WO2017149616A1 - Robot d'emballage de boîte et procédé de planification d'emballage de boîte - Google Patents

Robot d'emballage de boîte et procédé de planification d'emballage de boîte Download PDF

Info

Publication number
WO2017149616A1
WO2017149616A1 PCT/JP2016/056106 JP2016056106W WO2017149616A1 WO 2017149616 A1 WO2017149616 A1 WO 2017149616A1 JP 2016056106 W JP2016056106 W JP 2016056106W WO 2017149616 A1 WO2017149616 A1 WO 2017149616A1
Authority
WO
WIPO (PCT)
Prior art keywords
boxing
storage
area
manipulator
stored
Prior art date
Application number
PCT/JP2016/056106
Other languages
English (en)
Japanese (ja)
Inventor
佳奈子 江崎
宣隆 木村
潔人 伊藤
Original Assignee
株式会社日立製作所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社日立製作所 filed Critical 株式会社日立製作所
Priority to PCT/JP2016/056106 priority Critical patent/WO2017149616A1/fr
Publication of WO2017149616A1 publication Critical patent/WO2017149616A1/fr

Links

Images

Definitions

  • the present invention relates to a boxing robot and a boxing planning method using a boxing robot.
  • Patent Document 1 In the industrial fields such as manufacturing and logistics, technology that robots pick objects and pack them is used.
  • Patent Document 2 Patent Document 1
  • Non-Patent Document 1 As background art.
  • Patent Document 1 is an adsorber disposed in a picking head, which adsorbs packaged articles that are lined up on a carry-in conveyor and arranges them in a packing box that is partitioned by a sawtooth mounting surface and a partition plate. Is described.
  • Patent Document 2 a guide member is placed in a lower box placed on a pallet, a plate-like object is inserted between the outer edge of the lower box and the guide member, and thereafter, the guide member is supported by the guide member. The next plate-shaped object is brought close to the last plate-shaped object, the guide member is moved by the thickness of the plate-shaped object, and the next plate-shaped object is moved between the guide member and the last plate-shaped object.
  • a system for packing a plurality of plate-like objects by repeating insertion is described.
  • Non-Patent Document 1 assigns priorities based on the location of each storage space divided by a plane, the weight of cargo, etc., and optimizes the combination of each storage space and cargo according to the priorities. Describes how to generate a cargo packing plan.
  • Patent Document 1 and Patent Document 2 a storage auxiliary member such as a partition plate in the box or a guide member to be inserted in the box is used on the assumption that the same type of object is stored in the box.
  • a storage auxiliary member such as a partition plate in the box or a guide member to be inserted in the box is used on the assumption that the same type of object is stored in the box.
  • it is necessary to pack different types of products in a storage box, and it is difficult to prepare a storage auxiliary member in advance.
  • Non-Patent Document 1 pays attention to how the stability of an object is stacked as a constraint condition, and reduces the problem of maximizing the volumetric efficiency of packing, so that how to pack a plurality of different types of goods into a storage box can be achieved. Is planned. However, the interference with obstacles within the movement range of the robot manipulator, which is necessary for the robot to store the object in the box, is not considered.
  • the boxing robot to be disclosed uses a manipulator for storing an object to be stored in the storage container, a memory for storing the storage container and the geometric model data of the storage object, and a storage container using the geometric model data of the storage container. Calculate, hold the storage-scheduled object in the manipulator and move it within the empty area, and based on the result of the movement of the storage-scheduled object, find the corner of the storable area within the empty area and within the movable range of the manipulator, The found corner is determined as the storage position of the object to be stored, and the boxing plan data is updated based on the determined storage position.
  • the disclosed boxing robot can generate a boxing plan that avoids obstacles within the movement range of the manipulator.
  • Example 1 is a configuration example of a boxing robot according to a first embodiment.
  • 3 is an example of a movement simulation of an object to be stored according to the first embodiment.
  • It is a process flowchart of the boxing plan part of Example 1.
  • It is a process flowchart of the boxing plan part of Example 2.
  • It is a conceptual diagram showing the relationship between the boundary point of the storable area
  • It is a process flowchart of the boxing plan part of Example 4.
  • FIG. 1 is a configuration example of a boxing robot (hereinafter referred to as a robot) 101 in the present embodiment.
  • the robot 101 includes a memory 102, a boxing planning unit 103, a motion generation unit 104, and a manipulator 105.
  • the memory 102 stores the robot 101, the storage container 201, the storage planned object 202, the boxed planned object 203 and the geometric model data 106 of the stored object 204, and the boxing plan data 107 generated by the boxing planning unit 103.
  • the boxed planned object 203 is an object that has been stored in the storage container 201 in a simulation described later and has been boxed.
  • the stored object 204 is an object that is already stored in the storage container 201.
  • the geometric model data 106 includes boxing planning geometric model data 108 and state grasping geometric model data 109.
  • the boxing planning unit 103 executes a simulation of moving the storage object 202 by the manipulator 105, and generates a boxing plan of the storage object 202 in the storage container 201.
  • the boxing plan unit 103 stores the generated boxing plan in the memory 102 as boxing plan data 107.
  • the action generation unit 104 generates an action of the manipulator 105 using the state grasping geometric model data 109 and the boxing plan data 107 stored in the memory 102 in order to execute the generated boxing plan by the manipulator 105.
  • the manipulator 105 executes the operation generated by the operation generation unit 104.
  • FIG. 2 is an example of a movement simulation of the storage object 202 based on the geometric model data 106.
  • the geometric model data 106 of each object of the robot 101, the storage container 201, the storage planned object 202, the boxed planned object 203, and the stored object 204 includes position data, posture data, and shape data.
  • the position data and the posture data respectively represent the position and posture of the representative point of each object.
  • the representative point need not be one point.
  • the shape data is stored as 3D point cloud data.
  • the shape of the object may be approximated in advance to a primitive shape such as a rectangular parallelepiped or a cylinder, and stored as a combination of a primitive shape name and a parameter corresponding to the primitive shape such as the length of three sides of the rectangular parallelepiped. .
  • a primitive shape such as a rectangular parallelepiped or a cylinder
  • a parameter corresponding to the primitive shape such as the length of three sides of the rectangular parallelepiped.
  • FIG. 3 is a processing flowchart of the boxing planning unit 103.
  • the boxing planning unit 103 calculates a storage position candidate of the storage planned object 202 based on the geometric model data 106, and executes a simulation of moving the storage planned object 202 by the manipulator 105 toward one of the storage position candidates.
  • the storage position candidate that has been moved is set as the storage position of the storage object 202.
