WO2023211007A1

WO2023211007A1 - Robot for assembling furniture

Info

Publication number: WO2023211007A1
Application number: PCT/KR2023/004789
Authority: WO
Inventors: 박재흥; 박수한; 이해성
Original assignee: 서울대학교산학협력단
Priority date: 2022-04-25
Filing date: 2023-04-10
Publication date: 2023-11-02

Abstract

A robot for assembling furniture according to one embodiment of the present invention assembles furniture by assembling a plurality of frames and connection parts for connecting each of the plurality of frames, and comprises: a base stand which forms a working space in which the plurality of frames are assembled; a plurality of robot arms installed on the base stand; gripping parts which are installed on the robot arms and grip at least a portion of the plurality of frames; and a controller which controls the operations of the plurality of robot arms and the gripping parts installed on each of the plurality of robot arms, wherein the controller can control the operations of the plurality of robot arms and the gripping parts to assemble the frames according to a work template including assembly sequence information and motion information about the plurality of robot arms for assembling the furniture.

Description

furniture assembly robot

The present invention relates to a furniture assembly robot, and more specifically, to a furniture assembly robot for assembling furniture based on the connection relationship and assembly sequence of the furniture frame derived from the assembly instructions, and a method of automating the assembly of the robot using the same. .

In processes where mass production was the main focus, robots were mainly taught directly to perform given tasks. The direct teaching method allows a person to easily perform repetitive tasks by directly informing the robot of the joint settings and target positions at which the robot can move. However, the direct teaching method has the disadvantage of being less efficient due to increased human intervention in environments where the production process changes frequently, such as small quantity production of many types, and performing work without using the direct teaching method involves many uncertainties in object recognition, planning, and control. causes Therefore, the problem of allowing intelligent robots to perform tasks autonomously according to the situation rather than through direct teaching is considered a field with many problems to overcome, and much research has been actively conducted.

Research is also being conducted to solve tasks that were previously difficult to solve through direct teaching, such as furniture assembly and cooking, through autonomous planning and control of collaborative robots. A representative example is the precedent of assembling commercial chairs by a research team at Nanyang Technological University in Singapore. .

Robotic assembly automation has developed in two major areas. The first is the work planning area, which focuses on planning possible assembly sequences based on the constraints of the assembly task. In this field, there is a disadvantage that it is difficult to apply to actual situations where uncertainty exists, considering only feasibility without considering the physical limitations of the actual robot. The second is the field of assembly control and motion planning. Among control methods, the most representative method of overcoming uncertainty is peg-in-hole operation. Most studies on peg-in-hole work involve inserting pegs and holes made of metal, but peg-in-hole work on wood, unlike metal materials, requires overcoming more friction. Additionally, in order to perform numerous assembly tasks in a limited space, it is important to utilize the available assembly area. As a way to solve this problem, sampling-based motion planning technology has been developed. Since a solution that satisfies a given task is found in the robot's joint space, a variety of solutions can be calculated. However, the existing method does not consider work space control after following the path obtained from the motion plan, so there is a problem that assembly control in the work space may fail if the arrival position is near a singularity or joint limit.

The present invention is intended to solve the problems of the prior art described above, and includes a furniture assembly robot that successfully performs assembly work even in a narrow work space by utilizing a plurality of robot arms for assembly work that requires large force in an unstructured real environment. It is about a method of automating the assembly of robots using it.

In order to achieve the above object, the furniture assembly robot according to an embodiment of the present invention assembles furniture by assembling a plurality of frames and connecting parts connecting each of the plurality of frames, and provides a work space for assembling the plurality of frames. A base forming base, a plurality of robot arms installed on the base, a gripping unit installed on the robot arm and gripping at least a portion of the plurality of frames, and a controller controlling the operation of the plurality of robot arms and the gripping unit installed on each of the plurality of robot arms. It includes, and the controller may control the operation of the plurality of robot arms and the gripper to perform assembly of the frame by a work template including motion information and assembly order information of the plurality of robot arms for assembling furniture.

