JP4737099B2 - Robot and robot control apparatus and control method - Google Patents

Description

  The present invention relates to a robot. In particular, the present invention relates to a robot that includes a hand unit having at least two fingers and holds an object by the hand unit.

A robot has been developed that includes a hand unit having at least two fingers and holds an object by the hand unit. This type of robot is required to reliably hold various objects. Therefore, many techniques for recognizing the shape of an object to be grasped by a camera and determining a grasping operation according to the shape of the object have been proposed.
In addition, a technique for recognizing the shape of an object to be grasped without requiring a camera has been proposed (Patent Document 1). In this technique, a sensor that detects contact between each finger part and a gripping object and a sensor that detects an opening / closing angle of each finger part are used, and based on the opening / closing angle of each finger part when contacting the gripping object. To calculate the approximate shape of the object to be grasped. According to this technique, for example, even when the object is arranged in the dark, the gripping operation can be adjusted according to the shape of the object.

JP 2004-160594 A

In the conventional technique, although the shape of the object to be grasped can be recognized, the surface property of the object to be grasped cannot be recognized. Therefore, for example, it is necessary to grip an object having a rough surface and a smooth object by the same gripping operation. For example, regarding the gripping force when gripping an object, it is necessary to grip an object with a rough surface and a smooth object with the same gripping force, and even though an object with a rough surface can be gripped, You may not be able to grip. If the gripping force is determined based on an object with a smooth surface, an object with a smooth surface can be gripped, but an object with a rough surface will be gripped with an excessive gripping force, and unnecessary energy will be consumed. End up.
The present invention solves the above problems. The present invention provides a technique that makes it possible to appropriately grip various objects by adjusting the gripping operation according to the surface properties of the object.

  The technology of the present invention can be embodied in a robot that holds an object. This robot has a hand part having at least two finger parts, a contact force measuring means for measuring a contact force that the finger part receives from an object, and a contact force measuring means when the finger part slides on the object surface. Frequency measuring means for measuring the vibration frequency of the contact force to be applied, and gripping operation adjusting means for adjusting the gripping operation in which the hand unit grips the object based on the vibration frequency measured by the frequency measuring means.

When the finger is slid on the object surface, the contact force that the finger receives from the object vibrates periodically according to the roughness of the object surface. Therefore, the roughness of the object surface can be grasped by measuring the vibration frequency of the contact force that the finger receives from the object when the finger is slid on the object surface. And it becomes possible to implement | achieve the holding | grip operation | movement according to the roughness of the object surface by adjusting the holding | grip operation | movement of a hand part based on the measured vibration frequency. For example, when it is determined that the object surface is smooth, the gripping force for gripping the object can be increased, and the arrangement of each finger portion with respect to the object can be changed.
According to this robot, it is possible to appropriately hold objects having various surface properties.

In the robot described above, it is preferable that the gripping movement adjusting means sets a target value of the gripping force for gripping the object based on the measured vibration frequency of the contact force.
According to this robot, the object can be gripped more reliably by the gripping force according to the roughness of the object surface.

In the robot described above, it is preferable that the gripping motion adjusting means increases the target value of the gripping force as the measured vibration frequency of the contact force is small.
According to this robot, an object with a smooth surface can be gripped with a large gripping force, and an object with a rough surface can be gripped with a small gripping force.

The technology of the present invention is embodied in a robot control device that causes a robot to include a hand unit having at least two finger units and a contact force measuring unit that measures a contact force that the finger unit receives from an object. You can also. This robot control device is based on frequency measurement means for measuring the vibration frequency of the contact force measured by the contact force measurement means when the finger slides on the surface of the object, and on the vibration frequency measured by the frequency measurement means. And a gripping motion adjusting means for adjusting the gripping motion of the hand portion gripping the object. Here, it is preferable that the gripping movement adjusting unit increases the target value of the gripping force with which the hand unit grips the object as the measured vibration frequency is small.
According to this robot control apparatus, the robot can appropriately hold objects having various surface properties.

The technology of the present invention is also embodied in a robot control method that causes a robot to grip an object by including a hand unit having at least two finger units and a contact force measuring unit that measures a contact force that the finger unit receives from an object. Is done. This control method is based on the frequency measurement step of measuring the vibration frequency of the contact force measured by the contact force measurement means when the finger part slides on the object surface, and the vibration frequency measured by the frequency measurement step. A gripping operation adjusting step for adjusting a gripping operation in which the hand unit grips an object is provided. Here, in the gripping movement adjustment step, it is preferable that the target value of the gripping force with which the hand unit grips the object is increased as the measured vibration frequency is lower.
According to this robot control method, the robot can appropriately hold objects having various surface properties.