  • one storage target object 202 is selected in order from the plurality as a processing object, and the boxing planning unit 103 processes the selected processing object as a target. Execute.
  • the order in which the processing objects are selected from the plurality of objects to be stored 202 is set in advance according to the descending order of the size of the objects.
  • the boxing planning unit 103 initializes the boxing planning geometric model data 108 with the value of the state grasping geometric model data 109 (described later) (step 301).
  • the values of the variables in the state grasping geometric model data 109 are used as the boxing planning geometric model data 108. Is assigned to the corresponding variable.
  • the state grasping geometric model data 109 is converted into the data format of the boxing planning geometric model data 108. After conversion so as to match, it is substituted into the corresponding variable of the geometric model data 108 for boxing plan.
  • the state grasping geometric model data 109 is data acquired by a sensing unit (not shown) except for the data of the boxed planned object 203.
  • the data of the robot 101 is acquired from an encoder or the like mounted on the robot 101, and the data of the storage container 201, the storage scheduled object 202, and the stored object 204 are acquired from a camera or the like.
  • the data of the boxed planned object 203 is acquired from the boxed plan data 107 stored in the memory 102.
  • the data of the storage container 201, the storage scheduled object 202, and the stored object 204 may not be acquired by the sensing unit.
  • the relative positions of the robot 101, the storage container 201, and the storage object 202 are fixed values
  • the fixed values may be used as the position data of the storage container 201 and the storage object 202.
  • the data of the stored object 204 may be acquired from the boxing plan data 107 stored in the memory 102.
  • the boxing planning unit 103 acquires the boxing planning geometric model data 108 from the memory 102, and calculates storage position candidates (step 302).
  • the boxing planning unit 103 uses the boxing planning geometric model data 108 acquired from the memory 102 to project an image projected on a plane parallel to the bottom surface of the storage container 201 (in a figure representing the bottom surface of the storage container 201 An image obtained by projecting the bottom graphics of the object 203 and the stored object 204 is generated.
  • the boxing planning unit 103 performs corner detection processing on the generated image, and geometrically determines the corners of the area where the boxed planned object 203 and the stored object 204 are not present in the bottom surface of the storage container 201.
  • the corner is a storage position candidate.
  • the boxing planning unit 103 does not need to use the bottom image of the storage container 201 in order to obtain the corners of the area without the boxed planned object 203 and the stored object 204 in the storage container 201.
  • the boxing planning unit 103 executes corner detection processing on the three-dimensional point cloud data, and the storage container 201, By combining with the position data and the posture data of the boxed planned object 203 and the stored object 204, the corner of the area without the boxed planned object 203 and the stored object 204 in the storage container 201 is obtained.
  • the boxing planning unit 103 can obtain not only a two-dimensional corner viewed from a specific direction of the storage container 201 but also a three-dimensional corner by using the three-dimensional point cloud data. In addition, the boxing planning unit 103 may exclude the position determined to be unable to store the object by processing the previous object from the storage position candidates. The boxing plan unit 103 can speed up the generation of the boxing plan by narrowing down the storage position candidates using the processing result of the previous object.
  • the boxing planning unit 103 executes a simulation for holding the processing object with the manipulator 105 (step 303).
  • the manipulator 105 mounts a suction pad (not shown) on its hand (picking head) to hold an object.
  • a vacuum control device not shown
  • the holding operation of the manipulator 105 in the simulation by the boxing planning unit 103 means that the suction pad and the processing object move integrally. Further, the suction pad need not be used in order for the manipulator 105 to hold the object.
  • the processing object may be gripped with a multi-finger gripper.
  • the processing object When the processing object is actually held by the manipulator 105, the processing object can be gripped by the multi-finger gripper even when the processing object does not have a flat surface or when the processing object is heavy and cannot be sucked by the suction pad.
  • the holding operation of the manipulator 105 in the simulation by the boxing planning unit 103 means that the multi-finger gripper and the processing object move together as in the case of using the suction pad.
  • the boxing planning unit 103 selects one of the storage position candidates calculated in step 302 as a target storage position (step 304).
  • the boxing planning unit 103 assigns different numbers to the respective storage position candidates, and selects a storage position candidate corresponding to the number selected using a random number as the target storage position.
  • the boxing planning unit 103 does not have to randomly select the target storage position.
  • the boxing planning unit 103 may prioritize the storage position candidates according to a predetermined rule and select the storage position candidate with the highest priority as the target storage position.
  • the rules for assigning priorities increase the priority of storage position candidates that are in contact with the previously processed object, and decrease the priority of storage position candidates that are determined to be unable to be stored by processing the previous object.
  • the priority is increased in the ascending order of points projected on all the storage position candidates on a certain axis (for example, the axis parallel to one side of the storage container 201), and the priority is increased in the ascending order of the similarity of the shape with the processing object.
  • the target storage position selected by the boxing planning unit 103 may be plural. In that case, the boxing planning unit 103 executes the determination of whether or not the trajectory can be generated in step 305 for each target storage position.
  • the boxing planning unit 103 determines whether or not the manipulator 105 can generate a trajectory that moves the processing object toward the target storage position (step 305). For trajectory generation, methods such as Rapidly-exploring Random Trees (RRT), Probabilistic Roadmap Method (PRM), and linear interpolation are used.
  • RRT Rapidly-exploring Random Trees
  • PRM Probabilistic Roadmap Method
  • linear interpolation are used.
  • obstacles such as the storage container 201, the boxed planned object 203, the stored object 204, and the storage target object 202 other than the processing object in the region through which the processing object and the robot 101 pass
  • the boxing planning unit 103 determines that a trajectory can be generated, generates a trajectory, and executes Step 306. In other cases, the boxing planning unit 103 determines that the trajectory cannot be generated, and executes Step 308.
  • the boxing planning unit 103 executes a simulation in which the manipulator 105 moves the processing object according to the trajectory generated in step 305 (step 306).
