JP6036662B2 - Robot simulation apparatus, program, recording medium and method - Google Patents

Description

  The present invention relates to a technique for simulating a work in which a robot arm picks up a workpiece stacked in a virtual space.

  A conventional robot simulation device that arranges sensors, robots, and workpiece models, simulates actual loading and unloading work in a virtual space that creates a state where workpieces are randomly stacked, and estimates the working time, for example, It is disclosed in Patent Document 1. By estimating the work time using such a robot simulation device, it is possible to easily design and start up a system that includes actual sensors, robots, and workpieces, and to shorten the work time required for design and start-up. It becomes. In addition, when the interference of the robot hand with a workpiece other than the object to be gripped is relatively light, it is possible to change the size of the robot hand model in the virtual space and examine the robot hand with the optimum size. For example, it is disclosed in Patent Document 2.

JP2007-241857 JP2007-326160 (paragraph 0041)

  In the conventional robot simulation device, it is described that the robot hand having the optimum size in the virtual space is examined, but in the limited case that the interference of the robot hand with the workpiece other than the gripping object is relatively light Was only against. In addition, if the shape of the robot hand is not appropriate, the effect of increasing the gripping success rate by changing the dimensions is further limited. On the other hand, if the shape of the robot hand suitable for gripping can be selected according to the shape of the workpiece, the degree of freedom of selection can be expanded, and the gripping success rate can be increased in many cases. In the work for unloading and unloading workpieces by the robot, if the success rate of gripping the workpiece is increased, the tact time required for removal is dramatically improved. However, there has never been a robot simulation device that can select the shape of a robot hand that increases the success rate of gripping.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a robot simulation apparatus that can select the shape of a robot hand that increases the gripping success rate by calculation in a virtual space.

The robot simulation apparatus according to the present invention includes a stacking calculation unit that reproduces in a virtual space a state in which workpiece models obtained by modeling a workpiece are stacked, and a sensor model that models a depth measurement sensor in the virtual space. Depth measurement simulation unit that generates depth data consisting of depth information at each point by simulating depth measurement at each point of the work model using the sensor model, and hand shape model and depth data modeling the shape of multiple robot hands based on the bets, it calculates the gripping potential of the work model in the robot hand relative to the shape of a plurality of robot hand, and a hand selector for selecting the shape of the robot hand on the basis of the gripping possibility, the hand shape The model is the depth of depth data that indicates the area that should be in contact with the workpiece when gripping the workpiece. When there is a hand contact area that is a two-dimensional area projected on a plane perpendicular to the direction and an area that may collide with the work before the hand contact area touches the work when gripping the work The hand collision area is defined by a hand collision area that is a two-dimensional area obtained by projecting an area that may collide onto a vertical plane. Possibility of collision between the work model contact area, which is a candidate area for contacting the hand contact area when gripping the work, and the hand collision area before contacting the hand contact area with the work model contact area in the measurement area The work model collision area, which is a certain area, is obtained based on the depth data, and the hand contact area, hand collision area, work model contact area, And it calculates a gripping possibility based on Kumoderu impact area.
The robot simulation apparatus according to the present invention also includes a stacking unit that reproduces in a virtual space a work model obtained by modeling a work, and a sensor model that models a depth measurement sensor in the virtual space. A depth measurement simulation unit that arranges and generates depth data composed of depth information at each point by simulating depth measurement of each point of the work model by a sensor model; and a hand shape model that models the shapes of a plurality of robot hands A hand selection unit that calculates the gripping possibility of the work model in the robot hand based on the depth data for the shapes of the plurality of robot hands and selects the shape of the robot hand based on the gripping possibility. The selection unit has the highest gripping ability average value obtained for multiple types of workpiece models. There is to select the shape of the robot hand.
The robot simulation apparatus according to the present invention also includes a stacking unit that reproduces in a virtual space a work model obtained by modeling a work, and a sensor model that models a depth measurement sensor in the virtual space. A depth measurement simulation unit that arranges and generates depth data composed of depth information at each point by simulating depth measurement of each point of the work model by a sensor model; and a hand shape model that models the shapes of a plurality of robot hands A hand selection unit that calculates the gripping possibility of the work model in the robot hand based on the depth data for the shapes of the plurality of robot hands and selects the shape of the robot hand based on the gripping possibility. The stacking operation unit reproduces the stacked work model multiple times, and the hand selection unit In which the average value of the gripping possibilities determined for state reproduced workpiece model more than once stacked to select the shape of the highest robot hand.

The robot simulation apparatus according to the present invention includes a stacking calculation unit that reproduces in a virtual space a state in which workpiece models obtained by modeling a workpiece are stacked, and a sensor model that models a depth measurement sensor in the virtual space. Depth measurement simulation unit that generates depth data consisting of depth information at each point by simulating depth measurement at each point of the work model using the sensor model, and hand shape model and depth data modeling the shape of multiple robot hands based on the bets, it calculates the gripping potential of the work model in the robot hand relative to the shape of a plurality of robot hand, and a hand selector for selecting the shape of the robot hand on the basis of the gripping possibility, the hand shape The model is the depth of depth data that indicates the area that should be in contact with the workpiece when gripping the workpiece. When there is a hand contact area that is a two-dimensional area projected on a plane perpendicular to the direction and an area that may collide with the work before the hand contact area touches the work when gripping the work The hand collision area is defined by a hand collision area that is a two-dimensional area obtained by projecting an area that may collide onto a vertical plane. Possibility of collision between the work model contact area, which is a candidate area for contacting the hand contact area when gripping the work, and the hand collision area before contacting the hand contact area with the work model contact area in the measurement area The work model collision area, which is a certain area, is obtained based on the depth data, and the hand contact area, hand collision area, work model contact area, Since calculating the gripping possibilities based on Kumoderu impact area, the shape of the robot hand to increase the gripping success rate can be selected by calculation on the virtual space. Since a robot hand having a high gripping success rate is selected, the tact time required for taking out can be shortened.
The robot simulation apparatus according to the present invention also includes a stacking unit that reproduces in a virtual space a work model obtained by modeling a work, and a sensor model that models a depth measurement sensor in the virtual space. A depth measurement simulation unit that arranges and generates depth data composed of depth information at each point by simulating depth measurement of each point of the work model by a sensor model; and a hand shape model that models the shapes of a plurality of robot hands A hand selection unit that calculates the gripping possibility of the work model in the robot hand based on the depth data for the shapes of the plurality of robot hands and selects the shape of the robot hand based on the gripping possibility. The selection unit has the highest gripping ability average value obtained for multiple types of workpiece models. Since selecting the shape of the robot hand it has the shape of a robot hand to increase the gripping success rate can be selected by calculation on the virtual space. Since a robot hand having a high gripping success rate is selected, the tact time required for taking out can be shortened.
The robot simulation apparatus according to the present invention also includes a stacking unit that reproduces in a virtual space a work model obtained by modeling a work, and a sensor model that models a depth measurement sensor in the virtual space. A depth measurement simulation unit that arranges and generates depth data composed of depth information at each point by simulating depth measurement of each point of the work model by a sensor model; and a hand shape model that models the shapes of a plurality of robot hands A hand selection unit that calculates the gripping possibility of the work model in the robot hand based on the depth data for the shapes of the plurality of robot hands and selects the shape of the robot hand based on the gripping possibility. The stacking operation unit reproduces the stacked work model multiple times, and the hand selection unit Since the shape of the robot hand that has the highest average gripping probability obtained for the state where the workpiece models reproduced multiple times are stacked is selected, the shape of the robot hand that increases the gripping success rate is selected in the virtual space. It can be selected by calculating Since a robot hand having a high gripping success rate is selected, the tact time required for taking out can be shortened.

