JP2015104796A - Gripping method, transportation method and robot - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a gripping method capable of precisely gripping a body as an object to be gripped without moving an obstacle even when a part of the body as the object to be gripped is hidden from sight.SOLUTION: A gripping method recognizes and grips an object to be gripped by means of a gripping device even when a part of the object to be gripped is hidden from sight by an obstacle. The gripping method includes a step of gaining environmental information of the surrounding which contains the object to be gripped, a step of recognizing the object to be gripped from the surrounding environment by means of the gained environmental information, a step of determining that, when a part of the object to be gripped is recognized, the recognized part is the object to be gripped and a step of gripping the determined gripping region. The step of determining the gripping region determines the region recognized as one part of the object to be gripped in the step of recognizing the object to be gripped, as the gripping region.

Description

  The present invention relates to a gripping method, a transporting method, and a robot, and more particularly to a technique for recognizing and gripping an object.

  Patent Document 1 discloses a robot that can recognize and grasp a target by image recognition even when the target to be grasped is hidden behind an obstacle. When it is determined that an obstacle is present based on an image photographed by the camera, the robot moves the obstacle with the robot arm and performs a target re-recognition process. When the robot recognizes the target, the robot performs an operation of gripping the target.

JP 2007-290056 A

  However, the robot disclosed in Patent Document 1 has a problem in that it cannot grasp the target object without moving the obstacle and is not efficient. In addition, there may be a case where it is not desired to arbitrarily change the position of an obstacle.

  The present invention has been made to solve the above-described problems. Even when a part of an object to be grasped is hidden by an obstacle, the object can be grasped without moving the obstacle. It is an object of the present invention to provide a gripping method, a transporting method, and a robot that can accurately grip a target object.

  The gripping method according to the first aspect of the present invention acquires environment information that is three-dimensional information acquired by a gripping device, recognizes a gripping object from the acquired environment information, and recognizes the recognized gripping object. A gripping method for recognizing and gripping the gripping object by a gripping device even when part of the object is hidden by an obstacle, and acquiring surrounding environment information including the gripping object; A step of recognizing the gripping object from the surrounding environment based on the acquired environment information; a step of determining a gripping object when a part of the gripping object is recognized; and Determining the gripping area of the object determined to be, and gripping the determined gripping area, wherein the step of determining the gripping area is performed before the step of recognizing the gripping object. The recognized region as part of a grasped object is to determine as the gripping region.

  A transport method according to a second aspect of the present invention is a transport method for transporting a gripping object gripped by the gripping method according to the above-described aspect to a transport destination, and is recognized by the step of recognizing the entire gripping target object. The entire gripping object is provided with a step of determining a trajectory that avoids the obstacle and transporting the gripping object in the trajectory.

  The robot according to the third aspect of the present invention acquires environment information that is three-dimensional information, recognizes a gripping object from the acquired environment information, and a part of the recognized gripping object is an obstacle. An environment information acquisition unit that recognizes the object to be grasped and grasps it with a robot arm even if it is hidden, and obtains surrounding environment information including the object to be grasped, and the obtained environment Based on the information, when a part of the gripping object is recognized from the surrounding environment, an object recognition unit that determines that the object is a gripping object and a gripping area of the object that is determined to be the gripping object are determined. A gripping region determining unit, wherein the gripping region determining unit determines a region recognized as a part of the gripping object as the gripping region.

  When a part of the object is hidden by the obstacle, the part that is not hidden by the obstacle is recognized. Therefore, as in each aspect of the present invention described above, by determining an area that is recognized as a part of the object to be grasped as a grasping area, it is possible to grasp an area that is not hidden by an obstacle. It is possible to avoid contact between the gripping device (robot) and the obstacle when gripping an object.

  According to each aspect of the present invention described above, even when a part of an object to be grasped is hidden by an obstacle, the object to be grasped is accurately grasped without moving the obstacle. A gripping method, a transporting method, and a robot can be provided.