  • the boxing planning unit 103 executes a simulation for releasing the holding of the processing object by the manipulator 105 (step 307), and executes step 310.
  • the boxing planning unit 103 excludes the target storage position selected in step 304 from the storage position candidates (step 308), and determines whether storage position candidates remain (step 309). If the storage position candidates remain, the boxing planning unit 103 returns to step 304. If no storage position candidates remain, the boxing planning unit 103 determines that the processing object cannot be stored in the storage container 201 and executes step 310.
  • the boxing plan unit 103 updates the boxing plan data 107 stored in the memory 102 (step 310).
  • the boxing plan data 107 includes storage order data and storage geometric data of each boxed planned object 203.
  • the storage order data is an order for selecting a processing object set in advance. If it is determined in step 309 that the processing object cannot be stored in the storage container 201, the boxing planning unit 103 thins out the corresponding processing object from the storage order data.
  • the initial value of the storage geometric data is the geometric model data 109 for grasping the state of the storage planned object 202.
  • the boxing planning unit 103 When simulating the movement of the processing object in Step 306, the boxing planning unit 103 overwrites the storage geometric data of the processing object with the position data and orientation data of the processing object of the movement destination, and the processing object is replaced with the boxed planned object 203. To do. Further, the boxing plan data 107 may include the trajectory (data) generated in step 305. By including the trajectory in the boxing plan data 107, it is not necessary to generate the trajectory again by the motion generation unit 104, and the motion generation can be speeded up.
  • the operation generation unit 104 After the boxing plan unit 103 completes the generation of the boxing plan for the plurality of scheduled storage objects 202, the operation generation unit 104 generates an operation for executing the generated boxing plan by the manipulator 105. Based on the state grasping geometric model data 109 and the boxing plan data 107 stored in the memory 102, the manipulator 105 generates an operation for holding, moving, and storing the boxed planned objects 203 in order. When the boxing plan data 107 includes a trajectory for moving the boxed planned object 203, the motion generation unit 104 may generate a motion using the trajectory. In addition, every time one or a plurality of storage-scheduled objects 202 become boxed planned objects 203 in the boxing planning unit 103, the movement generation unit 104 may generate an operation of the manipulator 105.
  • the corner of the stowable area is found using the intersection of the hand area of the manipulator 105 when moved within each limit angle range as the stowable area.
  • the boxing planning unit 103 sets the hand of the manipulator 105 when the joint of the robot 101 is operated within the limit angle range without causing the hand of the manipulator 105 holding the storage object 202 to interfere with the obstacle.
  • the corner of the storable area is found with the area of the intersection of the movable area and the empty area as the storable area.
  • the configuration of the robot 101 of this embodiment is the same as that of the first embodiment.
  • FIG. 4 is a processing flowchart of the boxing planning unit 103.
  • the boxing planning unit 103 determines the moving direction of the storage object 202, executes a simulation of moving the storage object 202 with the manipulator 105, obtains a distance variation (described later) between the boundary points of the storage area, and calculates the distance variation. Find the corner of the stowable area based on.
  • one storage target object 202 is selected in order from the plurality as a processing object, and the boxing planning unit 103 displays a diagram for the selected processing object. 4 is executed.
  • the order in which the processing objects are selected from the plurality of objects to be stored 202 is set in advance according to the descending order of the size of the objects. Further, the boxing planning unit 103 may combine the processes as appropriate, such as executing the process shown in the flowchart of FIG. 4 when it is determined in step 309 of FIG. 3 that the processing object cannot be stored in the storage container 201. .
  • the boxing planning unit 103 executes the same process as step 301 in FIG. 3 (step 401).
  • the boxing planning unit 103 acquires the boxing planning geometric model data 108 from the memory 102, and calculates an area in the storage container 201 where the boxed planning object 203 and the stored object 204 are not present (step 402).
  • the boxing planning unit 103 executes the same process as step 303 in FIG. 3 (step 403).
  • the boxing planning unit 103 determines an initial value of the moving direction of the processing object (step 404). Different numbers are assigned to the entire 360 degrees of the processing object for each unit angle (for example, 1 degree), and the boxing planning unit 103 sets the direction corresponding to the number selected using a random number as the initial value of the moving direction. And The boxing planning unit 103 does not have to randomly select an initial value in the movement direction. For example, when processing the second and subsequent objects of the storage-scheduled object 202, the boxing planning unit 103 will be described later in the direction toward the representative point of the previously processed object and the processing of the previous object.
  • the moving direction when the distance fluctuation between the boundary points of the storable area becomes negative for the first time may be set as the initial value of the moving direction.
  • the boxing planning unit 103 determines whether or not a trajectory for moving the processing object by the moving unit by the manipulator 105 in the moving direction can be generated (step 405).
  • trajectory generation methods such as Rapidly-exploring Random Trees (RRT), Probabilistic Roadmap Method (PRM), and linear interpolation are used.
  • the boxing planning unit 103 determines that a trajectory can be generated, generates a trajectory, and executes Step 406. In other cases, the boxing planning unit 103 determines that the trajectory cannot be generated, and executes Step 407.
  • the boxing planning unit 103 executes a simulation of moving the processing object with the manipulator 105 according to the trajectory generated in step 405 (step 406), and returns to step 405.
  • the boxing planning unit 103 stores the current position of the processing object as the boundary point position of the storable area (step 407).
  • the boxing planning unit 103 determines the number of boundary points of the stored storage areas (step 408).
  • the boxing planning unit 103 executes step 409 when the number of boundary points is 3 or more, and executes step 413 when the number of boundary points is 2 or less.
  • the boxing planning unit 103 determines the distance variation between the boundary points of the storable area (step 409).
  • the distance variation (d) between the boundary points of the storable area is expressed by the following equation. However, the boundary point of the storage area that is stored most recently is the point P A , the boundary point of the storage area that is stored one time before is the point P B , and the boundary point of the storage area that is stored two times before is the boundary point of the storage area the point P C.
  • (Distance variation between boundary points of the housing area: d) (the point P A, the distance between the points P B) - (point P B, the point P C distance)