1 is a functional block diagram showing a robot simulation apparatus according to Embodiment 1 of the present invention. The flowchart for demonstrating operation | movement of the calculating part in the robot simulation apparatus by Embodiment 1 of this invention. The figure which shows an example of the display of the display part in the robot simulation apparatus by Embodiment 1 of this invention. The figure for demonstrating the graspable state of the robot hand in the robot simulation apparatus by Embodiment 1 of this invention. The figure for demonstrating the shape model of the robot hand defined in the contact area | region and the collision area | region in the robot simulation apparatus by Embodiment 1 of this invention. The figure which shows an example of the area | region depth data in the robot simulation apparatus by Embodiment 1 of this invention. The figure for demonstrating the contact area and collision area | region of a workpiece | work model in the robot simulation apparatus by Embodiment 1 of this invention. The figure for demonstrating the method of calculating | requiring the grippable area | region and grippability map in the robot simulation apparatus by Embodiment 1 of this invention. The functional block diagram which shows the robot simulation apparatus by Embodiment 2 of this invention. The flowchart for demonstrating operation | movement of the calculating part in the robot simulation apparatus by Embodiment 2 of this invention. The functional block diagram which shows the robot simulation apparatus by Embodiment 3 of this invention. The flowchart for demonstrating operation | movement of the calculating part in the robot simulation apparatus by Embodiment 3 of this invention. The functional block diagram which shows the robot simulation apparatus by Embodiment 4 of this invention. The functional block diagram which shows the robot simulation apparatus by Embodiment 5 of this invention.

Embodiment 1 FIG.
FIG. 1 is a functional block diagram showing a robot simulation apparatus according to Embodiment 1 of the present invention. Note that the robot here refers to, for example, a robot arm with 6 degrees of freedom for industrial use, but may be a robot arm with any degree of freedom as long as it is not 6 degrees of freedom. Further, a double-arm robot or a humanoid including a robot arm as a structure also corresponds to the robot here. The robot hand here refers to a gripping or clamping means for gripping a workpiece, and refers to a gripper connected to a flange portion at the tip of an industrial robot arm. Or a hand with a large number of fingers).

  Hereinafter, the configuration of the robot simulation apparatus according to the present embodiment will be described with reference to FIG. The robot simulation apparatus according to the present embodiment includes a storage unit 1, a calculation unit 2, a display unit 3, and an input unit 4. The display unit 3 may be omitted, but it is more convenient to prepare for confirming the simulation status and results. The calculation unit 2 includes a stacking calculation unit 21, a depth measurement simulation unit 22, a hand selection unit 23, and a control unit 24. The control unit 24 controls the overall operation of the calculation unit 2. Further, the depth measurement simulation unit 22 includes a sensor model placement unit 221 and a depth data generation unit 222, and the hand selection unit 23 includes a shape model conversion unit 231, a shape model rotation unit 232, a region extraction unit 233, and a graspability calculation. Unit 234 and evaluation selection unit 235.

  The storage unit 1 can input and output data. For example, any storage medium such as a random access memory, a hard disk drive, or a solid state drive may be used. The computing unit 2 may be realized only by hardware, or may be realized by introducing a program recorded in a recording medium such as a CD-ROM into an information processing apparatus such as a computer and executing the program. Furthermore, the input unit 4 may use any means such as a keyboard, a mouse, a touch pen, a touch pad, a gesture recognition device, a 3D (3 Dimensional) scanner that reads the shape of a component, or a camera. Multiple combinations may be used. The robot simulation apparatus of the present embodiment is configured as described above.

  Hereinafter, the operation of the robot simulation apparatus according to the present embodiment will be described with reference to FIG. Design information is input to the storage unit 1 from the input unit 4. The design information input by the input unit 4 includes a workpiece shape model that is a shape model of a workpiece to be picked up, a sensor shape model that is a shape model of a depth measurement sensor, and parameters for depth measurement in the depth measurement sensor. A first hand shape model group that is a shape model group of the robot hand, a robot shape model that is the shape model of the robot, and a surrounding environment shape model that is a shape model of the surrounding environment. Note that the shape of the surrounding environment includes the shape of the supply box that stores the workpiece, and also includes the restriction when the work space in which the robot operates has restrictions other than the supply box. The design information also includes arrangement information described later. Further, the design information may include a robot motion model.

  The design information will be described in more detail. The first hand shape model group is a set of first hand shape models that are shape models of the robot hand, and each first hand shape model has a different shape. These are candidates for the shape of the robot hand to be finally selected. The first hand shape model is associated with information for determining whether the shape model is a clamping method or a suction method, which will be described later. The workpiece shape model, sensor shape model, first hand shape model, and surrounding environment shape model have a three-dimensional shape, and also have a model inside / outside determination so that the shape models can collide with each other. It is CAD (Computer Aided Designing) data.

  The sensor parameter is a parameter necessary for reproducing the three-dimensional depth measurement by the depth measurement simulation unit 22. For example, if the depth measurement simulation unit 22 simulates an active stereo three-dimensional sensor having a camera and a light projecting unit, the focal length of the camera, the center position of the lens, the number of pixels, and the relative relationship between the camera and the light projecting unit. Includes information on various positions and postures. Note that the position and orientation represent a position and orientation. The depth sensor method simulated by the depth measurement simulation unit 22 may be a stereo view, a Time of Flight method, a view volume intersection method, a Depth from Focus method, a Depth from Defocus method, or a Shape from Shading method.

  Arrangement information indicates the relationship between the position and orientation of the robot, robot hand, supply box, and depth measurement sensor. For example, the base coordinate system of the robot, the coordinate system of the robot hand, the coordinate system of the supply box, and depth measurement Position and orientation information representing the coordinate system of the sensor and the relationship between each coordinate system is input. Further, the coordinate system of the workpiece is also input as the arrangement information. These coordinate systems are used as the coordinate system of each model when the robot, the robot hand, the supply box, and the depth measurement sensor are modeled and arranged in the virtual space. The relationship between coordinate systems is used to determine the position and orientation when each model is arranged in the virtual space. The above is a supplement regarding design information. In the present embodiment, the design information is input from the input unit 4, but may be stored in the storage unit 1 in advance. Moreover, you may comprise so that what was previously memorize | stored in the memory | storage part 1 may be corrected as needed. The design information stored in the storage unit 1 is used by the calculation unit 2.

  FIG. 2 is a flowchart for explaining the operation of the calculation unit 2. Hereinafter, the operation of the calculation unit 2 will be described with reference to FIG. 2 in addition to FIG. The operation of the calculation unit 2 is controlled by the control unit 24. First, in steps S211a to S211b, the stacking calculation unit 21 reads the workpiece shape model, the surrounding environment shape model, and the arrangement information from the storage unit 1, and uses these to indicate the state in which the workpieces are stacked in the supply box. Reproduce above. Hereinafter, an operation of the stacking calculation unit 21 in steps S211a to S211b will be described.

  In step S211a, the bulk stacking unit 21 arranges a supply box model obtained by modeling a supply box in the virtual space. Next, in step S211b, the unloading calculation unit 21 performs a simulation of physically dropping the work model obtained by modeling the work in the supply box model. The unloading calculation unit 21 arranges the work model above the supply box model so that the position and orientation are random, and drops the arranged work model onto the supply box model. By repeating this N times (N is a natural number), it is possible to reproduce a state in which N work models are stacked in a supply box model. At this time, the display unit 3 displays the reproduction of the workpiece model in a stacked state in a graphic so that the user can check whether the reproduction is appropriate. Furthermore, if reproduction is not appropriate, the user can redo the simulation after correcting the design information.

