JP2008272886A - Gripping candidate position selecting device, gripping candidate position selecting method, gripping passage forming device and gripping passage forming method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a gripping candidate position selecting device capable of shortening time required for teaching work. <P>SOLUTION: This gripping candidate position selecting device has a normal vector calculating means for respectively calculating a normal vector to a plurality of shape pieces of constituting a shape model of a work, and a candidate position selecting means for selecting a pair of shape pieces of setting the inner product of the calculated normal vector to a threshold value or less, and allowing a distance up to the other shape piece from a normal of one shape piece to fall within an allowable range, from the plurality of shape pieces, as a gripping candidate position by a robot hand. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a gripping candidate position selection device, a gripping candidate position selection method, a gripping path generation device, and a gripping path generation method.

  2. Description of the Related Art In recent years, lot picking is performed in a vehicle assembly process, in which parts supplied from a supplier are rearranged according to a production order and supplied while being synchronized with an assembly line.

  In such lot picking, a technique for picking various parts by a picking robot has been developed. As a technique for picking various parts by controlling a picking robot, a gripping position / orientation teaching device disclosed in Patent Document 1 is known.

The teaching device of Patent Literature 1 includes a component gripping data setting unit that sets three-dimensional position data of two gripping positions as component gripping data, and a gripping position / posture calculation unit that calculates a gripping position / posture of a hand based on the component gripping data. And comprising. According to the teaching device having such a configuration, since the gripping position and posture of the robot are calculated based on the two gripping positions designated for each part, the gripping position and posture are not actually operated. Can be taught.
Japanese Patent Laid-Open No. 11-58279

  However, in the above teaching apparatus, there is a problem that the teaching work requires a lot of time because the operator designates two gripping positions for each part on the image.

  The present invention has been made to solve the above problems. Therefore, an object of the present invention is to provide a gripping candidate position selection device and a gripping candidate position selection method that can reduce the time required for teaching work.

  Another object of the present invention is to provide a gripping path generation device and a gripping path generation method using the gripping candidate position selection device and the gripping candidate position selection method.

  The above object of the present invention is achieved by the following means.

  The gripping candidate position selection device of the present invention includes a normal vector calculation means for calculating a normal vector for each of a plurality of shape pieces constituting a workpiece shape model, and the calculation from the plurality of shape pieces. A pair of shape pieces whose inner product of the normal vectors to be processed is equal to or smaller than a threshold and whose distance from the normal of one shape piece to another shape piece is within an allowable range is determined as a gripping candidate position by the robot hand. And a candidate position selecting means for selecting as follows.

  The gripping candidate position selection method of the present invention includes a step of calculating a normal vector for each of a plurality of shape pieces constituting a shape model of a workpiece, and the calculated normal line from among the plurality of shape pieces. A step of selecting a pair of shape pieces whose vector inner product is equal to or smaller than a threshold and whose distance from the normal of one shape piece to another shape piece is within an allowable range as a gripping candidate position by the robot hand It is characterized by having.

  The grip path generation device of the present invention generates a grip path for the robot hand to approach the workpiece based on the grip candidate positions selected by the grip candidate position selection device.

  The grip path generation method of the present invention is characterized in that a grip path where the robot hand approaches the workpiece is generated based on the grip candidate positions selected by the grip candidate position selection method.

  According to the gripping candidate position selection device and the gripping candidate position selection method of the present invention, since the gripping candidate position is automatically selected from the workpiece shape data, the time required for teaching work is shortened.

  According to the grip path generation device and the grip path generation method of the present invention, since a grip path is generated based on a grip candidate position selected in advance, the time required for picking work is reduced.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In addition, the same code | symbol was used for the same member in the figure.

  FIG. 1 is a diagram showing a schematic configuration of a picking system according to an embodiment of the present invention. As shown in FIG. 1, the picking system according to the present embodiment includes a picking robot 100, a gripping candidate position selection device 200, and a gripping path generation device 300. Further, in the present embodiment, the rigid workpieces having the same shape are housed in the component box 400 in a state of being stacked in bulk.

  In the embodiment described below, the picking robot 100 grips the workpiece by offline processing in the gripping candidate position selection device 200 before the picking robot 100 grips the workpiece housed in the parts box 400. A plurality of gripping candidate positions on the workpiece that can be selected are selected. Then, one gripping candidate position is selected from among a plurality of gripping candidate positions by online processing in the gripping path generation device 300. The operation of the picking robot 100 is controlled so as to grip the workpiece at the selected gripping candidate position. Hereinafter, the picking robot 100, the gripping candidate position selection device 200, and the gripping path generation device 300 according to the present embodiment will be sequentially described.

<Picking robot>
The picking robot 100 takes out a workpiece from the parts box 400. The picking robot 100 includes a robot main body 110 and a hand unit 120 provided at the tip of the robot main body 110. The robot body 110 is controlled by the grip path generation device 300 and guides the hand unit 120 to the vicinity of the component box 400 and corrects the posture of the hand unit 120 while allowing the hand unit 120 to approach the workpiece. The hand unit 120 is an opposing two-finger type robot hand, and clamps a workpiece when the two finger units move in parallel. In addition, a camera 130 that acquires image information of a work housed in the component box 400 is provided in the vicinity of the hand unit 120. Since the picking robot 100 itself is a general articulated robot, detailed description thereof is omitted.

<Gripping candidate position selection device>
The gripping candidate position selection device 200 selects a plurality of gripping candidate positions for the workpiece by the hand unit 120 before the hand unit 120 of the picking robot 100 grips the workpiece. The gripping candidate position selection device 200 selects a pair of triangular patches that satisfy a gripping condition described later as gripping candidate positions by the hand unit 120 from among a plurality of triangular patches (shape pieces) constituting the workpiece shape model.

  Further, the gripping candidate position selection device 200 according to the present embodiment selects an approach candidate vector indicating the approach direction of the hand unit 120 to the workpiece for each gripping candidate position.