  • Packing planning unit 103 the absolute value of the distance variation between boundary points of the housing area is smaller than a preset threshold, and, when the distance variation between boundary points of the housing area is negative, it can accommodate the point P A It is determined that it is the corner of the area, and step 410 is executed.
  • the boxing planning unit 103 executes Step 412 when the absolute value of the distance fluctuation between the boundary points of the storable area is equal to or greater than a preset threshold value and the distance fluctuation between the boundary points of the storable area is negative. .
  • the boxing planning unit 103 executes Step 413 when the distance variation between the boundary points of the storable area is 0 or more.
  • the boxing planning unit 103 executes the same processing as step 307 in FIG. 3 (step 410).
  • the boxing plan unit 103 updates the boxing plan data 107 stored in the memory 102 (step 411).
  • the boxing plan data 107 includes storage order data and storage geometric data of each boxed planned object 203.
  • the storage order data is an order for selecting a processing object set in advance.
  • the initial value of the storage geometric data is the geometric model data 109 for grasping the state of the storage planned object 202.
  • the housing geometry data processing object is overwritten with the position data and orientation data of the processing object at the point P A, the processing object and packing planned object 203.
  • the boxing plan data 107 may include the trajectory (data) generated in step 405. By including the trajectory in the boxing plan data 107, the motion generation unit 104 can generate the movement trajectory of the object by using the trajectory included in the boxing plan data 107, so that the speed of motion generation can be increased.
  • the boxing planning unit 103 uses the boundary point position data of the storable area to calculate the moving direction in which the processing object is closer to the corner of the storable area (step 412).
  • FIG. 5 is a conceptual diagram showing the relationship between the boundary points of the storable area and the corner 502 of the storable area. As shown in FIG. 5, the distance fluctuation between the boundary points of the storable area is negative, that is, (the distance between the points P A and P B ) is smaller than the (the distance between the points P B and P C ) In some cases, the processing object 501 may be approaching the corner 502 of the retractable area.
  • the boxing planning unit 103 calculates the moving direction of the processing object, which is different from both the moving direction that has been processed (steps 405 and 407 to 409 have been executed) and the reverse direction (step 413).
  • the boxing planning unit 103 sets, as the movement direction, a direction in which the processed movement direction and the direction opposite to the processed movement direction are excluded from 360 degrees of the entire circumference of the processing object.
  • the method of selecting the movement direction may be random as in step 404, or, for example, the reflection direction of the processed movement direction with respect to the boundary surface of the storage area may be set as a new movement direction.
  • the boxing planning unit 103 returns to step 405 after calculating the moving direction.
  • the corner of the storage area can be found and the object can be stored.
  • the storage of the storage object 202 in the corner that allows the movement from a specific direction among the geometrically determined corners in the first embodiment will be described.
  • the configuration of the robot 101 of this embodiment is the same as that of the first embodiment.
  • FIG. 6 is a processing flowchart of the boxing planning unit 103.
  • the boxing planning unit 103 selects one point around the processing object as a target position, executes a simulation of moving the storage target object 202 by the manipulator 105, and stores the boxed planned object 203 and the stored object 204 in the storage container 201.
  • a position that is partially constrained by the boundary surface of the storable area with no storage is defined as the storage position of the storage object 202.
  • one storage target object 202 is selected in order from the plurality as a processing object, and the boxing planning unit 103 displays a diagram for the selected processing object.
  • the order in which the processing objects are selected from the plurality of objects to be stored 202 is set in advance according to the descending order of the size of the objects.
  • the boxing planning unit 103 executes the same processing as Step 401 to Step 403 in FIG. 4 (Step 601 to Step 603).
  • the boxing planning unit 103 selects one point around the processing object as a target position in the area calculated in step 602 (step 604).
  • a peripheral area of a predetermined size is divided into a predetermined number of small blocks and a number is assigned to each block, and the boxing planning unit 103 selects a representative point of the block corresponding to the number selected using a random number.
  • the position is the target position.
  • the boxing planning unit 103 distributes a plurality of target position candidates in the area calculated in step 602 before the execution of step 604, and sets the target position candidate closest to the current position of the processing object as the target. It is good also as a position.
  • the boxing planning unit 103 determines whether or not a trajectory for moving the object to be processed by the manipulator 105 toward the target position selected in Step 604 is determined (Step 605).
  • trajectory generation methods such as Rapidly-exploring Random Trees (RRT), Probabilistic Roadmap Method (PRM), and linear interpolation are used.
  • RRT Rapidly-exploring Random Trees
  • PRM Probabilistic Roadmap Method
  • linear interpolation linear interpolation
  • the boxing planning unit 103 determines that a trajectory can be generated, generates a trajectory (data), and executes Step 606. In other cases, the boxing planning unit 103 determines that the trajectory cannot be generated, and executes Step 607.
  • the boxing planning unit 103 executes a simulation of moving the processing object with the manipulator 105 according to the trajectory generated in step 605 (step 606), and returns to step 605.
  • the boxing planning unit 103 is a boundary surface of the storage area in the storage container 201 where the boxed planning object 203 and the storage object 204 are not present (empty area). Is determined to be partially constrained (step 607).
  • FIG. 7 shows an example of a state where the processing object is partially restricted.
  • FIG. 7A shows a case where the shape of the processing object 501 is a rectangular parallelepiped (bottom surface is rectangular), and
  • FIG. 7B shows a case where the shape of the processing object 501 is a cylinder (bottom surface is circular).
  • FIG. 7A since the two surfaces (lines in the drawing) of the processing object 501 are in contact with the storage container 201 and the stored object 204, the movement in the direction of the arrow 703 is performed. It is restrained.
  • FIG. 7B two lines 704 and 705 (two points in the drawing) of the processing object 501 are in contact with the storage container 201 and the stored object 204, and thus the arrow 706.
  • Directional movement is constrained.
  • the shape of the processing object 501 is a sphere, there are two contact points between the processing object 501 and the accommodable area.
  • step 607 determines in step 607 that the processing object is partially constrained, it executes step 608. If the boxing planning unit 103 determines that the processing object is not partially constrained, the process returns to step 604.