Additionally, during the furniture assembly process, the controller may set a restriction condition that prevents the plurality of robot arms from performing furniture assembly and modify or create a work template to avoid the restriction condition.

Additionally, the limiting condition may be set according to at least one of spatial limitations according to the size of the work space, the gripping state of each of the plurality of robot arms, and the type of gripper installed in each of the plurality of robot arms.

In addition, when a limiting condition is met, the controller releases at least a portion of the plurality of frames held by any one of the plurality of robot arms, then changes the gripping position and grips them again, or holds the plurality of frames. One of the robot arms can be controlled to grip the other robot arm.

Additionally, the controller may create a work template so that when the gripper grips at least a portion of the plurality of frames, as many connection parts as possible are inserted into the frame.

Additionally, when assembling different frames, if a hole formed in one of the different frames is not exposed to the outside, the controller may generate a work template that exposes the hole to the outside by rotating one of the frames.

When assembling different frames, the controller controls the operation of a plurality of robot arms to insert the connecting parts assembled in one of the different frames into the hole formed in the other frame using the peck-in-hole method, and the peck-in-hole method is used to insert the connecting parts assembled in one of the different frames In order to insert it into the hole, it is pressed and at the same time moved along a predetermined trajectory to search for the location of the hole.

Additionally, when connecting different frames, the controller may detect the number of connection parts assembled in any one of the different frames and control the operation of a plurality of robot arms according to the number of detected connection parts.

Additionally, when two connection parts are inserted into the hole, the controller can control the operation of the plurality of robot arms to perform a double pick-in-hole method in which the connection parts move in a predetermined trajectory while making a wiggle motion.

Additionally, when there are three or more connection parts inserted into the hole, the controller can control the operation of a plurality of robot arms to insert the connection parts into the hole by combining the pick-in-hole method and the double pick-in-hole method.

In addition, the connection part includes a screw, and the controller includes a plurality of robots so that at least one of the plurality of robot arms grips the drill module when the screw is fastened, and the other robot arm that does not grip the drill module grips and fixes the frames. The movement of the arm can be controlled.

In addition, each of the plurality of robot arms is an articulated robot arm, and the controller sets a restriction condition under which the plurality of robot arms cannot perform furniture assembly due to spatial limitations according to the size of the work space, and sets a restriction condition under which the plurality of robot arms cannot perform furniture assembly, and The joint solution of an articulated robot arm can be calculated.

In addition, the joint solution of the multi-joint robot arm that does not correspond to the limiting conditions can be calculated by gradient descent using the cost functions of manipulability and joint limit.

Additionally, the controller may determine whether the joint solution changed during operation of the articulated robot arm corresponds to the constraint condition, and if the joint solution does not correspond to the constraint condition, the controller may store and select the joint solution in the search stack.

Additionally, if the joint solution meets the constraint condition, the controller determines whether the joint solution stored in the search stack exceeds n (n is a natural number of 2 or more). If the joint solution stored in the search stack exceeds n, the controller determines whether the joint solution stored in the search stack exceeds n. Delete n recently stored joint solutions, select the most recent joint solution, and if the number of joint solutions stored in the search stack does not exceed n, the joint solution changed during the motion of the articulated robot arm corresponds to the constraint condition. You can re-evaluate whether you do it or not.

In addition, after assembly of different frames is performed, the controller calculates the force applied to the end of the robot arm while moving the robot arm holding the different frames to a predetermined trajectory, and determines the force applied to the end of the robot arm by a predetermined amount. If the reference value is not reached, it can be determined that the assembly of different frames has failed.

Additionally, the controller can re-perform the assembly of different frames if the assembly of different frames fails.

In order to achieve the above object, a furniture assembly method using a robot according to another embodiment of the present invention is a furniture assembly automation method of assembling a plurality of frames and connecting parts connecting each of the plurality of frames using a robot. , setting constraints under which furniture assembly cannot be performed during the assembly process; creating a task template to avoid the constraint condition; and performing, by the robot, assembling the frame according to the work template.