  According to the present invention, it is possible to embody a robot that appropriately holds various objects.

First, the main features of the embodiments described below are listed.
(Feature 1) The robot includes two cameras that capture an image of the object to be grasped.
(Feature 2) The robot includes means for calculating the position of the gripping object from images obtained by capturing the gripping object by two cameras.
(Characteristic 3) The finger part has a plurality of links, and a pressure sensor is provided on the most distal link.
(Feature 4) The robot control device includes a gripping motion control unit that controls the motion of the finger based on the set target value of the gripping force and the measured value of the contact force that the finger receives from the gripping object. I have.

FIG. 1 is a diagram schematically showing the appearance of a robot 10 embodying the present invention. The robot 10 of the present embodiment is a humanoid robot made by imitating a part of a person. The robot 10 can grip and move the object (grip target) 2.
As shown in FIG. 1, the robot 10 includes a robot body 20 and a control device 100 that controls the operation of the robot body 20. The robot body 20 includes a head 30, a trunk 40, an arm 50, and a hand 70. The head 30 is connected to the trunk 40 so as to be swingable. The head 30 is provided with two cameras 32 (one is not shown) for photographing the grasped object 2. The arm unit 50 is a robot arm having a plurality of links. The base end of the arm part 50 is connected to the body part 40, and the hand part 70 is connected to the tip of the arm part 50. The arm unit 50 includes a plurality of motors 52 (see FIG. 3) that are actuators, and changes the position and orientation of the hand unit 70 by changing its posture.

FIG. 2 schematically shows the configuration of the hand unit 70. As shown in FIG. 2, the hand unit 70 is a robot hand configured to imitate a human hand, and includes a palm unit 74 and five finger units 80 connected to the palm unit 74. ing. Each finger 80 includes a fingertip link 81, an intermediate link 83 in which the fingertip link 81 is connected via a uniaxial joint 82, and a proximal link 85 in which the intermediate link 83 is connected via a uniaxial joint 84. It has. The proximal end link 85 of each finger part 80 is connected to the palm part 74 via a biaxial joint 86. The palm 74 is provided with a plurality of motors 72 that move the joints 82, 84, 86. A wire 76 is used to connect the motor 72 and the finger part 80. The surface portions of the palm portion 74 and the finger portion 80 are made of an elastic material, and are elastically deformed in contact with the grasped object 2.
A pressure sensor 90 is provided on the fingertip link 81 of each finger portion 80. The pressure sensor 90 is provided at a position that comes into contact with the grasped object 2 when the hand unit 70 grasps the grasped object 2. Each pressure sensor 90 has a plurality of pressure-sensitive elements arranged in a plane, and can measure a pressure distribution applied to the surface of the fingertip link 81. The pressure sensor 90 is provided to measure a force applied to the surface of the fingertip link 81 from the grasped object 2 (hereinafter referred to as a contact force).

FIG. 3 is a block diagram showing an electrical configuration of the robot 10. As shown in FIG. 3, the control device 100 includes an image processing unit 102, an arm control unit 104, and a hand control unit 110.
The image processing unit 102 processes images captured by the two cameras 32 provided on the head 30 of the robot body 20 and calculates the position of the grasped object 2. The image processing unit 102 outputs the calculated position of the grasped object 2 to the arm control unit 104.
The arm control unit 104 controls the arm unit 50 of the robot body 20. The arm control unit 104 calculates a target position for moving the hand unit 70 based on the position of the grasped object 2 input from the image processing unit 102. Then, the arm control unit 104 controls the motor 52 of the arm unit 50 so as to be positioned at the target position calculated by the hand unit 70.

The hand control unit 110 controls the operation of the hand unit 70 of the robot body 20. The hand control unit 110 includes a sensor signal processing unit 112, a gripping force setting unit 114, a finger motion calculation unit 116, and a plurality of motor drivers 118.
The sensor signal processing unit 112 receives an output signal of the pressure sensor 90 provided on each finger 80 of the hand unit 70, and calculates the contact force received by each finger unit 80 from the grasped object 2. The contact force of each finger 80 calculated by the sensor signal processing unit 112 is output to the gripping force setting unit 114 and the finger motion calculation unit 116.
The gripping force setting unit 114 sets a target value of the gripping force when gripping the gripping target object 2 based on the contact force of the finger unit 80 calculated by the sensor signal processing unit 112. This target value setting process will be described in detail later. The target value of the gripping force set by the gripping force setting unit 114 is output to the finger motion calculation unit 116.
The finger motion calculation unit 116 calculates the target motion of each finger unit 80 based on the deviation between the target value of the gripping force set by the gripping force setting unit 114 and the actual contact force calculated by the sensor signal processing unit 112. Calculate the quantity.
The motor driver 118 controls each motor 72 of the hand unit 70 based on the target motion amount of each finger unit 80 calculated by the finger motion calculation unit 116.
As described above, the hand control unit 110 sets the target value of the gripping force when gripping the gripping object 2, and operates the hand unit 70 so as to grip the gripping object 2 with the set gripping force. Control.