  • the boxing planning unit 103 executes the same process as step 307 in FIG. 3 (step 608).
  • the boxing plan unit 103 updates the boxing plan data 107 stored in the memory 102 (step 609).
  • the boxing plan data 107 includes storage order data and storage geometric data of each boxed planned object 203.
  • the storage order data is an order for selecting a processing object set in advance.
  • the initial value of the storage geometric data is the geometric model data 109 for grasping the state of the storage planned object 202.
  • the storage geometric data of the processed object is overwritten with the position data and orientation data of the processed object being processed, and the processed object is set as the boxed planned object 203.
  • the boxing plan data 107 may include the trajectory (data) generated in step 605. By including the trajectory in the boxing plan data 107, the motion generation unit 104 can generate the movement trajectory of the object by using the trajectory included in the boxing plan data 107, so that the speed of motion generation can be increased.
  • the boxing planning unit 103 finds the boundary surface of the packable area in the area without the boxed planned object 203 and the stored object 204 in the storage container 201, the object is moved along the boundary surface.
  • the configuration of the robot 101 of this embodiment is the same as that of the first embodiment.
  • FIG. 8 is a processing flowchart of the boxing planning unit 103.
  • the boxing planning unit 103 executes a simulation of moving the storage target object 202 by the manipulator 105, and moves the storage target object 202 along the boundary surface after the storage target object 202 contacts the boundary surface of the storage area. The position of the storage object 202 when it comes into contact with the boundary surface again is taken as the storage position.
  • one storage target object 202 is selected in order from the plurality as a processing object, and the boxing planning unit 103 displays a diagram for the selected processing object.
  • the order in which the processing objects are selected from the plurality of objects to be stored 202 is set in advance according to the descending order of the size of the objects.
  • the boxing planning unit 103 executes the same processing as Step 401 to Step 404 in FIG. 4 (Step 801 to Step 804).
  • the area that can be stored in the storage container 201 calculated in step 802 is defined as area A.
  • the boxing planning unit 103 determines whether or not a trajectory for moving the processing object by the moving unit by the manipulator 105 in the moving direction can be generated (step 805).
  • trajectory generation methods such as Rapidly-exploring Random Trees (RRT), Probabilistic Roadmap Method (PRM), and linear interpolation are used.
  • the boxing planning unit 103 determines that a trajectory can be generated, generates a trajectory, and executes Step 806. In other cases, the boxing planning unit 103 determines that the trajectory cannot be generated, and executes Step 807.
  • the boxing planning unit 103 executes a simulation of moving the processing object with the manipulator 105 according to the trajectory generated in step 805 (step 806), and returns to step 805.
  • the boxing planning unit 103 determines the contact state between the processing object and the boundary surface of the storable area calculated in Step 802 (Step 807).
  • FIG. 9 shows a contact state between the processing object 501 and the boundary surface of the storable area (area A) calculated in step 802 when the shape of the processing object 501 is a rectangular parallelepiped.
  • 9A shows a state in which the processing object 501 is not in contact with the boundary surface of the area A calculated in step 802.
  • FIG. 9B shows a state in which the processing object 501 faces the boundary surface of the area A calculated in step 802.
  • FIG. 9C shows a state where the surface 902 and the surface 903 are in contact with each other on the boundary surface of the region A calculated in step 802.
  • FIG. 9C is the same as FIG. 7A and shows a state in which the processing object 501 is partially restrained.
  • the shape of the processing object 501 is not limited to a rectangular parallelepiped.
  • the contact state can be defined similarly to the case where the shape of the processing object 501 is a rectangular parallelepiped.
  • step 807 determines in step 807 that the processing object is not in contact with the boundary surface of the region A calculated in step 802
  • the boxing planning unit 103 returns to step 804. If the boxing planning unit 103 determines that it is in contact with the boundary surface of the area A, it executes Step 808. If the boxing planning unit 103 determines that the boundary surface of the region A is in contact with the two surfaces, step 809 is executed.
  • the boxing planning unit 103 uses the boxing planning geometric model data 108 to determine the moving direction along the boundary surface of the region A calculated in step 802 (step 808). After determining the moving direction, the boxing planning unit 103 returns to Step 805.
  • the boxing planning unit 103 executes the same process as step 307 in FIG. 3 (step 809).
  • the boxing plan unit 103 updates the boxing plan data 107 stored in the memory 102 (step 810).
  • the boxing plan data 107 includes storage order data and storage geometric data of each boxed planned object 203.
  • the storage order data is an order for selecting a processing object set in advance.
  • the initial value of the storage geometric data is the geometric model data 109 for grasping the state of the storage planned object 202.
  • the storage geometric data of the processed object is overwritten with the position data and orientation data of the processed object being processed, and the processed object is set as the boxed planned object 203.
  • the boxing plan data 107 may include the trajectory (data) generated in step 805. By including the trajectory in the boxing plan data 107, the motion generation unit 104 can generate the movement trajectory of the object using the trajectory included in the boxing plan data 107, so that the speed of motion generation can be increased.
  • the processing object can be efficiently moved based on the shape of the region A in which the boxed planned object 203 and the stored object 204 are not present in the storage container 201, so that the shape of the storable region is a convex polyhedron.
  • the generation time of the boxing plan can be shortened.
  • a boxing plan that does not depend on the storage order of the storage target objects 202 set in advance is generated by repeating the processing of the boxing planning unit 103 of any one of the first to fourth embodiments a plurality of times. The method will be described.
  • the configuration of the robot 101 of this embodiment is the same as that of the first embodiment.
  • the memory 102 stores the robot 101, the storage container 201, the storage planned object 202, the boxed planned object 203, the geometric model data 106 of the stored object 204, and the boxing plan data 107 generated by the boxing planning unit 103.
  • a data structure equivalent to the boxing plan data 107 is stored in a list as boxing plan candidate data.
  • FIG. 10 is a processing flowchart of the boxing planning unit 103.
  • the boxing planning unit 103 sets a plurality of different orders for selecting the processing objects from the plurality.
  • the boxing plan unit 103 repeatedly generates a boxing plan.
  • the boxing plan unit 103 stores the generated boxing plans in the memory 102 as boxing plan candidate data, and selects the boxing plan candidate data with the highest volumetric efficiency as the boxing plan data 107.