Additionally, the robot may be a furniture assembly robot of any one of the above-described embodiments.

Additionally, when assembling different frames, it may include inserting a connection part assembled to one of the different frames into a hole formed in the other frame using a peck-in-hole method.

In addition, the robot includes an articulated arm, and the process of creating a work template to avoid the constraint condition includes calculating the manipulability of the articulated robot arm and the cost function of the joint limitations using a gradient descent method. It may include a step of calculating joint solutions that do not meet the conditions.

The furniture assembly robot according to an embodiment of the present invention successfully performs furniture assembly even in a real environment with limitations in the work area by generating robot tasks from the recognition results of the assembly instructions for furniture assembly where uncertainty is very high and adaptive control is essential. make it possible

In addition, the furniture assembly robot according to an embodiment of the present invention automatically generates a robot work plan based on constraints between tasks, can select an assembly position considering assembly work in a narrow space, and requires great force for assembly. Assembly is necessary and can be performed successfully even in situations where uncertainty exists.

1 is a diagram illustrating a furniture assembly robot according to an embodiment of the present invention.

Figure 2 is an exemplary diagram showing a process of performing furniture assembly by a furniture assembly robot.

Figure 3 is an exemplary diagram showing the relationship between a work template of a furniture assembly robot and an assembly operation by the furniture assembly robot.

Figure 4 is a flowchart illustrating a process in which a furniture assembly robot inserts connecting parts into a frame by the pick-in-hole method according to an embodiment of the present invention.

FIG. 5 is a diagram illustrating an example of a work template for inserting the connection parts shown in FIG. 4 into a frame using the pick-in-hole method.

Figure 6 is an exemplary diagram showing the motion trajectory of the connecting part when the connecting part is inserted into the frame by the pick-in-hole method.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. When adding reference numerals to components in each drawing, it should be noted that the same components are given the same reference numerals as much as possible even if they are shown in different drawings. Additionally, when describing an embodiment of the present invention, if a detailed description of a related known configuration or function is judged to impede understanding of the embodiment of the present invention, the detailed description will be omitted.

Referring to Figure 1, the furniture assembly robot 1 assembles furniture by assembling a plurality of frames and connection parts connecting each of the plurality of frames. For this purpose, the furniture assembly robot 1 includes a base 10 forming a work space for assembling a plurality of frames 30, a plurality of

robot arms

20a, 20b, and 20c installed on the base 10, each Holding parts (21a, 21b, 21c) installed on the robot arms (20a, 20b, 20c) and holding at least a part of the plurality of frames (30), a plurality of robot arms (20a, 20b, 20c), and a plurality of robot arms (20a, 20b, 20c) includes a controller that controls the operation of the gripping parts (21a, 21b, 21c) installed on each.

The base 10 may be a flat member for forming a work space for assembling a plurality of frames 30. A plurality of robot arms (20a, 20b, 20c) are installed on one surface of the base 10, and the space between the plurality of robot arms (20a, 20b, 20c) can be a work space.

As in the embodiment shown in FIG. 1, the furniture assembly robot 1 is placed in a work space and may include an auxiliary box 11 that supports the frame 30 being assembled. The furniture assembly robot 1 may proceed with assembling the frame 30 while placing the frame 30 being assembled on the auxiliary box 11 as necessary during the process of assembling furniture.

The furniture assembly robot 1 includes a plurality of robot arms, and in the embodiment shown in FIG. 1, the furniture assembly robot 1 includes three

robot arms

20a, 20b, and 20c. The three

robot arms

20a, 20b, and 20c are articulated robot arms and can operate independently or in relation to each other by a controller. A driving servo motor and an encoder for detecting the rotation angle of the servo motor are installed at the joints of each robot arm (20a, 20b, and 20c) (all not shown). Each servo motor is servo-controlled by a controller so that the

robot arms

20a, 20b, and 20c move along a taught trajectory.