  The shape of the work model is defined by the work shape model, and the shape of the supply box model is defined by the surrounding environment shape model. At this time, a coordinate system is defined for each work model, and the position and orientation of each work model constituting the reproduced stacking is known. Furthermore, if the design information includes the robot motion model and the lifting direction of the workpiece gripped by the robot is known, it is possible to determine the number of surrounding interference objects that interfere when lifting by the workpiece interference judgment. it can. As described above, the unloading calculation unit 21 reproduces the unloading scene in a state where the work models are stacked on the virtual space. The bulk stacking operation unit 21 outputs the position and orientation of each work model in the bulk stacking scene, and the number of surrounding interference objects in each work model. The above is the operation of the calculation unit 2 in steps S211a to S211b.

  Next, in steps S221 to S222, the depth measurement simulation unit 22 arranges a sensor model obtained by modeling the depth measurement sensor in the virtual space, and simulates the depth measurement of the workpiece model stacked by the sensor model to each point. Depth data consisting of depth information is generated. The depth measurement simulation unit 22 includes a sensor model placement unit 221 and a depth data generation unit 222. Hereinafter, the operation of the depth measurement simulation unit 22 in steps S221 to S222 will be described.

  In step S221, the sensor model placement unit 221 reads the placement information and the sensor shape model from the storage unit 1, and places the sensor model in the virtual space. As a result, the position and orientation of the sensor model in the virtual space are determined. The shape of the sensor model is defined by the sensor shape model. At the same time, the sensor model placement unit 221 reads one of the first hand shape models, the robot shape model and the placement information from the storage unit 1, and a robot model that models the robot, and a hand model that models the robot hand. Is also placed in the virtual space. The shape of the robot model is defined by the robot shape model. The shape of the hand model is defined by the first hand shape model.

  Next, in step S222, the depth data generation unit 222 reads the sensor parameters from the storage unit 1, and simulates the measurement by the sensor model of the stacked workpiece models based on the read sensor parameters to obtain the depth data. get. The depth data is image data that expresses the distance from the sensor model to each point of the target work model in light and dark, acquired with the field-of-view range and resolution set by the sensor parameters. The depth data is expressed as an image that is brighter as it is closer to the sensor model and darker as it is farther away. The depth measurement simulation unit 22 operates as described above.

  At this time, the display unit 3 displays the arrangement of the robot model, the hand model, and the sensor model, and the depth data as a graphic, so that the user can check whether the arrangement and the depth data are appropriate. If the reproduction is not appropriate, the user can redo the simulation after correcting the design information. FIG. 3 is a diagram illustrating an example of display on the display unit 3 of the robot simulation apparatus according to the present embodiment. The display unit 3 includes a depth data display unit 31, a virtual space display unit 32, and a design information display unit 33. Depth data is displayed in the depth data display unit 31. The virtual space display unit 32 displays a supply box model, a robot model, a sensor model, and a hand model arranged in the virtual space. In addition, design information is displayed on the design information display unit 33.

  Next, in steps S231a to S235, the hand selection unit 23 calculates the gripping possibility based on the depth data and the hand shape model, and selects the shape of the robot hand based on the gripping possibility. The gripping possibility is an index indicating the possibility of gripping the workpiece with the robot hand, and details will be described later. The hand selection unit 23 includes a shape model conversion unit 231, a shape model rotation unit 232, a region extraction unit 233, a grip possibility calculation unit 234, and an evaluation selection unit 235. Here, before describing the operation of the hand selection unit 23, the shape of the robot hand selected by the robot simulation apparatus of the present embodiment will be described.

  Robot hands can be broadly divided into two methods, a pinching method and a suction method, and the shape of each method is different. Therefore, in the robot simulation apparatus of the present embodiment, selecting either of these two methods also selects the shape. In addition, the items that determine the shape of the holding robot hand include the number of parts (hereinafter referred to as nails) that sandwich the object of the robot hand, the position of each nail, the shape of each nail, and the size of each nail. Determination of at least one of these items is also included in the shape selection in the robot simulation apparatus of the present embodiment. On the other hand, items for determining the shape of the suction type robot hand include the number of suction pads, the position of each suction pad, the shape of each suction pad, and the size of each suction pad, and at least one of these items. Is also included in the shape selection in the robot simulation apparatus of the present embodiment.

  Furthermore, before describing the operation of the hand selection unit 23, the definition of the grippable state used for calculating the grippability is described. The grippable state is a state in which the robot hand can grip the workpiece. FIG. 4 is a diagram for explaining a grippable state in the robot hand. The gripping state in the pinch type robot hand means that the robot hand pinches the target object, there is nothing to hit the collision area (claw part) of the robot hand, and the contact area (gap between the claws) is the target object. Is in contact with. On the other hand, the grippable state in the suction type robot hand is a state in which no contact area (cross section of the suction pad) remains and is in contact with the surface of the object. Here, the shape of the robot hand can be defined by a contact area and a collision area.

  The contact area of the robot hand is a two-dimensional area obtained by projecting an area that should be in contact with the surface of the workpiece when the workpiece, which is an object, is gripped, onto a plane viewed from the workpiece side. Here, the region to be in contact with the surface of the workpiece is a region to be brought into contact when gripping the workpiece. Depending on the shape and gripping method of the workpiece, the entire contact area does not necessarily contact the workpiece. When the work is normally gripped, at least a part of the contact area of the robot hand comes into contact with the surface of the work.

  In addition, the collision area of the robot hand is an area perpendicular to the depth direction of the depth data, where the area that may collide with the workpiece or the supply box when gripping the workpiece and the contact area may interfere with the surface of the workpiece. This is a two-dimensional area projected onto a surface. An area that protrudes toward the workpiece as compared to the contact area, such as a claw of the robot hand, is a collision area. In another expression, the hand collision area is a two-dimensional area obtained by projecting, onto the vertical plane, an area where the hand contact area may collide with the work before the hand contact area touches the work. is there. Some robot hands have no hand collision area.

  FIG. 5 is a diagram for explaining a shape model of the robot hand defined by the contact area and the collision area. FIG. 5 shows that the shape of the hand can be defined by the contact area Ht and the collision area Hc even if the number of nails changes or the robot hand system changes. In the second hand shape model, the contact area and the collision area can be expressed by a combination of a circle and a rectangle. For example, in the case of a two-claw type robot hand of the clamping system shown in FIG. 5, the shape can be expressed by four parameters of the vertical and horizontal of the collision area, and the vertical and horizontal of the contact area. Therefore, the number of circles and rectangles, the position of each circle and rectangle, the radius of the circle, and the length of each side of the rectangle are the shape parameters for determining the shape of the robot hand. In the case of the suction type shown in FIG. 5, the shape can be expressed by a radius 1 parameter.

  Hereinafter, operation | movement of the hand selection part 23 in step S231a-S235 is demonstrated. In step S231a, the shape model conversion unit 231 reads one of the first hand shape models in the CAD data format from the storage unit 1. Subsequently, in step S231b, the shape model conversion unit 231 converts the read first hand shape model into a second hand shape model defined by the contact area and the collision area. The second hand shape model is two-dimensional data. The second hand shape model may be composed of two pieces of data in which the contact area and the collision area are separately provided, or may be one data obtained by combining the contact area and the collision area. Here, the contact area in the second hand shape model is referred to as a hand contact area, and the collision area in the second hand shape model is referred to as a hand collision area.