  FIG. 2 is a block diagram illustrating a schematic configuration of the gripping candidate position selection apparatus according to the present embodiment. As illustrated in FIG. 2, the gripping candidate position selection device 200 is a computer such as a personal computer or a workstation, and includes a CPU 210, a RAM 220, a ROM 230, a hard disk 240, an input unit 250, a display unit 260, and an interface 270. These units are connected to each other via a bus.

  The CPU 210 includes a normal vector calculation unit (normal vector calculation unit), a candidate position selection unit (candidate position selection unit), a comparison unit (comparison unit), a first limitation unit (first limitation unit), and a radiation vector calculation unit. (Radiation vector calculating means), second limiting section (second limiting means), approach candidate vector calculating section (approaching candidate vector calculating means), interference determining section (interference determining means), and approach candidate vector selecting section (approaching candidate) Functions as a vector selection means). Here, the normal vector calculation unit calculates a normal vector for a plurality of triangular patches constituting the workpiece shape model. In addition, the candidate position selection unit has a normal vector inner product that is equal to or smaller than a threshold value among a plurality of triangular patches, and the distance from the normal of one triangular patch to another triangular patch is within an allowable range. A pair of triangular patches included is selected as a gripping candidate position by the hand unit 120.

  The comparison unit compares the interval between the pair of triangular patches and the opening amount of the hand unit 120, and the first limiting unit selects the triangular patch selected as the gripping candidate position based on the comparison result in the comparison unit. The pair of is limited. The radiation vector calculation unit calculates a radiation vector from one triangular patch to another adjacent triangular patch on the same plane of the workpiece, and the second limiting unit is a sum of inner products of a plurality of radiation vectors. Based on the above, the pair of triangular patches selected as the gripping candidate positions is limited.

  The approach candidate vector calculation unit calculates an approach candidate vector indicating an approach direction to the workpiece when the hand unit 120 grips the grip candidate position, and the interference determination unit performs the hand unit 120 according to the approach candidate vector. When the hand approaches the work, it is determined whether or not the hand unit 120 interferes with the work. An approach candidate vector selection part excludes an approach candidate vector from a candidate, when a workpiece | work and the hand part 120 interfere.

  The RAM 220 temporarily stores shape data of the work and the hand unit 120, and the ROM 230 stores a control program, parameters, and the like in advance. The hard disk 240 stores various processing programs executed by the CPU 210.

  The input unit 250 is a pointing device such as a keyboard, a touch panel, and a mouse, for example. The display unit 260 is a liquid crystal display or a CRT display, for example.

  The interface 270 receives input of shape data of the workpiece and the hand unit 120 from an external CAD device or the like, and outputs grip position / posture data to the grip path generation device 300.

<Grip path generator>
The gripping path generation device 300 selects one gripping candidate position from among a plurality of gripping candidate positions selected by the gripping candidate position selection apparatus 200, and generates a gripping path for the hand unit 120 to approach the workpiece. is there.

  The grip path generation device 300 according to the present embodiment generates a grip path of the hand unit 120 by selecting one approach candidate vector from among a plurality of approach candidate vectors selected for each gripping candidate position.

  FIG. 3 is a block diagram illustrating a schematic configuration of the grip path generation device of the present embodiment. As illustrated in FIG. 3, the grip path generation device 300 is a computer such as a personal computer or a workstation, and includes a CPU 310, a RAM 320, a ROM 330, a hard disk 340, an input unit 350, a display unit 360, and an interface 370. These units are connected to each other via a bus.

  The CPU 310 performs an inner product calculation unit (inner product calculation unit) that calculates an inner product of an approach vector from the hand unit 120 toward the work and an approach candidate vector, and an approach candidate vector that maximizes an inner product of the approach vector. It functions as a route generation unit (route generation means) that generates a grip route of the unit 120.

  The RAM 320 temporarily stores approach candidate vectors and the like, and the ROM 330 stores a control program, parameters, and the like in advance. The hard disk 340 stores various processing programs executed by the CPU 310.

  The input unit 350 is a pointing device such as a keyboard, a touch panel, and a mouse, for example. The display unit 360 is a liquid crystal display or a CRT display, for example.

  The interface 370 receives input of gripping position / posture data from the gripping candidate position selection device 200 and outputs an operation command to the picking robot 100.

  As described above, according to the picking system of the present embodiment configured, the gripping condition described later is selected from among a plurality of triangular patches constituting the workpiece shape model before the picking robot 100 grips the workpiece. A plurality of pairs of triangular patches satisfying the above are selected as gripping candidate positions by the hand unit 120. Then, one approach candidate vector is selected from a plurality of approach candidate vectors selected for each gripping candidate position, and a gripping path of the hand unit 120 is generated based on the selected one approach candidate vector. Hereinafter, the processing in the picking system of the present embodiment will be described in detail with reference to FIGS.

  First, with reference to FIG. 4, a process in which the gripping candidate position selection device 200 shown in FIG. 1 selects a gripping candidate position from among triangular patches constituting the workpiece shape model will be described.

  FIG. 4 is a flowchart showing basic processing in the gripping candidate position selection apparatus shown in FIG. In the gripping candidate position selection processing in the present embodiment, first, gripping candidate positions are selected from among the triangular patches constituting the workpiece shape model. For each selected gripping candidate position, a plurality of approach candidate vectors for picking the workpiece by the picking robot 100 are selected.

  As shown in FIG. 4, in the gripping candidate position selection process according to the present embodiment, first, the shape data of the workpiece is read (step S101). In this embodiment, CAD data constituting a workpiece shape model is read from an external CAD device. The read CAD data is converted into a triangular patch according to, for example, STL (Stereo Lithography Interface Format).

  Next, a normal vector is calculated for the triangular patches constituting the workpiece shape model (step S102). More specifically, as shown in FIG. 5, an outward normal vector NV is calculated for each triangular patch constituting the workpiece shape model. The normal vector NV of each triangular patch is calculated from the coordinate values of three points constituting the triangular patch. In addition, the surface front / back information is inherited from the CAD data of the workpiece.