The

gripping units

21a, 21b, and 21c may be installed on each of the plurality of

robot arms

20a, 20b, and 20c to grip objects. The furniture assembly robot 1 according to an embodiment of the present invention can grip the frame and connecting parts by the

gripping parts

21a, 21b, and 21c, but is not limited by the above. If necessary, the

gripping parts

21a, 21b, and 21c may grip other objects, for example, a drill module for fastening connection parts.

The controller can control the overall operation of the furniture assembly robot 1. For example, the controller may control the operation of the plurality of robot arms (20a, 20b, 20c) and the gripping portions (21a, 21b, 21c) installed on each of the plurality of robot arms (20a, 20b, 20c), and also The controller is configured to perform assembly of the frame by a work template including motion information and assembly sequence information of the plurality of

robot arms

20a, 20b, 20c for assembling furniture. Movement can be controlled. Meanwhile, the object controlled by the controller is not limited by the above. For example, the controller may modify a task template related to motion information or create a new task template depending on the status of the

robot arms

20a, 20b, and 20c.

The controller includes at least one processor. The processor may be implemented as an array of multiple logic gates, or as a combination of a general-purpose microprocessor and a memory storing a program that can be executed on the microprocessor. Additionally, those skilled in the art can understand that the present embodiment may be implemented with other types of hardware.

The frame 30 is a major part that forms the shape of furniture, for example, a wooden board, a wooden leg, etc., and is an object in which holes are formed for assembly. Connecting parts are parts that connect the frame 30, such as wooden pins, brackets, pegs, screws, etc.

Referring to FIG. 2, the furniture assembly robot 1 provides tasks and motion planning, assembly control, and recovery methods required for furniture assembly in an integrated manner, from the connection relationship between each frame of furniture and the order of connecting each frame. do. To this end, the furniture assembly robot 1 reads the assembly instructions (S1), derives the connection relationship between each frame of the furniture and the order of connecting each frame through a recognition algorithm from the assembly instructions (S2), and runs a task compiler (Task). compiler) to create a work template for the assembly work indicated by the assembly instructions (S3). Each assembly task reflects the current assembly environment information to avoid collisions and performs a motion plan to move to the assembly position (S4), and after moving to the assembly position, assembly control is performed to perform assembly (S5). After assembly, check whether assembly is complete and then perform recovery work if it fails.

Meanwhile, the process (S2) in which the furniture assembly robot 1 reads the assembly instructions and derives the connection relationship between each frame of the furniture and the order of connecting each frame is outside the scope of the present invention, so detailed description thereof Let's omit it.

The controller converts the assembly information from the assembly instructions into connection relationships and assembly sequences into the robot's assembly work through a task compiler. At this time, each assembly task can be defined as a task template that classifies the tasks used to assemble furniture. For example, task templates may include motion planning techniques, assembly control techniques, failure recovery plans, etc. In the work template, the assembly control strategy is modularized and consists of macro motion and micro motion. For example, the work template for assembling a frame using the pick-in-hole method includes a first macro motion that combines linear motion and rotational motion, which are micro motions, and a second macro motion that combines spiral motion and pressing motion, etc. It can be configured modularly. The controller performs the assembly of the frame by controlling the operation of the plurality of robot arms (20a, 20b, 20c) and the gripping parts (21a, 21b, 21c) according to the work template.

As shown in the example shown in Figure 2, the controller analyzes the assembly sequence and connection relationships given through the assembly instructions through a task compiler and lists appropriate task templates. At this time, the controller performs the work plan by avoiding the constraints applied to the work template. The restriction condition refers to a condition in which furniture assembly cannot be performed using a plurality of

robot arms

20a, 20b, and 20c during the furniture assembly process. Restriction conditions may be set according to, for example, spatial limitations according to the size of the work space, the gripping state of each of the plurality of robot arms, and the type of gripper installed in each of the plurality of robot arms. However, such limiting conditions are not limited by the above, and may be set depending on the size, type, weight, etc. of the furniture being assembled.