  In the present embodiment, the hand contact area and the hand collision area are represented by different data, and the second hand shape model is composed of hand contact area data and hand collision area data. The hand contact area data is image data in which the value of a pixel corresponding to the hand contact area is 1 and the value of a pixel other than the hand contact area is 0. The hand collision area data is image data in which the value of the pixel corresponding to the hand collision area is 1 and the value of the pixels other than the hand collision area is 0. Since the image data can be considered as a function of the pixel value with respect to the pixel position, the hand contact area data and the hand collision area data can also be considered as a function of the pixel value with respect to the pixel position. A second hand shape model group, which is a set of second hand shape models having different shapes, may be prepared in advance and stored in the storage unit 1 as design information. In this case, the process of step S231b is unnecessary, and the shape model conversion unit 231 is also unnecessary.

  Next, in step S232, the shape model rotation unit 232 rotates the second hand shape model in-plane by the in-plane rotation angle given from the control unit 24 to generate a third hand shape model. Here, in-plane rotation refers to rotation within a plane in which a contact area and a collision area are defined in the second hand shape model. Further, the in-plane rotation angle, which is an in-plane rotation angle, has an initial value of 0 degrees, and sequentially changes by an angle obtained by dividing 360 degrees by a preset rotation division number for each processing. For example, if the number of rotation divisions is set to 12, the in-plane rotation angle is 0 degree (that is, not rotated) in the first process, and the in-plane rotation angle is 30 degrees in the second process, and the in-plane rotation is performed in the 12th process. The angle is 330 degrees. By generating the third hand shape model, it is possible to know how much the robot hand having the same shape can be gripped when the posture of the robot hand is changed. Here, the third hand shape model is also composed of hand contact area data and hand collision area data, like the second hand shape model.

  Next, in step S233a, the region extraction unit 233 extracts regions separated by edges in the depth direction from the depth data obtained by the depth measurement simulation unit 22 by segmentation generally performed in image processing. Generate depth data. The area depth data is depth data corresponding to the extracted area. In general, the region extraction unit 233 extracts a plurality of regions, so that a plurality of region depth data are generated. FIG. 6 is a diagram illustrating an example of region depth data. The left side of the arrow in FIG. 6 is an example of depth data, and the right side of the arrow is an example of seven area depth data generated from the depth data. Subsequently, in step S233b, the region extraction unit 233 selects and outputs one of the generated region depth data.

  Next, in step S234, the gripping possibility calculation unit 234 calculates the gripping possibility for each processing condition based on the third hand shape model and the area depth data. Here, the processing condition refers to a combination of the number of times of stacking reproduction, the shape of the robot hand represented by the first or second hand shape model, the in-plane rotation angle, and the selected area depth data. First, the gripping possibility calculation unit 234 obtains the contact area and the collision area of the work model, similarly to the robot hand, based on the area depth data. Here, the concept about the contact area | region and collision area | region of a workpiece | work model is demonstrated. The contact area of the work model is an area that is a candidate for contacting the hand contact area when the robot hand grips the work, in the measurement area that is a collection of points in the depth data. Further, the work model collision area is an area in which the hand collision area may collide before the hand contact area is brought into contact with the work model contact area during gripping in the measurement area.

  The contact area of the workpiece model is an area binarized by the distance h from the sensor to the surface of the workpiece to be grasped. For each position of the area depth data, if a contact area function Wt is defined such that the position is 1 when the position is a contact area and 0 when the position is not a contact area, the position (x, y) of the area depth data is defined. Wt (x, y), which is the value of Wt in (), can be expressed by equation (1). In Expression (1), W (x, y) represents the distance from the sensor at the position (x, y).

  On the other hand, the collision area of the workpiece model is an area binarized at a height where the collision area of the robot hand has entered deeper by an amount d for further grasping from the surface of the workpiece to be grasped. For each position of the area depth data, when a collision area function Wc is defined such that the position is 1 when the position is a collision area and 0 when the position is not a collision area, the position (x, y) of the area depth data is defined. Wc (x, y), which is the value of Wc in (), can be expressed by equation (2).

  FIG. 7 is a diagram for explaining the contact area and the collision area of the work model. The diagram on the left side of FIG. 7 is a side view of the relationship between the depth measurement sensor, the workpiece, and the robot hand in order to explain the contact area and the collision area of the workpiece model. In the figure on the left side of FIG. 7, the binarized area in the cross section A is the contact area, and the binarized area in the cross section B is the collision area. The diagram on the right side of FIG. 7 illustrates the contact area and the collision area corresponding to the diagram on the left side.

  In the robot simulation apparatus according to the present embodiment, gripping possibility calculation unit 234 uses an area extracted by area depth data as a contact area of a workpiece. In the area depth data, an area divided by the edge of the depth is extracted from the depth data, and the surface of each work stacked is extracted, so that it can be used as a contact area of the work. The collision area of the workpiece is obtained by binarizing the depth data at a height that has entered deeper by d from the depth of the area extracted from the area depth data. When the depth of the region extracted from the region depth data is not constant within the region, the depth at the position farthest from the sensor is used.

  The gripping possibility calculation unit 234 generates work model contact area data representing the contact area of the work model and work model collision area data representing the collision area of the work model. The work model contact area data is image data in which the value of a pixel corresponding to the contact area is 1 and the value of a pixel other than the contact area is 0. The work model collision area data is image data in which the value of a pixel corresponding to the collision area is 1 and the value of a pixel other than the collision area is 0. The work model contact area data and the work model collision area data can be considered as a function of the pixel value with respect to the pixel position.

  Next, the gripping possibility calculation unit 234 obtains a grippable area using the contact area and collision area of the third hand shape model and the contact area and collision area of the work model. The grippable area represents an area where the robot hand and the workpiece can contact without interfering with each other. Subsequently, the gripping possibility calculation unit 234 obtains a gripping possibility map using the grippable area. FIG. 8 is a diagram for explaining a method for obtaining a grippable region and a grippability map. In FIG. 8, the sandwiching type robot hand is described as an example, but the grippable region can be similarly calculated by the suction type robot hand. In the case of the suction method, the calculation shown in FIG. 8 is performed in the same manner, except that the hand model used is different and the collision area Hc is all zero.

  The gripping possibility calculation unit 234 obtains the first area by convolution (convolution calculation) using the hand contact area data of the third hand shape model and the work model contact area data. For a sandwich-type robot hand, the area where the value obtained by convolution is 1 or more is set as the first area. On the other hand, for the suction-type robot hand, a region where the value obtained by convolution is equal to or greater than St is set as the first region. Here, St is the area of the contact area of the third hand shape model.

  Further, the gripping possibility calculation unit 234 obtains the second area by convolution using the hand collision area data of the third hand shape model and the work model collision area data. For the holding type robot hand, the region where the value obtained by convolution is 1 or more is set as the second region. On the other hand, for the suction type robot hand, as shown in the suction type model example of FIG. 5, there is no collision area, and the hand collision area data is 0 at all positions. It becomes 0 at all positions, and there is no second area. Next, the gripping possibility calculation unit 234 subtracts the value of each pixel in the corresponding second region from the value of each pixel in the first region, and sets the portion where the value becomes 1 as the grippable region. Thus, grippable area data is generated. The grippable area data is image data in which the value of a pixel corresponding to the grippable area is 1 and the value of a pixel other than the grippable area is 0. The grippable area data can be considered as a function of the pixel value with respect to the pixel position.