  Next, a pair of triangular patches having a normal vector whose inner product is equal to or smaller than a threshold value is selected from the plurality of triangular patches whose normal vectors have been calculated (step S103). In the present embodiment, a pair of triangular patches whose inner product of normal vectors is equal to or less than a threshold value is selected from the plurality of triangular patches whose normal vectors have been calculated in the process shown in step S102, so that they are mutually reversed. A pair of triangular patches having a normal vector of orientation is selected. Details of the processing shown in step S103 will be described later.

  Next, a triangle whose distance from the normal of one triangular patch that forms a pair of triangular patches to another triangular patch is included in the allowable range from a pair of triangular patches whose inner product of normal vectors is equal to or less than a threshold value. A patch pair is selected (step S104). In the present embodiment, by calculating the inner product of the direction vector from one triangular patch constituting the triangular patch pair to the other triangular patch and the normal vector of the single triangular patch, A pair of triangular patches whose distance from the normal to another triangular patch is included within an allowable value is selected. Details of the processing shown in step S104 will be described later.

  As described above, according to the processing shown in steps S103 to S104, as shown in FIG. 6, the inner product of the normal vectors NV1 and NV2 is equal to or less than the threshold value from among a plurality of triangular patches constituting the workpiece shape model. In addition, a pair of triangular patches M1 and M2 whose distance from the normal line of one triangular patch to another triangular patch is included in the allowable range is selected as a gripping candidate position by the hand unit 120.

  Next, among the plurality of triangular patch pairs selected as gripping candidate positions in the processing shown in steps S101 to S104, the triangular patch pair whose interval between the triangular patch pairs is outside the range of the opening amount of the hand unit 120. Are excluded (step S105). In the present embodiment, a plurality of pairs of triangular patches selected as gripping candidate positions in the processing shown in steps S101 to S104 based on a comparison result between the interval between the pairs of triangular patches and the opening amount of the hand unit 120. It is limited to those included in the range of the opening amount of the hand unit 120. Details of the processing shown in step S105 will be described later.

  Next, the pair of triangular patches located at the ends of the pair of planes constituting the workpiece is excluded from the plurality of triangular patch pairs limited by the processing shown in step S105 (step S106). In the present embodiment, when there are more than a set number of triangular patch pairs on a pair of planes, a pair of triangular patches located at the end of the plane is selected from a plurality of triangular patch pairs present on the pair of planes. Deleted. In other words, out of a plurality of triangular patch pairs limited in the process shown in step S105, a plurality of triangular patch pairs included in a pair of planes are included in the center of the plane. The Details of the process shown in step S106 will be described later.

  As described above, according to the processing shown in steps S105 to S106, the plurality of triangular patches selected as the gripping candidate positions in the processing shown in steps S101 to S104 are limited based on the two gripping conditions described above.

  Next, when the picking robot 100 grips a pair of triangular patches selected as gripping candidate positions, it is determined whether or not the hand unit 120 and the workpiece interfere (step S107). In the present embodiment, first, a plurality of approach candidate vectors indicating the direction in which the hand unit 120 approaches the workpiece is selected for each gripping candidate position. Then, when the hand unit 120 approaches the work according to the selected approach candidate vector, it is determined whether or not the hand unit 120 interferes with the work. As a result, an approach candidate vector is selected for each workpiece gripping candidate position so that the workpiece and the hand unit 120 do not interfere with each other. Details of the processing shown in step S107 will be described later.

  Next, gripping position / orientation data is created based on the approach candidate vectors selected in the process shown in step S107 (step S108). In the present embodiment, a plurality of triangular patch pairs selected as gripping candidate positions in the processing shown in steps S101 to S106, and a plurality of approach candidate vectors selected for each triangular patch pair in the processing shown in step S107. Gripping position and orientation data including information is created.

  Then, the interval between the pairs of triangular patches calculated in the process shown in step S105 is attached as opening amount data indicating the opening amount of the hand unit 120, and gripping position / posture data is output (steps S109 and S110).

  As described above, according to the process of the flowchart shown in FIG. 4, first, a plurality of triangular patch pairs are selected as gripping candidate positions by the hand unit 120 from among a plurality of triangular patches constituting the workpiece shape model. . Then, a plurality of approach candidate vectors are selected for each selected gripping candidate position, and gripping position / posture data is created. Hereinafter, each process shown in the flowchart of FIG. 4 will be described in detail with reference to FIGS.

  FIG. 7 is a flowchart for explaining the inner product evaluation process shown in step S103 of FIG. As described above, in the process shown in the flowchart of FIG. 7, a pair of triangular patches whose inner product of normal vectors is equal to or less than the threshold is selected from among a plurality of triangular patches.

  As shown in FIG. 7, in the inner product evaluation process according to the present embodiment, first, a pair of triangular patches is selected from a plurality of triangular patches whose normal vectors have been calculated (step S201).

  Next, the inner product of the selected pair of triangular patches is calculated, and it is determined whether or not the calculated inner product is equal to or less than a threshold value (steps S202 and S203). In the present embodiment, as shown by a broken line V1 in FIG. 8A, for example, a unit vector is normalized so that a pair of triangular patches having normal vectors opposite to each other are selected as gripping candidate positions. It is determined whether or not the inner product of the normalized normal vectors NV1 and NV2 is −0.9 (NV1 · NV2 = −0.9) or less.

  When the inner product of the normal vectors NV1 and NV2 is larger than the threshold value (step S203: NO), the process proceeds to step S205 and subsequent steps. On the other hand, if the inner product is equal to or smaller than the threshold value (step S203: YES), a corresponding triangular patch pair is selected (step S204). Then, it is determined whether or not there is a next pair of triangular patches (step S205), and the processes shown in steps S201 to S204 are repeated until there is no next pair of triangular patches.

  As described above, according to the processing of the flowchart shown in FIG. 7, a plurality of pairs of triangular patches whose inner product of normal vectors is equal to or less than the threshold value are selected from among the triangular patches whose normal vectors have been calculated. As a result, according to the processing of the flowchart shown in FIG. 7, as indicated by the broken line V <b> 1 in FIG. 8A, a plurality of triangular patches having normal vectors NV <b> 1 and NV <b> 2 that are opposite to each other. Pair (pair group V1) is selected. The selected pair of triangular patches is stored in the RAM 220, for example.