The controller can modify or create a work template to avoid constraints even when the assembly sequence and type of assembly work are already determined by the work template. For example, when the assembly operation by the plurality of

robot arms

20a, 20b, and 20c meets a constraint condition, the controller additionally creates work templates such as [Grip], [Release], and [Transfer/Re-Grap]. , assembly work can be performed by adjusting the order of already established work templates.

For example, when a constraint condition is met, the controller releases at least some of the frames held by one of the plurality of robot arms, then changes the gripping position and grips them again, or changes the gripping position of the plurality of frames held by one of the plurality of robot arms. Another robot arm is controlled to grip at least some of the plurality of frames.

As another example, the connection work of a frame that requires multiple connection relationships (for example, when several pins must be inserted into the frame at the same time) depends on the number of connection parts that must be inserted into the frame to be assembled. The type is determined. For example, when the gripper grips at least a portion of a plurality of frames, the controller may create a work template so that as many connection parts as possible are inserted into the frame. Since a number of connected parts are generally inserted into a frame, furniture assembly can be efficiently performed by creating a work template as described above.

As another example, when assembling different frames, if a hole formed in one of the different frames is not exposed to the outside, the controller generates a work template to expose the hole to the outside by rotating one of the frames. can do. During the furniture assembly process, if the holes of the frame to be assembled (for example, holes where pegs are inserted, screw holes, etc.) are covered by the floor or the like and are not exposed to the outside, the assembly work cannot be performed. Therefore, the furniture assembly robot can perform assembly work by rotating the frame that is the object of assembly. At this time, because the furniture is not completely assembled (for example, the frames are assembled with only wooden pins and not screws fastened), the rotation work is performed by multiple robot arms on the frames to be assembled. It is performed while maintaining force and position to prevent breakage/separation.

To describe the work template in more detail with reference to FIG. 3, (a) in FIG. 3 is an example of a work template used in the process of assembling furniture by the furniture assembly robot 1, and (in FIG. 3) b) is an example of an assembly work performed through the work template shown in (a) of FIG. 3.

Assembly of furniture by the furniture assembly robot 1 can be performed by assembly operations including inserting connecting parts, connecting frames, fastening screws, and rotating the assembly object if necessary. If we classify the above-mentioned four assembly operations more specifically, the assembly operations are [insertion of connection parts], [connection of frame (including connection parts)], [connection of frame (without connection parts)], [(complete assembly) ) screw fastening], [(incomplete) screw fastening], and [rotation of assembled articles]. However, the above-described assembly operations are merely illustrative classifications, and various assembly operations may be classified as needed.

In the example shown in (a) of FIG. 3, the work template for the furniture assembly robot 1 to perform the assembly task consists of [grasp], [release], and [handover/re-grasp]. At this time, [transfer/re-grip] is, for example, releasing the item being held by any one of the

gripping parts

21a, 21b, and 21c of the furniture assembly robot 1 and then gripping it again. Alternatively, it means an operation in which another gripper (21b or 21c) grips the article.

Figure 4 is a flow chart showing a process in which a furniture assembly robot inserts connecting parts into a frame by the pick-in-hole method in an embodiment of the present invention, and Figure 5 is a work template for inserting the connecting parts into the frame by the pick-in-hole method. This is a diagram illustrating an example, and FIG. 6 is an exemplary diagram illustrating the motion trajectory of the connecting component when the connecting component is inserted into the frame by the pick-in-hole method.

Referring to FIG. 4, when assembling different frames during the furniture assembly process, the controller uses a plurality of robot arms to insert the connecting parts assembled in one of the different frames into a hole formed in the other frame in a peck-in-hole manner. Controls movement. The peck-in-hole method includes an access step (S11), a search step (S12), a confirmation step (S13), an insertion step (S15), a separation step (S16), and optionally a recovery step (S14) after the confirmation step (S13). do.

The approach step (S11) is the step in which the frame and connecting parts first come into contact for furniture assembly. For this purpose, the controller controls the robot arm that holds the frame and the connection parts respectively so that the frame and the connection parts come into contact.