  • the boxing planning unit 103 repeats steps 1002 to 1006 until the number of repetitions N reaches the preset number of repetitions N MAX from the initial value 1 (step 1001).
  • the boxing planning unit 103 determines the order of selecting the processing objects (step 1002).
  • the order in which the processing objects are selected is the order of numbers selected by assigning different numbers to the plurality of objects to be stored 202 and using random numbers.
  • the order of selecting the processing objects may not be determined randomly. For example, the order may be determined according to the descending order of the size of the objects.
  • the process executed in step 1004 to be described later includes a step of selecting using a random number (such as step 304 in FIG. 3), since the randomness is obtained even if the order of selecting the process objects is the same, the process object is selected.
  • the order does not have to be changed every time.
  • the boxing planning unit 103 selects a processing object according to the order determined in step 1002 (step 1003).
  • the boxing planning unit 103 executes one of the processes shown in the processing flowchart of the boxing planning unit 103 in FIGS. 3, 4, 6, and 8 (step 1004).
  • the boxing planning unit 103 determines whether the processing object being processed is the last storage target object 202 in the order determined in step 1002 (step 1005). If the processing object being processed is the storage scheduled object 202 at the end of the order, the boxing planning unit 103 executes Step 1006. If the processing object being processed is not the storage-scheduled object 202 at the end of the order, the boxing planning unit 103 returns to Step 1003.
  • the boxing plan unit 103 additionally stores the boxing plan data 107 stored in the memory 102 in the boxing plan candidate data in the memory 102 (step 1006).
  • the boxing plan unit 103 acquires boxing plan candidate data, and calculates the volumetric efficiency of boxing in each candidate (step 1007).
  • the boxing plan unit 103 stores the boxing plan candidate data having the highest volumetric efficiency of boxing in the memory 102 as the boxing plan data 107 (step 1008).
  • a plurality of boxing plan candidate data having randomness can be obtained, and the boxing plan candidate data having the highest boxing capacity efficiency (storage efficiency in the storage container) can be selected. It becomes difficult to fall into a local solution, and a boxing plan with high volumetric efficiency of boxing can be generated.
  • FIG. 11 is a configuration example of the robot 101 in this embodiment.
  • the robot 101 includes a memory 102, a detection unit 1101, a boxing planning unit 103, and a manipulator 105.
  • the memory 102 stores the geometric model data 106 of the robot 101, the storage container 201, the storage object 202 and the stored object 204, and the boxing plan data 107 generated by the boxing planning unit 103.
  • the geometric model data 106 includes boxing planning geometric model data 108 and state grasping geometric model data 109.
  • the geometric model data 106 is the same as in the first to fourth embodiments, but details of the shape data need not be acquired.
  • the detection unit 1101 detects an external force applied to the robot 101.
  • the boxing plan unit 103 uses the boxing plan geometric model data 108 stored in the memory 102 to generate a boxing plan for the storage container 201 of the storage target object 202.
  • the boxing plan unit 103 stores the generated boxing plan in the memory 102 as boxing plan data 107.
  • the boxing plan unit 103 generates an operation to be executed by the manipulator 105 along with the generation of the boxing plan.
  • the manipulator 105 executes the operation generated by the boxing planning unit 103.
  • FIG. 12 is a processing flowchart of the boxing planning unit 103.
  • This processing flowchart is the same as the processing flowchart (FIG. 6) of the third embodiment except that the operation of the manipulator 105 is replaced with the actual operation from the simulation, and the data used for the contact determination is replaced with the sensor data from the boxing planning geometric model data 108. Yes.
  • the present embodiment will be described with reference to FIG. 12 in which the flowchart (FIG. 6) of the third embodiment is replaced. However, the first embodiment (FIG. 3), the second embodiment (FIG. 4), and the fourth embodiment (FIG. 8). Each of the flowcharts can be similarly replaced.
  • the boxing planning unit 103 executes the same processing as Step 601 and Step 602 in FIG. 6 (Step 1201 and Step 1202).
  • the boxing planning unit 103 holds the processing object with the manipulator 105 (step 1203).
  • the boxing planning unit 103 executes the same process as in step 604 (step 1203).
  • the boxing planning unit 103 determines whether or not a trajectory for moving the processing object with the manipulator 105 toward the target position selected in Step 1204 can be generated (Step 1205).
  • trajectory generation methods such as Rapidly-exploring Random Trees (RRT), Probabilistic Roadmap Method (PRM), and linear interpolation are used.
  • the boxing planning unit 103 determines that a trajectory can be generated, generates a trajectory (data), and executes Step 1206. . In other cases, the boxing planning unit 103 determines that the trajectory cannot be generated, and executes Step 1207.
  • the boxing planning unit 103 moves the processing object with the manipulator 105 according to the trajectory generated in step 1205, and returns to step 1205 (step 1206).
  • the boxing planning unit 103 Based on the output result of the detection unit 1101 (external force applied to the robot 101), the boxing planning unit 103 sets the storage area of the storage container 201 in the storage container 201 in the area where the boxed planned object 203 and the stored object 204 are not present. It is determined whether or not the boundary surface is partially constrained (step 1207).
  • the detection unit 1101 is a force sensor or a contact sensor mounted on the hand or wrist of the manipulator 105. When the sensor value is larger than a preset threshold value, the processing object held by the manipulator 105 is in contact with either the storage container 201 or the stored object 204, and the boxing planning unit 103 is as shown in FIG. It is determined that the processing object is partially constrained, and step 1208 is executed.
  • the processing object held by the manipulator 105 is not in contact with either the storage container 201 or the stored object 204, and the boxing planning unit 103 determines that the processing object is partially constrained. If it is not determined, the process returns to step 1204.
  • the boxing planning unit 103 releases the holding of the processing object by the manipulator 105 (step 1208).
  • the boxing plan unit 103 updates the boxing plan data 107 stored in the memory 102 (step 1209).
  • the boxing plan data 107 includes storage order data and storage geometric data of each boxed planned object 203.
  • the storage order data is an order for selecting a processing object set in advance.
  • the initial value of the storage geometric data is the geometric model data 109 for grasping the state of the storage planned object 202.
  • the boxing planning unit 103 overwrites the storage geometric data of the processed object with the position data and orientation data of the processed object being processed, and sets the processed object as the stored object 204.
  • a boxing plan can be generated even when detailed shape data of the storage-scheduled object 202 cannot be acquired.