  Subsequently, the graspability calculation unit 234 obtains a graspability map by convolution using the obtained graspable region data and a Gaussian function generally used in image processing. The grip possibility map represents the grip possibility for each position. The gripping possibility map has a distribution in which the value increases toward the center of the grippable area. By obtaining the peak position of the gripping possibility map, the gripping position where the robot hand does not interfere with the surroundings and the object can be easily gripped can be calculated at high speed. The gripping possibility calculation unit 234 sets the maximum gripping possibility value for each position in the gripping possibility map as the gripping possibility for each processing condition. In the calculation of the graspability map, if the Gaussian function is too small, the distribution is not sufficiently smoothed and multiple maximum values are obtained, so the Gaussian function has the same size as the graspable area. By doing so, the maximum value can be calculated stably. The above is the operation of the gripping possibility calculation unit 234 in step S234.

  In step S <b> 901, the control unit 24 determines whether all the region depth data generated by the region extraction unit 233 has been selected. If there is area depth data that has not yet been selected, the control unit 24 returns the process to step S233b. In step S233b, the region extraction unit 233 selects one region depth data that has not yet been selected. On the other hand, if all the area depth data have been selected, the control unit 24 advances the processing to step S902.

  In step S902, the control unit 24 compares the number of processings in this step with a preset number of rotation divisions. If the number of processes in this step has not reached the number of rotation divisions, the control unit 24 adds the number of processes once, changes the in-plane rotation angle, and returns the process to step S232. On the other hand, if the number of processes in this step has reached the number of rotation divisions, the control unit 24 resets the number of processes and the in-plane rotation angle, and proceeds to step S903.

  In step S903, the control unit 24 determines whether all the first hand shape models stored in the storage unit 1 have been read out. If there is a first hand shape model that has not yet been read, the control unit 24 returns the process to step S231a. In step S231a, the shape model conversion unit 231 reads from the storage unit 1 one of the first hand shape models that has not been read yet. On the other hand, if all the first hand shape models have been read, the control unit 24 advances the processing to step S904.

  In step S904, the control unit 24 compares the number of processes in this step with a preset number of loose stacks. If the number of times of processing in this step has not reached the number of times of stacking, the control unit 24 adds the number of times of processing once, and returns the processing to step S211b. On the other hand, if the number of times of processing in this step has reached the number of times of stacking, the control unit 24 resets the number of times of processing and advances the process to step S235. The number of times of stacking may be one or more, but in general, the accuracy of selecting the shape of the robot hand is improved by setting the number of times of stacking multiple times. Moreover, it is also possible to change the kind of a workpiece | work with respect to multiple times of bulk loading. Select the optimal robot hand shape when handling multiple types of workpieces by reproducing the stacking of multiple types of workpieces and calculating the gripping possibility for these workpieces to determine the shape of the robot hand be able to. In the present embodiment, the number of loose stacks is a plurality of times.

  When the process proceeds to step S235, gripping possibility for each processing condition is required for the multiple stacking reproduction, various first hand shape models, in-plane rotation angle, and area depth data. . In step S235, the evaluation selecting unit 235 first determines the maximum possibility of gripping for each processing condition with respect to various in-plane rotation angles and region depth data for the same first stack shape reproduction and the same first hand shape model. Choose one. As a result, for each first hand shape model, the gripping possibility for each stacking reproduction is required. Subsequently, the evaluation selection unit 235 calculates an average value of the gripping possibility for each stacking reproduction for each first hand shape model. As a result, the gripping possibility for each hand shape with respect to each first hand shape model is required.

  Next, the evaluation selection unit 235 compares the gripping possibility for each hand shape, selects the first hand shape model having the highest gripping possibility, and outputs it together with the gripping possibility for each corresponding hand shape. The output destination may be the display unit 3 or the storage unit 1. Moreover, you may provide another output means. From the output result, the operator can obtain a desired hand design. If the output destination is the storage unit 1, a shape selected by another tool or apparatus such as software for designing a bulk pickup system can be called, and design time can be shortened.

The process for selecting the shape of the robot hand described above can be expressed by Expression (3). In Equation (3), G _i is a function that represents the gripping possibility for each position, and i is an index that represents how many times the stacking is reproduced. N represents the number of times of performing stacking reproduction. X, y, and d are parameters representing the position of the hand. x and y correspond to positions in the area depth data, and d corresponds to which area depth data. θ is a parameter representing the posture of the hand and corresponds to the in-plane rotation angle. M is an index for identifying the first hand shape model. Equation (3) means that the gripping possibility for each hand shape is obtained by averaging the gripping possibilities for each reproduction of bulk stacks, and the first hand shape model that maximizes the gripping possibility for each hand shape is obtained. To do. In Expression (3), the processing for obtaining the gripping possibility for each processing condition from the gripping possibility for each position and the processing for obtaining the gripping possibility for each stacking reproduction from the gripping possibility for each processing condition are one. It is expressed as a function.

  In the robot simulation apparatus according to the present embodiment, in step S234, the gripping possibility calculation unit 234 sets the maximum gripping possibility value for each position in the gripping possibility map as the gripping possibility for each processing condition. Although configured, other configurations are possible. For example, instead of the maximum value of the gripping possibility for each position, an integrated value of the gripping possibility for each position can be set as the gripping possibility for each processing condition. It is also possible to use the maximum value and the integral value together. In this case, for example, the maximum value and the integral value may be weighted and added. When the maximum value of the gripping possibility for each position is close to a plurality of processing conditions, it is effective to use the maximum value of the gripping possibility for each position and the integral value together.

  In addition, the gripping possibility calculation unit 234 can have another configuration. For example, the gripping possibility calculation unit 234 can more accurately calculate the gripping possibility by considering the stability when gripping. If the work model has the position information of the center of gravity, the position of the center of gravity of the work model in the state of being stacked can be calculated from the position and orientation of the work model obtained by the stacking calculation unit 21. The further the gripping position is away from the center of gravity position of the work model, the higher the risk of gripping failure, such as a moment being generated after gripping, causing it to fly or drop.

Therefore, a more stable design can be achieved by adding the distance from the center of gravity of the work model to the calculation of the gripping possibility. By calculating the gripping possibility for each position in the gripping possibility map for each processing condition by calculating the distance from the center of gravity of the workpiece model at each position, the gripping possibility is more accurate. Is required. For example, the gripping possibility for each position can be divided by the distance from the center of gravity of the work model, and the maximum value of the division result can be made the gripping possibility for each processing condition. At this time, the process of selecting the shape of the robot hand can be expressed by Expression (4). In Expression (4), L _i is a function representing the distance from the center of gravity of the work model, and the others are the same as in Expression (3).

Further, the gripping possibility calculation means 234 can be further configured. For example, the gripping possibility calculation means 234 can more accurately calculate the gripping possibility by taking into account the interference state with the surrounding workpiece model. If it interferes with surrounding workpieces when lifting, it becomes difficult to lift after gripping. Therefore, a more stable design can be achieved by adding the number of peripheral interferences obtained by the bulk stacking calculation unit 21 to the calculation of the gripping possibility. By calculating the gripping possibility for each position in the gripping possibility map for each processing condition by performing calculation in consideration of the number of surrounding interference objects, more accurate gripping possibility is obtained. For example, the gripping possibility for each position can be divided by the number of surrounding interference objects, and the maximum value of the division result can be made the gripping possibility for each processing condition. At this time, the process of selecting the shape of the robot hand can be expressed by Expression (5). In Equation (5), C _i is a function representing the number of surrounding interferers, and the others are the same as in Equation (3).