The search step (S12) is a step to find the assembly target while overcoming the positional error between the hole formed in the frame and the connecting parts. The controller can pressurize the connecting component to insert it into the hole formed in the frame and move it along a predetermined trajectory to search for the location of the hole. For example, as shown in the example shown in FIG. 6, the predetermined trajectory may be a spiral trajectory (T1), and, if necessary, insert axis rotation (yawing) or multi-axis rotation (e.g. wiggle) into the spiral trajectory. By adding wiggle) motion, the connected parts can move as shown in the trajectory (T2) of FIG. 6. Additionally, when assembling wooden parts such as furniture, stronger friction must be overcome compared to peck-in-hole work on metal parts. The furniture assembly robot according to the present invention moves the connecting parts in the above-described predetermined trajectory in the search step (S12), thereby preventing a stick state that causes static friction and overcoming uncertainty in inserting the connecting parts. .

The confirmation step (S13) is a step to determine whether assembly of the frame was successful or failed. For example, the controller controls the robot arm to move the connecting parts in a cross shape (eg, in the x-axis and y-axis directions) after the above-described search step (S13). At this time, if the connecting part is inserted into the hole of the frame, the connecting part cannot move in a cross shape, so the controller can determine that the assembly is successful. However, if the connecting part is not inserted into the hole of the frame, the connecting part has a movable direction in a cross shape, so the controller may determine that the assembly has failed. In an assembly failure situation, the controller can identify the relative positions between the frame and the connection parts by determining the directions in which the connection parts can and cannot move.

As another example, after assembly of different frames is performed, the controller calculates the force applied to the end of the robot arm while moving the robot arm holding the different frames in a predetermined trajectory, and calculates the force applied to the end of the robot arm If this predetermined reference value is not reached, it may be determined that the assembly of different frames has failed.

The recovery step (S14) is a step to recover if the insertion of the connection part into the frame fails in the confirmation step (S13). For example, if the controller recognizes a failure in the confirmation step (S13), it can return to the previous state and then re-execute the access step (S11) to re-assemble the frame.

The insertion step (S15) is a step of forming a strong connection between the frame and the connection component when insertion of the connection component into the frame is successful in the confirmation step (S13). For example, a controller can control a robot arm that holds the frame and connecting components to rotate while holding the frame still and pressing the connecting components toward the frame.

The separation step (S16) is a step of separating the holding part from the connection part that has been fully fastened to the frame. For example, the controller can control the release of the gripper that grips the connection part that has been fastened to the frame.

Referring to FIG. 5, an example of a work template implemented in micro/macro motion for each step of the pick-in-hole work is shown. For the pick-in-hole operation, the controller receives variables related to the target object, grasp pose, assembly index, connector, and manipulator, and then the furniture assembly robot Execute the peck-in-hole operation according to the steps described above.

The assembly operation by the above-described peck-in-hole method is explained by taking the case of inserting one connection part into a frame, and hereinafter, the case of inserting multiple connection parts into the frame will be explained.

When connecting different frames, the controller can detect the number of connected parts assembled in any one of the different frames and control the operation of a plurality of robot arms according to the number of detected connected parts.

For example, the controller performs the above-described peck-in-hole method when there is only one connection part inserted into the hole of the frame. When there are two connection parts inserted into the hole of the frame, the controller can control the operation of a plurality of robot arms to perform a double pick-in-hole method in which the connection parts move in a predetermined trajectory while making a wiggle motion. In addition, when there are three or more connection parts inserted into the holes of the frame, the controller can control the operation of a plurality of robot arms to insert the connection parts into the holes of the frame by combining the above-described peck-in-hole method and the double peck-in-hole method. .

If the connecting part is a screw when assembling furniture, the controller causes at least one of the plurality of robot arms to grip the drill module when the screw is fastened, and the other robot arm that does not grip the drill module grips and fixes the frames. The operation of a plurality of robot arms can be controlled to do so.