2 is an external configuration diagram of the robot according to Embodiment 1. FIG. 2 is an internal configuration diagram of a robot according to Embodiment 1. FIG. FIG. 3 is a process block diagram of the robot according to the first embodiment. 3 is a flowchart showing processing of the robot according to the first embodiment. It is a figure which shows an example of the object used as the holding object. FIG. 10 is a process block diagram of the robot according to the second embodiment. 6 is a flowchart showing processing of the robot according to the second embodiment. It is a figure which shows an example of an obstacle avoidance. FIG. 10 is a process block diagram of the robot according to the third embodiment. 10 is a flowchart illustrating processing of the robot according to the third embodiment. It is a figure which shows an example of arrangement | positioning to the target point of an object.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. Specific numerical values and the like shown in the following embodiments are merely examples for facilitating understanding of the invention, and are not limited thereto unless otherwise specified. In the following description and drawings, matters obvious to those skilled in the art are omitted or simplified as appropriate for the sake of clarity.

<Embodiment 1 of the Invention>
The robot 1 according to the first embodiment will be described. First, the external configuration of the robot 1 according to the first embodiment will be described with reference to FIG. FIG. 1 is an external configuration diagram of the robot 1 according to the first embodiment.

  The robot 1 includes a robot main body 10, a moving carriage 20, a gripping unit 30, and a head 40. The robot 1 can be used, for example, as an industrial robot acting on behalf of a work of carrying a load (object) performed by a person in a factory. The robot 1 can also be used as a service robot that performs an operation of picking up an object (object) to a person at home.

  The upper part of the movable carriage 20 is connected to the lower part of the robot body 10 so as to support the robot body 10 above it. The movable carriage 20 moves the robot 1 by driving wheels (not shown) provided in the lower part thereof.

  The grip portion 30 is connected to the front portion of the robot body 10 so as to protrude forward of the robot body 10. The gripping unit 30 is an articulated arm (robot arm). The grip part 30 includes a hand 60 that can grip an arbitrary object at the tip part. Thereby, the robot 1 can move the hand 60 to a desired position by adjusting the angle of each joint of the gripping unit 30, and can grip an object existing at that position.

  The head 40 is connected to the upper part of the robot body 10 so as to be supported by the robot body 10 above the robot body 10. The imaging unit 50 is attached to the front of the head 40 so that the front of the head 40 can be observed by the imaging unit 50. The imaging unit 50 observes the environment around the robot 1 and generates (acquires) environment information indicating the observed environment. The imaging unit 50 functions as an environment information acquisition unit.

  The imaging unit 50 can measure a predetermined angle range (view angle). On the other hand, the connecting portion between the head 40 and the robot body 10 has a multi-degree-of-freedom joint structure so that the head 40 can be operated. For example, by having at least two degrees of freedom as the joint structure, the head 40 can be moved up and down and left and right. Thereby, even if the angle range in which the imaging unit 50 can be measured is limited, it is possible to observe a wide range by operating the head 40.

  Here, the imaging unit 50 is specifically a camera that images the surrounding environment. Therefore, the imaging unit 50 images the periphery of the robot 1 and generates image information indicating an image of the periphery of the robot 1 as environment information. The camera may be an arbitrary camera among various cameras such as a monocular camera and a stereo camera.

  With the configuration described above, the robot 1 recognizes an object included in the surrounding environment based on the image indicated by the image information acquired by the imaging unit 50. Then, the robot 1 grips the recognized object by the grip portion 30 and transports the object by driving the wheels of the moving carriage 20 to the target point that is the transport destination of the object.

  Next, the internal configuration of the robot 1 according to the first embodiment will be described with reference to FIG. FIG. 2 is an internal configuration diagram of the robot 1 according to the first embodiment.

