predetermined path and actively responds to uncertainties in the actual work environment to optimize the safety and efficiency of the movement path in real time.

[0131] In S810, the end effector (1000) can acquire camera images of the path ahead from the camera of the end effector while moving along the path. For example, while the end effector (1000) is moving along the path, the control unit (950) can acquire camera images of the path ahead from the vision sensor (222) continuously or at preset intervals. This is to continuously monitor the actual surface conditions on the path during movement.

[0132] In S820, the end effector (1000) can determine whether a suspected curvature section different from the curvature information in the 3D design information is identified based on the optical flow between the acquired camera images. For example, the end effector (1000) can analyze the optical flow between consecutive camera images. Optical flow analysis may include calculating a vector field indicating how surface patterns move between frames in the image. If the surface of the actual workpiece has the same curvature as the 3D design information (CAD data), the calculated vector field may show a continuous and predictable pattern. However, if there is actually an abrupt change in curvature not present in the CAD data due to assembly errors or processing deformation, discontinuous changes in the optical flow vector field in that area, such as divergence or convergence of vectors, may be detected. The end effector (1000) can identify a 'curvature suspected section' where a difference between the 3D design information and the actual shape is suspected by detecting the aforementioned discontinuity.

[0133] In S830, when the suspected curvature section is identified, the end effector (1000) can verify the actual curvature information of the workpiece surface by activating displacement sensors located on the attachment surface of the attachment frame of the end effector. By using the activated displacement sensors, the end effector (1000) can precisely measure the distance to the surface while passing through the section and quantitatively verify the actual surface curvature information by analyzing the statistical deviation (e.g., standard deviation) of these measurements.

[0134] In S840, the end effector (1000) can modify the command point density pre-assigned to each section type based on the actual curvature information confirmed above. For example, if a section classified as a 'general section' (low command point density) in the offline stage is actually confirmed to be a section with a sharp curvature (Case 1), the end effector (1000) can dynamically increase the command point density of that section for safe and precise path following. Conversely, if a section classified as a 'preliminary interest section' (high command point density) is actually confirmed to have a gentle curvature (Case 3), the control unit (950) can dynamically decrease the command point density to prevent unnecessary deceleration and shorten travel time. Additionally, if the end effector (1000) confirms that a section classified as a 'preliminary interest section' is actually a section with a sharp change in curvature (Case 2), the existing high command point density can be maintained.

[0135] In S850, the end effector (1000) can move to the second work point along a global path containing the modified command point density. That is, the end effector (1000) according to the present disclosure can thus safely and efficiently complete the movement to the second work point along a global path containing the command point density that is modified and optimized in real time. The end effector (1000) according to the present disclosure can significantly improve the robustness and reliability of the robotic drilling process through closed-loop control that overcomes the difference between the digital model and the actual environment.

[0137] FIG. 9 is a block diagram of an end effector according to one embodiment.

[0138] FIG. 10 is a block diagram of an end effector according to another embodiment.

[0139] According to one embodiment, the end effector (1000) may include a frame (910), a head (920), a camera (930), a displacement sensor (940), and a control unit (950). However, it is not limited to the above-described example, and the end effector (1000) may include more components as shown in FIG. 10. According to another embodiment, the end effector (1000) may further include a drive system (1060), a position sensor (1022), a chip removal module (1072), and a force torque sensor (1024) in addition to the frame (1010), a head (1020), a camera (1020), a displacement sensor (1040), and a control unit (1050). According to yet another embodiment, the end effector (1000) may be provided with fewer components than those shown in FIG. 9.

[0140] According to one embodiment, the frame (1010) may include an outer frame (1012), an attachment frame (1014), and a robot mounting area (1016).