In order for a furniture assembly robot to automatically perform all assembly tasks, it is important to select an assembly location to perform the assembly task. Due to the physical limitations of the furniture assembly robot (for example, joint limitations and singularities of the robot arm, etc.), the success rate of assembly control is different depending on the assembly location. Therefore, it is necessary to select the location of the assembly work to minimize physical limitations in a work space mixed with various obstacles.

Furniture assembly robots perform motion planning with minimal constraints to perform furniture assembly in a limited work space. For example, when inserting a connecting part such as a pin into a frame, motion planning is performed by targeting only the constraints on the relative position of the assembly without separately specifying the location where the assembly takes place.

When selecting the target position of the motion plan, the controller can calculate the joint solution of the articulated robot arm so that it does not fall under the constraint conditions, and the joint solution of the articulated robot arm that does not fall under the constraint condition can be calculated by calculating the manipulability and The cost function of the joint limit can be calculated by gradient descent, and gradient descent can be calculated repeatedly as needed. For example, in a situation where a connection part is fastened to a frame, a random solution that avoids the constraint between the frame and the connection part, that is, the posture constraint condition of the robot arm, is obtained from a random initial position, and this solution is combined with the manipulability of the robot arm. The solution is calculated to be as far away from the joint limits as possible so that the previously mentioned pick-in-hole motion can be successfully performed.

The controller determines whether the changed joint solution avoids the constraint condition during operation of the articulated robot arm. At this time, if the joint solution avoids the constraint condition, the controller may store and select the joint solution that avoids the constraint condition in the search stack of the memory.

If the joint solution meets the constraint condition, the controller determines whether the joint solution stored in the search stack exceeds n (n is a predetermined natural number of 2 or more). If the number of joint solutions stored in the search stack exceeds n, the controller deletes n recently stored joint solutions from the search stack, selects the most recent joint solution, and makes sure that the joint solutions stored in the search stack do not exceed n. If not, it is determined again whether the changed joint solution avoids the limiting condition during operation of the articulated robot arm, and the above-described process is performed again. For example, during the process of changing the joint solution, if the changed joint solution is determined to be an unusable solution due to a limiting condition such as colliding with an obstacle, a solution close to the obstacle is not produced by removing n joint solutions stored in the search stack. Avoid doing so.

The above description is merely an illustrative explanation of the technical idea of the present invention, and those skilled in the art will be able to make various modifications and variations without departing from the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but are for illustrative purposes, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted in accordance with the claims below, and all technical ideas within the equivalent scope should be construed as being included in the scope of rights of the present invention.

Claims

A furniture assembly robot that assembles furniture by assembling a plurality of frames and connecting parts connecting each of the plurality of frames,

a base forming a work space for assembling the plurality of frames;

a plurality of robot arms installed on the base;

a gripper installed on the robot arm to grip at least a portion of the plurality of frames; and

Comprising a plurality of robot arms and a controller that controls the operation of a gripper installed on each of the plurality of robot arms,

The controller is,

A furniture assembly robot that controls the operation of the plurality of robot arms and the gripping portion to perform assembly of the frame by a work template including assembly sequence information and motion information of the plurality of robot arms for assembling the furniture. .
According to paragraph 1,

The controller sets a restriction condition under which the plurality of robot arms cannot perform the furniture assembly during the furniture assembly process, and modifies or creates the work template to avoid the restriction condition.
According to paragraph 2,

The limiting condition is set according to at least one of a spatial limitation according to the size of the work space, a gripping state of each of the plurality of robot arms, and a type of gripper installed in each of the plurality of robot arms. A furniture assembly robot.
According to paragraph 2,

When the limiting condition is met, the controller releases at least a portion of the plurality of frames held by one of the plurality of robot arms and then changes the gripping position to grip them again. , A furniture assembly robot that controls another robot arm among the plurality of robot arms to grip.
According to paragraph 1,

The controller is a furniture assembly robot that generates the work template so that as many of the connecting parts as possible are inserted into the frames when the gripping unit grips at least a portion of the plurality of frames.
According to paragraph 1,