FIG. 4 is a flowchart showing a flow of an operation in which the robot 10 grips the gripping target object 2. Hereinafter, along the operation flow shown in FIG. 4, the gripping operation of the gripping target object 2 by the robot 10 will be described in detail.
In step S <b> 2, the image processing unit 102 calculates the position and shape of the grasped object 2 based on the image captured by the camera 32.
In step S <b> 4, the arm control unit 104 moves the arm unit 50 based on the position of the gripping target object 2 and moves the hand unit 70 to the vicinity of the gripping target object 2. The position where the hand unit 70 is moved is not a position for gripping the gripping object 2 but an initial position for a sliding operation in the next step S6.

In step S6, as shown in FIG. 5, the fingertip link 81 of at least one finger part 80 is moved on the surface 2a of the grasping object 2 by moving the hand part 70 along the surface of the grasping object 2. Slide. FIG. 5 shows an example of the sliding operation in step S6, and shows how the hand unit 70 is moved from position (a) to position (c) through position (b). During this sliding operation, the contact force applied to the fingertip link 81 from the surface 2 a of the grasped object 2 is measured by the pressure sensor 90 and the sensor signal processing unit 112. Time series data of the contact force measured during the sliding operation is input to the gripping force setting unit 114.
FIG. 6 shows an example of the contact force measured during the sliding operation. In FIG. 6, time t <b> 1 indicates a contact start time point between the finger 80 and the surface 2 a of the grasped object 2, and time t <b> 2 indicates a contact end time point between the finger 80 and the surface 2 a of the grasped object 2. ing. That is, the period from time t1 to time t2 indicates a period during which the finger 80 is sliding (boiled) on the surface 2a of the grasped object 2. The contact force measured while the finger 80 is sliding on the surface 2a of the grasped object 2 vibrates with time according to the roughness of the surface 2a of the grasped object 2.

In step S8 of FIG. 4, the gripping force setting unit 114 sets a target value of the gripping force when gripping the gripping target object 2 based on the time series data of the contact force measured during the sliding operation. First, the gripping force setting unit 114 calculates the vibration frequency of the contact force during the sliding operation. As described above, the contact force measured during the sliding operation oscillates with time according to the roughness of the surface 2a of the grasped object 2. The gripping force setting unit 114 extracts a peak point from the measured contact force time-series data, and calculates the vibration frequency of the contact force by counting the number of peak points included per unit time. Next, the gripping force setting unit 114 determines a target value of the gripping force based on the calculated vibration frequency of the contact force. Thereby, the target value of the gripping force is determined according to the roughness of the surface 2a of the gripping object 2.
FIG. 7 shows the relationship between the vibration frequency (horizontal axis) of the contact force and the target value (vertical axis) of the gripping force determined by the gripping force setting unit 114. As shown in FIG. 7, the gripping force setting unit 114 sets the target value of the gripping force to be smaller as the measured vibration frequency of the contact force is larger. Thereby, when the surface 2a of the gripping object 2 is rough, the target value of the gripping force is set to be small, and when the surface 2a of the gripping object 2 is smooth, the target value of the gripping force is set to be large.

  When the target value of the gripping force is set, in the subsequent step S12 (FIG. 4), an operation of actually gripping the gripping target object 2 is started. In step S12, first, the arm control unit 52 controls the arm unit 50 to move the hand unit 70 to a position where the gripping target object 2 can be gripped. Next, the hand control unit 110 controls each motor 72 of the hand unit 70 to bend each finger unit 80. As each finger 80 is bent, each finger 80 comes into contact with the grasped object 2 in due course. When each finger 80 comes into contact with the grasped object 2, the contact force applied from the grasped object 2 to each finger 80 is measured by the pressure sensor 90 of each finger 80. The measured contact force of each finger unit 80 is input to the finger motion calculation unit 116.