[0141] For example, the frame (1010) serves as a skeleton forming the overall structure of the end effector (1000) and may include an outer frame (1012), an attachment frame (1014), and a robot mounting area (1016). The outer frame (1012) may serve to protect and support internal precision components, such as a head and a drive system, from external impact. The robot mounting area (1016) may provide an interface (e.g., a mounting part) through which the end effector (1000) is physically coupled to the robot arm of the robot (2000). The attachment frame (1014) forms an attachment surface that contacts or approaches the workpiece during drilling and may be an important part in which a plurality of displacement sensors (1040) are mounted to serve as a reference point for calculating the surface normal vector (S620) for verticality adjustment.

[0142] According to one embodiment, the head (1020) may include a spindle (1022), a machining tool (1024), a tool change module (1028), and a pressure foot pad (1026). The head (1020) is supported by a frame and may include a drill bit, a spindle (1022) for driving the drill bit, and a machining tool (1029). The spindle (1022) may rotate the machining tool (1029) at high speed by a built-in motor. Alternatively, the spindle (1022) may perform the function of moving axially by a drive system (1060).

[0143] The machining tool (1029) may include, for example, a multi-stage drill bit (1025) having several stages of different diameters or a gun drill bit (1027) suitable for deep hole machining.

[0144] In particular, according to one embodiment, the multi-stage drill bit (1025) can be designed to increase process efficiency by simultaneously performing drilling and countersinking. To optimize hole quality and chip evacuation performance, the shape of the margin and flute (chip evacuation passage) of this multi-stage drill bit (1025) can be designed differently according to the diameter of each stage. For example, the leading end (1st stage), which has a relatively small diameter, may have two margins to enhance the straightness of the hole and the guiding function, and the trailing end (2nd stage), which has a large diameter, may have one margin to reduce the cutting load. Additionally, both the leading end and the trailing end may be designed to have one flute each for smooth chip evacuation. When using a multi-stage drill bit having such a complex shape, a problem may arise where chip evacuation is not smooth due to the difference in diameter between each stage. However, the present disclosure can effectively solve this problem by detecting chip jamming in real time through frequency analysis of torque data as described above and automatically performing peck drilling operations.

[0145] Additionally, the head (1020) may optionally include a tool change module (1028) capable of automatically replacing various types of processing tools to increase the level of process automation. In another embodiment, a pressure foot pad (1026) may be provided to stably support the workpiece during drilling and to apply constant pressure to the surface, which may also be utilized to measure the rigidity of the surface. According to another example, the end effector (1000) may use a drill bit in the form of a composite multi-stage drill bit that combines a multi-stage drill bit and a gun drill bit.

[0146] According to one embodiment, the end effector (1000) may include multiple sensor units for intelligent operation. For example, a camera (1020) may be attached to a frame to capture an image of a workpiece. For example, by using the image captured by the camera (1020), the end effector (1000) may perform the role of a multi-purpose 'eye', such as identifying a pilot hole in the image to perform center alignment (S520), detecting obstacles ahead of the path when moving between work points (S450), or detecting anomalies in the curvature of the actual surface (S820).

[0147] Three or more displacement sensors (1040) are placed on the attachment frame (1014) to measure the distance to the surface of the workpiece, which can provide data essential for calculating a normal vector (S620) to secure verticality and verifying actual curvature information (S830). The displacement sensors (1040) can acquire displacement values ​​for the surface of the workpiece by being positioned on the attachment surface.

[0148] The force / torque sensor (1024) may be embedded in the spindle motor or located between the robot mounting part and the frame, and may perform the role of providing sensor data to detect minute changes in force occurring during the drilling process, such as initial contact detection (S610), chip jamming detection, and penetration detection (S730).

[0149] According to another embodiment, the end effector (1000) may further include a projection unit (not shown) for non-contactually measuring the three-dimensional shape of the surface. The projection unit may serve to project structured light of a specific pattern (e.g., lines, grids, random dots) generated from, for example, a laser or LED light source onto the surface of the workpiece. A camera (1020) captures the shape of the projected pattern distorted by the curvature of the surface, and a control unit (950) may use this distortion information to analyze and calculate the normal vector of the surface.

[0150] The position sensor (1022) can be used to control the linear movement range of the spindle (1022) and to determine the origin.