When assembling different frames, if a hole formed in one of the different frames is not exposed to the outside, the controller generates a work template that exposes the hole to the outside by rotating the one frame. , furniture assembly robot.
According to paragraph 1,

When assembling different frames, the controller controls the operation of the plurality of robot arms to insert a connection part assembled in one of the different frames into a hole formed in the other frame in a peck-in-hole manner,

The pick-in-hole method is a furniture assembly robot that presses the connecting part to insert it into the hole and moves it along a predetermined trajectory to search for the location of the hole.
In clause 7,

The controller detects the number of connection parts assembled in any one of the different frames when connecting different frames, and controls the operation of the plurality of robot arms according to the number of the detected connection parts, furniture assembly. robot.
According to clause 8,

The controller controls the operation of the plurality of robot arms to perform a double pick-in-hole method in which, when there are two connecting parts inserted into the hole, the connecting parts move along the predetermined trajectory while making a wiggle motion. A furniture assembly robot.
According to clause 9,

The controller controls the operation of the plurality of robot arms to insert the connecting parts into the hole by combining the pecking-in-hole method and the double pecking-in-hole method when there are three or more connecting parts inserted into the hole. Furniture assembly robot.
In clause 7,

The connecting part includes a screw,

The controller operates the plurality of robot arms so that when the screw is tightened, at least one of the plurality of robot arms grasps the drill module, and the other robot arm that does not grip the drill module grasps and fixes the frames. Controlling a furniture assembly robot.
According to paragraph 1,

Each of the plurality of robot arms is an articulated robot arm,

The controller sets a limiting condition under which the plurality of robot arms cannot perform the furniture assembly due to spatial limitations according to the size of the work space, and the joint solution of the multi-joint robot arm that does not correspond to the limiting condition A furniture assembly robot that calculates.
According to clause 12,

The joint solution of the articulated robot arm that does not meet the limiting conditions is calculated by gradient descent using the cost function of manipulability and joint limit.
According to clause 12,

The controller determines whether the changed joint solution avoids the limiting condition during operation of the articulated robot arm,

A furniture assembly robot that stores and selects the joint solution in a search stack if the joint solution avoids the constraint condition.
According to clause 14,

The controller determines whether the joint solution stored in the search stack exceeds n (where n is a predetermined natural number of 2 or more) if the joint solution satisfies the constraint condition,

If the number of joint solutions stored in the search stack exceeds n, delete n recently stored joint solutions from the search stack and select the most recent joint solution,

A furniture assembly robot that re-determines whether a joint solution changed during operation of the articulated robot arm meets the limitation condition when the number of joint solutions stored in the search stack does not exceed n.
According to paragraph 1,

The controller calculates a force applied to the end of the robot arm while moving the robot arm holding the different frames along a predetermined trajectory after assembly of the different frames is performed,

A furniture assembly robot that determines that the assembly of the different frames has failed when the force applied to the end of the robot arm does not reach a predetermined reference value.
According to clause 16,

The controller is a furniture assembly robot that re-performs the assembly of the different frames when the assembly of the different frames fails.
A furniture assembly automation method that uses a robot to assemble a plurality of frames and connection parts connecting each of the plurality of frames,

Setting constraints under which furniture assembly cannot be performed during the assembly process;

creating a task template to avoid the constraint condition; and

Comprising: the robot performing assembly of the frame according to the work template,

How to assemble furniture using a robot.
According to clause 18,

When assembling different frames, it includes inserting the connecting parts assembled in one of the different frames into a hole formed in the other frame in a peck-in-hole manner,

How to assemble furniture using a robot.
According to clause 18,

The robot includes an articulated arm,

Creating a task template to avoid the above constraints involves:

Comprising the step of calculating the manipulability of the articulated robot arm and the cost function of the joint limitations by gradient descent to calculate a joint solution that does not correspond to the limitation conditions,

How to assemble furniture using a robot.