  The robot 1 includes a control unit 11, a storage unit 12, actuators 21a and 21b, wheels 22a and 22b, and actuators 31a and 31b. The control unit 11 and the storage unit 12 are included in the robot body 10. The actuators 21 a and 21 b and the wheels 22 a and 22 b are included in the moving carriage 20. The actuators 31 a and 31 b are included in the grip portion 30.

  The control unit 11 comprehensively controls the robot 1. The control unit 11 has a CPU (Central Processing Unit), and the CPU executes various programs as the control unit 11 by executing a program stored in the storage unit 12. That is, this program includes code for causing the CPU of the control unit 11 to execute various processes as the control unit 11.

  The storage unit 12 stores the program and various information used by the other control unit 11. The memory | storage part 12 contains at least 1 of the memory | storage device which can store the said program and various information. For example, at least one of a memory and a hard disk may be arbitrarily used as the storage device.

  The actuators 21 a and 21 b drive the wheels 22 a and 22 b of the robot 1. Actuator 21a is driven by control from control part 11, and rotates wheel 22a. Actuator 21b is driven by control from control part 11, and rotates wheel 22b.

  The wheel 22a and the wheel 22b correspond to the wheel attached to the lower part of the moving carriage 20 described above. That is, the control unit 11 controls the actuators 21a and 21b to move the robot 1 by rotating the wheels 22a and 22b.

  The actuators 31 a and 31 b drive the grip unit 30 of the robot 1. The actuators 31 a and 31 b are driven by control from the control unit 11, and each of the actuators 31 a and 31 b that operate the grip unit 30 is provided as a joint of the grip unit 30. Here, an example in which the gripping unit 30 has two joints of the actuator 31a and the actuator 31b will be described, but the number of joints of the gripping unit 30 is not limited to this.

  The hand 60 is driven by the control from the control unit 11 and grips or releases the object. When gripping an object to be gripped by the hand 60, the control unit 11 drives at least one of the actuators 31a, 31b, 21a, and 21b, and recognizes based on the image indicated by the image information output from the imaging unit 50. The hand 60 is guided to the finished object.

  Next, processing blocks of the robot 1 according to the first embodiment will be described with reference to FIG. FIG. 3 is a processing block diagram of the robot 1 according to the first embodiment.

  The control unit 11 functions as a feature extraction unit 111, an inter-feature association unit 112, an object recognition unit 113, and a grippable unit detection unit 114.

  The storage unit 12 functions as the object information storage unit 121. The object information storage unit 121 stores object recognition feature information, object 3D model information (object shape information), feature point position information, and an object ID for each of a plurality of objects existing in the environment in which the robot 1 operates. Has been. The object recognition feature information, the object 3D model information, the feature point position information, and the object ID are associated with each other for the same object. Such information may be prepared for each type of object.

  The feature information for object recognition is information indicating an object model by feature points (feature point group). The object 3D model information is information indicating a 3D model (three-dimensional shape as a model) of the object. The feature point position information is information indicating the positional relationship of each feature point indicated by the object recognition feature information with respect to the 3D model of the object indicated by the object 3D model information. The object ID is information that uniquely indicates each object.

  The feature extraction unit 111 extracts feature points (feature point group) in the image indicated by the image information output from the imaging unit 50. It should be noted that as the image feature used for extracting the feature point, an arbitrary feature among the features such as the SURF feature, the edge feature, and the HOG (Histogram of Oriented Gradient) feature may be used.

  The feature point association unit 112 associates the feature points extracted by the feature extraction unit 111 with the feature points of the object model indicated by the feature recognition feature information. That is, the feature point association unit 112 matches (matches) the extracted feature points with the feature points of the object model, thereby matching the feature points of the object model from the extracted feature points. Identify feature points and associate them with each other. That is, the feature point of the object is recognized from the feature points of the image. In other words, the object in the image is recognized by the object model.