  As described above, the gripping possibility calculation unit 234 can have various configurations, and can also be configured by combining them. A configuration that is optimal for the use conditions may be selected in consideration of the amount of calculation and the calculated graspability characteristics. Furthermore, the gripping possibility calculation method described in the present embodiment can also be used when obtaining the gripping position of the robot hand in an actual picking operation. In this case, the robot hand may be controlled using the position and in-plane rotation angle with the highest gripping possibility as the gripping position and gripping posture.

  The robot simulation apparatus according to the present embodiment operates as described above. Therefore, it is possible to select the shape of the robot hand that is highly likely to be gripped and that is unlikely to cause gripping failure. That is, it is possible to select the shape of the robot hand having a high gripping success rate. Since the bulk loading and unloading system is used to increase the production efficiency at the production site, it is important how the working time can be shortened. Conventional robot simulation devices only estimate work time from given robots, sensors, and workpieces, and can actively change the layout, configuration, shape of system components, etc. to reduce work time. could not.

  Among the components of the system, there is a robot hand for holding a workpiece, particularly related to work time. If a robot hand that is easy to grasp is selected according to the shape of the workpiece and the success rate of grasping the workpiece is increased, the tact time required for taking out is dramatically improved. However, in the conventional robot simulation apparatus, the robot hand only handles a predetermined shape, and there is no robot simulation apparatus that selects a robot hand that increases the success rate of gripping.

  Furthermore, system integrators who are in the position of designing systems to determine the robot hand using the actual product often have no knowledge of the relationship between the robot hand and the workpiece during unloading, and only the devices on the site. The adjustment of the arrangement using the or the robot hand already used. According to the robot simulation apparatus of the present embodiment, the design of the robot hand can be determined before the system construction, so that the design of the system for picking up bulk workpieces becomes efficient. Further, in the actual system operation, a robot hand having a high work success rate is selected, which has the effect of shortening the actual work time.

  As described above, the robot simulation apparatus according to the present embodiment is a sensor that models a stacking calculation unit that reproduces in a virtual space a work model obtained by modeling a work, and a depth measurement sensor. The model is placed in a virtual space, the depth measurement simulation unit that generates depth data consisting of depth information at each point by simulating the depth measurement of each point of the work model with the sensor model, and the shape of multiple robot hands A hand that calculates the gripping possibility of a work model in a robot hand based on the converted hand shape model and depth data for the shapes of a plurality of robot hands, and selects the shape of the robot hand based on the gripping possibility Since the selection part is provided, the shape of the robot hand that increases the gripping possibility is measured in the virtual space. It can be selected by. Since a robot hand having a high gripping possibility is selected, the tact time required for taking out can be shortened.

Embodiment 2. FIG.
In the first embodiment, the shape of the robot hand is selected from the shape model of the robot hand included in the design information. On the other hand, a shape parameter that can be changed may be associated with the shape model of the robot hand so that the robot simulation apparatus adjusts the shape parameter. As a result, while reducing the shape of the robot hand included in the design information, it is possible to select the shape of the robot hand that increases the gripping possibility from various shapes.

  FIG. 9 is a functional block diagram showing a robot simulation apparatus according to the second embodiment of the present invention. The robot simulation apparatus according to the present embodiment is different from that shown in FIG. 1 of the first embodiment in that the shape model adjustment unit 236 is provided in the hand selection unit 23. Hereinafter, the operation of the robot simulation apparatus according to the present embodiment will be described with reference to FIG. As in the first embodiment, design information is input to the storage unit 1 from the input unit 4. However, the first hand shape model included in the design information is associated with a shape parameter that can be changed, a changeable range, and a changeable step size for the shape model. The operations of the storage unit 1, the display unit 3, and the input unit 4 are the same as those in the first embodiment, and a description thereof will be omitted.

  FIG. 10 is a flowchart for explaining the operation of the calculation unit 2 in the robot simulation apparatus of the present embodiment. This flowchart is different from that in FIG. 2 of Embodiment 1 in that step 236 is provided between step S231b and step S232, and step S905 is provided between step S902 and step S903. Hereinafter, the operation of the calculation unit 2 will be described with reference to FIGS. 9 and 10. 9 and 10, the same reference numerals as those in FIGS. 1 and 2 are the same, and the description thereof is omitted.

  In step S236, the shape model adjustment unit 236 uses the changeable shape parameter associated with the first hand model to adjust the shape of the second hand shape model to generate a fourth hand shape model. To do. As described in the first embodiment, in the second hand shape model, the shape of the hand can be defined in the contact area and the collision area even if the number of claws changes or the robot hand system changes. In the second hand shape model, the contact area and the collision area can be expressed by a combination of a circle and a rectangle. For example, in the case of a two-claw type robot hand of a pinching method, the vertical, horizontal, The shape can be expressed by the vertical and horizontal parameters of the contact area. In the case of a suction-type robot hand, the shape can be expressed with a radius 1 parameter. Therefore, the number of circles and rectangles, the position of each circle and rectangle, the radius of the circle, and the length of each side of the rectangle are the shape parameters for determining the shape of the robot hand.

  In step S236, the shape model adjusting unit 236 adjusts the shape of the second hand shape model by adjusting the shape parameters that can be changed within the changeable range among these shape parameters, and the fourth hand. Generate a shape model. For example, if the first hand shape model is a pinching two-claw type and the shape parameters that can be changed are four parameters of the vertical and horizontal of the collision area, the vertical and horizontal of the contact area, the shape model adjustment unit 236 adjusts these four parameters. Next, in step S232, the shape model rotation unit 232 generates the third hand shape model by rotating the fourth hand shape model in-plane. Thereafter, the operations up to step S902 are the same as those in the first embodiment.

  In step S905, the control unit 24 determines whether the gripping possibility for each processing condition has been calculated for all combinations of shape parameters that can be changed. If there is a combination for which the gripping possibility for each processing condition has not yet been calculated, the control unit 24 returns the process to step S236. In step S236, the shape model adjustment unit 236 adjusts the shape parameter to one of the combinations for which the gripping possibility for each processing condition has not yet been calculated, and generates a fourth hand shape model. On the other hand, if the gripping possibility for each processing condition is calculated for all combinations of changeable shape parameters, the control unit 24 advances the process to step S903. Subsequent operations are the same as those in the first embodiment. The robot simulation apparatus according to the present embodiment operates as described above. In addition to the processing conditions described in the first embodiment, a combination of shape parameters is added to the processing conditions in the present embodiment.

  According to the robot simulation apparatus of the present embodiment, the number of first hand shape models included in the design information can be reduced. As a result, the number of first hand shape models prepared in advance can be reduced, and preparation work is reduced. On the other hand, the possibility of gripping various shapes can be simulated for one first hand shape model, and the possibility of selecting an optimally shaped robot hand increases. For example, for a two-claw type robot hand of a clamping method, it is only necessary to prepare one first hand shape model having a representative shape as design information. The gripping possibility for various shapes of the two-claw type of the clamping method can be calculated by variously changing the four shape parameters of the width, the length of the contact area, and the width. Further, the robot simulation apparatus according to the present embodiment also has the same effect as that in the first embodiment.

Embodiment 3 FIG.
In the first embodiment, the shape of the robot hand is selected from the shape model of the robot hand included in the design information. On the other hand, the robot simulation apparatus can be configured to generate the shape of the robot hand based on the shape parameter. As a result, the shape information only needs to be included in the design information, and it is not necessary to include the shape model of the robot hand.