  Next, the positional relationship evaluation process shown in step S104 of FIG. 4 will be described in detail with reference to FIG. As described above, in the process shown in the flowchart of FIG. 9, as shown in FIG. 8A, a pair of triangular patches is selected from a plurality of pairs of triangular patches (pair group V1) having reverse normal vectors. A pair of triangular patches (pair group V2) in which the distance from one triangular patch to the other triangular patch is within the allowable range is selected.

  As shown in FIG. 9, in the positional relationship evaluation process in the present embodiment, first, one of a plurality of triangular patch pairs having reverse normal vectors selected in the process of the flowchart shown in FIG. A triangular patch pair is selected (step S301).

  Next, a direction vector from one triangular patch constituting the selected triangular patch pair to another triangular patch is calculated (step S302). In the present embodiment, as shown in FIG. 8B, the direction vector MP from the midpoint (center of gravity) of one triangular patch M1 to the midpoint of another triangular patch M2 is calculated.

  Next, the inner product of the direction vector MP and the normal vector NV1 is calculated, and it is determined whether or not the calculated inner product is equal to or less than a threshold value (steps S303 and S304). In the present embodiment, for example, a normal vector NV1 normalized to a unit vector and a direction so as to select a pair of triangular patches having normal vectors located on substantially the same straight line as a gripping candidate position. It is determined whether or not the inner product with the vector MP is −0.9 (MP · NV1 = −0.9) or less.

  If the inner product of the vectors is less than or equal to the threshold (step S304: YES), the process proceeds to step S306 and subsequent steps. On the other hand, when the inner product of the vectors is larger than the threshold (step S304: NO), the selected triangular patch pair is excluded from the candidates (step S305). Then, it is determined whether or not there is a next pair of triangular patches (step S306), and the processes shown in steps S301 to S305 are repeated until there is no next pair of triangular patches.

  As described above, according to the processing of the flowchart shown in FIG. 9, as shown in FIG. 8A, one triangle is obtained from a pair (pair group V <b> 1) of a plurality of triangular patches having reverse normal vectors. A pair (pair group V2) of triangular patches M1 and M2b whose distance from the normal of the patch to another triangular patch is included in the allowable range is selected.

  Next, the interval evaluation process shown in step S105 of FIG. 4 will be described in detail with reference to FIG. As described above, in the processing shown in the flowchart of FIG. 10, as shown in FIG. 11, a plurality of triangular patch pairs (pair group V <b> 2) selected as gripping candidate positions are included within the range of the opening amount of the hand unit 120. Limited to a pair of triangular patches (pair group V3).

  As shown in FIG. 10, in the interval evaluation process according to the present embodiment, first, a pair of triangular patches is selected from a plurality of triangular patch pairs selected in the processes of the flowcharts shown in FIGS. (Step S401).

  Next, an interval between one triangular patch forming a pair of the selected triangular patch and another triangular patch is calculated (step S402). In the present embodiment, as shown in FIG. 11A, the distance LH between the midpoint of one triangular patch M1 and the midpoint of another triangular patch M2 is calculated.

  Next, it is determined whether or not the calculated distance LH is less than the upper limit (max) of the opening amount (step S403). If the distance LH is greater than or equal to the upper limit (step S403: NO), the triangular patch pair is excluded from the gripping candidate positions (step S404), and the process proceeds to step S407 and subsequent steps. On the other hand, when the distance LH is less than the upper limit value (step S403: YES), it is determined whether the distance LH is equal to or greater than the lower limit value (min) of the opening amount (step S405).

  When the distance LH is less than the lower limit (step S405: NO), the triangular patch pair is excluded from the gripping candidate positions (step S404), and the process proceeds to step S407 and subsequent steps. On the other hand, when the distance LH is greater than or equal to the lower limit (step S405: YES), the distance LH is stored as the opening amount data of the hand unit 120 when controlling the picking robot 100 (step S406).

  Then, it is determined whether or not there is a next pair of triangular patches (step S407), and the processes shown in steps S401 to S406 are repeated until there is no next triangular patch.

  As described above, according to the processing of the flowchart shown in FIG. 10, as shown in FIG. 11A, a plurality of triangular patch pairs (pair group V <b> 2) selected as gripping candidate positions are Is limited to a pair of triangular patches (pair group V3) included in the range.

  Next, with reference to FIG. 12, the edge candidate exclusion process shown in step S106 of FIG. 4 will be described in detail. As described above, in the process shown in the flowchart of FIG. 12, as shown in FIG. 13, a plurality of triangular patches selected as gripping candidate positions, a plurality of which exists on a pair of planes constituting the workpiece, at a set number or more. The triangular patch pair located at the end of the plane is excluded from the triangular patch pair (pair group V4).

  As shown in FIG. 12, in the edge candidate exclusion process in the present embodiment, first, a pair of triangular patches is selected from a plurality of triangular patch pairs limited by the process of the flowchart shown in FIG. (Step S501).

  Next, among the remaining triangular patch pairs excluding the single triangular patch pair selected in the process shown in S501, the same patch interval as the triangular patch pair selected in the process shown in Step S501 (that is, , Distance LH) and the number of triangular patch pairs having the same normal vector is calculated (step S502). That is, according to the process shown in step S502, as indicated by a broken line V4 in FIG. 13A, one triangular patch and another triangular patch forming a pair of triangular patches exist on a pair of parallel planes. The number of triangular patch pairs is calculated.

  Next, it is determined whether or not the calculated number of triangular patch pairs is equal to or greater than the set number (step S503). In other words, it is determined whether or not a pair of triangular patches equal to or greater than the set number are selected as gripping candidate positions on a pair of planes.