  The object recognition unit 113 recognizes the position / orientation of the object based on the feature point correspondence specified by the feature point association unit 112. The feature recognizing unit 113 determines that an object exists at the position when a part of the object is also recognized. For example, the object recognition unit 113 has a predetermined ratio of the feature points associated (matched) between the extracted feature points and the feature points of the object model with respect to all the feature points of the object model. May be recognized as being present at that position. In addition, the object recognition unit 113 identifies the posture (tilt) of the object from the extracted feature points of the object based on the positional relationship between the 3D model of the object indicated by the feature point position information and the feature points of the object. Note that the position of the object includes the distance from the robot 1 to the object. Here, if the imaging unit 50 is a monocular camera, the distance from the robot 1 to the object may be estimated based on the distance between the feature points, and may be a stereo camera that captures two or more images. For example, the calculation may be performed based on the shift amount of the same point between two images (for example, the shift amount of the same feature point).

  The grippable portion detection unit 114 detects a portion that can be gripped by the gripping unit 30 based on the recognition result of the object. Specifically, the grippable part detection unit 114 determines an area that can be recognized as a part of the object as an area that can be gripped by the gripper 30. That is, the grippable part detection unit 114 determines an area where the extracted feature point of the object matches the feature point of the model of the object as an area that can be gripped by the gripping unit 30. Then, the grippable part detection unit 114 determines any part of the region as a part that can be gripped by the gripping part 30 (a part to be gripped). For example, among the extracted feature points of the object, a portion that is the center of gravity of the feature point that matches the feature point group of the object model is determined as a grippable portion. The object recognition unit 113 and the grippable part detection unit 114 function as a recognition unit.

  Here, since the feature point of the obstacle is extracted in the region where the object is hidden by the obstacle, the feature point of the extracted object does not match the feature point of the object model. On the other hand, since the feature point of the object is extracted in the region where the object is not obstructed by the obstacle, the feature point of the extracted object matches the feature point of the object model. Therefore, as described above, the part that is not hidden by the obstacle is determined by determining the part that grips the object in the region where the extracted feature point of the object matches the feature point of the object model. Can be determined as the part that holds the object.

  Then, the grippable portion detection unit 114 drives the gripping unit 30 so as to grip the determined portion and grips the object. Specifically, the grippable portion detection unit 114 drives the actuators 31a and 31b to approach the center of the hand 60 of the gripping unit 30 to the portion determined to be gripped. Then, the grippable part detection unit 114 grips the object by driving the hand 60.

  Here, a plurality of grip patterns are generated in advance for each object, and a grip pattern that grips a portion determined to be gripped is selected from the plurality of grip patterns, and the object is gripped with the selected grip pattern. It may be. The gripping pattern defines a gripping method that can grip an object in consideration of the size and shape of the hand 60. For example, a portion to be gripped in the object, a tilt of the hand 60 with respect to the object, and the like are defined. For example, information indicating the grip pattern may be stored in the storage unit 12 in advance, and the grippable portion detection unit 114 may be realized by driving the actuators 31a and 31b based on the information.

  According to the processing described above, even if an unobserved region (region hidden by an obstacle) exists in a part of an object, the observation region (not hidden by the obstacle) Area) is determined. Therefore, when gripping an object, the grip portion 30 can be accurately gripped so as not to contact an obstacle.

  Subsequently, processing of the robot 1 according to the first embodiment will be described with reference to FIGS. 4 and 5. FIG. 4 is a flowchart showing processing of the robot 1 according to the first embodiment. Here, as shown in FIG. 5, a case where a part of an object to be grasped is hidden behind an obstacle will be described.

  The imaging unit 50 images the surrounding environment within the angle of view, generates image information indicating an image (camera image) of the captured environment, and outputs the image information to the feature extraction unit 111 (S1). The feature extraction unit 111 extracts feature points from the image indicated by the image information output from the imaging unit 50 (S2).

  The feature correlation unit 112 matches the feature points extracted by the feature extraction unit 111 with the feature points of the model of the object indicated by the feature information for object recognition stored in the object information storage unit 121 and matches each other. The feature point to be identified is specified (S3).