  FIG. 11 is a functional block diagram showing a robot simulation apparatus according to the third embodiment of the present invention. The robot simulation apparatus according to the present embodiment differs from that shown in FIG. 1 according to the first embodiment in that a shape model generation unit 237 is provided in the hand selection unit 23 instead of the shape model conversion unit 231. Hereinafter, the operation of the robot simulation apparatus according to the present embodiment will be described with reference to FIG. As in the first embodiment, design information is input to the storage unit 1 from the input unit 4. However, the design information does not include the first hand shape model, but includes a shape parameter, a changeable range of each parameter, and a changeable step size. The operations of the storage unit 1, the display unit 3, and the input unit 4 are the same as those in the first embodiment, and a description thereof will be omitted.

  FIG. 12 is a flowchart for explaining the operation of the calculation unit 2 in the robot simulation apparatus of the present embodiment. This flowchart is different from that in FIG. 2 of the first embodiment in that step S237 is provided instead of steps S231a and S231b, and step S906 is provided instead of step S903. Hereinafter, the operation of the calculation unit 2 will be described with reference to FIGS. 11 and 12. 11 and 12, the same reference numerals as those in FIGS. 1 and 2 are the same, and the description thereof is omitted.

  In step S237, the shape model generation unit 237 sets shape parameters included in the design model, and generates second hand shape models having various shapes. As described in the first embodiment and the second embodiment, in the second hand shape model, the shape of the robot hand can be expressed using the shape parameter. In step S237, the shape model generation unit 237 generates second hand shape models of various shapes by adjusting the shape parameters within a changeable range. Next, in step S232, the shape model rotation unit 232 generates a third hand shape model by rotating the second hand shape model in-plane. Thereafter, the operations up to step S902 are the same as those in the first embodiment.

  In step S906, the control unit 24 determines whether the gripping possibility for each processing condition has been calculated for all combinations of shape parameters. If there is a combination for which the gripping possibility for each processing condition has not yet been calculated, the control unit 24 returns the process to step S237. In step S237, the shape model generation unit 237 adjusts the shape parameter to one of the combinations for which the gripping possibility for each processing condition has not yet been calculated, and generates the second hand shape model. On the other hand, if the gripping possibility for each processing condition is calculated for all combinations of the shape parameters, the control unit 24 advances the processing to step S904. Subsequent operations are the same as those in the first embodiment. The robot simulation apparatus according to the present embodiment operates as described above. At this time, the process of selecting the shape of the robot hand can be expressed by Expression (6). In Expression (6), P1,..., Ph represent shape parameters, and h represents the maximum number of shape parameters. Others are the same as Formula (3).

  If there are restrictions on the shape of the robot hand because there are shapes that are difficult to obtain, and there are some shape parameters that you do not want to change, changeable shape parameters and non-changeable shape parameters. It is desirable to configure so that it can be set. Further, in the robot simulation apparatus of the present embodiment, as a technique for obtaining a combination of shape parameters that increase the gripping possibility, the simplest technique for obtaining the gripping possibility for all combinations is illustrated. It is not limited to this. The same applies to the other embodiments.

  According to the robot simulation apparatus of the present embodiment, it is not necessary to include the first hand shape model in the design information. As a result, it is not necessary to prepare the first hand shape model in advance, and preparation work is reduced. On the other hand, it becomes possible to simulate the gripping possibility of various shapes of robot hands, and the possibility of selecting an optimally shaped robot hand is further increased. Further, the robot simulation apparatus according to the present embodiment also has the same effect as that in the first embodiment.

Embodiment 4 FIG.
The robot simulation apparatus according to the first embodiment is configured to change the control parameter of the robot hand when the gripping possibility corresponding to the shape of the selected robot hand is smaller than a preset threshold value. It is also possible. As a result, it is possible to realize a design with an increased gripping success rate in actual system operation.

  FIG. 13 is a functional block diagram showing a robot simulation apparatus according to the fourth embodiment of the present invention. The robot simulation apparatus according to the present embodiment is different from that shown in FIG. 1 according to the first embodiment in that the calculation unit 2 includes a control parameter changing unit 25. In FIG. 13, the same reference numerals as those in FIG. 1 are the same, and the description thereof is omitted. Hereinafter, the operation of the calculation unit 2 will be described with reference to FIG. Since the operation flow of the calculation unit 2 is only the operation of the control parameter changing unit 25 added to the end of the first embodiment, the flow diagram is omitted.

  In the calculation unit 2, the processing until the hand selection unit 23 calculates the gripping possibility for each hand shape and selects the shape of the robot hand based on this is the same as that in the first embodiment. That is, the operations up to step S235 in FIG. 2 are the same in the robot simulation apparatus of the present embodiment. In the robot simulation apparatus of the present embodiment, the operation of the control parameter changing unit 25 described below is added after the operation of step S235 in FIG. The control parameter changing unit 25 compares the gripping possibility for the selected hand shape with a preset threshold value. If the gripping possibility is equal to or higher than the threshold value, the control parameter changing unit 25 ends the operation without changing the control parameter. On the other hand, if the gripping possibility is less than the threshold value, the control parameter changing unit 25 changes the control parameter.

  Here, the control parameter is a parameter related to the operation of the robot hand when gripping the workpiece, such as the operation speed or stop time of the robot hand, and is included in the design information. When the gripping possibility is less than the threshold, the control parameter changing unit 25 sets the control parameter so as to decrease the approach speed of the robot hand during gripping or the robot operation speed after gripping, or lengthen the stop time during gripping. change. According to the robot simulation apparatus of the present embodiment, it is possible to realize a design that further increases the gripping success rate by changing the control parameter according to the gripping possibility of the selected robot hand. Further, the robot simulation apparatus according to the present embodiment also has the same effect as that in the first embodiment.

Embodiment 5. FIG.
The robot simulation apparatus according to the first embodiment is configured to select the shape of the robot hand based on the gripping possibility for each hand shape. In addition to the possibility of gripping each hand shape, it is also possible to simulate the work of unloading workpieces by the robot hand and select the shape of the robot hand taking into account the estimated tact time is there. Thereby, the possibility that the shape of the robot hand that can shorten the tact time is selected can be further increased.

  FIG. 14 is a functional block diagram showing a robot simulation apparatus according to the fifth embodiment of the present invention. The robot simulation apparatus according to the present embodiment differs from that in FIG. 1 according to the first embodiment in that the arithmetic unit 2 includes a tact time calculation unit 26. In FIG. 14, the same reference numerals as those in FIG. 1 are the same, and the description thereof is omitted. Hereinafter, the operation of the calculation unit 2 will be described with reference to FIG.

  In the calculation unit 2, the processing until the hand selection unit 23b calculates the gripping possibility for each hand shape and selects the shape of the robot hand based on this is the same as that in the first embodiment. However, unlike in the first embodiment, the evaluation selection unit 235b included in the hand selection unit 23b selects a plurality of hand shapes in descending order of the gripping possibility for each hand shape. That is, as compared with the flowchart shown in FIG. 2, the operation of the robot simulation apparatus according to the present embodiment is based on the fact that a plurality of hand shapes are selected in step S235 and the tact time calculation unit after the operation in step S235. The difference is that 26 operations have been added.

  The tact time calculation unit 26 simulates the work of unloading workpieces using the shapes of the plurality of robot hands selected by the hand selection unit 23b, and estimates the tact time for each robot hand. Subsequently, the tact time calculation unit 26 converts the estimated tact time into a score. This score becomes higher as the tact time is shorter. Furthermore, the tact time calculation unit 26 weights and adds the gripping possibility for each hand shape and the score for the corresponding hand shape, and selects the hand shape having the highest value. The robot simulation apparatus according to the present embodiment operates as described above. According to the robot simulation apparatus of the present embodiment, it is possible to further increase the possibility of selecting the shape of the robot hand that can shorten the tact time. Further, the robot simulation apparatus according to the present embodiment also has the same effect as that in the first embodiment.