  If the number of triangular patch pairs is less than the set number (step S503: NO), the process proceeds to step S508 and subsequent steps. On the other hand, if the number of pairs of triangular patches is equal to or greater than the set number (step S503: YES), a radiation vector from one triangular patch of the triangular patch pair selected in the process shown in step S501 to the adjacent triangular patch is calculated. (Step S504). In the present embodiment, as shown in FIGS. 13B and 13C, a triangular patch positioned within a predetermined distance from the triangular patches Ma, Mx on the same plane with respect to one triangular patch Ma, Mx. Mb to Mk are selected, and a plurality of radiation vectors from the midpoints of the triangular patches Ma and Mx to the adjacent triangular patches Mb to Me and Mf to Mk are calculated.

  Then, the inner product of a plurality of radiation vectors is calculated (step S505), and it is determined whether the calculated inner product sum is less than a set value (step S506). More specifically, for example, the inner product of six radiation vector pairs selected from four radiation vectors from the triangular patch Ma to the triangular patches Mb to Me is calculated, and the six radiation vector pairs are calculated. It is determined whether the sum of the inner products is less than a set value. Note that when the triangular patches are adjacent only in a specific direction of the reference triangular patch, the sum of the inner products of such radiation vectors becomes large.

  When the sum of the inner products is less than the set value (step S506: YES), as shown in FIG. 13B, it is assumed that the pair of triangular patches Mx is adjacent to the pair of triangular patches Mx in step S508. The process proceeds to the following process. On the other hand, when the sum of the inner products is equal to or larger than the set value (step S506: NO), as shown in FIG. 13C, it is assumed that a pair of triangular patches is adjacent to the selected pair of triangular patches Ma only in a specific direction. The selected pair of triangular patches is excluded from the gripping candidate positions (step S507). As a result, the pair of triangular patches Ma positioned at the ends of the pair of planes is excluded from the gripping candidate positions as a pair of triangular patches adjacent to each other in a specific direction.

  Then, it is determined whether or not there is a next triangular patch pair (step S508), and the processes shown in steps S501 to S507 are repeated until there is no next triangular patch.

  As described above, according to the processing of the flowchart shown in FIG. 12, first, a plurality of triangular patch pairs (pair group V3) selected as gripping candidate positions are positioned on a pair of the same plane by a predetermined number or more. Of pairs (pair group V4) and other triangular patch pairs (pair group V5). After that, a pair of triangular patches positioned at the end of the plane is excluded from a plurality of triangular patch pairs (pair group V4) positioned on the same plane, and as shown in FIG. The pair of triangular patches (pair group V4) located on the plane is limited to the pair of triangular patches (pair group V6) located at the center of the plane. As a result, a plurality of triangular patch pairs (pair group V3) selected as gripping candidate positions are combined with a pair of triangular patches (pair group V6) positioned at the center on a pair of planes constituting the workpiece. It is limited to a pair of triangular patches that is a combination of a pair of triangular patches (pair group V5) located on the curved surface to be configured.

  14 to 16 are diagrams for explaining a pair of triangular patches selected by the processing of the flowchart described above. In the workpiece shape models shown in FIGS. 14 to 16, as an example, a pair of triangular patches in which the inner product of the normal vectors is −1 and the normal vectors are located on substantially the same straight line is used as a gripping candidate position. The case where it is elected is shown.

  FIG. 14A is a diagram for explaining a pair of triangular patches selected as gripping candidate positions of a deformed spherical workpiece. As shown in FIG. 14A, in a deformed spherical workpiece, for example, three pairs of triangular patches respectively corresponding to three pairs of normal vectors are selected as gripping candidate positions. On the other hand, as shown in FIG. 14B, a pair of triangular patches having normal vectors in opposite directions, and the distance from one triangular patch to another triangular patch forming the triangular patch pair is out of the allowable range. Is not selected as a gripping candidate position.

  FIG. 15A is a diagram for explaining a pair of triangular patches selected as a gripping candidate position of a drum-shaped workpiece. As shown in FIG. 15A, in the drum-shaped workpiece, for example, a pair of upper and lower triangular patches and triangular patches around the recess corresponding to six pairs of normal vectors are selected as gripping candidate positions. On the other hand, as shown in FIG. 15B, a pair of triangular patches having normal vectors that are opposite to each other, and the distance from one triangular patch to another triangular patch forming the triangular patch pair is outside the allowable range. Is not selected as a gripping candidate position.

  FIG. 16 is a diagram for explaining a pair of triangular patches selected as a gripping candidate position of a floating ring-shaped workpiece. As shown in FIG. 16A, in the floating ring-shaped workpiece, for example, pairs of triangular patches on the side surface respectively corresponding to six pairs of normal vectors in the radial direction of the floating ring-shaped workpiece are selected as gripping candidate positions. Also, as shown in FIG. 16B, for example, pairs of upper and lower triangular patches respectively corresponding to six pairs of normal vectors in the axial direction of the floating ring-shaped workpiece are selected as gripping candidate positions.

  As described above, according to the processing of the flowchart described above, a plurality of triangular patch pairs satisfying the above-described gripping conditions are selected as gripping candidate positions from among the triangular patches constituting the workpiece shape model. Hereinafter, with reference to FIGS. 17 to 19, for each pair of triangular patches selected as gripping candidate positions, processing for selecting an approach candidate vector that allows the hand unit 120 to approach the workpiece without interfering with the workpiece. Will be described.

  FIG. 17 is a flowchart for explaining the interference evaluation process shown in step S107 of FIG. As described above, according to the processing of the flowchart shown in FIG. 17, first, for each pair of triangular patches selected as the gripping candidate position, as shown in FIG. 18, when the picking robot 100 grips the pair of triangular patches. A plurality of approach candidate vectors AP indicating the approach direction of the hand unit 120 are selected. Then, according to the approach candidate vector AP, interference between the hand unit 120 and the work when the hand unit 120 approaches the work is evaluated.

  As shown in FIG. 17, in the interference evaluation process in the present embodiment, first, the shape data of the hand unit 120 of the picking robot 100 is read (step S601).