  The object recognizing unit 113 derives the position / orientation of the object from the matched feature points based on the feature point correspondence specified by the feature association unit 112 (S4). Specifically, as shown in FIG. 5, the position / posture of the object is recognized from the feature points of the object not hidden behind the obstacle. In other words, if at least part of the feature point matches the feature point of the object model, the feature model of the object model is applied when the feature point of the object model is applied based on the matching feature point. It can be estimated that an object exists at the position of. Further, from the feature points of the model of the object applied with the matching feature points as a reference, the region where the object exists (the entire object) based on the positional relationship between the object 3D model indicated by the feature point position information and the feature points of the object Can also be specified. Therefore, for example, the position of the object can be recognized as the center of gravity of each feature point in the feature point of the model of the applied object or the center of gravity of the region where the object exists.

  Further, the posture (inclination) of the object can be specified as the inclination of the distribution of the feature point of the extracted object with respect to the distribution of the feature point in the model of the feature point of the object with respect to the matching feature point. For example, this inclination may be specified by defining a direction in a state where the object is placed on a horizontal plane without inclination in a model of a feature point of the object. According to this, based on the feature point position information, the posture of the object can be specified as the inclination of the distribution of the feature point of the extracted object with respect to the distribution of the feature point at the feature point of the model of the object.

  The grippable part detecting unit 114 derives the position / tilt of the hand 60 that can grip the object according to the position / posture of the object recognized by the object recognizing unit 113, so that the hand 60 becomes the position / tilt. The trajectory of the grip part 30 is derived.

  At this time, in order to avoid contact with an obstacle, the grippable part detection unit 114 specifies a grippable part (gripable place) from the matched feature points, and sets a trajectory for gripping the grippable part. Derived (S5). Specifically, as described above, the grippable portion detection unit 114 uses the gripping unit 30 to detect an area in which the feature point of the extracted object and the feature point of the object model match among the objects. Determine as grippable part. That is, for example, among the extracted feature points of the object, a portion serving as the center of gravity of the feature point that matches the feature point group of the object model is determined as a grippable portion.

  Then, the grippable part detection unit 114 grips the determined grippable part (S6). In other words, the grippable portion detection unit 114 drives the actuators 31a and 31b and the hand 60 to operate the gripping unit 30 in the derived trajectory and grip the object.

  As described above, in the first embodiment, the area recognized as a part of the grasped object is determined as the grasping area. Here, when a part of the object is hidden by the obstacle, the part that is not hidden by the obstacle is recognized. Therefore, even when an object is recognized from a part of the gripping object, an area that is recognized as a part of the gripping object among the entire object is determined as the gripping area, so that the area that is not hidden by the obstacle Can be gripped. Therefore, it is possible to avoid contact between the gripping device (robot) and the obstacle when gripping an object to be gripped.

<Embodiment 2 of the Invention>
Next, the robot 1 according to the second embodiment will be described. Since the external configuration and internal configuration of the robot 1 according to the second embodiment are the same as those in FIGS. 1 and 2, the description thereof is omitted. A processing block of the robot 1 according to the second embodiment will be described with reference to FIG. FIG. 6 is a processing block diagram of the robot 1 according to the second embodiment.

  Compared with the control unit 11 according to the first embodiment, the control unit 11 according to the second embodiment further functions as an unobserved region deriving unit 115 and a moving direction calculation unit 116.

  The unobserved area deriving unit 115 derives an unobserved area of the object (area hidden by the obstacle) based on the 3D model of the object indicated by the object 3D model information. Specifically, the unobserved region deriving unit 115 determines a region (observation region) identified as a feature point by the feature association unit 112 from the entire object region identified by the object recognition unit 13 as described above. ) Is derived as an unconfirmed region of the object.