  In step S14, the finger motion calculation unit 116 and the motor driver 118 feed back the operation of each finger unit 80 of the hand unit 70 based on the measured contact force of each finger unit 80 and the set target value of the gripping force. Control. Thereby, the operation of each finger unit 80 is adjusted so that the gripping force for gripping the gripping object 2 of the hand unit 70 becomes equal to the target value. When the gripping force for gripping the gripping object 2 of the hand unit 70 becomes equal to the target value (YES in step S16), the finger motion calculation unit 116 determines that the gripping of the gripping object 2 has been completed. When the gripping of the gripping target object 2 by the hand unit 70 is completed, the arm control unit 102 moves the arm unit 50 and moves the hand unit 70 (that is, the gripping target object 2) to the instructed position.

As described above, the robot 10 according to the present embodiment can measure the roughness (smoothness) of the surface 2a of the grasped object 2 and grasp the grasped object 2 with a grasping force according to the measured roughness. it can. For example, when the surface 2a of the gripping object 2 is smooth, the gripping object 2 that is slippery can be reliably gripped by gripping the gripping object 2 with a relatively large gripping force. On the contrary, when the surface 2a of the gripping object 2 is rough, unnecessary energy consumption and damage to the gripping object 2 can be prevented by gripping the gripping object 2 with a relatively small gripping force.
In the present embodiment, the pressure sensor 90 provided on the fingertip link 81 of the finger part 80 measures the surface roughness of the grasped object 2 and measures the grasping force applied by the finger part 80 to the grasped object 2. ing. Thereby, it is not necessary to separately provide a dedicated pressure sensor for measuring the surface roughness of the grasped object 2. However, a dedicated pressure sensor for measuring the surface roughness of the grasped object 2 may be separately provided, for example, on the back side of the palm part 74 or the hand part 70 (position corresponding to the back of the hand).

When setting the gripping force for gripping the gripping target object 2, it is also effective to consider the surface hardness of the gripping target object 2 in addition to the surface roughness of the gripping target object 2. For example, if the gripping force set based on the surface roughness of the gripping object 2 is excessive with respect to the surface hardness of the gripping object 2, the gripping object 2 may be destroyed. Therefore, it is preferable to limit the gripping force for gripping the gripping object 2 according to the surface hardness of the gripping object 2. Accordingly, when the gripping force set based on the surface roughness of the gripping object 2 is excessive with respect to the surface hardness of the gripping object 2, the gripping of the gripping object 2 is stopped by The destruction of the object 2 can be prevented in advance.
The surface hardness of the grasped object 2 is determined by, for example, pressing the surface of the grasped object 2 with the fingertip link 81, and the amount of deformation of the grasped object 2 at that time (that is, the amount of movement of the fingertip link 81) It can be calculated from the applied contact force. That is, when the gripping object 2 is gripped by the hand unit 70 (step S14 in FIG. 4), the surface hardness of the gripping object 2 is measured by measuring the movement amount of the fingertip link 81 and the contact force applied to the fingertip link 81. Can be calculated. If the target value of the gripping force is excessive with respect to the calculated surface hardness of the gripping object 2, the gripping of the gripping object 2 can be stopped.

  Alternatively, both the surface roughness and the surface hardness of the grasped object 2 can be measured by the method described below. When the surface roughness and surface hardness of the grasped object 2 are different, the contact force (force perpendicular to the surface 2a) when the fingertip link 81 is slid on the surface of the grasped object 2 and the friction force (on the surface 2a) Parallel force) and the vibration frequency of the contact force or frictional force change. In other words, the surface roughness and surface hardness of the grasped object 2 can be calculated by measuring the contact force and frictional force when the fingertip link 81 is slid on the surface of the grasped object 2 and the vibration frequency of one of them. It becomes possible to do. FIG. 8 shows an example of this, which uses a neural network. The neural network 200 shown in FIG. 8 is configured to input the contact force, friction force, and vibration frequency described above and output the surface roughness and surface hardness of the grasped object 2. The neural network 200 can learn (update the connection load between the neurons 202) by being given the actual surface roughness and surface hardness of various gripping objects 2 as teacher signals. By sufficiently learning the neural network 200, a neural network that accurately outputs the surface roughness and surface hardness of the grasped object 2 can be constructed.