  Next, one triangular patch pair is selected from a plurality of triangular patch pairs selected as gripping candidate positions in the processing of the flowchart shown in FIG. 12 (step S602).

  Next, an approach candidate vector of the hand unit 120 when gripping the selected pair of triangular patches is calculated (step S603). In the present embodiment, as shown in FIG. 18A, first, a line segment L connecting a midpoint of one triangular patch M1 and a midpoint of another triangular patch M2 forming a pair of triangular patches is used as an axis. A circle R having a center C at the midpoint of the line segment L is set. The radius AP of the circle R is determined by the shape data of the hand unit 120. Then, a plurality of approach candidate vectors AP are calculated on the circumference of the circle R toward the center C at regular intervals at a predetermined angle. In the present embodiment, an orientation vector parallel to the line segment L is calculated together with the approach candidate vector AP, and a plurality of homogeneous transformation matrices H indicating the gripping posture of the workpiece having the approach candidate vector and the orientation vector are calculated.

  Next, one approach candidate vector is selected from the plurality of approach candidate vectors calculated in the process shown in step S603 (step S604). Then, when the hand unit 120 approaches the work according to the selected one approach candidate vector, it is determined whether or not the hand unit 120 and the work interfere with each other (step S605). In the present embodiment, as shown in FIGS. 18B and 18C, first, the shape model of the hand unit 120 is arranged on a circle R, and the shape model of the hand unit 120 and the shape model of the workpiece are The presence or absence of interference is evaluated. Next, the shape model of the hand unit 120 is slid by a set amount in the approach candidate vector direction, and the interference between the shape model of the hand unit 120 and the shape model of the workpiece is evaluated. In addition, since the technique itself which evaluates interference using the shape model of a workpiece | work and a hand part is a general interference evaluation technique, detailed description is abbreviate | omitted.

  FIG. 19 is a diagram for explaining an example of interference evaluation between the floating ring-shaped workpiece and the hand unit illustrated in FIG. 16. As shown in FIG. 19A, when the hand unit 120 approaches the workpiece according to the approach candidate vector parallel to the radial direction of the floating ring-shaped workpiece, the hand unit 120 contacts the workpiece at two points. It is determined that there is no interference with the workpiece. On the other hand, as shown in FIG. 19B, when the hand unit 120 approaches the workpiece according to the approach candidate vector parallel to the axial direction of the floating ring-shaped workpiece, the hand unit 120 comes into contact with the workpiece at three points. When the contact at is set as an error, it is determined that the hand unit 120 and the workpiece interfere with each other.

  In the process shown in step S605, when the hand unit 120 and the workpiece do not interfere with each other (step S605: NO), the process proceeds to step S607 and subsequent steps. On the other hand, when the hand unit 120 interferes with the work (step S605: YES), the selected approach candidate vector is excluded from the candidates (step S606).

  Then, it is determined whether or not there is a next approach candidate vector (step S607), and the processes shown in steps S604 to S606 are repeated until there is no next approach candidate vector. As a result, as shown in FIG. 18C, for example, a matrix Ha having approach candidate vectors in which the hand unit 120 and the workpiece interfere with each other is deleted from the candidates.

  Then, it is determined whether or not there is a next pair of triangular patches (step S608), and the processes shown in steps S602 to S607 are repeated until there is no next pair of triangular patches.

  As described above, according to the processing of the flowchart shown in FIG. 17, for each pair of triangular patches selected as gripping candidate positions, a plurality of approach candidate vectors that do not interfere with the hand unit and the workpiece are selected.

  As described above, according to the gripping candidate position selection process of the present embodiment described above, a pair of triangular patches satisfying the gripping condition described above is selected as a gripping candidate position from among a plurality of triangular patches constituting the workpiece shape model. Multiple elected. Then, a plurality of approach candidate vectors are selected for each selected gripping candidate position, and gripping position / posture data is created. Hereinafter, with reference to FIG. 20 and FIG. 21, a process in which the grip path generation device 300 generates a grip path based on the grip position / posture data generated by the grip candidate position selection apparatus 200 will be described.

  FIG. 20 is a flowchart showing a grip path generation process in the grip path generation apparatus shown in FIG. As described above, in the grip path generation process according to the present embodiment, one approach candidate vector is selected as an approach vector from among a plurality of approach candidate vectors included in the gripping position and orientation data, and the selected approach vector is used. The picking robot 100 is controlled so that the hand unit 120 holds the workpiece.

  As shown in FIG. 20, in the grip route generation process according to the present embodiment, first, grip position / posture data is read from the gripping candidate position selection device 200 (step S701). As described above, the gripping position / orientation data in the present embodiment includes information on the gripping candidate positions of the workpiece by the hand unit 120 and a plurality of approach candidate vectors selected for each gripping candidate position.

  Next, the image information of the workpiece accommodated in the parts box 400 is acquired (step S702). In the present embodiment, workpiece image information is acquired by the camera 130 and is matched with workpiece shape data for image recognition, thereby creating workpiece recognition position and orientation data.

  Next, an approach candidate vector is applied to the work for which image information has been acquired in the process shown in step S702 (step S703). More specifically, the workpiece recognition position / orientation data and the gripping position / orientation data are matched, and a plurality of approach candidate vectors are applied to the workpiece to be gripped accommodated in the component box 400 in a stacked state. The

  Next, inner products of a plurality of approach candidate vectors applied to the work in the process shown in step S703 and an approach vector from the hand unit 120 toward the work are calculated (step S704). In the present embodiment, for example, the imaging direction of camera 130 is set as the approach vector of hand unit 120.

  Next, one approach candidate vector having the maximum inner product with the approach vector is determined as an approach vector of the hand unit 120 from among a plurality of approach candidate vectors (step S705). In the present embodiment, by selecting an approach candidate vector having the maximum inner product with the approach vector of the hand unit 120, the approach candidate vector having the smallest angle with the approach vector is determined as the approach vector of the hand unit 120. . That is, the approach candidate vector having the direction most similar to the direction of the approach vector of the hand unit 120 is used as the approach vector of the hand unit 120 so that the posture correction of the hand unit 120 when approaching the workpiece can be minimized. It is determined.