  The movement direction calculation unit 116 calculates the movement direction of the object grasped by the grasping unit 30 so that the unobserved region derived by the unobserved region deriving unit 115 disappears. Specifically, the movement direction calculation unit 116 calculates a direction from the unobserved region of the object specified by the unobserved region deriving unit 115 toward the observation region of the object as the moving direction of the object.

  Here, the moving direction of the object may be determined by an arbitrary method as long as it is a direction from the unobserved area of the object to the observed area of the object. For example, the direction may be a direction from the center of gravity of the unobserved area toward the center of gravity of the observed area. Here, the centroid of the unobserved region may be the centroid of the feature point of the unobserved region when the feature point of the object model is applied to the unobserved region. Further, the center of gravity of the observation region may be the center of gravity of the feature points (matching feature points) in the observation region.

  Then, the movement direction calculation unit 116 moves the grasped object in the calculated movement direction. Specifically, the grippable portion detection unit 114 drives the actuators 31a and 31b to operate the gripping unit 30 in the calculated movement direction and move the object. Accordingly, it is possible to recognize more parts of the object in a state where the unobserved area of the object is not hidden by the obstacle. Therefore, even when an object is transported after this, it is possible to derive a trajectory that can avoid an obstacle more reliably.

  Subsequently, processing of the robot 1 according to the second embodiment will be described with reference to FIGS. 7 and 8. FIG. 7 is a flowchart showing processing of the robot 1 according to the second embodiment. Here, the continuation of the processing described in the first embodiment will be described. That is, since steps S1 to S6 are the same as those described in the first embodiment, the description thereof is omitted. As shown in FIG. 8, a case where the grasped object is moved so as to avoid the obstacle will be described.

  The unobserved region deriving unit 115 derives an unobserved region of the object (region hidden by the obstacle). The moving direction calculation unit 116 determines the direction in which the unobserved area derived by the unobserved area deriving unit 115 decreases as the moving direction of the object. That is, as described above, the movement direction calculation unit 116 calculates the direction from the unobserved area of the object specified by the unobserved area deriving unit 115 to the observation area of the object as the movement direction of the object. The calculation of the moving direction is performed, for example, in a state before gripping as shown in FIG. The movement direction calculation unit 116 moves the object in the determined direction by driving the actuators 31a and 31b (S7). As a result, as shown in FIG. 8, the object is moved in a direction in which the unobserved area of the object decreases.

  On the other hand, the movement direction calculation unit 116 periodically determines whether or not the number of feature points hidden by the obstacle is equal to or less than a predetermined threshold during the movement of the object (S8). That is, the imaging unit 50, the feature extraction unit 111, the inter-feature association unit 112, and the object recognition unit 113 also periodically operate to update the recognition content of the object. Specifically, the movement direction calculation unit 116 calculates the feature points associated by the feature association unit 112 based on the number of feature points of the object model as the number of feature points hidden by the obstacle. The number obtained by subtracting the number of is calculated.

  When it is determined that the number of feature points hidden by the obstacle is not equal to or less than the predetermined threshold (S8: No), the movement direction calculation unit 116 moves the object in a direction in which the unobserved area decreases. Continue (S7).

  When it is determined that the number of feature points hidden by the obstacle is equal to or less than the predetermined threshold (S8: No), the movement direction calculation unit 116 stops moving the object in the direction in which the unobserved area decreases. Then, the movement of the object to the target point as the object transport destination is started (S9).

  Here, theoretically, when the whole object can be recognized, the number of feature points hidden by the obstacle is zero. However, in practice, depending on the recognition status of the object, even if the entire object is recognized, the number of feature points of the extracted object does not necessarily match the number of feature points of the object model. In some cases. Further, as shown in FIG. 8, there may be a feature point that is hidden by the hand 60 when an object is gripped. Therefore, as described above, when the number of feature points hidden by the obstacle is equal to or less than a threshold value greater than 0, it is assumed that the entire object has been recognized and the movement in the calculated movement direction is stopped. Good.