  The neural network 200 described above can be mounted on the gripping force setting unit 114 in the robot 10 of this embodiment. Each input value to the neural work 200 can be obtained as follows. First, for the contact force and the vibration frequency when the fingertip link 81 is slid on the surface of the grasped object 2, the time series data (see FIG. 6) of the contact force measured in step S6 of FIG. 4 is used. Can be calculated. The frictional force when the fingertip link 81 is slid on the surface of the grasped object 2 is calculated from the load applied to the arm unit 50 during the sliding operation in step S6 (this is equal to the output of the arm unit 50). be able to. Alternatively, if the fingertip link 81 is provided with a strain sensor capable of measuring the frictional force, it can be measured more directly. By using the neural network 200, the gripping force setting unit 114 can accurately grasp the surface roughness and surface hardness of the grasped object 2. Based on the surface roughness and surface hardness of the gripping object 2, it is possible to appropriately determine a target value of the gripping force and determine whether or not the gripping object 2 can be gripped.

Specific examples of the present invention have been described in detail above, but these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes of the specific examples illustrated above.
For example, the contact force that the finger 80 receives from the grasped object 2 can be calculated by measuring the torque applied to the joints 82, 84, 86 of the finger 80 without using the pressure sensor 90.
Further, in the robot 10 of this embodiment, the posture of each finger unit 80 when gripping the gripping object 2 may be changed based on the surface roughness of the gripping object 2. For example, when the surface 2a of the grasped object 2 is smooth (that is, when the vibration frequency of the contact force is small), each finger unit 80 is in a posture that makes contact with the grasped object 2 in a wide area. The operation of the finger unit 80 may be adjusted. For this purpose, it is preferable to prepare a plurality of types of gripping motion data describing the motion of each finger unit 80 when gripping the gripping target object 2 and store them in the hand control unit 110 in advance. Accordingly, the hand control unit 110 can select gripping operation data to be used based on the determined surface roughness of the gripping object 2.
The technical elements described in this specification or the drawings exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology exemplified in this specification or the drawings achieves a plurality of objects at the same time, and has technical utility by achieving one of the objects.

The figure which shows the structure of a robot typically. The figure which shows the structure of a hand part typically. The block diagram which shows the electric constitution of a robot. The flowchart which shows the flow of the holding | grip operation | movement by a robot. The figure which shows sliding operation | movement. The figure which illustrates the contact force measured at the time of sliding operation. The figure which shows the relationship between a vibration frequency and the target value of gripping force. The figure which illustrates the neural network which can be utilized for a robot.

Explanation of symbols

10: Robot 20: Robot body 30: Head 32: Camera 40: Body 50: Arm unit 52: Arm unit motor 70: Hand unit 72: Hand unit motor 80: Finger unit 81: Fingertip link 90: Pressure sensor 100: control device 102: image processing unit 104: arm control unit 110: hand control unit 112: sensor signal processing unit 114: gripping force setting unit 116: finger motion calculation unit 118: motor driver 200: neural network

Claims (3)

A robot that grips an object,
A hand portion having at least two fingers;
Contact force measuring means for measuring the contact force received by the finger from the object;
Frequency measuring means for measuring the vibration frequency of the contact force measured by the contact force measuring means when the finger part slides on the object surface;
Based on the vibration frequency measured by the frequency measuring means, the gripping movement adjusting means for adjusting the gripping movement in which the hand unit grips the object;
Equipped with a,
The gripping action adjusting means, as said measured vibration frequency is low, the hand portion characterized in that increasing the target value of the gripping force for gripping the object the robot. A robot control device that causes a robot to grip an object, comprising a hand unit having at least two finger units, and a contact force measuring unit that measures a contact force that the finger unit receives from an object,
Frequency measuring means for measuring the vibration frequency of the contact force measured by the contact force measuring means when the finger part slides on the object surface;
Based on the vibration frequency measured by the frequency measuring means, the gripping movement adjusting means for adjusting the gripping movement in which the hand unit grips the object;
Equipped with a,
The gripping action adjusting means, as said measured vibration frequency is low, the hand portion characterized in that increasing the target value of the gripping force for gripping the object control device. A robot control method for causing a robot to grip an object by a robot including a hand unit having at least two finger units and a contact force measuring unit that measures a contact force received by the finger unit from the object,
A frequency measurement step for measuring the vibration frequency of the contact force measured by the contact force measurement means when the finger part slides on the object surface;
Based on the vibration frequency measured by the frequency measurement step, a gripping operation adjustment step for adjusting a gripping operation in which the hand unit grips an object;
Equipped with a,
The gripping operation by adjusting process, said measured as vibration frequency is low, control how to characterized by increasing the target value of the gripping force which the hand unit to grip an object.

Images