  Next, based on the determined approach vector, robot motion data indicating the grip path of the hand unit 120 is created, and the created robot motion data is output to the picking robot 100 (steps S707 and S708). In the present embodiment, the correction amount for the current position and orientation of the hand unit 120 is output to the picking robot 100 as robot operation data. As a result, the hand unit 120 can approach the workpiece and grip a corresponding pair of triangular patches while the posture is corrected so as to grip the workpiece with the determined orientation of the approach vector. At this time, the opening amount of the finger portion of the hand unit 120 is adjusted based on the distance LH between the pair of triangular patches attached as opening amount data in the process shown in step S109 of the flowchart of FIG.

  The robot motion data described above is created using coordinate transformation among the camera coordinate system Σc of the camera 130, the hand coordinate system Σh of the hand unit 120, and the workpiece model coordinate system Σm (see FIG. 21). . Processing for calculating a homogeneous transformation matrix from the hand coordinate system Σh of the hand unit 120 to the gripping position and orientation on the model coordinate system Σm based on the camera coordinate system Σc and the hand coordinate system Σh whose relative positional relationships are known Since this is a general coordinate transformation technique, a detailed description thereof will be omitted.

  As described above, according to the processing of the flowchart shown in FIG. 20, the approach candidate vector in the direction most similar to the approach vector of the hand unit 120 is determined as the approach vector for generating the grip route of the hand unit 120. Therefore, the posture correction of the hand unit 120 when gripping the workpiece is minimized, and the time required for the picking process is shortened.

  As described above, the described embodiment has the following effects.

  (A) The gripping candidate position selection apparatus according to the present embodiment includes a normal vector calculation unit that calculates a normal vector for each of a plurality of triangular patches constituting a workpiece shape model, and a plurality of triangular patches. From the triangle patch pair in which the inner product of the calculated normal vectors is equal to or less than the threshold and the distance from the normal line of one triangular patch to the other triangular patch is within the allowable range. A candidate position selection unit that selects as a gripping candidate position. Therefore, since the gripping candidate position is automatically selected from the workpiece shape data, the time required for the teaching work is shortened.

  (B) The candidate position selection unit calculates a pair of triangular patches whose inner product of a direction vector from one triangular patch to another triangular patch and a normal vector of the one triangular patch is equal to or less than an allowable value. The distance from the normal to the other triangular patch is selected as a pair of triangular patches that fall within the allowable range. Therefore, by calculating the inner product of the vectors, it is possible to easily obtain a pair of triangular patches whose distance from the normal line of one triangular patch to another triangular patch is within an allowable range. That is, it is possible to easily obtain a pair of triangular patches that can be held by two fingers of the hand.

  (C) The gripping candidate position selection apparatus according to the present embodiment is based on a comparison unit that compares the interval between the pair of triangular patches and the opening amount of the hand unit, and a comparison result between the interval and the opening amount of the pair of triangular patches. And a first limiting unit that limits a pair of triangular patches selected as gripping candidate positions. Accordingly, it is possible to more precisely select a workpiece gripping candidate position by offline processing. Further, by using the interval between the pairs of triangular patches as the opening data of the hand unit, the opening amount of the hand unit when gripping the workpiece can be controlled.

  (D) The gripping candidate position selection device according to the present embodiment includes a radiation vector calculation unit that calculates a radiation vector from one triangular patch to another adjacent triangular patch on the same plane of the workpiece, and a plurality of radiation vectors. A second limiting unit that limits a pair of triangular patches selected as gripping candidate positions based on a sum of inner products; Therefore, a pair of triangular patches positioned at the ends of a pair of planes constituting the workpiece can be excluded from the gripping candidate positions. As a result, the gripping candidate position at the workpiece end that is relatively difficult for the hand unit 120 to grip is excluded, thereby improving the picking success rate. In addition, the amount of information is reduced by limiting the pairs of triangular patches selected as gripping candidate positions, and the calculation load in subsequent processing is reduced.

  (E) The gripping candidate position selection device according to the present embodiment includes an approach candidate vector calculation unit that calculates an approach candidate vector indicating an approach direction of the hand unit to the workpiece when the hand unit grips the gripping candidate position, and an approach When the hand unit approaches the workpiece according to the candidate vector, if the hand unit interferes with the interference judgment unit that determines whether or not the hand unit interferes with the workpiece, the approach candidate vector is excluded from the candidates. And an approach candidate vector selection unit. Therefore, it is possible to easily select a direction in which the hand unit can approach the workpiece without interfering with the workpiece by offline processing. In the subsequent process, the approach vector of the hand unit can be easily calculated based on the approach candidate vector.

  (F) The grip path generation device according to the present embodiment generates a grip path for the hand unit to approach the workpiece based on the grip candidate positions selected by the grip candidate position selection device. Therefore, since a grip route is generated based on the grip candidate positions selected in advance, the time required for the picking operation is reduced.

  (G) The grip path generation device of the present embodiment is calculated by an inner product calculation unit that calculates an inner product of an approach candidate vector calculated based on the grip candidate position and an approach vector from the hand unit toward the workpiece. A path generation unit that generates a grip path of the hand unit according to an approach candidate vector that maximizes the inner product. Therefore, it is possible to select an approach vector that can minimize the correction amount of the hand unit posture from a plurality of approach candidate vectors. In addition, the approach vector can be easily selected by calculating the inner product of the vectors.

  (H) The gripping candidate position selection method of the present embodiment is calculated from the step of calculating the normal vector for each of the plurality of triangular patches constituting the workpiece shape model and the plurality of triangular patches. A pair of triangular patches whose inner product of the normal vectors is equal to or smaller than a threshold and whose distance from the normal of one triangular patch to another triangular patch is within an allowable range is set as a gripping candidate position by the hand unit. Selecting. Therefore, since the gripping candidate position is automatically selected from the workpiece shape data, the time required for the teaching work is shortened.