[0151] According to one embodiment, the drive system (1060) may include a spindle shaft (1062), a motor (1064), and a precision drive system (1066). For example, the drive system (1060) serves to precisely drive the head (1020) according to the command of the control unit (1050) and may include a motor (1064) for linear motion of the spindle shaft (1062) and a precision drive system (1066) composed of a ball screw, an LM guide, etc.

[0152] According to one embodiment, the control unit (1050) may include a memory (1052), a network interface (1054), and a processor (1056). For example, the control unit (1050) may identify a global path containing curvature information for each section of the workpiece based on three-dimensional design information set for the workpiece, identify that the end effector has entered a first work point along the global path, align the center of the drill bit of the end effector with respect to the first work point, and when the center of the drill bit is aligned, adjust the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece containing the first work point, generate drill control information based on torque sensor data of the spindle motor of the end effector, and control the operation of the end effector based on the drill control information.

[0153] For example, the memory (1052) may include one or more instructions. According to one embodiment, the memory (1052) may include one or more instructions for controlling the operation method of the end effector.

[0154] The memory (1700) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM (Random Access Memory), SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, and optical disk.

[0155] According to one embodiment, the network interface (1054) can receive control information for controlling the operation of the end effector from an external device or server, or transmit various sensor data measured by the end effector to the outside.

[0157] According to one embodiment, the processor (1056) can control the operation of the end effector by executing one or more instructions stored in memory. For example, the processor (1056) can perform all or part of the operation method of the end effector described in FIGS. 1 to 8 by controlling the network interface (1054) and memory (1052). According to another example, the processor (1056) may receive control information from a robot arm connected to the end effector or from an external server connected to the end effector, and perform the operation method of the end effector by receiving the received control information.

[0159] The method of operation of an end effector according to the present disclosure may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software.

[0160] Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0161] Although embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concept of the present invention as defined in the following claims also fall within the scope of the present invention.