Landscapes

  • Manipulator (AREA)

Abstract

La présente invention concerne un robot d'emballage de boîte qui comprend : un manipulateur pour stocker des objets devant être stockés dans un récipient de stockage ; une mémoire pour stocker des données de modèle géométrique pour le récipient de stockage et les objets devant être stockés ; et une unité de planification d'emballage de boîte pour calculer la surface vide du récipient de stockage au moyen des données de modèle géométrique pour le récipient de stockage, amener le manipulateur à tenir un objet devant être stocké et déplacer celui-ci dans la zone vide, trouver un coin d'une zone dans laquelle le stockage est possible qui est dans la zone vide et dans la portée dans laquelle le manipulateur peut se déplacer sur la base des résultats du déplacement de l'objet étant stocké, déterminer le coin trouvé en tant que position de stockage pour l'objet étant stocké et mettre à jour les données de planification d'emballage de boîte sur la base de la position de stockage déterminée.
PCT/JP2016/056106 2016-02-29 2016-02-29 Robot d'emballage de boîte et procédé de planification d'emballage de boîte WO2017149616A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/JP2016/056106 WO2017149616A1 (fr) 2016-02-29 2016-02-29 Robot d'emballage de boîte et procédé de planification d'emballage de boîte

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2016/056106 WO2017149616A1 (fr) 2016-02-29 2016-02-29 Robot d'emballage de boîte et procédé de planification d'emballage de boîte

Publications (1)

Publication Number Publication Date
WO2017149616A1 true WO2017149616A1 (fr) 2017-09-08

Family

ID=59743659

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2016/056106 WO2017149616A1 (fr) 2016-02-29 2016-02-29 Robot d'emballage de boîte et procédé de planification d'emballage de boîte

Country Status (1)

Country Link
WO (1) WO2017149616A1 (fr)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107977756A (zh) * 2017-12-21 2018-05-01 厦门大学嘉庚学院 解决三维装箱问题的三叉树规划计算方法
CN109835706A (zh) * 2017-11-24 2019-06-04 发那科株式会社 工件配置系统
WO2019206924A1 (fr) * 2018-04-23 2019-10-31 Franka Emika Gmbh Procédé d'introduction d'objets dans un réceptacle d'objet commun
CN110615227A (zh) * 2019-09-10 2019-12-27 灵动科技（北京）有限公司 自动搬运系统
CN110884715A (zh) * 2018-09-07 2020-03-17 株式会社日立物流 机器人系统及其控制方法
US11045946B2 (en) 2018-03-19 2021-06-29 Kabushiki Kaisha Toshiba Holding device, transport system, and controller
DE112021002430T5 (de) 2020-04-21 2023-02-16 Fanuc Corporation Positionseinstellvorrichtung zum einstellen einer werkstückstapelposition und robotervorrichtung, die mit der positionseinstellvorrichtung bereitgestellt ist