  (I) The grip path generation method according to the present embodiment generates a grip path for the hand unit to approach the workpiece based on the grip candidate positions selected by the grip candidate position selection method. Therefore, since a grip route is generated based on the grip candidate positions selected in advance, the time required for the picking operation is reduced.

  As described above, in the embodiment described above, the gripping candidate position selection device, the gripping candidate position selection method, the gripping path generation device, and the gripping path generation method according to the present invention have been described. However, it goes without saying that the present invention can be appropriately added, modified, and omitted by those skilled in the art within the scope of the technical idea.

  For example, in the embodiment described above, after selecting a pair of triangular patches having normal vectors in the reverse direction by calculating the inner product of the normal vectors of the triangular patches constituting the workpiece shape model, one triangle is selected. A pair of triangular patches whose distance from the patch normal to another triangular patch was within the allowable range was selected. However, the selection procedure of the gripping candidate position is not limited to the above-described embodiment, and a pair of triangular patches whose distance from the normal line of one triangular patch to another triangular patch is within an allowable range is selected. Then, by calculating the inner product of the normal vectors, a pair of triangular patches having reverse normal vectors can be selected.

It is a figure which shows schematic structure of the picking system in one embodiment of this invention. It is a block diagram which shows schematic structure of the holding | grip candidate position selection apparatus shown in FIG. It is a block diagram which shows schematic structure of the holding | grip path | route production | generation apparatus shown in FIG. It is a flowchart which shows the basic process in the holding | grip candidate position selection apparatus shown in FIG. It is a figure for demonstrating the normal vector calculation process shown to step S102 of FIG. FIG. 5 is a diagram for explaining a pair of triangular patches selected by the process shown in the flowchart of FIG. 4. It is a flowchart for demonstrating the inner product evaluation process shown to step S103 of FIG. FIG. 5 is a diagram for explaining a pair of triangular patches selected by the process shown in the flowchart of FIG. 4. It is a flowchart for demonstrating the positional relationship evaluation process shown to FIG.4 S104. It is a flowchart for demonstrating the space | interval evaluation process shown to step S105 of FIG. It is a figure for demonstrating the process shown to the flowchart of FIG. It is a flowchart for demonstrating the edge part candidate exclusion process shown to FIG.4 S106. It is a figure for demonstrating the process shown to the flowchart of FIG. It is a figure for demonstrating the process which selects a holding | grip candidate position with respect to a deformation | transformation spherical shape workpiece | work. It is a figure for demonstrating the process which selects a holding | grip candidate position with respect to a drum-shaped workpiece. It is a figure for demonstrating the process which selects a holding | grip candidate position with respect to a floating ring-shaped workpiece | work. It is a flowchart for demonstrating the interference evaluation process shown to step S107 of FIG. It is a figure for demonstrating the process shown to the flowchart of FIG. It is a figure for demonstrating an example of interference evaluation with the floating ring-shaped workpiece | work shown in FIG. 16, and a hand part. It is a flowchart which shows the grip path | route production | generation process in the grip path | route production | generation apparatus shown in FIG. It is a figure for demonstrating the process shown to the flowchart of FIG.

Explanation of symbols

100 picking robot,
200 gripping candidate position selection device,
300 gripping path generation device,
400 Parts box.

Claims (9)

Normal vector calculation means for calculating normal vectors for a plurality of shape pieces constituting the workpiece shape model,
Among the plurality of shape pieces, a shape in which the inner product of the calculated normal vectors is equal to or less than a threshold value, and the distance from the normal of one shape piece to another shape piece is included in an allowable range A gripping candidate position selection device comprising candidate position selection means for selecting a pair of pieces as a gripping candidate position by a robot hand.   The candidate position selection means includes a pair of shape pieces in which an inner product of a direction vector from one shape piece to another shape piece and a normal vector of the one shape piece is an allowable value or less. The gripping candidate position selecting device according to claim 1, wherein a distance from the normal line to the other shape piece is selected as a pair of shape pieces within an allowable range. Comparison means for comparing the distance between the pair of shape pieces and the opening amount of the robot hand;
The apparatus further comprises first limiting means for limiting a pair of shape pieces selected as the gripping candidate positions based on a comparison result between the interval between the pair of shape pieces and the opening amount. Item 2. The gripping candidate position selection device according to Item 1. Radiation vector calculation means for calculating a radiation vector directed from one shape piece to another adjacent shape piece on the same plane of the workpiece;
2. The gripping device according to claim 1, further comprising: a second limiting unit configured to limit a pair of shape pieces selected as the gripping candidate positions based on a sum of inner products of a plurality of the radiation vectors. Candidate position selection device. An approach candidate vector calculating means for calculating an approach candidate vector indicating an approach direction of the robot hand to the workpiece when the robot hand grips the grip candidate position;
Interference determining means for determining whether or not the robot hand interferes with the workpiece when the robot hand approaches the workpiece according to the approach candidate vector;
The grasping candidate position selecting device according to claim 1, further comprising approach candidate vector selecting means for excluding the approach candidate vector from candidates when the workpiece and the robot hand interfere with each other.   A grip path generation device that generates a grip path for a robot hand to approach a workpiece based on the grip candidate positions selected by the grip candidate position selection device according to claim 1. Inner product calculating means for calculating an inner product of an approach candidate vector calculated based on the gripping candidate position and an approach vector from the robot hand toward the workpiece;
The grip path generation device according to claim 6, further comprising: a path generation unit that generates a grip path of the robot hand according to an approach candidate vector that maximizes the calculated inner product. Calculating a normal vector for each of a plurality of shape pieces constituting the shape model of the workpiece;
Among the plurality of shape pieces, a shape in which an inner product of the calculated normal vectors is equal to or less than a threshold value, and a distance from the normal of one shape piece to another shape piece is included in an allowable range. Selecting a pair of pieces as a candidate gripping position by a robot hand.   9. A grip path generation method, comprising: generating a grip path for the robot hand to approach a workpiece based on the grip candidate positions selected by the grip candidate position selection method according to claim 8.

Images