Claims

Claim 1 A method of operation for an end effector attached to and detached from a robot comprises: a step of identifying a global path including segment-by-segment curvature information of a workpiece based on preset three-dimensional design information of a workpiece; a step of aligning the center of a drill bit of the end effector with respect to the first workpiece when it is identified that the end effector has entered a first workpiece according to the global path; a step of adjusting the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first workpiece when the center of the drill bit is aligned; and a step of performing drilling on the first workpiece according to drill control information determined based on torque sensor data of the spindle motor of the end effector; wherein, when the work on the first workpiece is completed, the method further comprises a step of moving to the second workpiece based on a movement path connecting the first workpiece and the second workpiece identified on the global path. A method comprising: further including, the step of adjusting the verticality of the central axis, the step of detecting the initial contact of the drill bit with the surface of the workpiece based on the rate of change per hour of torque sensor data of the spindle motor after the step of aligning the center of the drill bit; the step of identifying the normal vector of the surface of the workpiece based on the displacement values ​​of a plurality of displacement sensors located on the attachment surface of the attachment frame of the end effector while the transfer of the drill bit is stopped when the initial contact of the drill bit with the surface of the workpiece is detected; and the step of adjusting the verticality of the central axis of the drill bit according to the direction of the identified normal vector. Claim 2 delete Claim 3 The method according to claim 1, wherein the step of verifying the global path comprises: a step of verifying a curvature value predetermined for each section of the surface of the workpiece based on the three-dimensional design information; a step of verifying a section type of the surface of the workpiece predetermined based on whether the curvature value verified for each section exceeds a threshold; and a step of verifying the global path including information regarding the curvature value for each section and the verified section type. Claim 4 In claim 3, the step of aligning the center of the drill bit comprises: a step of identifying that entry has been made to the first work point based on coordinate information for the first work point appearing in the global path and encoder coordinate information according to the encoding data of the servo motor of the end effector; and, when entry to the first work point is identified, a plurality of candidate groups representing the edges of the pilot hole centered on the center coordinates of the pilot hole from a camera image of the first work point, and aligning the center of the drill bit such that the error between the center of one of the candidate groups and the center of the drill bit is less than or equal to a threshold value. Claim 5 In claim 4, the step of aligning the center of the drill bit comprises: acquiring a camera image for the first work point; binarizing the acquired camera image; detecting edges from pixel regions for the pilot hole from the binarized camera image; determining one candidate among the plurality of candidates by applying a Hough One transform algorithm to the detected edges; and aligning the center of the drill bit by repeatedly performing the steps of acquiring the camera image and determining the one candidate such that the error between the center of the determined candidate and the center of the drill bit is less than or equal to a threshold. Claim 6 delete Claim 7 A method according to claim 1, wherein the step of performing the drilling comprises: a step of performing soft touch control to gradually increase the feed speed of the drill bit to start initial cutting when the adjustment of the verticality of the center axis is completed while contact of the drill bit is detected. Claim 8 A method according to claim 7, wherein the step of performing the drilling comprises: converting torque sensor data of the spindle motor into the frequency domain; identifying a preset chip jamming pattern in the converted frequency domain; and, when the chip jamming pattern is identified, performing peck drilling by stopping the drill bit or retracting and advancing it in a rotating state. Claim 9 An end effector that is detachably attached to a robot, comprising: a frame having a mounting portion that is detachably attached to the robot and an attachment surface that contacts the surface of a workpiece; a head supported by the frame and including a drill bit and a spindle for driving the drill bit; a camera attached to the frame to acquire an image by photographing the workpiece; at least one displacement sensor positioned on the attachment surface to acquire displacement values ​​for the surface of the workpiece; and a control unit that identifies a global path including curvature information for each section of the workpiece based on pre-set 3D design information for the workpiece, identifies that the end effector has entered a first work point along the global path, aligns the center of the drill bit of the end effector with respect to the first work point, adjusts the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece containing the first work point when the center of the drill bit is aligned, generates drill control information based on torque sensor data of the spindle motor of the end effector, and controls the operation of the end effector based on the drill control information. An end effector comprising: a control unit, wherein when work is completed for the first work point, the control unit moves to the second work point based on a movement path connecting the first work point and the second work point identified on the global path; wherein the control unit detects the initial contact of the drill bit with the surface of the work object based on the rate of change per hour of the torque sensor data of the spindle motor after the step of aligning the center of the drill bit; and when the initial contact of the drill bit with the surface of the work object is detected, while the transfer of the drill bit is stopped, the control unit identifies the normal vector of the surface of the work object based on the displacement values ​​of a plurality of displacement sensors located on the attachment surface of the attachment frame of the end effector, and adjusts the perpendicularity of the center axis of the drill bit according to the direction of the identified normal vector. Claim 10 A method of operation for an end effector attached to and detached from a robot comprises: a step of identifying a global path including segment-by-segment curvature information of a workpiece based on preset three-dimensional design information of a workpiece; a step of aligning the center of a drill bit of the end effector with respect to the first workpiece when it is identified that the end effector has entered a first workpiece according to the global path; a step of adjusting the perpendicularity of the center axis of the drill bit with respect to the surface of the workpiece including the first workpiece when the center of the drill bit is aligned; and a step of performing drilling on the first workpiece according to drill control information determined based on torque sensor data of the spindle motor of the end effector; wherein, when the work on the first workpiece is completed, the method further comprises a step of moving to the second workpiece based on a movement path connecting the first workpiece and the second workpiece identified on the global path. A computer-readable recording medium storing a program to perform a method, wherein the step of adjusting the verticality of the central axis comprises: detecting the initial contact of the drill bit with the surface of the workpiece based on the rate of change of torque sensor data of the spindle motor after the step of aligning the center of the drill bit; identifying the normal vector of the surface of the workpiece based on the displacement values ​​of a plurality of displacement sensors located on the attachment surface of the attachment frame of the end effector while the transfer of the drill bit is stopped when the initial contact of the drill bit with the surface of the workpiece is detected; and adjusting the verticality of the central axis of the drill bit according to the direction of the identified normal vector.

Images