  DESCRIPTION OF SYMBOLS 1 Memory | storage part, 2 calculating part, 3 Display part, 4 Input part, 21 Stacking calculating part, 22 Depth measurement simulation part, 23, 23b Hand selection part, 24 Control part, 25 Control parameter change part, 26 Tact time calculation part 221 sensor model placement unit, 222 depth data generation unit, 231 shape model conversion unit, 232 shape model rotation unit, 233 region extraction unit, 234 graspability calculation unit, 235, 235b evaluation selection unit, 236 shape model adjustment unit, 237 Shape model generation part, 31 Depth data display part, 32 Virtual space display part, 33 Design information display part.

Claims (12)

A bulk calculation unit that reproduces in a virtual space a state in which work models obtained by modeling a work are stacked;
Depth measurement in which a sensor model obtained by modeling a depth measurement sensor is arranged in the virtual space, and depth data including depth information at each point is generated by simulating depth measurement at each point of the work model using the sensor model. A simulation section,
Based on the hand shape model obtained by modeling the shapes of a plurality of robot hands and the depth data, the gripping possibility of the work model in the robot hands is calculated for the shapes of the plurality of robot hands, and the grip A hand selection unit that selects the shape of the robot hand based on the possibility ,
The hand shape model is
A hand contact area that is a two-dimensional area obtained by projecting an area to be in contact with the work when gripping the work onto a plane perpendicular to the depth direction of the depth data;
When there is a region that may collide with the workpiece before the hand contact region contacts the workpiece when gripping the workpiece, the region that may collide with the vertical surface. It is defined by a hand collision area that is a projected two-dimensional area,
The hand selection unit
A work model contact area that is a candidate area for contacting the hand contact area when the robot hand grips the work out of a measurement area that is a collection of points in the depth data;
A workpiece model collision area that is a region where the hand collision area may collide before the hand contact area is brought into contact with the workpiece model contact area of the measurement area;
Is obtained based on the depth data,
The robot simulation apparatus characterized by calculating a gripping possibility based on the hand contact area, the hand collision area, the work model contact area, and the work model collision area . The workpiece model has position information of the center of gravity of the workpiece,
The hand selection unit, the robot simulation apparatus according to claim 1, characterized in that calculating said gripping possibilities by using the position information of the center of gravity. The hand selection unit
Robot simulation apparatus according to claim 1, characterized in that calculating said gripping possibilities by using the interference the number of the peripheral when the robot hand is lifted by gripping the workpiece model. The hand selection unit
Robot simulation apparatus according to any one of claims 1 to 3, characterized in that the gripping possibility to select the shape of the highest the robot hand. A bulk calculation unit that reproduces in a virtual space a state in which work models obtained by modeling a work are stacked;
Depth measurement in which a sensor model obtained by modeling a depth measurement sensor is arranged in the virtual space, and depth data including depth information at each point is generated by simulating depth measurement at each point of the work model using the sensor model. A simulation section,
Based on the hand shape model obtained by modeling the shapes of a plurality of robot hands and the depth data, the gripping possibility of the work model in the robot hands is calculated for the shapes of the plurality of robot hands, and the grip A hand selection unit that selects the shape of the robot hand based on the possibility ,
The hand selection unit
The robot simulation apparatus , wherein the shape of the robot hand having the highest average value of the gripping possibilities obtained for a plurality of types of the work models is selected . A bulk calculation unit that reproduces in a virtual space a state in which work models obtained by modeling a work are stacked;
Depth measurement in which a sensor model obtained by modeling a depth measurement sensor is arranged in the virtual space, and depth data including depth information at each point is generated by simulating depth measurement at each point of the work model using the sensor model. A simulation section,
Based on the hand shape model obtained by modeling the shapes of a plurality of robot hands and the depth data, the gripping possibility of the work model in the robot hands is calculated for the shapes of the plurality of robot hands, and the grip A hand selection unit that selects the shape of the robot hand based on the possibility ,
The unloading unit reproduces a state in which the work models are stacked a plurality of times,
The hand selection unit selects the shape of the robot hand having the highest average value of the gripping possibility obtained for a state where the work models reproduced a plurality of times are stacked. A robot simulation device. The said form of a plurality of the robot hand, the robot simulation apparatus according to any one of claims 6 to include different the shapes of the gripping system from claim 1, wherein. The robot simulation apparatus according to claim 7 , wherein the gripping method includes a pinching method and a suction method. Based on the grip may correspond to the shape selected by the hand selection unit, according claim 1, characterized in that it comprises the control parameter changing unit for changing a control parameter for controlling the operation of the robot hand Item 9. The robot simulation apparatus according to any one of items 8 to 9 . A bulk stacking unit that reproduces in a virtual space a state in which work models obtained by modeling information processing devices are stacked;
Depth measurement in which a sensor model obtained by modeling a depth measurement sensor is arranged in the virtual space, and depth data including depth information at each point is generated by simulating depth measurement at each point of the work model using the sensor model. A simulation section,
Based on the hand shape model obtained by modeling the shapes of a plurality of robot hands and the depth data, the gripping possibility of the work model in the robot hands is calculated for the shapes of the plurality of robot hands, and the grip Function as a hand selection unit that selects the shape of the robot hand based on the possibility ,
The hand shape model is
A hand contact area that is a two-dimensional area obtained by projecting an area to be in contact with the work when gripping the work onto a plane perpendicular to the depth direction of the depth data;
When there is a region that may collide with the workpiece before the hand contact region contacts the workpiece when gripping the workpiece, the region that may collide with the vertical surface. It is defined by a hand collision area that is a projected two-dimensional area,
The hand selection unit
A work model contact area that is a candidate area for contacting the hand contact area when the robot hand grips the work out of a measurement area that is a collection of points in the depth data;
A workpiece model collision area that is a region where the hand collision area may collide before the hand contact area is brought into contact with the workpiece model contact area of the measurement area;
Is obtained based on the depth data,
Grasping possibility is calculated based on the hand contact area, the hand collision area, the work model contact area, and the work model collision area.
A robot simulation program characterized by this . A recording medium on which the robot simulation program according to claim 10 is recorded. A step in which the bulk calculation unit reproduces in a virtual space a state in which workpiece models obtained by modeling workpieces are stacked;
A depth measurement simulation unit arranges a sensor model obtained by modeling a depth measurement sensor in the virtual space, simulates depth measurement of each point of the work model by the sensor model, and includes depth information at each point. Generating data; and
Based on the hand shape model obtained by modeling the shape of a plurality of robot hands and the depth data, the hand selection unit determines the gripping possibility of the work model in the robot hand with respect to the shapes of the plurality of robot hands. Calculating and selecting the shape of the robot hand based on the gripping possibility , and
The hand shape model is
A hand contact area that is a two-dimensional area obtained by projecting an area to be in contact with the work when gripping the work onto a plane perpendicular to the depth direction of the depth data;
When there is a region that may collide with the workpiece before the hand contact region contacts the workpiece when gripping the workpiece, the region that may collide with the vertical surface. It is defined by a hand collision area that is a projected two-dimensional area,
The step of selecting the shape of the robot hand includes
A work model contact area that is a candidate area for contacting the hand contact area when the robot hand grips the work out of a measurement area that is a collection of points in the depth data;
A workpiece model collision area that is a region where the hand collision area may collide before the hand contact area is brought into contact with the workpiece model contact area of the measurement area;
Obtaining based on the depth data;
A robot simulation method comprising: calculating a gripping possibility based on the hand contact area, the hand collision area, the work model contact area, and the work model collision area .