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62251811A (ja) * 1986-04-25 1987-11-02 Hitachi Ltd 貨物自動積付システム
JPH0217577A (ja) * 1988-07-06 1990-01-22 Mitsui Eng & Shipbuild Co Ltd 混載パレットの積付けパターン決定方式
JPH06182684A (ja) * 1992-12-18 1994-07-05 Hitachi Eng Co Ltd 積載ロボット用制御デ−タ生成装置及びその方法
JP2014205209A (ja) * 2013-04-11 2014-10-30 三菱電機株式会社 ロボットシステム、及びロボットシステムの制御方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62251811A (ja) * 1986-04-25 1987-11-02 Hitachi Ltd 貨物自動積付システム
JPH0217577A (ja) * 1988-07-06 1990-01-22 Mitsui Eng & Shipbuild Co Ltd 混載パレットの積付けパターン決定方式
JPH06182684A (ja) * 1992-12-18 1994-07-05 Hitachi Eng Co Ltd 積載ロボット用制御デ−タ生成装置及びその方法
JP2014205209A (ja) * 2013-04-11 2014-10-30 三菱電機株式会社 ロボットシステム、及びロボットシステムの制御方法

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10722992B2 (en) 2017-11-24 2020-07-28 Fanuc Corporation Workpiece placement system for placing workpiece in containment area or on jig
CN109835706A (zh) * 2017-11-24 2019-06-04 发那科株式会社 工件配置系统
JP2019093495A (ja) * 2017-11-24 2019-06-20 ファナック株式会社 収容領域又は治具にワークを配置するワーク配置システム
DE102018128784B4 (de) 2017-11-24 2023-06-22 Fanuc Corporation Werkstückanordnungssystem, das Werkstücke in einem Aufnahmebereich oder an einer Einspanneinrichtung anordnet
CN109835706B (zh) * 2017-11-24 2020-11-20 发那科株式会社 工件配置系统
CN107977756B (zh) * 2017-12-21 2022-03-11 厦门大学嘉庚学院 解决三维装箱问题的三叉树规划计算方法
CN107977756A (zh) * 2017-12-21 2018-05-01 厦门大学嘉庚学院 解决三维装箱问题的三叉树规划计算方法
US11045946B2 (en) 2018-03-19 2021-06-29 Kabushiki Kaisha Toshiba Holding device, transport system, and controller
WO2019206924A1 (fr) * 2018-04-23 2019-10-31 Franka Emika Gmbh Procédé d'introduction d'objets dans un réceptacle d'objet commun
EP3620272A3 (fr) * 2018-09-07 2020-06-17 Hitachi Transport System, Ltd. Système de robot et son procédé de commande
JP2020040149A (ja) * 2018-09-07 2020-03-19 株式会社日立物流 ロボットシステム及びその制御方法
CN110884715A (zh) * 2018-09-07 2020-03-17 株式会社日立物流 机器人系统及其控制方法
JP7145702B2 (ja) 2018-09-07 2022-10-03 株式会社日立物流 ロボットシステム及びその制御方法
CN110884715B (zh) * 2018-09-07 2022-05-10 株式会社日立物流 机器人系统及其控制方法
US11279504B2 (en) 2018-09-07 2022-03-22 Hitachi Transport System, Ltd. Robot system and control method thereof
CN110615227A (zh) * 2019-09-10 2019-12-27 灵动科技（北京）有限公司 自动搬运系统
CN110615227B (zh) * 2019-09-10 2021-08-06 灵动科技（北京）有限公司 自动搬运系统
WO2021047289A1 (fr) * 2019-09-10 2021-03-18 灵动科技（北京）有限公司 Système de transport automatique
DE112021002430T5 (de) 2020-04-21 2023-02-16 Fanuc Corporation Positionseinstellvorrichtung zum einstellen einer werkstückstapelposition und robotervorrichtung, die mit der positionseinstellvorrichtung bereitgestellt ist

Similar Documents

Publication Publication Date Title
WO2017149616A1 (fr) Robot d'emballage de boîte et procédé de planification d'emballage de boîte
KR102325417B1 (ko) 패킹 메커니즘을 구비한 로봇 시스템
KR102332603B1 (ko) 실시간 배치 시뮬레이션을 사용하여 패키지를 팔레트화하기 위한 로봇 시스템
EP3650181B1 (fr) Procédé de génération d'itinéraire, système de génération d'itinéraire et programme de génération d'itinéraire
US11446824B2 (en) Palletizing control device, system and method and storage medium
US11701777B2 (en) Adaptive grasp planning for bin picking
US20200078938A1 (en) Object Pickup Strategies for a Robotic Device
KR20200138076A (ko) 오류 검출 및 동적 패킹 메커니즘을 구비한 로봇 시스템
JP5144785B2 (ja) ロボットの着目部位と周辺物との干渉を予測する方法及び装置
WO2021118702A1 (fr) Procédé et système informatique de planification de mouvement sur la base d'informations d'image générées par un appareil de prise de vues
JPWO2018092860A1 (ja) 干渉回避装置
JP2022042972A (ja) オンラインロボット動作計画のフレームワーク
Kaldestad et al. Collision avoidance with potential fields based on parallel processing of 3D-point cloud data on the GPU
JP2010179443A (ja) 把持位置計算装置及び把持位置計算方法
CN114044369A (zh) 一种基于自适应巡航技术的码垛机械手的控制方法
CN111605938B (zh) 用于使用实时放置模拟来码垛包裹的机器人系统
Weng et al. A framework for robotic bin packing with a dual-arm configuration
US11883962B2 (en) Object manipulation with collision avoidance using complementarity constraints
Esaki et al. Placement-position search technique for packing various objects with robot manipulator
Romero et al. Time-optimal trajectory planning for industrial robots with end-effector acceleration constraints
WO2024075394A1 (fr) Dispositif de commande et procédé de commande
Shi et al. Real-time grasp planning with environment constraints
Costa et al. Cooperative robotic exploration and transport of unknown objects
CN117872973A (zh) 一种货物码垛的路径规划方法、系统和工控机
WO2022250659A1 (fr) Auto-génération de contraintes de trajet pour une stabilité de préhension

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16892461

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 16892461

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: JP