  Further, the determination that the entire object has been recognized is not limited to the determination based on the above-described threshold value. For example, the movement direction calculation unit 116 may estimate the movement amount by which the entire object is recognized from the unobserved area derived by the unobserved area deriving unit 115, and move the object by the estimated movement amount.

  As described above, in the second embodiment, the gripping object is moved in the direction from the region not recognized as a part of the gripping object toward the region recognized as a part of the gripping object, and more of the gripping object is detected. The part is recognized. According to this, it is possible to recognize more parts of the object in a state where the unobserved region of the object is not hidden by the obstacle. Therefore, even when an object is transported after this, it is possible to derive a trajectory that can avoid an obstacle more reliably.

  Therefore, for example, when the object model associated by the feature association unit 112 is incorrect before the object is moved, the object is associated with the correct object model after the object is moved. Therefore, it is possible to recognize the correct position / posture of the object by the object recognition unit 113, and it is possible to derive a trajectory that can avoid the obstacle more reliably when the object is subsequently transported.

<Third Embodiment of the Invention>
Next, the robot 1 according to Embodiment 3 will be described. Since the external configuration and internal configuration of the robot 1 according to the third embodiment are the same as those in FIGS. 1 and 2, the description thereof is omitted. With reference to FIG. 9, a processing block of the robot 1 according to the third embodiment will be described. FIG. 9 is a processing block diagram of the robot 1 according to the third embodiment.

  The control unit 11 according to the second embodiment further functions as a posture correction amount calculation unit 117 as compared with the control unit 11 according to the second embodiment.

  When the posture of the object recognized by the object recognition unit 113 is tilted with respect to the installation surface of the target point, the posture correction amount calculation unit 117 corrects the posture of the object so that the tilt becomes smaller. Preferably, the inclination of the bottom surface of the object is calculated based on the posture of the object recognized by the object recognition unit 113 and the 3D model of the object, and a correction amount that makes the bottom surface of the object parallel to the installation surface is calculated.

  Subsequently, processing of the robot 1 according to the third embodiment will be described with reference to FIGS. 10 and 11. FIG. 11 is a flowchart showing processing of the robot 1 according to the third embodiment. Here, the continuation of the processing described in the second embodiment will be described. That is, steps S1 to S9 are the same as those described in the second embodiment, and thus the description thereof is omitted. As shown in FIG. 11, the case where the movement of the object to the target point is almost completed and the object is placed at the target point will be described.

  The object recognizing unit 113 recognizes the position / orientation of the object from the feature points not hidden by the grasping unit 30 when placing the object at the target point (S10). As described above, similarly to the case where an object is hidden by an obstacle, the object is recognized by the feature point associating unit 112, and the position / posture thereof can also be recognized by the object recognizing unit 113.

  Based on the posture of the object recognized by the object recognition unit 113, the posture correction amount calculation unit 117 is configured to reduce the inclination of the object when the posture of the object is inclined with respect to the installation surface of the target point. After correcting the posture, the object is placed at the target point (S11).

  As described above, the posture of the gripping object is identified from the recognizable portion of the gripping object that is not hidden by the gripping unit 30 (gripping device), and the installation surface of the transport destination with the identified posture of the gripping object The posture of the grasped object is corrected so that the inclination with respect to is smaller. According to this, the object can be stably placed without falling down.

<Other embodiments of the present invention>
In the above-described embodiment, the case where a camera that generates image information is used as the imaging unit 50 has been described. However, the present invention is not limited to this. As the imaging unit 50, a sensor that generates distance information indicating the distance (three-dimensional point group) to each point in the surrounding environment may be used. As the sensor, for example, an arbitrary sensor among sensors such as a laser range finder and a distance image sensor may be used. In this case, the feature extraction unit 111 extracts feature points from the three-dimensional point group indicated by the distance information. It should be noted that as a feature used for extracting feature points, an arbitrary feature may be used among features such as Point Pair Feature.

  Note that the present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention.

DESCRIPTION OF SYMBOLS 1 Robot 10 Robot main body 20 Moving cart 30 Grasping part 40 Head 50 Imaging part 60 Hand 11 Control part 12 Storage part 21a, 21b Actuator 22a, 22b Wheel 31a, 31b Actuator 111 Feature extraction part 112 Feature correlation part 113 Object recognition Unit 114 grippable unit detection unit 115 unobserved region deriving unit 116 moving direction calculation unit 117 posture correction amount calculation unit 121 object information storage unit

Claims (9)

This is a case where environmental information that is three-dimensional information acquired by the gripping device is acquired, a gripping object is recognized from the acquired environmental information, and a part of the recognized gripping object is hidden by an obstacle. However, it is a gripping method for recognizing and gripping the gripping object by a gripping device,
Obtaining surrounding environmental information including the gripping object;
Recognizing the grasped object from the surrounding environment based on the acquired environment information;
Determining a gripping object when a portion of the gripping object is recognized;
Determining a gripping area of an object determined to be the gripping object;
Gripping the determined gripping area, and
The step of determining the gripping area includes
A gripping method in which an area recognized as a part of the gripping object in the step of recognizing the gripping object is determined as the gripping area. The step of recognizing the gripping object extracts a feature point from the environment information, and executes a matching process between the extracted feature point and a feature point of a model stored in advance as a feature point of the gripping object. By recognizing the gripping object,
The step of determining the gripping area determines, as the gripping area, an area where the feature point of the model matches the feature point of the gripping object among the gripping object.
The gripping method according to claim 1, wherein: The step of determining the gripping region determines, as the gripping region, centroids of a plurality of feature points that match feature points of the model among the gripping objects.
The gripping method according to claim 2, wherein: The gripping method further includes moving the gripping object in a direction from an unobserved area that is not recognized as a part of the gripping object toward an observation area that is recognized as a part of the gripping object. The grasping method according to claim 2, further comprising a step of recognizing a larger part of the object. Recognizing more parts of the grasped object,
Periodically, the surrounding environment information is acquired, and the feature points of the gripping object are extracted based on the acquired environment information,
Moving the gripping object until the number of feature points that do not match between the feature points of the gripping object and the feature points of the model is equal to or less than a predetermined threshold;
The gripping method according to claim 4. The step of recognizing more parts of the grasped object is the center of gravity of the unobserved region, or the center of gravity of the feature point of the unobserved region when the feature point of the model is applied to the unobserved region, Moving the gripping object in a direction toward the center of gravity of the observation region or the center of the feature point of the observation region;
6. The gripping method according to claim 4 or 5, wherein: A transportation method for transporting a gripping object gripped by the gripping method according to any one of claims 4 to 6 to a transport destination,
Determining a trajectory in which a larger part of the grasped object recognized by the step of recognizing a larger part of the grasped object avoids an obstacle, and transporting the grasped object in the trajectory. A transportation method characterized by The step of transporting the gripping object includes:
Acquire surrounding environment information, identify the posture of the gripping object from the recognizable portion of the gripping object that is not hidden by the gripping device in the state gripped by the gripping device, based on the acquired environment information,
Correct the posture of the gripping object so that the inclination of the posture of the identified gripping object with respect to the installation surface of the transport destination is less;
The transportation method according to claim 7, wherein: Even if the gripping object is recognized from the acquired environment information and a part of the recognized gripping object is hidden by an obstacle, the gripping target is acquired. A robot that recognizes an object and grips it with a robot arm,
An environmental information acquisition unit that acquires environmental information around the gripping object;
A recognition unit that determines a gripping object when a part of the gripping object is recognized from the surrounding environment based on the acquired environment information;
A gripping area determination unit that determines a gripping area of an object determined to be the gripping object,
The robot according to claim 1, wherein the gripping area determination unit determines an area recognized as a part of the gripping object as the gripping area.

Images